Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Clear cell ovarian carcinoma and emergence of the novel tumour suppressor gene ARID1A Wiegand, Kimberly Charlotte 2013

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

Item Metadata

Download

Media
24-ubc_2013_fall_wiegand_kimberly.pdf [ 173.91MB ]
Metadata
JSON: 24-1.0073992.json
JSON-LD: 24-1.0073992-ld.json
RDF/XML (Pretty): 24-1.0073992-rdf.xml
RDF/JSON: 24-1.0073992-rdf.json
Turtle: 24-1.0073992-turtle.txt
N-Triples: 24-1.0073992-rdf-ntriples.txt
Original Record: 24-1.0073992-source.json
Full Text
24-1.0073992-fulltext.txt
Citation
24-1.0073992.ris

Full Text

CLEAR CELL OVARIAN CARCINOMA AND EMERGENCE OF THE NOVEL TUMOUR SUPPRESSOR GENE ARID1A  by  Kimberly Charlotte Wiegand  B.Sc., The University of Victoria, 2000  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  Doctor of Philosophy  in  THE FACULTY OF GRADUATE STUDIES  (Pathology and Laboratory Medicine)  THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)  July 2013  © Kimberly Charlotte Wiegand, 2013      ii Abstract Clear cell carcinomas (CCCs) are a subtype of ovarian cancer that is understudied, and which does not respond well to conventional therapeutic strategies.  There is a desperate need to clarify the genetic mechanisms of CCC to allow for the development of subtype specific therapeutics.  To determine genetic changes responsible for the development of CCC, whole transcriptomes of 18 ovarian CCCs and one ovarian CCC cell line were sequenced.  Somatic mutations were found in ARID1A in 6 samples. ARID1A encodes BAF250a, a key component of the SWI/SNF chromatin remodeling complex. ARID1A was sequenced in an additional 210 ovarian carcinomas and a second CCC cell line, and BAF250a expression was measured by immunohistochemistry (IHC) in an additional 455 ovarian carcinomas and over 3000 non- ovarian malignancies. Overall, ARID1A mutations were seen in 46% of CCCs and 30% of endometrioid carcinomas (EC) implicating ARID1A as a tumour suppressor frequently disrupted in CCC and EC. Loss of the BAF250a protein correlated strongly with the presence of ARID1A mutations. In two patients, ARID1A mutations and loss of BAF250a expression were evident in the tumour and contiguous atypical endometriosis, implicating ARID1A loss as an early event in tumourigenesis.  Screening 3000 cases of different malignancies by IHC showed loss of BAF250a was is most frequent in cancers of endometrial origin.  Reverse phase protein array (RPPA) for tumour samples with known ARID1A mutation status revealed a notable change pAKT-Thr308 (FDR < 1%), which may allow for targeted therapeutic strategies in ARID1A deficient tumours. To identify compounds synthetic-lethal with ARID1A deficiency, two nearly isogenic cell lines were screened along with 18 other cell lines, including ten CCC cell lines of known ARID1A status against a panel of kinase inhibitors, revealing several promising therapeutic compounds effective at inhibiting the growth of CCC cell lines, including GSK461364, Pelitinib, and Tovok, however these did not appear to be dependent on ARID1A deficiency.  Collectively, these studies provided the first report of ARID1A as a major tumour suppressor in cancers of endometrial origin and several other tumour types; these studies have now been validated by multiple groups in diverse tumour types.  iii Preface This work was conducted under the supervision of Dr. David G. Huntsman.  Ethical approval for this work is covered by the UBC BC Cancer Agency Research Ethics Board Protocol #H05-60119, entitled:  Formation of the Gynecological Cancer Tissue Bank and Protocol #H02-61375, entitled: Immunohistochemical and Fluorescent In-Situ Hybridization (FISH) Studies of Cancer.  The chapters and appendices of this dissertation are presently in various stages of publication; details of these manuscripts and the contributions of all co-authors are described below.  A version of the work presented in Chapter 2 has been published: Wiegand KC, Shah SP, Al-Agha OM,  Zaho Y, Tse K, Zeng T, Senz J, McConechy MK, Anglesio MS, Kalloger SE, Yang W, Heravi-Moussavi A, Giuliany R, Chow C, Fee J, Zayed A, Prentice L, Melnyk N, Turashvili G, Delaney AD, Madore J, Yip S, McPherson AW, Ha G, Bell L, Fereday S, Tam A, Galletta L, Tonin PN, Provencher D, Miller D, Jones SJM, Moore RA, Morin GB, Oloumi A, Boyd N, Aparicio SA, Shih IeM, Mes-Masson AM, Bowtell DD, Hirst, M, Gilks B, Marra MA, Huntsman DG.  ARID1A Mutations in Endometriosis-Associated Ovarian Carcinomas. New England Journal of  Medicine. 2010; Oct 14; 363(16):1532-43. Epub 2010 Sep 8. Copyright © 2010, Massachusetts Medical Society.  Reprinted with permission.  The study presented in Chapter 2 was designed by the candidate KCW, with SPS and DGH. The candidate, KCW was responsible for identification of the recurrent ARID1A mutations in the dataset, conducted all preliminary validation work and Sanger sequencing, as well as LCM for the precursor lesions, cloning, and supervised all aspects of the project.   Data was gathered by KCW, SPS, OA, YZ, KT, TZ, JS, MM, WY, CC, JF, AZ, NM, GT, AD, JM, SY, LB, SF, AT, LG, PNT, DP, DM, RAM, GBM, AO, IS, AM, DB, MH, CBG, MAM, and DGH.  Data analysis was performed by KCW, SPS, OA, MA, SEK, AHM, RG, AWM, GH, SJ, NB, SA, MAM, and DGH.  All bioinformatic work was conducted by the team of SPS. Statistical analysis was performed by SEK. The first draft of the manuscript was prepared by KCW and subsequent drafts of the manuscript were prepared by KCW, SPS, MA, SEK, NB, CBG, and DGH who also vouch for the accuracy of the data.  The decision to publish this work was made by KCW, CBG, MAM, and DGH.  A version of the work presented in the first section of Chapter 3 has been published: Wiegand KC, Lee AF, Al-Agha OM, Chow C, Kalloger SE, Scott DW, Steidl C, Wiseman SM, Gascoyne RD, Gilks B, Huntsman DG.  Loss of BAF250a (ARID1A) is frequent in high- grade endometrial carcinomas.  Journal of Pathology. 2011; Jul; 224(3):328-33  Epub ahead of print April 1, 2011. Copyright © 2011, Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.  Reprinted with permission.  KCW, DGH and BG  were responsible for the study design and data interpretation presented in Chapter 3.  AFL and OMA collected the data; KCW compiled all data, wrote the first draft of the manuscript, and prepared all images and figures; CC optimized conditions for the IHC; SEK provided statistical consultation and analysis; and DWS, CS, RDG and SMW provided clinical material and consultation. All authors were involved in the writing of the manuscript and had final approval of the submitted and published versions.  iv The work presented in the second half of Chapter 3 is in preparation for publication: Wiegand KC, Sy K, Kalloger SE, Woods R, Kumar A, Streutker CJ, Hafezi-Bakhtiari S, Zhou C, Lim HJ, Huntsman DG, Clarke B, Schaeffer DF.  ARID1A/BAF250a as a prognostic marker for gastric carcinoma is dependent on stage: a study of two cohorts.  KCW conceived of the study in the second half of Chapter 3 with KS, BC, DGH, DFS, and wrote the first draft of the manuscript.  KS Scored the IHC for the Toronto cohort and contributed to the writing of the manuscript. SEK Compiled all of the data from Toronto and Vancouver, ran all of the statistical analysis, and contributed to the writing and review of the manuscript.  RW provided review and quality control for clinical data at the BCCA. AK assisted with follow up for the clinical database in Vancouver. CJS provided pathology review and selection of the cases for the Toronto cohort and contributed to the text of the manuscript.  SHB helped score TMAs for the Toronto cohort.  CZ provided pathology review and helped with selection of the cases for the Vancouver.  HJL contributed to selection of the cases for the Vancouver cohort and pathology review.  DGH Conceived of the study with KCW, DFS and helped with the preparation of the manuscript.  BC conceived of the study with KCW, DGH and DFS, scored the IHC on the TMAs, and contributed to the content of the figures and the text of the manuscript.  DFS conceived of the study with KCW, DGH and BC, scored the IHC on the TMAs, and contributed to the content of the figures and the text of the manuscript.  A modified version of the work in Chapter 4 is in preparation for publication: Wiegand KC, Hennessy BT, Leung S, Ju Z, McGahren M, Gilks B, Kalloger SE, Finlayson S, Stemke-Hale K, Lu Y, Zhang F, Huntsman DG,  Mills GB, Carey MS. Functional Proteomic Analysis of Endometrioid and Clear Cell Carcinomas:  Protein expression patterns are histotype-specific and tumours with loss of the ARID1A/BAF250a  are associated with AKT phosphorylation.  In Chapter 4, KCW was responsible for designing the project with MKC, drafting the manuscript with MKC, and helping to verify all data.  SL, SEK, and MC were responsible for statistical analysis of the RPPA data . BG and SF assisted with clinical verification of the cases, BTH, KSH, YL, and FZ were responsible for design and optimization of the RPPA. DGH and GBM contributed to the study design and drafting of the manuscript.  MKC was responsible for compiling the first draft of the manuscript, and experimental work including all NanoPro work and western blotting.  The kinase inhibitor screening work in Chapter 5 is in preparation for publication.  The work in Chapter 5 was conducted in collaboration with the lab of Dr. T. Michael Underhill under technical supervision from Dr. Arthur V. Sampaio.  KCW was responsible for most of the experiments and analysis in this chapter, with the exception of the lentiviral infections performed by Adrian Wan at the BCCRC.  Clustering analysis  was performed by Steve Kalloger. Table 5.1 is modified with permission from Dr. Mike Anglesio, and an alternate version is in preparation for a publication related to all work on the ovarian cell line models. Technical assistance for qPCR and western blotting was provided by Winnie Yang and Sarah Maines-Bandiera, respectively.   v The work presented in Appendix A has been published: McAlpine JN, Wiegand KC, Vang R, Ronnett BM, Adamiak A, Köbel M, Kalloger SE, Swenerton KD, Huntsman DG, Gilks CB, Miller DM.  HER2 overexpression and amplification is present in a subset of ovarian mucinous carcinomas and can be targeted with trastuzumab therapy.  BMC Cancer.  2009; 9:433. Copyright © 2009, Biomed Central. Reprinted with permission.  JNM, DMM and KS provided clinical care and pertinent clinical information on the involved patients. KCW interpreted the FISH results. CBG, DGH, and MK reviewed the pathology, scored the IHC, and CBG and DGH confirmed discrepant FISH results. AA performed the IHC. BMR and RV provided cases and clinical histories from their institution, and shared their expertise in mucinous ovarian tumours. JNM, DGH, DMM, KCW and CBG participated in the design, and coordination of the manuscript with JNM, KCW, and CBG principally involved in its draft. All authors read and approved the final version of the manuscript.            vi Table of Contents  Abstract.................................................................................................................................... ii Preface..................................................................................................................................... iii Table of Contents ................................................................................................................... vi List of Tables ........................................................................................................................ viii List of Figures......................................................................................................................... ix List of Abbreviations ............................................................................................................. xi Acknowledgements .............................................................................................................. xiv Dedication .............................................................................................................................. xv  Chapter 1:  Introduction ........................................................................................................ 1 1.1 Background- epithelial ovarian cancer ........................................................................... 1 1.2 Clear cell and endometrioid carcinomas......................................................................... 2 1.2.1 Histological and molecular features of CCC and EC .............................................. 2 1.2.2 Clinical features of CCC and EC ............................................................................. 3 1.2.3 Origins of CCC and EC ........................................................................................... 4 1.3 Role of SWI/SNF complexes in cancer .......................................................................... 7 1.4 The SWI/SNF complex................................................................................................... 7     1.5 The SWI/SNF gene ARID1A .......................................................................................... 8 1.6 ARID1A in cancer ......................................................................................................... 12 1.7 Hypotheses and summary of objectives........................................................................ 14 1.7.1 Hypotheses ............................................................................................................. 14 1.7.2 Rationale and objectives ........................................................................................ 14 Chapter 2:  ARID1A gene mutations in endometriosis associated carcinomas............... 18 2.1 Introduction................................................................................................................... 18 2.2 Methods......................................................................................................................... 19 2.3 Results........................................................................................................................... 29 2.4 Discussion and conclusions .......................................................................................... 40  vii Chapter 3:  Examining the role of BAF250a (ARID1A) in other malignancies .............. 45 3.1 Introduction................................................................................................................... 45 3.2 Methods......................................................................................................................... 47 3.2.1 Materials and methods for BAF250a loss in diverse malignancies ....................... 47 3.2.2 Materials and methods for focused study of gastric carcinoma............................. 49 3.3 Results........................................................................................................................... 53 3.3.1 Results from the initial IHC screening of BAF250a on over 3000 cases .............. 53 3.3.2 Results for BAF250a IHC in additional cases of gastric carcinoma .................... 58 3.4 Discussion and conclusions .......................................................................................... 70 Chapter 4:  Functional proteomic analysis of endometrioid and clear cell carcinomas and associations with ARID1A/BAF250a............................................................................ 75 4.1 Introduction................................................................................................................... 75 4.2 Methods......................................................................................................................... 76 4.3 Results........................................................................................................................... 79 4.4 Discussion and conclusions .......................................................................................... 91 Chapter 5:  In vitro models and novel therapeutic targets for CCC ................................ 96 5.1 Introduction................................................................................................................... 96 5.2 Methods......................................................................................................................... 98 5.3 Results......................................................................................................................... 105 5.4 Discussion and conclusions ........................................................................................ 125 Chapter 6:  Overall summary, conclusions, and future directions ................................ 135 6.1 Overall summary and conclusions .............................................................................. 135 6.2 Future directions ......................................................................................................... 140  Bibliography ........................................................................................................................ 147 Supplementary Appendix................................................................................................... 160 Appendix A:  Supplemental material for Chapter 1 ......................................................... 160 Appendix B:  Supplemental material for Chapter 2.......................................................... 189 Appendix C:  Supplemental material for Chapter 4.......................................................... 203 Appendix D:  Supplemental material for Chapter 5 ......................................................... 215  viii  List of Tables Table 2.1 Results of Illumina RNA sequencing and exon resequencing ............................... 31 Table 2.2 ARID1A mutation status in the discovery and validation cohorts.......................... 34 Table 3.1 BAF250a IHC results for diverse malignancies..................................................... 54 Table 3.2 Clinicopathologic characteristics for two cohorts of gastric cancer ...................... 61 Table 3.3 Summary of IHC and ISH staining for biomarkers vs. clinical variables.............. 62 Table 3.4 Heatmap statistics................................................................................................... 67 Table 4.1 RPPA patient sample characteristics by histotype ................................................. 84 Table 4.2 Effect of PIK3CA mutations, ARID1A mutations, and BAF250a expression on AKT phosphorylation ............................................................................................................ 85 Table 5.1 Molecular features of cell lines used in kinase inhibitor screening ..................... 112 Table 5.2 Secondary screening results ................................................................................. 116 Table 5.3 IC50 values of top four kinase inhibitors ............................................................. 117 Table 5.4 Summary of cytostatic and cytotoxic effects ....................................................... 118          ix List of Figures Figure 1.1 Endometrioid and clear cell carcinoma H&E......................................................... 4 Figure 1.2 Progression model of endometriosis to ovarian carcinoma.................................... 6 Figure 1.3 The SWI/SNF complex and BAF250a protein (ARID1A gene) ........................... 10 Figure 1.4 ARID1A cDNA (from ATG start to TGA stop) and BAF250a protein ............... 11 Figure 2.1 Mutations discovered in ARID1A and the BAF250a protein .............................. 30 Figure 2.2 Illumina based resequencing of individual ARID1A exons ................................. 32 Figure 2.3 ARID1A-ZDHHC18 fusion prediction ............................................................... 33 Figure 2.4 Results of immunohistochemical analyses of BAF250a expression ................... 37 Figure 2.5 Clear cell carcinoma and associated endometriosis ............................................. 39 Figure 3.1 Immunostaining for BAF250a expression in diverse malignancies .................... 55 Figure 3.2 High-grade malignancies of the endometrium .................................................... 56 Figure 3.3 Biopsy of atypical endometriosis with BAF250a IHC......................................... 57 Figure 3.4 Distribution of clinical variables in two gastric carcinoma cohorts ..................... 63 Figure 3.5 Representative images of invasive gastric carcinoma .......................................... 64 Figure 3.6 Loss of BAF250a (anti-ARID1A) expression in gastric carcinoma ..................... 65 Figure 3.7 Heatmap clustering by biomarker status in gastric carcinoma ............................. 66 Figure 3.8 Survival curves for BAF250a and HER2 in gastric carcinoma............................ 69 Figure 4.1 Hierarchical clustering of samples and proteins analyzed by RPPA ................... 80 Figure 4.2 Up-regulated and down-regulated proteins in CCC and EC ............................... 82 Figure 4.3 Heat map of RPPA samples ................................................................................. 87 Figure 4.4 Western blot results from siRNA knockdown of BAF250a ................................ 90 Figure 5.1 Lentiviral knockdown of ARID1A in RMG1 and HCT116 ............................... 106  x Figure 5.2 Transient transfection of ARID1A mutants ........................................................ 108 Figure 5.3 Overview of kinase inhibitor screening on 19 ovarian cancer cell lines ........... 111 Figure 5.4 Hierarchical clustering of kinase inhibitor hits .................................................. 114 Figure 5.5 Dose response titration of GSK461364, Pelitinib and Tovok ........................... 121 Figure 5.6 Apoptosis assay for Caspase-3/7 ....................................................................... 122 Figure 5.7 GSK461364, Pelitinib and Tovok treatment of CCC cell lines ......................... 123                 xi List of Abbreviations: ARID1A- AT rich interactive domain 1A ATRA- All-trans retinoic acid BAF- BRG/BRM associated factors BAF250a- BRG/BRM associated factor 250a BRCA1/2- Breast cancer associated gene 1/2 BRG1- Bhrama-related gene 1 BRM- Bhrama CCC- Clear cell carcinoma CCLE- Cancer Cell Line Encyclopedia cDNA- complementary DNA aCGH- array-comparative genomic hybridization cDNA- complementary DNA ChIP- Chromatin immunoprecipitation CNV- Copy number variation COSP- Calculator for Ovarian Carcinoma Subtype Prediction CTNNB1- Catenin (cadherin) associated protein beta-1 (beta-catenin) DAB- Diaminobenzene DLBCL- Diffuse large B-cell lymphoma EBV- Epstein-Barr virus EC- Endometrioid carcinoma EGC- Early gastric cancer EGFP- Enhanced green fluorescent protein  xii EGF- Epidermal growth factor EGFR- Epidermal growth factor receptor EOC- Epithelial ovarian cancer FACS- Fluorescence activated cell sorting FIGO- International Federation of Gynecology and Obstetrics FISH- Fluorescence in situ hybridization FFPE- Formalin-fixed paraffin-embedded GIST- Gastrointestinal stromal tumour H&E- Hematoxylin and eosin stain HER2- Human epidermal growth factor receptor 2 HGS- High-grade serous carcinoma HIC1- Hypermethylated in cancer 1 HNF1B- Hepatocyte nuclear factor 1B GC- Gastric cancer IC50- Half maximal inhibitory concentration ICC- Immunocytochemistry IHC- Immunohistochemistry LCM- Laser-capture microdissection LGSC- Low-grade serous carcinoma LOH- Loss of heterozygosity MCL- Mantle cell lymphoma MEF- Mouse embryonic fibroblasts MMR- Mismatch repair  xiii MMTV- Mouse mammary tumour virus MOC- Mucinous ovarian carcinoma MSI- Microsatellite instability NMD- Nonsense mediated decay ORF- Open reading frame PARP- Poly (ADP-ribose) polymerase PCR- Polymerase chain reaction PFSI- Progression free survival interval PIK3CA- Phosphatidylinositol 3-kinase PLK1- Polo-like kinase 1 PTEN- Phosphatase and tensin homolog RA- Retinoic acid RARA- Retinoic acid receptor alpha REB- Research ethics board RECIST- Response evaluation criteria in solid tumours RPPA- Reverse phase protein array RTK- Receptor tyrosine kinase RTTA- Reverse tetracycline-controlled trans-activator SISH- Silver in situ hybridization SWI/SNF- SWItch/Sucrose NonFermentable SNV- Single nucleotide variant TMA- Tissue microarrray WGA- Whole-genome amplified  xiv  Acknowledgements  I am very grateful to have had the opportunity to work with Dr. David Huntsman and the OvCaRe team.  David has been a great mentor, and his enthusiasm, eternal optimism, and dedication to ovarian cancer research has been inspiring.  I would sincerely like to thank my committee members, Dr. Blake Gilks, Dr. Michael Underhill, Dr. Jessica McAlpine, and Dr. Wan Lam (my committee chair) for all of the support and guidance they have provided over the course of my project.  During the past five years, I was extremely lucky to work with an exceptional group of technologists, scientists, clinicians, statisticians, and bioinformaticians from the Huntsman lab and OvCaRe, the Underhill lab, GPEC, CTAG, UBC, the BCCA and the BCCRC- thank you all for your help, support and encouragement.  Thank you also to the MSFHR foundation for the support provided by the award in honour of Dr. Nelly Auersperg; her research legacy continues to inspire the ovarian cancer research community.          xv        To my parents Heino Michael Wiegand and Joyce Caroline Wiegand (1944-1999) Thank you for giving me a lifetime of unconditional love, support, encouragement, and friendship.  1 Chapter 1:  Introduction 1.1 Background- epithelial ovarian cancer Ovarian cancer is the eighth most common cancer diagnosed in women worldwide, and epithelial ovarian cancer (EOC) is the fifth leading cause of cancer death in the United States (1, 2). Recent evidence suggests that instead of being just one disease, EOC is actually a heterogeneous group of at least five different subtypes, which includes high-grade serous (HGS), endometrioid carcinoma (EC), clear cell carcinoma (CCC), mucinous ovarian carcinoma (MOC) and low-grade serous carcinoma (LGSC) (3).  These subtypes of epithelial ovarian cancer have been divided into two categories, Type I and Type II (4). The Type I tumours are relatively genetically stable and include endometrioid, clear cell, mucinous, and low-grade serous carcinomas (4).  Endometrioid and clear cell carcinomas are thought to originate from atypical endometriosis (5, 6), while mucinous tumours are of unknown origin (7) with approximately 19% having amplification of HER2 (Appendix A (8), and (9)), and low-grade serous carcinomas are thought to originate from serous borderline tumours (7). The majority of Type II cancers are the genetically unstable high grade serous carcinomas accounting for approximately 70% of all EOC cases (10).  The Type II ovarian cancers are under intense investigation by The Cancer Genome Atlas Project (11). High grade serous carcinomas frequently harbor mutations in BRCA1, BRCA2, and TP53 (4, 12), and the majority are thought to originate from the fallopian tube (13, 14). Clear Cell Carcinomas (CCC) and the other Type I cancers, on the other hand, are still relatively understudied.  CCC is the second most common subtype (approximately 10-12% of cases) (3, 10), with a higher rate of 20-25% in Japan (15, 16) and is the second leading cause of ovarian cancer associated  2 death. ECs account for approximately 10% of all EOC cases diagnosed, while MOCs account for only 3-4% of ovarian carcinomas, and LGSCs account for approximately 5% (3). 1.2 Clear cell  and endometrioid carcinoma 1.2.1 Histological and molecular features of CCC and EC CCC, shown in Figure 1.1A, can often be identified histologically by the presence of clear cells (i.e. cells with abundant clear cytoplasm due to glycogen accumulation) and hobnail cells (17).  It has been found that CCC frequently expresses a homeobox transcription factor that functions during embryogenesis, hepatocyte nuclear factor–1beta (HNF-1β) (18). CCC rarely expresses any of the biomarkers commonly associated with high grade serous or other ovarian cancers (19), and the distinctive CCC immunophenotype can be helpful in cases that are diagnostically problematic (20). CCCs typically have a low mitotic rate (19), and are relatively genetically stable.  They do not exhibit the complex karyotypes or chromosomal instability associated with high grade serous cancers (21, 22). Before the work presented in this dissertation began, the gene identified as most frequently mutated in CCC was PIK3CA, a well-known oncogene involved in tumour progression (present in 14%-50% of cases) (12, 23-26).  Mutations in the tumour suppressor genes BRCA1, BRCA2, and TP53 that are common in high-grade serous cancers are typically absent in CCCs (4, 12).  Endometrioid carcinoma (EC), shown in Figure 1.1B, resembles carcinoma of the endometrium (3).  EC frequently can be identified by the presence of round or tubular glands, or more rarely by the presence of small rosette like glands (known as a microglandular pattern) (3). Clear cells may be present in ECs, however they can be distinguished from  3 CCCs based on their architectural pattern and immunostaining profiles (3, 19). Before the work presented in this dissertation was undertaken, the most commonly mutated genes found in EC were in CTNNB1 (β-catenin) in nearly 60% of cases, PTEN and PIK3CA mutation in 20%, MSI (microsatellite instability) in 12-19%, and less frequently mutations in KRAS and BRAF (approximately 10% of cases) (4, 27, 28). Greater than 80% of ECs have nuclear β- catenin detectable by IHC, which may aid in diagnosis (3).  1.2.2 Clinical features of CCC and EC Most ECs are grade 1 or 2, and confined to the pelvis at the time of diagnosis (early stage) with excellent prognosis.  When they are diagnosed correctly, they are actually the type of EOC with the best prognosis (3). CCCs are often diagnosed at an early stage, with 80% of cases presenting with stage I or II carcinoma (29, 30), however survival rates for stage I/II CCC are significantly lower (60%) compared to patients with other ovarian cancer subtypes presenting with stage I/II disease (15, 31).  It is becoming more widely accepted that ovarian carcinoma subtypes are essentially different diseases (3, 19), however it is still common practice to treat them all with platinum/taxane chemotherapy.  Unfortunately, CCCs respond extremely poorly to this treatment (15, 32, 33), showing response rates of 15% compared to 80% for high grade serous carcinomas (19). The lack of complex karyotypes or chromosomal instability associated with high-grade serous cancers (21, 22) may factor into the chemoresistance exhibited by CCCs.  Unfortunately, there are currently no effective anti- cancer agents for CCCs; there is much need to elucidate the mechanisms of this disease and establish appropriate subtype specific treatment strategies.  4  Figure 1.1 (A) Clear cell carcinoma (CCC) and (B) endometrioid carcinoma (EC) with H&E staining. The EC is characterized by a glandular pattern resembling the endometrium. The CCC exhibits frequent clear cells and hobnail cells.  1.2.3 Origins of clear cell and endometrioid carcinoma It has been reported that 42-43% of EC and 54-70% of CCCs occur in the presence of endometriosis (34, 35), although it is suggested that this number is likely much higher due to difficulties in sampling. Endometriosis occurs in 7-15% of women (36), typically during their reproductive years, and is characterized by endometrial-like tissue implants that occur in areas outside of the uterus, most commonly on the ovaries (37). These ectopic endometrial cells respond to hormonal changes in the same way as the cells in the endometrium, therefore symptoms are often worse during the menstrual cycle (38).  The main but not universal symptoms of endometriosis are pelvic pain, dyspareunia, and dysmenorrhoea (36).  The shed endometriotic tissue has no way to exit the body, so it results in inflammation, prostaglandin production, and pain (39). Endometriosis is seen in up to 50% of women with infertility (40).  5 Several theories exist for the origins of endometriosis; the most commonly accepted is the theory of retrograde flow, proposed by Sampson in 1925, whereby menstrual endometrium effluxes retrograde through the fallopian tubes, then becomes implanted on the ovaries and other surfaces (41).  However, only a small percentage of women that have retrograde flow actually develop endometriosis (42).  An alternate theory is that viable endometrial cells spread through the circulatory or lymphatic system then implant at distant sites (43), which would explain the implantation of endometrial tissue at such distant sites as the lung and pericardium.  Endometriosis provides fertile ground for the development of cancers, as it displays several of the key characteristics seen in cancer development; it manages to escape apoptosis and immune surveillance, it attaches and invades normal tissue, it creates a new vascular supply, and it proliferates profusely (42, 44).  Currently, little is known about the molecular transitions that cause the progression from normal endometriosis to atypical endometriosis, and to either clear cell or endometrioid carcinoma.  Mutations in the PTEN gene have been described in 20% of endometriotic cysts (45) and conditional expression of either oncogenic K-ras or deletion of the PTEN tumour suppressor in the ovarian surface epithelium has been found to induce endometriosis in a mouse model (46).  However, KRAS mutations are not seen in human endometriosis and are rare in endometriosis associated ovarian cancers (46). Good evidence for the molecular basis of transformation comes from cases of ovarian cancer adjacent to endometriosis or arising from endometriosis that have shown common genetic LOH alterations in the endometriosis and cancer, indicating a possible genetic transition spectrum (47) (Figure 1.2). Progression model (circa 2009) of endometriosis to ovarian endometrioid and clear cell carcinoma (4, 11, 21-26, 45)Figure 1.2  7 1.3 Role of SWI/SNF complexes in cancer The association between the SWI/SNF complex and cancer has also been the subject of several recent reviews (48-50).  Before this project was undertaken, several SWI/SNF complex proteins had been implicated in cancer including BAF47/SNF5, a tumour suppressor gene located on chromosome 22 (22q11) in a region that is often rearranged in pediatric rhabdoid tumours (51).  Additionally, the SWI/SNF associated ATPases (BRG1 and BRM) both showed loss of expression in 15-20% of primary non-small cell lung cancers (49, 52), and the complex had been shown to regulate key cancer genes, including TP53, BRCA1, and E2F1 (53).  The BRG1 ATPase was shown to directly interact with BRCA1 (54), and also RB1 to silence S-phase-inducing transcription factor E2F1, resulting in inhibited proliferation (55).  1.4 The SWI/SNF complex Chromosomal DNA is highly organized.  At the most basic level of genomic organization, DNA is wound around proteins called histones to form a complex structure called chromatin. The chromatin consists of nucleosomes, which contain 147bp of DNA wrapped tightly around an octamer of histone proteins (two molecules each of H2A, H2B, H3, and H4). These nucleosomes are connected to each other by linker DNA, creating an arrangement typically known as “beads-on-a-string”.  Supercoiling of chromatin creates a higher order structure known as heterochromatin, which is inaccessible to transcriptional machinery. Chromatin remodeling, either through covalent modification of histones or through the mobilization of nucleosomes, is required before DNA can be accessed for transcriptional initiation or other nuclear processes such as DNA replication or repair.  8  The SWI/SNF (SWItch/Sucrose NonFermentable), initially identified in the yeast Saccharomyces cerevisiae, is present in all eukaryotes and is essential for many cellular processes including development, differentiation, proliferation, DNA repair, and tumour suppression (49).  The complex is comprised of one of two mutually exclusive core catalytic ATPases, BRM (Brahma) or BRG1 (Brahma-Related Gene 1) (56, 57), together with conserved core subunits and accessory proteins termed BAFs (BRM- or BRG1- associated factors) (Figure 1.3). This protein complex uses the energy provided by ATP hydrolysis to mobilize nucleosomes, which modulates accessibility to transcription and DNA repair machinery.  It is typically associated with transcriptional activation or repression and functions at promoter regions.  The specific combination of proteins within different SWI/SNF complexes is believed to confer specificity with respect to gene regulation.  1.5 The SWI/SNF gene ARID1A The ARID1A gene (AT-rich interactive domain 1A), which encodes the protein BAF250a (also known as p270), is the human equivalent of a highly evolutionarily conserved family of proteins first identified by co-purification with the SWI/SNF complex in the yeast Saccharomyces cerevisiae (58, 59).  This protein family, with more than a dozen members, share a similar non-sequence-specific DNA binding domain known as ARID (AT-rich interactive domain) (60, 61). The ARID1A gene located at 1p36.11 has two different transcript variants encoding different isoforms, the longer and most commonly studied transcript is NM_005015.4, with twenty coding exons, and a transcript length of 8,577 bp (for a translated protein length of 2,285 amino acid residues) (Figure 1.4), which encodes the  9 protein BAF250a. In addition to the conserved ARID domain, ARID1A (BAF250a) contains a glucocorticoid receptor binding domain at its C-terminus (62), and a HIC1 binding domain (63) (see Figure 1.4).  The tumour suppressor gene HIC1 (hypermethylated in cancer 1) encodes a transcriptional repressor that is epigenetically silenced or deleted in many cancers (64). There is evidence that the two repression domains of HIC1 interact with BAF250a (through residues 1355-1451), facilitated by the presence of the core catalytic ATPase BRG1 but not the mutually exclusive counterpart BRM (63).  BRG1 containing SWI/SNF complexes contain either BAF250 or BAF180, while BRM complexes contain only BAF250 (Figure 1.3).  There are two BAF250 proteins, which are encoded by separate genes:  BAF250a is encoded by the ARID1A gene, and BAF250b is encoded by the ARID1B gene.  These proteins are mutually exclusive within BRG1 or BRM containing SWI/SNF complexes (65).  Co-immunoprecipitation studies indicate that BAF250a and BAF250b interact with BRG1 and BRM through their C-terminal domains (66), and the interaction between BAF250a and BRG1 has been shown to be required for transactivation of the steroid hormone responsive MMTV (mouse mammary tumour virus) promoter (67).        10        Figure 1.3 The SWI/SNF complex and BAF250a protein (ARID1A gene). (A) The 15 SWI/SNF complex subunits and their associated genes. (B) An example of a SWI/SNF complex containing BAF250a. The arrow indicates that either BAF250a or the mutually exclusive subunit BAF250b may be in the complex. The PBAF SWI/SNF complex (not shown) has BRG1 and contains BAF200 and BAF180 instead of BAF250a/b.  Constant core components of the complex are indicated in blue, the two ATPase subunits are shown in green, and BAF250a is highlighted in pink.     11       Figure 1.4   ARID1A cDNA (from ATG start to TGA stop) and BAF250a protein. The ARID1A gene is homologous to the Osa gene in Drosophila melanogaster and the SWI1 gene in Saccharomyces cerevisiae. BAF250a has a DNA binding or ARID domain of approximately 100 amino acids, and multiple LXXLL (where L is leucine and X is any amino acid) motifs, which potentially interact with nuclear hormone receptors. The 20 exons of ARID1A are shown (grey boxes) above a schematic of the BAF250a protein (blue). In BAF250a, the ARID DNA binding domain is shown (pink), as well as the putative HIC1 binding domain (green), and four LXXLL motifs (yellow), which facilitate interaction with glucocorticoid receptor.  12 1.6 ARID1A in cancer Prior to the work presented in Chapter 2, there was some indirect evidence supporting a role for ARID1A in cancer. The region 1p36 has long been suspected to contain tumour suppressor genes (68, 69). Rearrangements and deletions of ARID1A had been identified in a primary breast cancer and a lung cancer cell line, respectively (70).  BAF250a loss had also been observed in cervical and breast carcinoma cell lines (71).  A screen of 241 tumours had demonstrated that ARID1A transcript levels were decreased in approximately 6% of cancers in general, and specifically in 30% of renal carcinomas and 10% of breast carcinomas; of the 14 ovarian cancers studied, all had normal ARID1A levels (65), however this may be because the ovarian cancers studied were of the high grade serous subtype.  In the final paragraph of a paper by Normal Nagl in 2007 the author expressed surprise that ARID1A had not yet been identified as a bona-fide tumour suppressor (72).  BAF250a and BAF250b typically co-exist within cells (as opposed to being tissue-specific variants), and their inclusion in SWI/SNF complex may confer specificity of function. In several studies of ARID1A/BAF250a function, it was found that ARID1A was associated with cell cycle control regulation, and seemed to act as a counterbalance to the actions of ARID1B- where ARID1A associated SWI/SNF complexes acted as the brakes for cell cycle control and ARID1B associated complexes acted as the accelerator (72).  Using an siRNA- knockdown approach to clarify the roles of ARID1A (BAF250 and ARID1B (BAF250b), Nagl et al. demonstrated that ARID1A is required for differentiation associated cell cycle arrest (73, 74), while cells deficient in ARID1B exhibit delayed re-entry into the cell cycle but still undergo cell cycle arrest (72).  Through co-immunoprecipitation (CoIP) assays, the  13 Nagl group demonstrated that ARID1A associates with repressive transcription factor on genes important for cell cycle progression, including E2F4 and E2F5, as well as Sin3a/HDAC1/2, while ARID1B is associated with the two repressive transcription factors, plus HDAC3, and plays a minor role in transcriptional repression, but exclusively associates with the activating transcription factor E2F1 (72). ARID1B function is required for the activation of cyclin A, cyclin E, cdc2, and plays a critical role in the activation of the oncogene c-Myc (72).  Additional early functional evidence for a possible role of ARID1A as a tumour suppressor came through CoIP experiments where BAF250a was found to be associated with HIC1 (Hypermethylated in Cancer 1- a known sequence specific transcriptional repressor and bona-fide fide tumour suppressor), in a BRG dependent manner (63).  BAF250a was associated with promoters of two of the repressor HIC1’s known target genes E2F1 (an activating transcription factor associated with cell-cycle regulated promoters) and ATOH1 (a transcription factor frequently up-regulated in medulloblastoma) (63).  ARID1A and ARID1B have been associated with human embryonic stem cell pluripotency and self-renewal (75, 76).  Chromatin immunoprecipitation coupled with oligo-microarrays of known human promoter regions demonstrated that the repressor E2F4 is found at the ARID1A promoter while OCT4, SOX2, and NANOG, key regulators of human embryonic stem cell pluripotency and self-renewal, co-occupy the ARID1B promoter (75).  The study by Gao et al. confirmed that BAF250a is essential for ES cell pluripotency, with ablation of BAF250a causing the down-regulation of genes associated with self renewal (such as FGF4,  14 SOX2, OCT4 and UTF1), while genes associated with early development were found to be up-regulated (GATA4, GATA6, CITED1, CITED2, DAB2, and SOX17) (76). Interestingly, it was necessary to create conditional knockout models for BAF250a as BAF250a-null embryos arrested at E6.5, indicating that BAF250a is essential for early embryogenesis (76).  1.7 Hypotheses and summary of objectives  1.7.1 Hypotheses 1. Clear cell and endometrioid carcinomas, being relatively genetically stable, may have characteristic mutations that define them.  These characteristic mutations should be identifiable by interrogating a limited number of cases. 2. The mutations defining clear cell and endometrioid carcinomas would likely be early events during oncogenesis. These specific genetic events could lead the progression of benign endometriosis to atypical endometriosis capable of transforming into ovarian clear cell and endometrioid carcinomas. 3. Clear cell carcinomas may have specific vulnerabilities created by their characteristic genetic deficiencies that could render them sensitive to targeted therapeutic strategies.  1.7.2 Rationale and objectives The strategy for determining the genetic changes underlying clear cell and endometrioid carcinomas was to utilize next generation sequencing technology to interrogate the transcriptomes of patient tumour samples.  Our group had successfully utilized this approach to detect pathognomonic mutations in FOXL2 in granulosa-cell tumours of the ovary (77).  15 This technology can fully interrogate genomes or transcriptomes at a single base resolution for single nucleotide variants, insertions or deletions, splice variants, copy number changes, and genomic rearrangements.  In the case of paired-end sequencing, next generation sequencing technology generates millions of randomly fragmented, short sequenced reads that flank longer unsequenced regions.  Data is generated using a four-color DNA “sequencing-by-synthesis” technology followed by fluorescence detection. After completion of the first read, templates are regenerated in situ to allow for a second read from the opposite end of the fragments, producing end-sequence pairs. Resulting paired-end reads are then aligned to a reference sequence (e.g. NCBI build 36.1, hg18), which produces relevant data for each read (such as location within the transcriptome, quality of read, number of mismatches, and paired-end flags).  Single nucleotide variants (SNVs) are predicted based on discrepancies between the reference genome and the aligned mapped reads.  Fusion transcripts and other rearrangements can be recognized by identifying all paired-ends that did not align in pairs to the human genome.  It would have been possible to use this next- generation sequencing technology for whole genome analysis, however this was much more costly than RNA-seq (i.e. whole transcriptome analysis, which sequences cDNAs generated from total mRNA). We therefore chose to use the latter approach, which is described in detail in Chapter 2, with the discovery of frequent mutations in the ARID1A gene in clear cell and endometrioid carcinomas.  To determine if the mutations in ARID1A discovered in clear cell and endometrioid carcinomas were early events during oncogenesis, pathology review was performed on all of the cases of clear cell and endometrioid carcinomas in the tumour bank to identify cases with  16 the suspected precursor lesion, atypical endometriosis, as well as distant (normal) endometriosis that could be used for laser capture microdissection (LCM) to obtain DNA that could be Sanger sequenced for the same mutations observed in the tumours. The results from this are described in Chapter 2.  The objective of the work presented in Chapter 3 was to determine whether ARID1A/BAF250a loss is common in other malignancies besides ovarian cancer. Immunohistochemistry (IHC) for BAF250a was performed on tissue microarrays (TMAs) in more than 3000 cancers, including carcinomas of breast, lung, thyroid, endometrium, kidney, stomach, oral cavity, cervix, pancreas, colon, and rectum, as well as endometrial stromal sarcomas, gastrointestinal stromal tumours (GIST), sex cord-stromal tumours and four major types of lymphoma (diffuse large B-cell lymphoma [DLBCL], primary mediastinal B-cell lymphoma [PMBCL], mantle cell lymphoma [MCL], and follicular lymphoma).  In the second part of Chapter 3 the loss of ARID1A in gastric cancer is further studied in a large series of gastric cancer cases from the BC Cancer Agency (n=173) and a second set from Toronto (n=80).  The objective of the work in Chapter 4 was to try and determine functional consequences of ARID1A/BAF250a loss in clear cell and endometrioid carcinoma.  To accomplish this, the proteomic patterns of expression of the three major ovarian cancer subtypes (HGSC, CCC and EC) were examined using the reverse phase protein array (RPPA) platform. Whole tumour lysates from 127 ovarian carcinomas, including 34 CCC, 28 EC and 65 HGSC were  17 profiled by RPPA. The ARID1A mutation and IHC status was previously defined for 96 of the ovarian cancers in the RPPA series.  The objective of the work presented in Chapter 5 was to determine whether clear cell carcinomas with defects in ARID1A possess an Achilles heel that may allow for a synthetic- lethal treatment strategy, in a similar way that cell lines and tumours with BRCA mutations are defective in the homologous recombination pathway and show sensitivity to PARP inhibitors. Several approaches were taken to try and generate an appropriate model system for screening, i.e. an isogenic cell line pair that could be used together in parallel to screen for therapeutic targets or compounds preferentially effective with ARID1A deficiency.   The cell line pairs most closely resembling suitable isogenic cell line pairs were the RMG1 and HCT116 cell lines with knockdown of BAF250a achieved through lentiviral delivery of an shRNAmir targeting ARID1A.  A kinase inhibitor screen was performed on these nearly isogenic cell line pairs together with an additional 18 ovarian cancer cell lines, including ten clear cell carcinoma lines with known ARID1A mutation status.  The kinase inhibitor panel consisted of over 340 compounds, with many that were successfully used in at least Phase 1 clinical trials, as well as a number of target specific kinase inhibitor tool compounds.  From this screen, several compounds were identified that effectively inhibited the growth of CCC cell lines.  These compounds included GSK461364, an inhibitor of PLK1 as well as Pelitinib and Tovok, both inhibitors of EGFR.     18 Chapter 2:  ARID1A gene mutations in endometriosis associated ovarian carcinomas  2.1 Introduction Although ovarian clear cell carcinoma (CCC) is the second most common histologic subtype in North America (12% of cases and more frequent in Japan (78)) and the second leading cause of ovarian cancer associated death (10), it is relatively understudied. Since CCCs have a low mitotic rate (19, 21), are relatively genetically stable, and do not exhibit the complex karyotypes or chromosomal instability associated with high grade serous cancers (21, 22, 79, 80), we hypothesized it would be possible to find recurrent mutations characteristic of the disease by sequencing the whole transcriptomes of a relatively small number of cases.  As mentioned in Chapter 1, CCCs exhibit a lack of sensitivity to platinum chemotherapy; however, this remains the adjuvant treatment of choice, as effective alternatives have not been identified (15, 32, 33).  Both CCC and ovarian endometrioid carcinoma (EC) are commonly associated with endometriosis (81, 82), but the genetic events associated with the transformation of endometriosis into CCC and EC are largely unknown. We utilized Illumina for whole-transcriptome sequencing (RNA sequencing) for 18 CCCs and the CCC derived cell line TOV21G, with the same methods and strategy as previously described (77). Through this, variants in the ARID1A gene were identified in our initial ‘discovery’ cohort, which was then confirmed in a larger cohort of ovarian carcinomas and associated endometriosis. BAF250a, the protein encoded by ARID1A (56, 57), is one of the accessory subunits of the SWI/SNF complex that is believed to confer specificity with respect to regulation of gene expression.  The SWI/SNF chromatin remodeling complex is present in all eukaryotes, and is involved in the regulation of many cellular processes including  19 development, differentiation, proliferation, DNA repair, and tumour suppression (49). The complex utilizes ATP to mobilize nucleosomes, thereby modulating the accessibility of promoters to transcriptional activation or repression. Prior to this study, ARID1A rearrangement had been identified in a breast cancer and a lung cancer cell line, respectively, and it had been suggested that ARID1A is a potential tumour suppressor gene (70). Mutations or other aberrations of this gene had not been described in ovarian carcinomas, and this was the first study to suggest that ARID1A is a tumour suppressor in CCC and EC. In parallel with our discovery, another group found mutations in ARID1A in 57% of their CCC cohort, and the journals to which we had submitted manuscripts kindly arranged to publish the studies of ARID1A in ovarian clear cell carcinomas back-to-back (83, 84).  2.2 Methods Patients and samples Eighteen ovarian CCCs from the OvCaRe (Ovarian Cancer Research) frozen tumour bank and one CCC cell line (TOV21G) (85) were selected for RNA sequencing (discovery cohort). The 18 tumour specimens were collected via a BCCA prospective tumour banking protocols for ovarian tumours (Formation of Gynecological Tumour Bank). Patients were approached for written informed consent before undergoing surgery, to donate tissue surplus to diagnostic requirements plus a blood sample, for use in a research ethics board (REB) approved research protocol. All patients were informed at the time of consent about the potential risks of loss of confidentiality arising from use of their samples in research and that none of the research study data would ever form part of the clinical record or be reported back to the care physicians. Specimen materials were released to investigators only with an  20 REB approved study certificate, via a privacy guardian who anonymizes samples such that research investigators have no access to patient identifiers or information that could be used to directly identify a subject. The REB ethics protocol for genome-scale datasets stipulates that primary datasets from the tumour transcriptomes will not be released into the public domain, but can be made available via a tiered access mechanism to named investigators of institutions agreeing by a materials transfer agreement to honor the same ethical and privacy principles as the BCCA investigators. To determine the frequency of ARID1A mutations in CCC and other ovarian cancer subtypes we used Illumina based targeted exon resequencing to interrogate the DNA sequence of 101 CCCs (in addition to the 19 cases for RNA sequencing, described above), 33 ECs, 76 HGS carcinomas and the CCC derived cell line ES2 (86) (mutation validation cohort).  Pathological review All tumour samples were independently reviewed by a gynecologic pathologist before mutational analysis was performed. In cases in which the review diagnosis differed from the source diagnosis, the samples were reviewed by a second gynecologic pathologist, who acted as an arbiter. Both review pathologists were unaware of the results of genomic studies. From our cohort of 119 CCCs (both discovery cohort and mutation validation cohort) and 33 ECs (mutation validation cohort), 86 CCCs and all 33 ECs were examined to determine if endometriosis was present at the time of surgery (a summary of the endometriosis status of the cases is provided in Appendix B Supplemental Table 2.5)    21 DNA and RNA extraction from patient samples DNA and RNA were extracted using standard methodologies, as previously described (77). In cases for which insufficient DNA for ARID1A resequencing was available whole genome amplification (WGA) was used to extend the DNA template, however mutations were all confirmed using non-WGA treated DNA.  Paired-end RNA sequencing and analysis (RNA-sequencing) RNA sequencing was performed as previously described (77). Genomic DNA for cases from both discovery and mutation validation cohorts were subjected to Illumina-based exon resequencing.  First, double stranded cDNA was synthesized from polyadenylated RNA, and the resulting cDNA was sheared. The 190- 210bp DNA fraction was then isolated and PCR amplified to generate the sequencing library, as per the Illumina Genome Analyzer paired end library protocol (Illumina Inc., Hayward, CA). The resulting libraries were sequenced on an Illumina GAii. Short read sequences obtained from the Illumina GAii were mapped to the reference human genome (NBCI build 36.1, hg18) plus a database of known exon junctions (87)  using MAQ (88) in paired end mode. The data from RNA sequencing is summarized in Appendix B Supplemental Table 2.1. Single nucleotide variants (SNVs) were predicted using a Bayesian mixture model, SNVmix, as previously described (77, 89). Only bases with > Q20 base quality were considered to minimize errors. SNVs were cross-referenced against dbSNP version 129 and published genomes in order to eliminate any previously described germline variants (77). In addition to variants in ARID1A, CTNNB1 (C110G (S37C), NM_001904.3) somatic mutations were detected in CCC02 and CCC03 and validated by PCR amplification and Sanger sequencing in both tumour and germline DNA from these  22 cases. Additionally, two variants were predicted based on RNA sequencing data in the TOV21G cell line in PIK3CA (C3139T (H1047Y), NM_006218.2) and KRAS (G37T (G13C), NM_004985.3), which were validated by PCR amplification and Sanger sequencing. Though variants in BRAF were observed in the RNA sequencing data, none of these predictions passed validation by Sanger sequencing in tumour DNA.  Illumina-based targeted exon resequencing of ARID1A Genomic DNA for the cases from both the discovery and mutation validation cohorts were subjected to Illumina based targeted exon resequencing. All ARID1A exons were PCR amplified and individual amplicons were indexed, pooled, and sequenced. Automated primer design was performed using Primer3 (90) and custom scripting. Primers were designed to span annotated exons of ARID1A (UCSC build hg18) with an average PCR product size of 2067bp. Primers were synthesized by Integrated DNA Technologies at a 25nmol scale with standard desalting (IDT Coralville, IA) and tested in PCR using control human genomic DNA. Primer pairs that failed to generate a product of the expected size were redesigned. The sequences for the primers are provided in Appendix B Supplemental Table 2.2. Polymerase cycling reactions were set up in 96-well plates and comprised of 0.5 µM forward primer, 0.5 µM reverse primer, 1ng of gDNA template or 1ng of gDNA that was whole genome amplified using the REPLI-g® Mini/Midi (QIAGEN, Valencia, CA), 5X Phusion HF Buffer, 0.2 µM dNTPs, 3% DMSO, and 0.4 units of Phusion DNA polymerase (NEB, Ipswich, MA, USA). Reaction plates were cycled on a MJR Peltier Thermocycler (model PTC-225) with cycling conditions of a denaturation step at 98°C for 30 sec, followed by 35 cycles of [98°C for 10 sec, 69°C for 15 sec, 72°C for 15 sec] and a final extension step at  23 72°C for 10 min. PCR reactions were visualized by SybrGreen (Life Technologies, Carlsbad, CA, USA) in 1.2% agarose (SeaKem LE, Cambrex, NJ, USA) gels run for 90min at 170V to assess PCR success. Reactions were pooled (4µl per well) by template and sheared to an average size of 200bp using a Covaris E210 ultrasonic 96 well sonication platform (75 seconds, duty cycle 20, intensity 5, cycles/burst 200; Covaris Inc. Woburn, MA) and subjected to plate based library construction on a BioMek FX Laboratory Automation Workstation (Beckman Coulter, Brea, CA) using a modified paired-end protocol (Illumina, Hayward, CA). This involved end-repair and A-tailing of sheared amplicons followed by ligation to Illumina PE adapters and PCR amplification. At each step in the process, reactions were purified using solid phase reversible immobilization paramagnetic beads (Agencourt AMPure, Beckman Coulter, Brea, CA) in 96 well plates on the BioMek FX platform using custom in house programs. Purified adapter-ligated amplicons were PCR-amplified using Phusion DNA polymerase (NEB, Ipswich, MA) in 10 cycles using PE primer 1.0 (Illumina) and a custom multiplexing PCR Primer [5'CAAGCAGAAGACGGCATACGAGATNNNNNNCGGTCTCGGCATT CCTGCTGAACCG CTCTTCCGATCT-3’] where “NNNNNN” was replaced with 96 unique hexamer barcodes. Individual amplicons were indexed and pooled by plate and the 200-400bp size range purified away from adapter ligation artifacts on an 8% Novex TBE PAGE gel (Invitrogen, Carlsbad, CA, USA). Individual indexes enabled the deconvolution of reads deriving from individual samples concurrently sequenced from the same library. DNA quality was assessed and quantified using an Agilent DNA 1000 series II assay (Agilent, Santa Clara CA) and Nanodrop 7500 spectrophotometer (Nanodrop, Wilmington, DE) and subsequently diluted to 10nM. The final concentration was confirmed using a Quant-iT  24 dsDNA HS assay kit and Qubit fluorometer (Invitrogen, Carlsbad, CA). For sequencing, clusters were generated on the Illumina cluster station using v4 cluster reagents and paired- end 75bp reads generated using v4 sequencing reagents on the Illumina GAiix platform following the manufacturer’s instructions. Between the paired 75bp reads a third 7 base pair read was performed using the following custom sequencing primer [5’ GATCGGAAGAGCGGTTCAGCAGGAATGCCGAGACCG] to sequence the hexamer barcode. Image analysis, base-calling and error calibration was performed using v1.60 of Illumina’s Genome analysis pipeline. The exon-by-exon coverage of the ARID1A gene from Illumina based targeted exon resequencing is shown in Figure 2.2.  Data processing for ARID1A Illumina-based targeted exon resequencing Sequence reads from the ARID1A targeted exon resequencing experiment were aligned to the genomic regions targeted by the PCR primers using MAQ version 0.7.1. Each exon was assessed for coverage by enumerating all uniquely aligning reads to the targeted space. SNVs were determined by computing the allelic counts for each genomic position within the complete targeted space. All positions exhibiting an allelic ratio of at least 10% variant were considered for validation by Sanger sequencing. Insertions and deletions were predicted using the Maq indelpe program using 10% allelic ratio criteria for selection for experimental follow up. In addition, to determine a confidence measure for each SNV prediction, we applied a one-tailed Binomial exact test to each position covered as described in Shah et al (77), using all aligned reads to compute the expected distribution. Benjamini-Hochberg correction for multiple comparisons was applied to the resultant Binomial-test p-values to yield q-values for each position (91).  25 RNA-sequencing fusion prediction Gene fusions were predicted using deFuse (92). deFuse predicts gene fusions by searching paired end RNA-sequencing data for reads that harbor fusion boundaries. Spanning reads harbor a fusion boundary in the unsequenced region in the middle of the read, whereas split reads harbor a fusion boundary in the sequence of one end. deFuse searches for spanning reads with reads ends that align to different genes. Approximate fusion boundaries implied by spanning reads are then resolved to nucleotide level using dynamic programming based alignment of candidate split reads. See Figure 2.3.  Copy number analysis of Affymetrix SNP 6.0 arrays The Affymetrix SNP 6.0 arrays were normalized using CRMAv2 (93) using the default settings for performing allelic-crosstalk calibration, probe sequence effects normalization, probe-level summarization, and PCR fragment length normalization. Log ratios were then computed by normalizing against a reference generated using a normal dataset of 270 HapMap samples obtained from Affymetrix. Segmentation is performed using an 11-state hidden Markov model. This approach simultaneously detects and discriminates somatic and germline DNA copy number changes in cancer genomes. The hidden Markov model performs segmentation of the log ratio intensity data and predicts discrete copy number status for each resulting segment from the set of five somatic states (homozygous deletion, hemizygous deletion, gain, amplification, and high-level amplification), five analogous germline states, and neutral copy number. The boundaries of the segments provide candidate breakpoints in the genome as a result of copy number alteration events.  In all cases with Affymetrix SNP 6.0 data, only CCC04 contained a breakpoint in ARID1A. The segment  26 (chr1:26898389-27000523) is a homozygous deletion that breaks the gene near the 5’ end and truncates it. The published CNV map from 450 HapMap individuals (94) was studied to see whether any regions overlapping ARID1A were reported and none were found. Based on this, we predict that this is a somatic change.  Gene expression analysis For gene expression analysis, the RNA-sequencing reads initially were mapped to the genome (NCBI36/hg 18) using MAQ (0.7.1). We used the Sequence Alignment/Map (SAMtools 0.1.7) for downstream processing. Up to five mismatches was allowed. Raw expression values (read counts) were obtained by summing the number of reads that mapped to human genes based on the Ensembl database (Release 51). The initial gene expression values were normalized using a quantile normalization procedure using aroma.light (1.16.0.) package in R (2.11.1). Results for the 50 genes with the greatest differential expression with respect to ARID1A mutation status are shown in Appendix B Supplemental Table 4.  Immunohistochemical analysis of BAF250a protein Immunohistochemical (IHC) staining for BAF250a was performed in all cases in both the discovery and mutation validation cohorts, with the exception of the 42 CCCs from the AOCS and 4 samples from JHU. Four hundred and fifty-five ovarian carcinomas from a previously described tissue microarray (IHC validation cohort) (19), including 132 CCCs, 125 ECs, and 198 HGS were also analyzed for BAF250a expression. All normal gynecologic tissues showed moderate to intense nuclear immunoreactivity for BAF250a. Tumours were scored positive for BAF250a if tumour cells showed definite nuclear staining, or negative if  27 tumour nuclei had no immunoreactivity but endothelial and other non-tumour cells from the same samples showed immunoreactivity. Cases in which neither normal cells in the stroma nor tumour cells were immunoreactive were considered technical failures. Additional IHC staining for hepatocyte nuclear factor (HNF)-1β, and estrogen receptor (ER) was performed on whole sections for two cases with contiguous atypical endometriosis as previously described (20). Immunohistochemical analysis was performed on 4µm thick paraffin sections on the semi-automated Ventana Discovery® XT instrument (Ventana Medical Systems, Tucson, AZ). ARID1A and HNF-1β was stained using the Ventana ChromoMapTM DAB kit. Antigen retrieval was standard CC1 with a two-hour primary incubation. ARID1A mouse clone 3H2 (Abgent, San Diego, CA) was applied at 1:25 followed by a 16-minute secondary incubation of pre-diluted UltraMapTM Mouse HRP (Ventana). HNF-1β goat polyclonal (Santa Cruz Biotechnology, Santa Cruz, CA) was applied at 1:200 dilution followed by a 32-minute incubation of unconjugated rabbit anti- goat secondary at 1:500 (Jackson ImmunoResearch Labs Inc., West Grove, PA). Afterwards the tertiary antibody was incubated for 16 minutes with the prediluted Ventana UltraMapTM Rabbit HRP. ER immunostaining was done using the Ventana DABMapTM kit with standard CC1. The rabbit clone SP1 (Thermo Scientific, Fremont, CA) was incubated at 1:25 for 60 minutes with heat followed by a 32-minute secondary incubation with the pre-diluted Ventana Universal Secondary. Histologic images were obtained with the use of a ScanScope XT digital scanning system (Aperio Technologies Inc., Vista, CA).     28 Laser-capture microdissection (LCM), DNA isolation, and cloning In two cases with identified ARID1A mutations, atypical (contiguous) and distant endometriosis sections were identified by a gynecological pathologist. Laser capture microdissection was used to isolate endometriotic epithelium. DNA extracted from these cells was analyzed by sequencing for the mutations seen in each case.  Atypical (adjacent) and distant endometriosis sections were identified in CCC13 and CCC23 by a gynecological pathologist. For microdissection, formalin-fixed paraffin- embedded (FFPE) sections (5µM) were cut on a Tissue-Tek® Cryo3® cryostat (Sakura Finetek, Dublin, OH) onto clean uncharged slides. FFPE sections were deparaffinized and rehydrated, stained with Arcturus® HistoGene® Staining Solution (Molecular Devices, Inc., Sunnyvale, CA), then dehydrated in alcohol and xylene. All reagents were prepared with nuclease-free water and all steps were performed using nuclease-free techniques.  Atypical or distant endometriotic cells were microdissected from prepared FFPE sections using the VeritasTM Laser Capture Microdissection System (Arcturus Bioscience, Inc., Mountain View, CA) according to the manufacturer’s standard protocols. LCM caps with captured cells were placed directly in 15µl of lysis buffer with 10 µl of Proteinase K, and DNA was isolated using the QIAamp® DNA Micro kit (QIAGEN, Hilden, Germany). DNA was subsequently quantified on a NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, DE). PCR was performed, followed by gel extraction of PCR products using the QIAquick Gel Extraction Kit (QIAGEN), PCR products were cloned using the Topo® TA Cloning® Kit following manufacturer’s instructions (Invitrogen Corp., Carlsbad, CA). Inserts from individual clones were PCR amplified and Sanger sequenced to determine mutation frequency.   29 2.3 Results ARID1A mutations in ovarian cancers The RNA-sequencing data, including the number of mapped sequencing reads and potential non-synonymous sequence variants, are summarized in Appendix B Supplemental Table 2.1. RNA-sequencing of the 19 samples in the discovery cohort resulted in detection of 3 somatic truncating mutations (C4201T (Q1401*) in CCC06, C5164T (R1722*) in CCC09, and C1680A (Y560*) in CCC14), 2 somatic indels (6018-6020delGCT in CCC01 and 5541insG in CCC13), 1 somatic missense mutation (T5953C (S1989P) in CCC13), and 1 gene rearrangement involving ARID1A and the neighboring gene ZDHHC18 (in CCC04) (Table 2.1 and Figure 2.1 (mutations listed above gene in ‘Discovery’ phase)).  The fusion ends of this rearrangement map to a homozygous deletion involving most of the ARID1A gene which is shown as Figure 2.3. All predicted variants of ARID1A were validated by Sanger sequencing in DNA from the source tumours, except for the deletion/rearrangement detected in CCC04, which was validated by Affymetrix SNP 6.0 data (Appendix B Supplemental Table 2.3).  The finding of multiple types of mutations in a single gene in CCCs led us to further explore ARID1A in this cancer type. As mutations in PIK3CA, CTNNB1, KRAS, and TP53 are recurrent in CCC(26) and other ovarian tumours, we further analyzed the RNA-sequencing data and performed PCR for the presence of variants in these genes (Table 2.1).  Genome sequence data from the 19 cases has been deposited at the European-Genome-Phenome Archive (accession number EGAS00000000075).  30 Figure 2.1. Mutations found in ARID1A and the BAF250a protein it encodes.  The 20 exons of ARID1A are represented (as numbered gray boxes) above a schematic of the BAF250a protein (the blue segment, with the ARID [AT-rich interactive domain] DNA-binding domain in pink, the HIC1 [hypermethylated in cancer 1] binding domain in green, and the three C-terminal leucine-rich LXXLL motifs that facilitate interaction with glucocorticoid receptor in yellow). The nucleotide mutations (with corresponding amino acid mutations in parentheses) listed above the schematic are those identified by means of transcriptome sequencing (RNA sequencing) of the 18 samples of ovarian clear-cell carcinoma and the TOV21G cell line in the discovery cohort, and those listed below the schematic were identified in subsequent validation efforts with the use of targeted exon resequencing and Sanger sequencing of genomic DNA from the 210 ovarian-cancer samples in the mutation-validation cohort. All unique somatic mutations detected in samples of ovarian clear- cell carcinoma, endometrioid carcinoma, and high-grade serous carcinoma are shown. Numbers 1 through 6858 below the schematic indicate the nucleotide (nt) position, starting with the A in the ATG start codon for ARID1A in position 1 (based on the sequence given in record number NM_006015.4 in Entrez Gene; also see Table 2.1 in the Supplementary Appendix B). UTR denotes untranslated region. Wiegand KC, et al. NEJM. 2010; Oct 14; 363(16):1532-43. Copyright © 2010, Massachusetts Medical Society.  Reprinted with permission.  31 Table 2.1 Results of RNA sequencing and exon resequencing of the discovery cohort of 19 specimens of CCC and CCC cell line TOV21G.  Wiegand KC, et al. NEJM. 2010; Oct 14; 363(16):1532-43. Copyright © 2010, Massachusetts Medical Society.  Reprinted with permission.  32                                  Exon (denoted by genomic coordinates on chromosome 1)  Figure 2.2 Illumina based resequencing of individual ARID1A exons. Boxplots represent read coverage from each exon for every sample, open circles denote outlier samples, and the bold line represents the median coverage. Exons 1-20 are denoted by their chromosomal coordinates on chromosome 1 along the x-axis. Note that exon 1 resequencing was largely ineffective by this method and was completed by Sanger sequencing (see also Methods and Supplemental Methods).       Wiegand KC, et al. NEJM. 2010; Oct 14; 363(16):1532-43. Copyright © 2010, Massachusetts Medical Society.  Reprinted with permission. N um be r of  S eq ue nc in g R ea ds   33    Figure 2.3 ARID1A-ZDHHC18 fusion prediction.  Alignment of the predicted fusion sequence showing alignment of one end to ARID1A and the other end to ZDHHC18 at 1p36. Vertical bars in the “fusion” alignment represent exon alignment over the genomic locus. Evidence for the ARID1A-ZDHHC18 fusion transcript includes 5 spanning reads and 1 read split by the fusion boundary.  Copy number analysis from SNP 6.0 array shows a large homozygous deletion encompassing almost the entirety of ARID1A with the exception of the 5’ end, and extending over the neighboring gene PIGV.      Wiegand KC, et al. NEJM. 2010; Oct 14; 363(16):1532-43. Copyright © 2010, Massachusetts Medical Society.  Reprinted with permission.  34 Table 2.2 ARID1A mutation status in the discovery and validation cohorts according to carcinoma type (cell lines excluded)    Wiegand KC, et al. NEJM. 2010; Oct 14; 363(16):1532-43. Copyright © 2010, Massachusetts Medical Society.  Reprinted with permission.  35 ARID1A mutation frequency in CCCs and other ovarian cancer subtypes was established through Illumina based targeted exon resequencing of the mutation validation cohort of 210 samples and one CCC cell line, along with the original discovery cohort of 18 CCCs and CCC cell line described above. The total frequencies of CCCs, ECs, and HGS carcinomas with significant ARID1A mutations are 55/119 (46%), 10/33 (30%), and 0/76 respectively (for details see Table 2.2 and Appendix B Supplemental Table 2.3).  Seventeen cases (12 CCCs and 5 ECs) had two validated ARID1A mutations. In addition, the CCC cell line, TOV21G had a truncating mutation in ARID1A (1645insC).  We analyzed germline DNA from 55 cases (47 CCCs and 8 ECs) for the presence of 65 (53 in CCCs and 12 in ECs) truncating mutations and in all cases found mutations to be somatic. Based on this, we made the assumption that 12 (10 in CCC and 2 in EC) subsequent truncating mutations would be somatic (i.e. predicted somatic without germline DNA testing in Appendix B Supplemental Table 2.3). The presence of ARID1A mutations shows a strong association (Fisher Exact p<0.0001) with endometriosis associated ovarian cancer subtypes (CCC or EC).  BAF250a protein expression The correlation between ARID1A mutations and BAF250a expression was evaluated by IHC staining for BAF250a in 73 CCCs, 33 ECs and 76 HGS carcinomas from both the discovery and mutation validation cohorts. These results are summarized in Table 2.2 and Figure 2.4. The presence of truncating mutations in ARID1A was significantly associated with BAF250a loss in endometriosis- associated cancers (Fisher Exact p<0.0001). Within CCCs, and ECs,  36 27/37 (73%) and 5/10 (50%) cases with mutations respectively showed loss compared to 4/36 (11%) and 2/23 (9%) of mutation negative cases respectively (Figure 2.4 A).  Loss of BAF250a expression is strongly associated with endometriosis-related ovarian cancers (31/73 (42%) of CCCs and 7/33 (21%) of ECs) compared to high-grade serous cancers (1/76 (1%)) (Fisher Exact p<0.0001) (Figure 2.4 B).  There was no association between ARID1A mutations and the presence of endometriosis at time of surgery in the 86 CCCs and 33 ECs that had such data available (Appendix B Supplemental Table 2.5).  An additional 132 CCCs, 125 ECs, and 198 HGS from the IHC validation cohort were assessed for BAF250a expression (Figure 2.4 B).  Results from this analysis show that 55/132 (42%) CCCs, 39/125 (31%) ECs, and 12/198 (6%) HGS cases lack BAF250a expression, which agrees with the proportions observed in the discovery and mutation validation cohorts above. There were, however, no significant associations between age of presentation, stage of disease (low or high), or disease-specific survival with BAF250a expression within any of the subtypes as assessed by Welch’s ANOVA, Fisher’s Exact Test, and the Log Rank statistic respectively (p>0.05 in all cases).  37  Figure 2.4. Results of immunohistochemical analyses of BAF250a expression.  The percentages of tumours (with number and total number in parentheses) from three subtypes of ovarian cancer: clear-cell carcinoma (CCC), endometrioid carcinoma (EC), and high-grade serous (HGS) carcinoma from the discovery and mutation-validation cohorts that showed loss of BAF250a expression are shown in (A) for samples with and samples without ARID1A mutations and in (B) for samples in the discovery and mutation-validation cohorts and samples in the immunohistochemical validation cohort. The rate of BAF250a loss was higher among CCC specimens with an ARID1A mutation than among those without an ARID1A mutation (P<0.001); the same was true for EC specimens (P=0.02). The loss of expression was also consistently more common in CCC and EC (the two endometriosis-associated carcinomas) than in HGS carcinoma when assessed in the discovery and mutation-validation cohorts and again in the immunohistochemical validation cohort (B), with P<0.001 for all comparisons. All P values were calculated with the use of Fisher's exact test.   Wiegand KC, et al. NEJM. 2010; Oct 14; 363(16):1532-43. Copyright © 2010, Massachusetts Medical Society.  Reprinted with permission.  38  Analysis of ARID1A in associated endometriosis Two CCCs (CCC13 and CCC23) with ARID1A mutations had atypical (contiguous) endometriosis available (Figure 2.5 and Appendix B Supplemental Figure 2.1).  CCC23 was heterozygous for an ARID1A truncating mutation (G6139T (E2047*)) in exon 20. This mutation was also found in 17/42 clones derived from atypical endometriosis and 0/52 clones from a distant endometriotic lesion (Fisher Exact p<0.001) (Figure 3C).  Both the CCC and atypical endometriotic epithelium had BAF250a expression loss, while the distant endometriotic lesion maintained BAF250a expression (Figure 2.5 B).  HNF-1β was expressed in the CCC but not in the contiguous or distant endometriosis while ER was expressed in both the atypical/contiguous and distant endometriotic epithelium but not in the CCC, as expected.(20) Thus, atypical endometrium could only be distinguished from the distant endometrium based on loss of BAF250a expression, which correlated with the presence of the ARID1A mutation.  CCC13 had two somatic mutations in ARID1A (and loss of BAF250a expression): both these mutations along with a CTNNB1 missense mutation were present in the tumour and the contiguous atypical endometriosis but not in a distant endometriotic lesion (Appendix B Supplemental Figure 2.1 C and D)         39 Figure 2.5  Wiegand KC, et al. NEJM. 2010; Oct 14; 363(16):1532-43. Copyright © 2010, Massachusetts Medical Society.  Reprinted with permission.  40 Figure 2.5 (previous page). Analysis of ovarian clear-cell carcinoma and associated endometriosis in a study patient. (A) An H&E section (hematoxylin and eosin) on which a clear-cell carcinoma (black arrow) has arisen in an endometriotic cyst (white arrow). The same section, viewed at a higher magnification, shows regions of the clear-cell carcinoma and contiguous atypical endometriosis. A region of distant endometriosis from the same patient is also shown.  (B) Results of immunohistochemical staining of the epithelial portions of tissue specimens shown in Panel A for expression of BAF250a, hepatocyte nuclear factor 1β (HNF-1β), and estrogen receptor (ER). BAF250a immunoreactivity is lost in both the clear-cell carcinoma and the contiguous atypical endometriosis but is maintained in the distant endometriosis. Both regions of endometriosis differ from the carcinoma in their lack of HNF-1β expression (with weak expression in the contiguous atypical endometriosis) and maintenance of estrogen-receptor expression.  (C) Sequencing chromatograms for the clear- cell carcinoma and polymerase-chain-reaction (PCR) clones of microdissected material from the contiguous atypical endometriosis and distant endometriosis, from which DNA was extracted. The carcinoma and contiguous atypical endometriosis show nucleotide variation corresponding to G6139T (as indicated with the dashed box); the tumour shows a heterozygous peak at that location, whereas the atypical endometriosis is homozygous for the substitution (in 17 of 42 clones). In contrast, the distant endometriosis shows wild-type sequence (in all 52 clones analyzed). None of the PCR clones from the distant endometriosis showed variation from the wild-type sequence.   2.4 Discussion and conclusions Overall, 46% of CCC and 30% of EC had somatic truncating or missense mutations in ARID1A; no ARID1A mutations were found in any of the 76 HGS cancers analyzed. Loss of ARID1A expression was also subtype specific with loss of nuclear BAF250a seen in 36% of CCCs and ECs but only 1% of HGS carcinomas.  Our original screen using RNA sequencing identified seven somatic mutations in ARID1A in 19 samples (discovery cohort), however four additional mutations were subsequently identified when these cases were analyzed by exon resequencing.  The additional mutations were likely not seen in RNA-sequencing data due to transcripts being rapidly targeted for nonsense mediated decay (NMD) (95), or due to a decreased sensitivity to mutations at the 5’ end of transcripts inherent in the methods. Thus, while RNA sequencing is a useful discovery tool, targeted exon resequencing may be more appropriate for determination of true mutation frequency.  41 ARID1A is located at 1p36.11 (96). This region is commonly deleted in cancers, and it has been suggested that deletion regions encompassing 1p36 could contain tumour suppressor genes (68). Rearrangements and deletions of ARID1A were identified in a primary breast cancer and a lung cancer cell line (70), respectively and loss of BAF250a has also been observed in cervical and breast carcinoma cell lines (71). A screen of 241 tumours has demonstrated that ARID1A transcript levels are decreased in approximately 6% of cancers in general and, specifically, in 30% of renal carcinomas and 10% of breast carcinomas, however none of the 14 ovarian cancers studied by Wang et al. showed loss of expression, potentially because they were predominantly of high grade serous subtype (65).  The ARID1A mutations identified in this study were mostly truncating mutations; these were evenly distributed across the gene. The presence of mutations is strongly correlated with loss of BAF250a protein (see Table 2.2 and Figure 2.4 A).  In CCCs and ECs, loss of BAF250a expression was seen in 73% and 50% of mutation positive cases, respectively, and in only 11% and 9% of mutation negative cases, respectively. Seventeen cases had two mutations in ARID1A.  In all but one of these cases where IHC data was available, BAF250a expression was not detected; the single case that had BAF250a expression (EC13) had both a C-terminal truncating and a missense mutation and either of these changes could produce detectable protein.  A single case (CCC04) had ARID1A loss and rearrangement resulting in homozygous deletion of ARID1A.  Three cases (CCC09, CCC23, CCC54) also appear to have loss of heterozygosity based on the frequency ratio between mutant and wildtype alleles (Appendix B Supplemental Table 3) and subsequent loss of BAF250a expression. However, the majority of cases with somatic ARID1A mutations and loss of BAF250a  42 expression, appear to have a wildtype allele present.  Data from exon reseqencing and RNA- sequencing show similar agreement between fraction of mutant and wildtype alleles at both the DNA and RNA levels (see Table 2.1) suggesting that epigenetic silencing is not a significant factor.  Post-transcriptional/post-translational regulation or dominant-negative effects of the mutations are possible, though untested, explanations for the lack of protein expression in these cases.  The presence of BAF250a immunoreactivity in 15 mutation positive cases (all but one of which have truncating mutations) may indicate that haploinsufficiency is pathogenic, as reported in mice (76). Alternatively this may be due to immunohistochemical detection of truncated but non-functional BAF250a protein. The antibody used targets a 111 amino acid region in the middle of the protein (amino acids 1216-1326) and 7 of the 15 IHC positive cases had mutations that would result in truncations C-terminal to this epitope.  The mutations are present at a high frequency in endometriosis associated ovarian carcinomas (CCC and EC) but not the unrelated HGS carcinoma, strongly suggesting that they are highly relevant in the former, and not random events.  Mutations in the PTEN gene have been described in 20% of endometriotic cysts (45), and conditional expression of either oncogenic K-ras or deletion of the PTEN tumour suppressor in the ovarian surface epithelium of mice was found to induce endometriosis (46). Expression of K-Ras accompanied by simultaneous loss of PTEN resulted in widely metastatic ovarian carcinoma, however KRAS mutations are not seen in human endometriosis and are rare in endometriosis associated ovarian cancers.  By comparing CCCs to their contiguous atypical endometriotic lesions, we  43 show that the same mutations are present in the putative precursor lesions as the tumours. In contrast, the distant endometriotic lesions do not have ARID1A mutations. In CCC23 shown in Figure 2.5, the mutation (G6139T (E2047*)) is present before the atypical endometriosis has developed the immunophenoptype associated with the cancer (ER negative, HNF-1β positive) (20) suggesting that the mutation is a very early event in neoplastic transformation. Taken together, this suggests that ARID1A is a classic tumour suppressor gene. Unlike BRCA or TP53 mutations, which can be found in the germline DNA, all truncating ARID1A mutations tested in germline DNA were somatic.  This is not surprising as deletion of ARID1A on one allele results in embryonic lethality in mice (76).  Gaining an understanding of initiating events for CCC and EC subtypes could lead to the development of new therapeutic approaches and enable the creation of identification tools for endometriotic lesions that are at risk for neoplastic transformation. Mutations in ARID1A and loss of BAF250a expression were preferentially seen in CCCs and ECs, cancers that do not feature the genomic chaos, near ubiquitous TP53 mutations, and frequent BRCA abnormalities seen in HGS carcinomas (21, 97). It is possible that defects in genes, such as ARID1A, that alter the accessibility of transcription factors to chromatin, along with previously described WNT and PI3 kinase pathway mutations (26); will define CCCs and ECs. If such a model is correct, other abnormalities impacting the ARIDIA locus or dysregulation of other chromatin remodeling genes will be found in the ARID1A mutation negative CCCs and ECs.  This idea is supported by the clinical similarity between ARID1A and mutation positive and mutation negative CCCs.   44 The mechanism by which somatic mutations in ARID1A enables the progression of the benign condition of endometriosis to carcinoma has yet to be elucidated, however, our findings strongly suggest a critical role for ARID1A mutation in the genesis of both CCC and EC.  Although its functional and therapeutic significance, along with the frequency of mutations in other cancer types is unknown, the loss of ARID1A in endometriotic epithelium appears to be of fundamental importance in malignant transformation in this tissue type.    45 Chapter 3:  Examining the role of BAF250a (ARID1A) loss in other malignancies  3.1 Introduction In Chapter 2, mutation of the ARID1A gene and loss of the corresponding protein BAF250a was described as a frequent event in clear cell and endometrioid carcinomas of the ovary with 46% of clear cell carcinomas and 30% of endometrioid carcinomas showing mutations in the ARID1A gene.  To determine whether BAF250a loss is common in other malignancies, immunohistochemistry (IHC) for BAF250a was performed on tissue microarrays (TMAs) in more than 3000 cancers, including carcinomas of breast, lung, thyroid, endometrium, kidney, stomach, oral cavity, cervix, pancreas, colon, and rectum, as well as endometrial stromal sarcomas, gastrointestinal stromal tumours (GIST), sex cord-stromal tumours and four major types of lymphoma (diffuse large B-cell lymphoma [DLBCL], primary mediastinal B-cell lymphoma [PMBCL], mantle cell lymphoma [MCL], and follicular lymphoma). We found BAF250a loss is frequent in endometrial carcinomas, but infrequent in other types of malignancies, with loss observed in 29% of Grade 1 or 2, and 39% of Grade 3 endometrioid carcinomas of the endometrium, 18% of endometrial high grade serous, and 26% of endometrial clear cell carcinomas. Since endometrial cancers showed BAF250a loss, we stained whole tissue sections for BAF250a expression in 9 cases of atypical hyperplasia and 10 cases of atypical endometriosis.  Of the 9 cases of complex atypical endometrial hyperplasia, all showed BAF250a expression, however of 10 cases of atypical endometriosis (the putative precursor lesion for clear cell and ovarian carcinoma), one case showed loss of staining for BAF250a in the atypical areas with retention of staining in areas of non-atypical endometriosis; this was the sole case that recurred as an endometrioid carcinoma, indicating  46 that BAF250a loss may be an early event in carcinogenesis.  Since BAF250a loss is seen in endometrial carcinomas at a rate similar to that seen in ovarian carcinomas of clear cell and endometrioid type and is uncommon in other malignancies, we conclude loss of BAF250a is a particular feature of carcinomas arising from endometrial glandular epithelium.  Following the publication of this large initial survey examining ARID1A protein expression by IHC in over 3000 cases of different cancer types (98), sequencing of the ARID1A gene in a large panel of cancers detected mutations in the ARID1A gene in 10% of gastric cancers (99). Additional reports identified mutation or loss of ARID1A in gastric cancer (100-103). Several of the studies of ARID1A in gastric cancer reported that ARID1A loss is associated with microsatellite instability (MSI) (99-101), and Epstein Barr Virus (EBV) infection (101, 102).  Mutations in ARID1A in gastric cancer were also shown to be negatively associated with mutations in TP53 (101). Interestingly, reports of the prognostic significance of ARID1A mutation and loss were conflicting, with one report showing ARID1A mutation to be associated with better prognosis (101), and two reports of ARID1A being a negative prognostic marker (102, 103). Since reports of the clinical relevance of ARID1A loss were conflicting in gastric cancer, we wanted to clarify whether ARID1A is indeed of prognostic and potentially therapeutic significance in this disease, and whether similar associations could be found with EBV infection.      47 3.2 Materials and methods  3.2.1 Materials and methods for BAF250a loss in diverse malignancies Sample collection Cases from the archives of Vancouver General Hospital, St. Paul’s Hospital, and the British Columbia Cancer Agency were used to construct tissue microarrays from duplicate 0.6 mm cores, as described previously (104). The follicular lymphoma TMA was constructed using duplicate 1.0 mm cores. For the studies of atypical hyperplasia of the endometrium, hysterectomy cases where there was no co-existent carcinoma were used and full sections were immunostained. Immunostaining on the cases of atypical endometriosis was also performed on full sections. All prospectively collected patient samples were collected with informed patient consent under a research ethics board (REB)-approved protocol, and analysis of archived samples was covered by pre-existing REB approvals.  Immunohistochemical (IHC) staining of BAF250a Immunohistochemical (IHC) staining for BAF250a was performed on all cases included in this study. IHC was performed on 4µm thick paraffin sections of tissue microarrays or whole tissue sections on the semi-automated Ventana Discovery® XT instrument (Ventana Medical Systems, Tucson, AZ) using the Ventana ChromoMap™ DAB kit. Antigen retrieval was standard CC1 with a two-hour primary incubation. BAF250a mouse clone 3H2 (Abgent, San Diego, CA) was applied at 1:50 followed by a 16-minute secondary incubation of pre-diluted UltraMap™ Mouse HRP (Ventana). Histologic images were obtained with the use of a ScanScope XT digital scanning system (Aperio Technologies Inc.,Vista, CA).  48 IHC staining of p53 For the BC Cohort, IHC for p53 was performed on the Ventana Discovery XT (Ventana Medical Systems Inc., AZ).  The p53 antibody was purchased from Dako (Catalog number M7001, Clone DO-7), and was used at a dilution of 1:400 with standard CC1 pretreatment, followed by incubation with Universal Secondary, and detection with the DAB MAP Detection Kit (Ventana).  For the Toronto Cohort, p53 IHC was performed on Ventana Benchmark®XT, with standard CC1 pretreatment, and incubation with prediluted p53 antibody (clone D07, Ventana), followed by XT ultraView™ Universal DAB detection (Ventana).  IHC scoring The scoring for BAF250a was performed as previously described (83).  Non-neoplastic cells, including endothelial cells, fibroblasts, and lymphocytes, normally show BAF250a nuclear staining and served as positive internal controls.  Positively scored tissue cores were ones that contained any positive tumour cell nuclear staining, regardless of intensity.  Negatively scored tissue cores were ones that showed completely absent tumour cell nuclear staining, as well as positive non-neoplastic cell nuclear staining.  Tissue cores lacking tumour cells were not scored. Cases in which neither normal cells in the stroma nor tumour cells were immunoreactive were considered to be the result of technical failure.  Each case on a tissue microarray was represented as duplicate cores; one positive core in a duplicate was sufficient to count the case as positive.    49 3.2.2 Materials and methods for focused study of gastric carcinoma Sample collection from Vancouver and Toronto Two hundred and fifty-three surgically resected specimens of gastric carcinoma were selected from the British Columbia Cancer Agency in Vancouver (n=173) and the St. Michael’s Hospital in Toronto (n=80).  The BC Cancer Agency study cohort consisted of patients with gastric cancer of any stage that had previously undergone resection of their primary tumour. Cases of  “gastric carcinoma” or carcinoma of “gastroesophageal junction” (n=80) were identified through the pathology database system at St. Michael’s Hospital.  All cases at either site were reviewed by 2 pathologists.  Staging was revised to match the staging system of the 7th edition of the AJCC (105). Clinical information for both cohorts included age, gender, date of resection, date of last follow-up, and date of death if applicable. Ethnicity (Asian vs. non-Asian) was available for the cases from Toronto only. For IHC, tissue microarrays were constructed from a selected paraffin tissue block for each case, as previously described (104).  IHC staining of BAF250a (ARID1A) for additional gastric cancer cohorts from Toronto and Vancouver Immunohistochemical (IHC) staining for BAF250a was performed in all cases included in this study on 4μm thick paraffin sections of tissue microarrays or whole tissue sections on the semi-automated Ventana Discovery® XT instrument (Ventana Medical Systems, Tucson, AZ). Antigen retrieval was standard CC1: Anti-ARID1A (BAF250a) rabbit polyclonal (Sigma) was applied at 1:75, followed by incubation with pre-diluted UltraMap™ anti-rabbit HRP secondary, and detection with Ventana UltraMap™ DAB kit (Ventana).  50 Immunoreactivity was considered positive in tissue cores that contained any positive tumour cell nuclear staining, regardless of intensity. Negative scores were assigned to tissue cores that showed complete absence of tumour cell nuclear staining, as well as positive non- neoplastic cell nuclear staining, as previously described (83).  IHC Staining of Her2/neu For the Toronto Cohort, IHC testing was carried out using the Ventana BenchMark ULTRA instrument (Ventana Medical Systems, Tucson, AZ, USA) and the PATHWAY HER-2/neu (4B5) antibody (Ventana Medical Systems). Cases where the IHC was scored as 0 or 1+ were considered to be negative, while 3+ staining was considered to be positive. In cases with equivocal (2+) IHC staining for HER2, dual ISH testing was performed according to the INFORM HER2 dual ISH kit protocol (Ventana Medical Systems), on a single slide with the digoxigenin-labeled probe for HER2 visualized by silver in-situ hybridization, and the chromosome 17 centromere (CEP17) targeted with a digoxigenin-labeled probe and visualized with the ultraView Red ISH DIG detection kit. To limit effects of heterogeneous staining, a re-cut 4 µm slide from the original block from these tumours was utilized for scoring. A minimum of 20 nuclei were counted, with an additional 40 nuclei counted in cases where the HER2/CEP17 ratio was 1.8 to 2.2.  HER2 SISH on gastric carcinomas For Vancouver, SISH (silver in situ hybridization) was utilized to detect HER2 amplification. HER2 and Chr 17 were used according to manufacturer’s protocol (Ventana). Both probes were labeled with dinitrophenol (DNP) and optimally formulated for use with the Ventana  51 Benchmark® XT automatic IHC/ISH staining module. The slides were conditioned with CC2, treated with protease, hybridized with HER2 DNA probe, and counterstained with Silver C. Slides were then hybridized with Cep17 and counterstained with haematoxylin.  EBER-ISH on gastric carcinomas For Vancouver, 4μm thick paraffin TMA sections or whole tissue sections were deparaffinized and treated with ISH protease 2, followed by staining with INFORM® EBER probe (Ventana) on the Ventana Benchmark® XT (Ventana), and Nuclear Fast Red counterstain (Ventana). For Toronto, 4μm TMA sections were pretreated with CC2 (mild) followed by ISH protease 3, stained with INFORM® EBER probe (Ventana) on the Ventana Benchmark® XT, and counterstained with Nuclear Fast Red (Ventana).  IHC staining of MSI Markers MLH1, MSH2, MSH6 and PMS2 on gastric carcinomas For the BC cohort, the MMR markers MLH1, MSH2, MSH6 and PMS2 were assessed for all cases included in the study. IHC was performed on 4μm thick formalin-fixed paraffin embedded (FFPE) sections of tissue microarrays or whole tissue sections.  MLH1, MSH2 and MSH6 were all stained on automated Ventana Benchmark XT using UltraView "multimer" detection kit.  PMS2 was stained manually using Dako's EnVision FLEX polymer detection kit.  The MLH1 antibody was purchased from Leica (manufactured by Novocastra) (Catalog number NCL-L-MLH1, Clone ES05), and was used at a dilution of 1:25.  The MSH2 antibody was purchased from Leica (manufactured by Novocastra) (Catalog number NCL-MSH2, Clone 25D12) and used at a dilution of 1:5.  The MSH6 antibody was purchased from BD Biosciences (Catalog number 610918, clone 44/MSH6),  52 and was used at a dilution of 1:300. The PMS2 antibody was purchased from BD Biosciences (Catalog number 556415, clone A16-4), and was used at a dilution of 1:100.   For the Toronto cohort, TMAs were sectioned at 4μm, followed by microwave antigen retrieval. Sections were then incubated with mouse monoclonal antibodies against hMLH1 at a dilution of 1:150, ES05, (Novocastra, Vector Labs), hMSH2 (1:100, 25D12, Novocastra), hMSH6 (1:600, PU29, Novocastra), and hPMS2 (1:150, MOR4G, Novocastra) for 1 hr. The antibodies were detected with the VECTASTAIN Elite ABC Kit (Vector Laboratories). The TMA stains were scored independently, and blinded to the diagnosis by 2 pathologists.  Statistics Associations between categorical clinico-pathologic variables and biomarkers were quantified with contingency analysis and determined to be significant by the use of Fisher’s Exact Test for 2 by 2 comparisons and Pearson’s 힆2 statistic for comparisons which exceed the aforementioned conditions. Associations between continuous clinico-pathologic variables and biomarkers were, after ensuring equal variances, quantified with a two-sample t-test. Prognostic associations were derived using overall survival as the outcome variable (defined as the time from first diagnosis to the date of last follow-up or death from any cause). Uinvariable survival analyses were carried out using Kaplan-Meier Survival Analysis and significant differences were quantified using the Log-Rank Statistic.  Multivariable survival analyses were performed using the Cox Proportional Hazards Model.  All p-values are uncorrected for multiple comparisons and values less than 0.05 was considered significant for all analyses.  All statistical analyses were computed with JMP v. 10.0.2 (SAS Institute, Cary, NC, U.S.A).  53 3.3 Results 3.3.1 Results from the initial screening of BAF250a on over 3000 cases For the initial screen, overall, loss of BAF250a expression measured by IHC was not a common event in non-gynecological malignancies (Table 3.1 and Figure 3.1), with loss of BAF250a in more than 10% of cases of a given tumour type only seen in gastric cancer (14%) and anaplastic thyroid carcinoma (14%).   Cancers of endometrial origin showed the highest frequency of BAF250a loss, with 29% of Grade 1 or 2 endometrioid, 39% of Grade 3 endometrioid, 26% of clear cell, and 18% of high grade serous cancers of the endometrium showing BAF250a expression loss (Table 3.1 and Figure 3.2), while 14% of uterine carcinosarcomas showed BAF250a loss.  Nine cases of complex atypical hyperplasia of the endometrium were stained for BAF250a, and all nine showed the same pattern of staining as adjacent normal endometrium (i.e. moderate to intense nuclear positivity).  Of the ten cases of atypical endometriosis, all but one showed retention of BAF250a (i.e. normal staining pattern). A single case showed of loss of staining in the cytologically atypical areas with retention of staining in non-atypical endometriosis (Figure 3.3). This patient developed frank carcinoma of endometrioid type at this site (cul-de-sac) 2 years later.      54 Table 3.1 Results of BAF250a IHC on TMAs for diverse malignancies Cancer Type Negative Cases/Total % Negative DLBCL* 1/137 0.7 PMBCL‡ 0/19 0 MCL¥ 0/78 0 Follicular Lymphoma 1/73 1 Oral 0/53 0 Gastric 26/180 14 Colon 1/130 0.8 Rectal 1/120 0.8 Lung 12/523 2 Benign Thyroid 0/95 0 Differentiated Thyroid 2/146 1 Anaplastic Thyroid 5/35 14 Renal 1/58 2 Pancreatic 5/85 6 GIST§ 37/623 6 Breast 11/315 3 Cervical 1/133 0.8 Endometrial Stromal Sarcoma 3/44 7 Sex Cord 8/87 9 Grade 1 or 2 Endometrioid of Endometrium 29/101 29 Grade 3 Endometrioid of Endometrium 44/113 39 Serous of Endometrium 17/95 18 Clear Cell of Endometrium 6/23 26 Uterine Carcinosarcoma 18/127 14  Legend: *DLBCL, diffuse large B cell lymphoma; ‡PMBCL, primary mediastinal B cell lymphoma; ¥MCL, mantle cell lymphoma; §GIST, gastrointestinal stromal tumour.  Wiegand KC, et al. JPath. 2011; Jul; 224(3):328-33. Copyright © 2011, John Wiley & Sons, Ltd.  Reprinted with permission.   55  Figure 3.1 Immunostaining for BAF250a expression in diverse malignancies, including: (A) DLBCL, (B) MCL, (C) follicular lymphoma, (D) oral cancer, (E) gastric cancer, (F) anaplastic thyroid cancer, (G) renal cancer, (H) pancreatic cancer, (I) GIST, (J) breast cancer, (K) cervical cancer, and (L) sex cord-stromal tumours. Loss of BAF250a is demonstrated in gastric cancer (E), as shown by lack of tumour cell staining and positive stromal staining; all other panels demonstrate positive BAF250a staining.  Images were captured at 20X magnification.  Wiegand KC, et al. JPath. 2011; Jul; 224(3):328-33. Copyright © 2011, John Wiley & Sons, Ltd.  Reprinted with permission.  56    Figure 3.2 High-grade malignancies of the endometrium show loss of BAF250a expression. Tissue cores of (A) high-grade endometroid carcinoma, (B) clear cell carcinoma, (C) high- grade serous carcinoma, and (D) carcinosarcoma.  For all panels, note the lack of BAF250a immunostaining in the tumour cells, while the adjacent non-neoplastic stromal cells show positive BAF250a nuclear staining.  Original magnification for all panels, 20X.     Wiegand KC, et al. JPath. 2011; Jul; 224(3):328-33. Copyright © 2011, John Wiley & Sons, Ltd.  Reprinted with permission.  57    Figure 3.3 Biopsy from cul-de-sac showing endometriosis, with endometrial-type glands and stroma. (A) On H&E staining there is focal cytological atypia of the glandular epithelium (arrowhead), while other glandular epithelial cells do not show atypia (arrow). (B) Immunostaining for BAF250a shows loss of expression in the glandular epithelial cells of atypical endometriosis (arrowhead), with expression in non-atypical glandular epithelial cells and in endometrial stromal cells (arrow). Panels (A) and (B) were captured at 20X.  Panels (C) and (D) show 40X magnification for the H&E and BAF250a IHC, respectively, for the non-atypical glandular epithelial cells.  Panels (E) and (F) show 40X magnification for the H&E and BAF250a IHC, respectively, for the atypical endometriosis.  Wiegand KC, et al. JPath. 2011; Jul; 224(3):328-33. Copyright © 2011, John Wiley & Sons, Ltd.  Reprinted with permission.  58   3.3.2 Results for ARID1A in additional cases of gastric carcinoma Case selection and clinicopathologic variables One hundred and seventy-three cases were selected from the British Columbia Cancer Agency in Vancouver and eighty cases from St. Michael’s Hospital in Toronto (total n = 253).  The set of clinicopathologic variables available for both cohorts included gender, age at diagnosis, Lauren’s classification (diffuse, intestinal or mixed), site of presentation, tumour stage, lymph node status, and disease stage (summarized in Table 3.2, and Figure 3.4).  Correlation of BAF250a loss with MSI, EBV, and P53 IHC staining of BAF250a (Anti-ARID1A) showed an overall loss of BAF250a expression in 22.5% of gastric adenocarcinomas from the Vancouver group and 20% for Toronto, for an overall rate of loss of 22% in the combined set of 253 cases. Representative images of invasive gastric adenocarcinoma stained with BAF250a (anti-ARID1A) are shown in Figure 3.5. Interestingly, there appears to be a grade dependent loss of nuclear immunoreactivity (Figure 3.5).  In both Vancouver and Toronto cohorts, the loss of BAF250a was significantly correlated with loss of mismatch repair protein expression (p < 0.0001 and p = 0.0347 respectively; Fisher’s exact 2-Tailed Test), where loss of any of the standard microsatellite markers (MLH1, MSH2, MSH6 and PMS2) was considered to render the case mismatch repair deficient (representative images are shown in Figure 3.6).  For the combined cohorts overall, 47.8% of cases with loss of mismatch repair protein expression were found to have loss of BAF250a compared with only 18.7% of cases with intact mismatch repair protein  59 expression.   No significant association of BAF250a loss with EBV status was found in either cohort. Abnormal p53 staining was associated with normal BAF250a in the Vancouver cohort both when p53 mutants were analyzed separately (gain of p53 vs. loss of p53) (p = 0.0018; Pearson 힆2) or combined (p = 0.0031; Fisher’s exact 2-Tailed Test), however this association was not observed for Toronto.  Relationship between BAF250a, MMR, p53 or HER2 status and overall survival All 253 cases were subjected to two-way hierarchical clustering using Ward’s algorithm, the results of which are shown in Figure 3.7 and summarized in Table 3.4.   This exercise resulted in 9 clusters based on biomarker status.  These clusters show a wide range in the 5- year survival rate, with the cluster of exclusively HER2 amplified cases showing the worst survival rate (0% 5 year survival), while cluster 2 displayed the best 5-year survival fraction (P53 aberrant/BAF250a loss/MMR deficient).  Cluster 2 failed to differentiate itself from any cluster, which is explained due to a late death that occurred at the 6.37- year time point where the previous death to this occurred at the 1.51-year time point. We surmise that this death was almost assuredly from a cause other than gastric cancer given that the patient was 74 years of age at the time of diagnosis.  Loss of BAF250a expression was significantly associated with poor overall survival in the Toronto cohort (log- rank p=0.0046), whereas no significant association with overall survival was observed in the Vancouver cohort (Figure 3.8A and 3.8B).  In the Vancouver Cohort, MMR deficiency was associated with poor overall survival (log-rank p=0.0019) The alteration of p53 did not show any significant association with survival in either cohort,  60 whereas HER2 was found to be a negative prognostic indicator in both cohorts. (Figure 3.8 C and D).  Clinical parameters associated with BAF250a, MMR, p53, and HER2 The clinical variables taken into account for the multivariate analysis included sex, age at diagnosis, Lauren’s classification, differentiation, lymph node status, tumour stage, and site of presentation (Figure 3.4 and Table 3.3).  By multivariate analysis, BAF250a loss was not significantly associated with any clinical parameters in either cohort.  The clinical parameters associated with MMR deficiency in the multivariate analysis for Toronto were tumour stage and lymph node status (p=0.018, and p=0.026), whereas lymph node status was the only clinical parameter associated with MMR deficiency in the Vancouver cohort (p=0.004) (all Fisher’s exact 2-tail. The p53 status for Toronto was not significantly associated with any clinical parameters, however for the Vancouver cohort p53 mutation status was significantly associated with site of presentation (p=0.022, Pearson’s).  For Vancouver, HER2 amplification was significantly associated with the intestinal subtype (p=0.012, Pearson’s), while in the Toronto cohort HER2 was significantly associated with tumour differentiation, (p=0.002, Pearson’s).  For the combined cohort of 253 cases, MMR deficiency was associated with positive lymph nodes (p =0.001, Fisher’s exact 2-Tailed), as well as with the distal site of presentation (p=0.022; Pearson), and earlier stage (p=0.027; Fisher).  The p53 status of the combined cohort was associated with site of presentation (p=0.039; Pearson).  HER2 status (i.e. amplification) of the combined cohort was associated with several clinical parameters,  61 including subtype (intestinal) (p=0.004; Pearson), as well as with differentiation (p = 0.020; Pearson), and also with positive lymph node status (p=0.017; Fisher’s exact 2-Tailed).  Table 3.2 Clinicopathologic characteristics of the gastric cancer cohorts   62                 Table 3.3 Summary of IHC and ISH staining results!for!biomarkers!vs.!clinical!variables.!(A)!Comparison!of!biomarker!status!for!BAF250a,!MMR,!p53,!and!HER2!against!common!clinicoCpathologic!markers!in!gastric!cancer.!!Comparisons!for!clinicopathologic!variables!with!two!levels!were!computed!with!Fisher’s!Exact!Test!while!clinicopathologic!variables!with!three!levels!were!assessed!with!Pearson’s!훘2!statistic.$(B)$Comparison!of!biomarker!status!for!BAF250a,!MMR,!p53,!and!HER2!against!age.!Means!and!ranges!are!reported!and!differences!are!quantified!using!analysis!of!variance!(ANOVA)!after!the!confirmation!of!equal!variances.$  63   Figure 3.4 Distribution of clinical variables in two gastric carcinoma cohorts:  Toronto (Tor) and Vancouver (Van). For each part, graphical representations and associated statistical comparisons are located at the left and right respectively. Comparisons include (A) Age at diagnosis, (B) Gender, (C) Lauren’s classification, (D) Differentiation, (E) Nodal status (F) Stage (where ‘Early’ is defined as T-Stage equal to 1 and ‘Late’ is defined as T-Stage ≥ 2), and (G) Site of malignancy.  64     Figure 3.5 Representative images of invasive gastric adenocarcinoma stained with BAF250a (Anti-ARID1A). There is a grade dependent loss of nuclear immunoreactivity from well differentiated (A) to moderately (B) and poorly (C) differentiated tumours. Interestingly, all signet ring cell carcinomata (D) in the cohort also demonstrated loss of BAF250a staining. [Magnification: 100x]       65  Figure 3.6 Loss of BAF250a (Anti-ARID1A) expression (A&B) in poorly differentiated gastric adenocarcinomata was also significantly associated with evidence of mismatch repair deficiency. The represented example shows nuclear loss of hMLH1 (C) and hPMS2 (F), while there is normal expression of hMSH2 (D) and hMSH6 (E). Stromal lymphocytes serve as an internal control. [Magnification: 100x]   66   Figure 3.7 Heatmap clustering by biomarker status.  All 253 cases were subjected to two- way hierarchical clustering using Ward’s algorithm.  The presumed biologically important biomarker statuses are coloured red and the inverse status is coloured blue.  The number of clusters were derived manually in order to minimize the number of clusters with impure biomarker phenotypes while still keeping the N of each cluster as large as possible for the purpose of comparative analysis.  The resultant clusters are described in Table 3.4.  67 Table 3.4. Heatmap statistics (to accompany Figure 3.7)    68 Legend accompanying Table 3.4 (previous page). The nine clusters generated from the biomarker specific clustering exercise are displayed.  For the four biomarkers, cells coloured blue indicate a complete absence of the phenomenon while cells coloured red indicate a uniform presence of the phenomenon.  Accordingly, cells coloured purple indicate and impure condition of the phenomenon.  The distribution of binarized T-Stage, with the cut- point between 1 (Early) and ≥ 2 (Late), and Nodal Status are displayed for each cluster.  The 5-year overall survival fraction is also shown for each cluster.  Kaplan-Meir survival curves and the Log-Rank statistic were computed for each cluster-by-cluster comparison and only the statistically significant findings are presented.  Note: While cluster 2 has the best 5-year survival fraction, it fails to differentiate itself from any cluster due to a late death, which occurred at the 6.37 year time point where the previous death to this occurred at the 1.51 year time point. We surmise that this death was almost assuredly from a cause other than gastric cancer given that the patient was 74 years of age at the time of diagnosis.  69      Figure 3.8 Kaplan-Meier survival curves for ARID1A (BAF250a) and HER2 for Toronto (A,C) and Vancouver (B,D).    70 3.4 Discussion and conclusions This chapter establishes that loss of BAF250a is characteristic of a wide range of tumours arising from eutopic as well as ectopic endometrium, but is less frequent in other tumour types studied.  The carcinomas of the endometrium, particularly those of higher grade, show the most frequent loss of BAF250a.   This is perhaps not surprising, since they are related to clear cell and endometrioid carcinomas of the ovary, which have BAF250a loss and ARID1A mutation of 46-57% and 30%, respectively (83, 84).  In the carcinomas of the endometrium that showed BAF250a loss, we do not know the mutational status of the ARID1A gene, however in the clear cell and endometrioid carcinomas of the ovary, mutation of ARID1A correlated well, although not perfectly, with BAF250a expression (83).  Therefore, we hypothesize that in carcinomas of the endometrium with BAF250a loss, most will harbor mutations in the ARID1A gene.  In cases that do not show BAF250a loss, it is possible that other components of the SWI/SNF chromatin remodeling complex will show loss of function. Additionally, since the deletion of ARID1A on one allele (haploinsufficiency) results in embryonic lethality in mice, it is possible that partial loss of BAF250a expression could have a biologic effect in tumours and thus we could be underestimating the effect of ARID1A by screening for total BAF250a loss by IHC (76). The measurement of partial loss would require a nuanced approach to scoring or the use of multiplexed immunofluorescence.  BAF250a loss in precursors of endometrial, extrauterine endometrioid and clear cell carcinomas We studied BAF250a expression in complex atypical hyperplasia of the endometrium, and atypical endometriosis, considered to be precursors of endometrial carcinoma and  71 extrauterine carcinomas of endometrioid and clear cell type, respectively. We hoped to gain some insight into the timing of ARID1A mutations during tumour development. We had previously shown that ARID1A mutations were present in areas of atypical endometriosis adjacent to clear cell carcinoma, but not in non-atypical endometriosis from the same patients, suggesting that ARID1A mutations occurred early during oncogenesis. In this study we did not identify BAF250a loss in any of the nine cases of atypical endometrial hyperplasia. One of the ten cases of atypical endometriosis had loss of BAF250a expression. This patient returned two years later with an endometrioid carcinoma at the location of the atypical endometriosis. This finding could be interpreted in two ways. Firstly BAF250a loss and thus ARID1A mutation is a late event in the progression of precursor lesions to cancer or that the particular lesion studied was already fully malignant, although not recognized as such on morphological grounds. Either way, this case along with the frequency of BAF250a loss in frank carcinomas, the rarity (or absence) of loss in normal tissue and precursor lesions suggest that loss of BAF250a expression is a feature highly indicative of malignancy.  In conclusion, loss of BAF250a expression is relatively common in carcinomas arising from the endometrium, but much less frequent in non-gynecological cancers, suggesting that the search for ARID1A mutations should predominantly focus on the former subset of tumours and their precursor lesions. BAF250a loss has potential as a marker for malignancy derived from endometrial epithelium and ARID1A loss as a targetable feature in such cancers.  Gastric cancer (GC) is the second leading cause of cancer death worldwide, with over 700,000 deaths occurring annually (106). Following the publication of the work in the first  72 part of this chapter, other reports have shown ARID1A to be mutated or lost in between 8- 27% of gastric cancer (100-102), with one study reporting loss as high as 51% (103). In the current study, we observe a similar rate of BAF250a loss in two separate cohorts, with IHC staining of BAF250a in one cohort  (Vancouver) showing 22.5% loss and the other (Toronto) 20% loss, for an overall frequency of BAF250a loss of 22% in the combined set of 253 cases. The set of clinicopathologic variables available for both the Vancouver and Toronto cohorts is summarized in Table 3.2 and Figure 3.4. By multivariate analysis, BAF250a loss was not significantly associated with any clinical parameters in either the Toronto or Vancouver cohort.  In the largest study to date on ARID1A loss in gastric cancer of 857 cases, ARID1A loss correlated with multiple parameters in EBV-negative MSI-normal gastric carcinoma, including larger tumour size, advanced invasion depth, and lymph node metastasis (102). Unfortunately, the number of cases in our study does not allow us to make such comparisons.  Consistent with previous publications of HER2 amplification as a negative prognostic marker in gastric cancer (107), we similarly found HER2 to be a negative prognostic indicator in both cohorts, irrespective of the difference in the sampling bias.  P53 loss is similarly frequent in gastric cancer, and reports suggest TP53 mutation may be responsible for alternate pathways of carcinogenesis to ARID1A loss, as these mutations seem to be almost mutually exclusive. (99, 101)  In our present study, we observed that abnormal p53 status was associated with normal BAF250a expression in the Vancouver cohort, however this relationship was not observed for the Toronto cohort.   73 Microsatellite instability is commonly found at a rate of approximately 8-45% in sporadic gastric cancer depending on the study and the definition of MSI (108, 109).  We stained for a standard panel of four microsatellite markers including MLH1, MSH2, MSH6 and PMS2, and found the loss of BAF250a significantly correlated to loss of mismatch repair protein expression in both cohorts (p < 0.0001 and p = 0.0347 respectively; Fisher’s exact 2-Tailed Test).  As suggested previously, it is possible that the mutation of ARID1A in gastric cancer cases with MSI could be due to defective mismatch repair machinery (99, 101).  In contrast to two other reports (101, 102), we did not find an association between BAF250a loss and EBV infection, however the rate of EBV infection in both cohorts of our study was lower than expected (3.5% for Vancouver and 2.2% for Toronto, by EBER-ISH), in contrast to other published reports of EBV in gastric cancer which suggest a rate of 10% (110).  Since it has been controversial whether or not loss of ARID1A (BAF250a) is of prognostic value, with one report showing ARID1A mutation to be associated with better prognosis (101), and two reports of ARID1A being a negative prognostic marker (101, 102), we sought to clarify whether ARID1A is indeed of prognostic and potentially therapeutic significance in this disease.  In the analysis of two separate cohorts from different institutions, BAF250a expression was significantly associated with poor overall survival in one cohort (Toronto) (log rank p=0.0046), whereas no significant association with overall survival was observed in the Vancouver group (Figure 3.8). A possible explanation for the centre observed difference may be explained by the fact that the Toronto cohort encompassed an increased frequency of earlier stage disease with more than double the percentage of T1 patients  (Figure 3.4). This may reflect a referral bias between the two centres: while the Vancouver cohort is comprised  74 of resection specimens from a provincial cancer care centre, the Toronto cohort has a greater number of early stage cancers due to the availability of endoscopic mucosal resection for intramucosal (T1a malignancies) at that centre: cases which are found to be unresectable by endoscopic means (pathologic submucosal invasion or poor prognostic features) progress to surgery, giving a higher proportion of early stage cancers.  To investigate the relationship of overall survival with P53, HER2, BAF250a, and MMR, biomarker specific clustering was performed (Figure 3.7), resulting in the identification of nine biomarker specific clusters.  Interestingly, a drastic difference in patient survival is observed with this biomarker stratified clustering approach, with HER2 predicting the worst outcome with 0% 5-year survival, while the aberrant P53/BAF250a loss/MMR-deficient cases as a group showed the highest 5-year survival at 66%. It will be of importance future studies to explore these relationships in other cohorts.  Overall, our findings confirm that ARID1A appears to have an important prognostic value within gastric cancer, particularly in patients with earlier stage disease and may thus serve as a prognostic biomarker in gastric carcinogenesis. This observation becomes especially important, bearing in mind that the incidence of EGC is increasing due to early detection. In addition we confirmed the correlation between loss of expression of mismatch repair and ARID1A proteins.  While the results presented certainly highlight the potential value of ARID1A specific therapeutics and merit further study, our study highlights another, albeit more practical point: pathological assessment of gastric cancer has to move away from the single biomarker approach and increasingly employ a multi-marker approach.  75 Chapter 4:  Functional proteomic analysis of endometrioid and clear cell carcinomas and associations with ARID1A/BAF250a  4.1 Introduction The purpose of this work presented in this chapter was to examine proteomic patterns of expression of the three major ovarian cancer subtypes (HGSC, CCC and EC) using the reverse phase protein array (RPPA) platform. The genomic landscape of high-grade serous ovarian cancers has been defined by the Cancer Genome Atlas (TCGA), however, the TCGA was restricted to high grade serous ovarian cancer at the genomic level and did not consider the proteome of EOCs (11).  Since approximately 50% of CCCs and 30% of ECs harbor mutations in the chromatin-remodeling gene ARID1A (83, 84), we attempted to determine if ARID1A mutation status and/or BAF250a expression would reveal potential therapeutic targets in EOC using the RPPA profiles. We examined the association of ARID1A mutation, BAF250a protein expression, PTEN, and PIK3CA mutations on AKT phosphorylation (pAKT) in EOCs. PIK3CA mutations are also common in endometriosis-associated ovarian cancers, as well as endometrial and breast cancers (23, 111, 112). The PI3K/AKT signaling pathway plays a key role in mediating growth and many regulatory functions pertaining to the cancer phenotype including drug resistance. PIK3CA mutations are frequent in clear cell carcinomas (26), therefore activating mutations of PIK3CA may also provide a potential target for anticancer therapy in this chemotherapy resistant patient population (113, 114), as there are a number of novel protein kinase inhibitors that have been developed to target this pathway (113). For the proteomic profiling, whole tumour lysates were prepared from 127 ovarian carcinomas, including 34 CCC, 28 EC and 65 HGSC, and were profiled by RPPA.  76 ARID1A mutation and IHC status was previously defined for 90 of the ovarian cancers in our RPPA series, and of these cases we analyzed 31 CCCs and 24 ECs for common oncogenic mutations (including PIK3CA) by MALDI-TOF mass spectrometry using the Sequenom® platform. Although ARID1A may function as a tumour suppressor gene (115) there are no proven clinical features or differences in outcomes associated with expression of its associated protein (BAF250a) by immunohistochemistry (IHC) in clear cell carcinoma (83, 116). However, a better understanding of the molecular and cellular aberrations in endometriosis-associated cancers engendered by aberrant ARID1A/BAF250a function could lead to identification of novel therapeutic approaches in these two subtypes.  4.2 Methods  Reverse phase protein array We performed RPPA on 127 ovarian cancer whole tumour lysates as previously described (114, 117). Approximately 116 antibodies to cell surface growth factor receptors, common signaling pathway proteins, steroid hormone receptors, and others proteins involved in proliferation and apoptosis were assessed.  A list of antibodies used for RPPA is included in Appendix C.  Collection of tumour samples and IHC Tumour samples were obtained from the gynecology tumour bank at Vancouver General Hospital and the British Columbia Cancer Agency (BCCA).  The research was conducted with approval from the University of British Columbia (UBC) institutional review board.  77 Tumour samples were collected at the time of primary surgery and snap frozen within 60 minutes after collection.  Patients treated by neoadjuvant chemotherapy were excluded from the study.  All patient samples were subjected to pathology review (BG) to confirm the histological subtype and site of origin.  Clinical data was accessed through the Cheryl Brown Ovarian Cancer Outcomes Unit and is updated on a regular basis. Tumours were studied by immunohistochemistry (IHC) as part of a tumour bank tissue microarray (TMA).  Methods for preparation, staining, and scoring of the EOC cases for BAF250a have been previously described (83); tumours with any degree of nuclear staining for BAF250a protein expression was considered a positive result for purposes of this study.  BAF250a knockdown experiments in clear cell lines ES-2, JHOC5, and RMG1 cell lines were cultured in 6-well plates in 5-10% FBS.  Cells were grown in RPMI with the exception of the ES-2 that were grown in McCoy’s media.  Cells were treated with pooled siRNA (Dharmacon) to ARID1A at 20nM using RNAiMax™ as a transfection reagent (Invitrogen) according to the manufacturer’s recommended protocol. Lysates were prepared 60 hours following transfection using Bicine/Chaps lysis buffer.  EGF stimulation was performed using 20ng/ml EGF for 15- 20 minutes, and protein levels were assessed by Western blotting (PAGE/SDS).  Antibodies used were as follows: BAF250a (Sigma), Aktp473 and 308 (Cell signaling and New England BioLabs), PDK1 (Cell Signaling), p70S6 Kinase (Cell Signaling), B-actin.     78 Isoelectric focusing for AKT and ERK expression Native capillary isoelectric point focusing (Protein Simple™) was used to assess AKT and ERK expression according to recommended protocols using a NanoPro™ 2000. An AKT1 antibody (Santa Cruz (C-20): sc1618) was used for these experiments.  DNA sequencing of tumours and mutational analyses ARID1A mutations were determined primarily by exome sequencing using next generation technologies as previously described (83).  As exon capture was not successful for exon 1 due to high GC content, Sanger sequencing was used for this exon (83). Common oncogenic mutations were determined by MALDI-TOF mass spectrometry (MassARRAY®, Sequenom Inc.) (118). Mutations of PIK3CA were categorized as kinase domain mutation H1047R, and helical domain mutations (E542K, E545K, and also E545D and Q546R).  In addition, there were several other less common mutations included (N345K, R88Q, E110K).  Statistical methods To obtain a global visualization and assessment of tumour protein expression profiles, unsupervised and supervised cluster analysis was performed using TreeView software (University of Glasgow, Scotland).  X-cluster software was used to generate heat maps and cluster groups (Stanford).  Data was analyzed using SPSS software (Version 20, Chicago, Illinois). Proteins that were differentially expressed in clear cell cancers relative to the other subtypes were determined by t-testing and significance analysis of microarray data (SAM) (119). Median protein expression levels were used as cut-points for multivariate logistic regression models except for PTEN expression.  A cut-point of the lowest 20% of values on  79 RPPA was chosen for PTEN loss based on previous work showing a 28% incidence of PTEN loss in CCCs (120), and IHC data from our own centre showing a 12% incidence of PTEN loss in 42 cases of CCC/ECs.  4.3 Results Protein expression is histotype specific There were 127 ovarian cancers in total used for the proteomic cluster analysis:  33 clear cell carcinomas (CCCs), 29 endometrioid carcinomas (ECs) and 65 high-grade serous carcinomas (HGSCs).  As expected, a greater proportion of CCC and EC (73%) presented with early- stage disease (stage I or II) compared to HGS cancers (17%).  Hierarchical clustering of samples and proteins analyzed by RPPA are shown in Figure 4.1. Clustering is clearly driven by histological subtype.  CCCs and ECs form distinct clusters separated by two HGS cancer subgroups.            80 Figure 4.1 Hierarchical clustering of samples and proteins analyzed by RPPA.  Differential expression of proteins on each array was examined by significance analysis of microarray data (SAM analysis- typically used for gene expression studies) for both CCCs and ECs compared to HGS cancers as the reference group.  Figure 4.2 shows the differentially expressed proteins according to histological subtype. A number of proteins are under-expressed in both EC and CCC relative to the HGS group.  Notable proteins are GAB2, YAP, Cyclin B1, CofilinpSer3, and Caveolin1, 4EBP1pThr37, STAT5, and c-Myc.  81 CCCs clearly have lower ERα, AR, and PR expression compared to HGS.  The complete lists of protein comparisons by SAM analysis and histotype is included in Appendix C. Other proteins with low expression in CCC include β-Catenin, fibronectin, PTEN, and 4EBP1, and Stathmin.  EC tumours are characterized by lower levels of Cyclin E1, phosphorylated S6, 14-3-3 Zeta, and BIM, and STAT3.  CCC and EC shared four proteins that were overexpressed compared to HGS.  Notably pAKT-Thr308  is higher in both histologies though pAKT-Ser473 was of borderline significance. (FDR= 5.2 and 6.9).  Other overexpressed proteins in both CCC and EC include p21, HSP70, and 4EBP1-pSer65.  EC as expected have higher expression of ERα, ERαpSer118 and PR compared to HGS, while IRS1, CHK1, EGFR, and Cyclin D1 are also elevated. Some of the overexpressed proteins that characterize CCC include α-Catenin, Cyclin E1, HSP27, E-cadherin, p38, p38pThr180Tyr182, SMAD3, and GSK proteins.            82      Figure 4.2  (A) Up-regulated and (B) Down-regulated proteins in clear cell and endometrioid carcinoma by SAM analysis on RPPA data, with high grade serous carcinoma (HGSC) as the reference group.  Proteins with Q-values <5% are shown. Proteins that are over or under expressed in both CCC and EC are shown in the center between the two subtypes.     83 Investigating the effect of PIK3CA and ARID1A mutation on protein expression We then examined more carefully the effect of ARID1A mutation/BAF250a loss on changes in protein expression.  For 96 samples, ARID1A mutation status, BAF250a expression by IHC, and PIK3CA mutation status was determined.  Table 4.1 shows the characteristics of this study population.  Serous tumours more frequently presented with advanced stage disease (83%) whereas the CCC and EC cancers more commonly presented with stage I/II disease (59% and 87% respectively).  ARID1A mutations were present in 17 of 31 CCCs in this study (55%), and 5 of 24 (21%) cases of ECs.  These cases were previously reported as part of a larger cohort used for the initial determination of the frequency of ARID1A mutation status in endometriosis associated cancers (83).  Sequenom MassARRAY® analysis of these tumours showed that PIK3CA mutations were present in 45% of the CCCs and 46% of the ECs. Fifty-nine percent of tumours (13/22) with ARID1A mutations had PIK3CA mutations versus 36% (12/33) percent of tumours wild type for ARID1A (p = 0.1; Chi-square test). While 8/11 (73%) PIK3CA helical domain mutants were wild type for ARID1A, kinase/other domain mutants were more common in tumours with ARID1A mutations (10/14; 71%), (p = 0.05; Fisher’s exact test).  Figure 4.3 shows a heat map with samples categorized according to BAF250a IHC result and PIK3CA mutation status.  It is evident that there are no obvious clustering patterns due PIK3CA mutation status, however the small number of clear cell cancers with BAF250a expression loss do form a cluster.    84 Table 4.1 Patient/sample characteristics according to histotype                   85 Table 4.2 Effect of PIK3CA mutations, ARID1A mutations, and BAF250a expression on AKT phosphorylation   Note: univariate mean and SD values represent Log(2) relative protein expression measured by RPPA * significant p-value (< 0.05) § lowest 20% of samples by RPPA                86 Impact of PIK3CA, ARID1A, and PTEN alterations on pAKT In terms of clarifying the impact of the different PIK3CA mutations on pAKT, the distributions (box plots) of pAKT-Thr308 and pAKT-Ser473 were compared (Appendix C). By ANOVA, mutation status was not associated with significantly different pAKT levels by RPPA.   Helical domain mutants were not associated with an increase in pAKT levels whereas kinase domain mutants and other mutants resulted in higher mean levels of pAKT (both Thr308 and Ser473) that were not statistically significant.  However, when the 11 low PTEN cases were excluded, kinase domain and other less common mutants had significantly higher pAKT-Thr308 (p < 0.05; t-test Bonferroni corrected) than wild type, while helical domain mutants did not.  For pAKT-Ser473, only the uncommon mutants had significantly higher levels by RPPA (t-test, Bonferroni corrected) compared to wild type or helical domain mutants.  Table 4.2 shows the uni- and multivariate analyses of AKT phosphorylation based on ARID1A mutation status, BAF250a IHC expression, PIK3CA mutation status, and PTEN levels by RPPA.  On univariate testing, pAKT-Thr308 expression by RPPA was significantly higher in tumours that were BAF250a negative by IHC (p = 0.007), and those with low PTEN (p = 0.025).  In contrast, neither ARID1A mutation status nor the presence of a PIK3CA mutation was associated with a change in pAKT-Thr308 levels. Changes in pAKT- Ser473 were not associated with any of the above factors on univariate testing.   Figure 4.3 Figure 4.3. Heat map with samples categorized according to BAF250a IHC scores and PIK3CA mutation status.  No obvious clustering patterns due to either BAF250a by IHC or PIK3CA mutation status are present.  87  88 Multivariate logistic regression showed that pAKT-Thr308 levels by RPPA are significantly higher in tumours with loss of BAF250a expression (p = 0.002), those with low PTEN (p= 0.003), and cancers with activating PIK3CA mutations (p = 0.02).   As with the univariate testing, pAKT-Thr308 levels were not associated with ARID1A mutation status.  On multivariate analysis, increases in pAKT-Ser473 were only associated with BAF250a expression (p = 0.04). Subgroup analysis of AKT phosphorylation in EOC devoid of PIK3CA mutations (18 CCC and 12 EC) showed statistically significant increases in both pAKT-Ser473 and pAKT-Thr308 in the 8 tumours lacking BAF250a expression on IHC (p = 0.05 and 0.008 respectively; t-test).  Multivariate logistic regression analysis on only the 31 clear cell cancers showed that both pAKT-Thr308 and pAKT-Ser473 were significantly higher in tumours with loss of expression of BAF250a (p = 0.05 and 0.005 respectively).  The EC subgroup was not similarly analyzed due to the fact that there were too few tumours with ARID1A mutations/BAF250a negative for comparison. SAM was performed looking at proteins differentially expressed in CCC BAF250a-IHCnegative versus CCC BAF250a-IHCpositive and the only differentially expressed protein with a low false discovery rate (FDR) was pAKT-Thr308 (FDR < 1%).  The top 4 other proteins differentially expressed were as follows: Bcl-2, p27, phosphorylated p38, and pAKT-Ser473.  FDR rates by SAM for these 5 proteins varied from 16-20%.  It is worth noting however that all but one of these proteins (p38) are primary signaling proteins in the PI3K pathway.      89 siRNA knockdown of BAF250a does not effect AKT expression in CCC cell lines In terms of trying to investigate the relationship between BAF250a expression and pAKT we performed siRNA knockdown experiments of BAF250a.  Results from these experiments are shown in Figure 4.4.  Despite good knockdown of BAF250a we did not see any change in AKT phosphorylation or levels of p70S6K, a downstream signaling protein of pAKT in any of the three cell lines tested.  None of these cell lines are known to have a PIK3CA mutation. JHOC5 cells demonstrated much higher baseline levels of pAKT.  Although pAKT-Thr308 was difficult to detect in the RMG1 cell lines, with EGF stimulation all cell lines showed an increase in pAKT phosphorylation except for pAKT-Ser473 in the JHOC5 cells.  Changes in pAKT with EGF stimulation were not altered by BAF250a knockdown.  We also observed that PDK1, PTEN levels did not change with BAF250a knockdown in any of the cell lines. Native protein AKT profiles were also assessed using capillary tube isoelectric point focusing (Figure 4.4 B-D).  Although we do not know which AKT isoforms and their corresponding phosphorylated counterparts are represented by specific peaks, this technology is very sensitive to changes in peak levels due to phosphorylation events with a corresponding reduction in total protein.  In general, phosphorylation events are seen as peak increases at a lower isoelectric point.  Native AKT profiles are in keeping with the western blot findings in that there is little or no change in profiles with BAF250a knockdown in all cell lines tested (Figure 4.4 C). Taken together, these findings indicate that BAF250a knockdown has little to no effect in AKT phosphorylation in the cell lines tested.    90   Figure 4.4  (A) Western blot results from siRNA knockdown of BAF250a on cell lines ES2, JHOC5, and RMG1 to clarify the interaction between BAF250a expression and pAKT. Despite good knockdown of BAF250a no change in AKT phosphorylation or levels of p70S6K, a downstream signaling protein of pAKT can be seen in the ES2 and RMG1 cell lines.  The baseline levels of pAKT are much higher in the JHOC5 cell line, and there is a suggestion of an increase in pAKT-Thr308 with BAF250a knockdown without obvious similar changes in pAKT-Ser473.  PDK1 and PTEN levels did not change with BAF250a knockdown in any of the cell lines.  (B-D)  Native protein AKT profiles using capillary tube isoelectric point focusing.  Native AKT profiles are consistent with the western blot result in (A), as little change occurs in AKT following siRNA mediated BAF250a knockdown.          91 Clinical outcomes Due to the size of our patient population and representation of tumours across all stages, survival outcomes were not examined.  Instead, FIGO stage distribution at presentation was analyzed to determine if this was influenced by BAF250a loss, PIK3CA, or ARID1A mutation status.  For this comparison, FIGO stage was classified as low-stage (stage 1/2) or advanced-stage (stage 3/4).  Those patients with BAF250a loss were more likely to have advanced stage (5/11 = 45% vs. 9/42 = 21%), however this difference was not statistically significant.  Eighty-four percent (21/25) of patients with a PIK3CA mutation presented with low-stage disease, compared to 64% (18/28; 2 cases with missing stage) without a mutation. Again, this difference was not statistically significant.  4.4 Discussion We used functional proteomics (RPPA) to assess protein expression in 3 major histotypes of ovarian cancer.  It is apparent that protein expression is clearly related to histotype.  It has been previously shown that IHC classifications can distinguish histotype as well (19).  From the standpoint of classifying ovarian cancers, proteomic assessment similarly may show some promise in determining histotype in cases that are difficult to assign based on light microscopy using H&E stains.   Further validation in a separate cohort would be required to prove the usefulness of such an approach.  In our study approximately 28% of HGS carcinomas are distinct by unsupervised clustering showing low expression of AR and ER.   This has been noted previously using some of the same tumours however significance analysis was not performed in this previous study (121).  92 This subset of HGS carcinomas is also characterized by lower levels of PR, β-catenin, α- catenin, Cyclin B1, E-cadherin, PTEN, and Stathmin.  In contrast, this subset of HGS carcinomas have higher levels of caveolin, HSP70, and transglutaminase expression, along with higher levels of stromal markers such as fibronectin, collagen VI, and CD31 relative to the larger group of HGS carcinomas.  This subgroup is of interest to study further.  Our study also provides evidence that BAF250a expression identifies a subgroup of CCC and ECs with higher AKT-Thr308 and AKT-Ser 473 phosphorylation that could contribute to changes in treatment responsiveness or provide a novel approach to target tumours with ARID1A/BAF250a aberrations.  Tumours lacking BAF250a expression show higher levels of AKT phosphorylation independent of PIK3CA mutation and PTEN loss suggesting a novel mechanism for activation of the PI3K/AKT pathway.  Remarkably similar findings have been published recently in a large cohort of endometrial cancers (122). For the subset of CCCs, logistic regression analysis showed that both pAKT-Thr308 and pAKT-Ser473 were associated with BAF250a expression loss. Although the mechanism is as yet unclear, this study strongly suggests that ARID1A/BAF250a aberrations may modulate PI3K/pAKT pathway activation in tumour samples.  Unlike previous findings in endometrial cancers (122), we have been unable to demonstrate a mechanism by which AKT phosphorylation occurs in tumours using cell line models.   We performed siRNA knockdown in 3 ovarian clear cell cancer lines (RMG1, ES2, and JHOC5) and despite good knockdown of BAF250a in these lines, we could not demonstrate an increase in AKT phosphorylation, or downstream signaling, although this may be time  93 sensitive (it might be worthwhile to check additional time points).   In addition, we confirmed this by using native protein assessment by capillary tube isoelectric point focusing.  Our findings suggest that cell line models may not accurately reflect the signaling changes in tumour samples; one possible explanation being that changes in AKT phosphorylation (pAKT) may be modulated by tumour/stromal interactions.  These findings are not in keeping with the previous reported effects of BAF250a knockdown in endometrial cancer cells lines and as such these observations may be lineage specific.  The biological explanation for pAKT changes in endometriosis associated ovarian cancers remains to be elucidated.  It could be that BAF250a expression is associated with other PI3K pathway abnormalities such as regulatory domain mutations as they have also been shown to alter pAKT (123).  We previously reported that ARID1A mutation status correlates well, although not perfectly with BAF250a loss by IHC (83). In this work we focused on a subset of cases from our original study, and here loss of BAF250a protein showed a stronger association with PI3K pathway activation than ARID1A mutation status, which may argue that assessment of BAF250a by IHC is preferable to sequencing ARID1A in tumour samples to identify associations with PI3K signaling in EOCs.  Similarly, IHC has some advantages over sequencing for the assessment of PTEN function in endometrial cancers though a much greater proportion of cases (44%) show PTEN loss in the presence of a normal PTEN sequence (123). This finding should be validated in a larger cohort as our sample size is small and we cannot exclude the importance of mutation status as a predictor of pAKT.  94 Mutations in ARID1A are more commonly heterozygous and it is not known whether these mutations truly have a dominant negative effect.  Proteomic assessment of CCCs and ECs identified a number of pathways that may be important as potential targets.  We identified over 50 proteins that are differentially expressed in CCC compared HGS.  This information is helpful to direct future studies.  In contrast to the many differentially expressed proteins associated with histotype, differential expression of proteins associated with BAF250a on IHC was limited primarily to the members of the PI3K pathway.  By SAM analysis, only changes in pAKT-Thr308 levels had a false discovery rate of <1%.   We could not prove that other downstream signaling events were being mediated by these changes in AKT phosphorylation though all of the top candidate proteins but one were key signaling proteins in the PI3K pathway.   Because CCCs and ECs are relatively uncommon compared to HGS carcinoma, it is difficult to gather large numbers of patient samples in order to have the power to detect multiple pathway aberrations.  For this reason it would be useful to study a larger population of patients to determine if these findings are reproducible and see if other downstream signaling candidates can be validated as well.  There is evidence indicating that different PIK3CA mutations have differing effects on PIK3CA signaling and AKT phosphorylation, and therefore may determine prognosis.  In cell lines, PIK3CA helical domain mutants are not consistently associated with increases in pAKT and appear to act through alternate signaling mechanisms potentially involving SGK3  95 (124). Similarly, we were not able to identify obvious increases in pAKT signaling in EOC patients with helical domain PIK3CA mutants.  CCCs are less sensitive to chemotherapy Mackay, 2010 #124} and it has not been possible to show that other chemotherapy agents such as the addition of irinotecan improve outcomes (125). In breast cancer patients, a recent study suggests that patients with PIK3CA mutations may have a more favorable prognosis (126), and patients with PIK3CA mutations irrespective of cancer type might have improved response rates to PI3K-directed therapies. Recently Rahman et al. studied 56 CCC showing a possible improved outcome associated with PIK3CA mutations (127). While the findings in this study are of interest, all PIK3CA mutations were analyzed together and the impact of PTEN loss was not examined. There are numerous regulatory events that impact on activation of the PI3K pathway and thus detailed assessments of pathway aberrations are required to better understand the mechanism of action of novel PI3K directed therapies (111).  Although ARID1A mutations are common in these EC and CCC subtypes, previous studies in ovarian cancer have failed to show any distinguishing clinical features or outcomes related to the presence of an ARID1A mutation (116). These findings do not negate the importance of future study on signaling pathway aberrations as they might have impact on treatment outcomes or represent therapeutic opportunities.  While histotype designation is key to classifying ovarian cancers, work should continue to identify subgroups or markers that identify patients who will benefit from novel therapies.   96 Chapter 5:  In vitro models and novel therapeutic targets for clear cell carcinoma  5.1 Introduction With the increased understanding of the genetic defects present in the different subtypes of ovarian cancer, an opportunity now exists to exploit these defects using targeted therapeutic strategies. Clear cell ovarian carcinomas (CCCs), a subtype of ovarian cancer with a typically poor prognosis and resistance to standard platinum-based chemotherapy (15, 32, 33), have inactivating mutations of the tumour suppressor gene ARID1A in approximately 50% of cases (84, 98). Cell lines and tumours harboring defects in ARID1A may have an Achilles heel that would render them preferentially sensitive to specific compounds in a similar synthetic-lethal way that cell lines and tumours with BRCA mutations are defective in the homologous recombination pathway and show sensitivity to PARP inhibitors (128).  This synthetic-lethal approach is attractive, as directly targeting known tumour suppressor genes and proteins has otherwise proven to be intractable. Screening for therapeutic compounds directed at tumours harboring mutations or alterations in a specific gene of interest utilizing in vitro systems requires the choice of proper control cells.  This may be problematic, as “normal” control cells may not actually correspond well to the gene and tumour of interest.  To address this issue, one solution that is commonly utilized is to generate an isogenic cell line pair that differs only in the single target gene of interest, which can then be used together for screening therapeutic targets or compounds. To try and create such a set of isogenic cell lines for functional studies of ARID1A, several approaches were taken.  First, lentiviral vectors expressing shRNAmir were used to try and  97 knockdown ARID1A in two cell lines containing normal levels of the gene- RMG1 (a clear cell carcinoma line), HCT116 (a well established and frequently studied colon cancer line), and JHOC5 (clear cell carcinoma cell line). Next, a lentiviral approach was used in a complementary strategy to try create an isogenic cell line pair with restored ARID1A function using the inducible RTTA system and the CCC cell lines TOV21G and JHOC9, both known to have mutations in ARID1A and complete loss of the BAF250a protein by IHC and western blot (summarized in Table 5.1).  Finally, transient transfection for the reintroduction of ARID1A and two truncated ARID1A mutants, C1570T (Q524*) and C1680A (Y560*), were performed with two BAF250a-null cell lines, TOV21G and OVTOKO and the BAF250a- positive cell line JHOC5 using an N-terminally tagged GFP vector (pCDNA-TM6-2-N- EMGFP-DEST) and an IRES-GFP vector (pSG5-IRES-EGFP), to assess the tolerance of the cells for reintroduction of either full length or mutant ARID1A. Overall, the cell line pairs most closely resembling suitable isogenic cell line pairs to carry forward were the RMG1 and HCT116 cell lines with substantial, knockdown of BAF250a with ARID1A shRNAmir (Figure 5.1 a and b). To identify novel therapeutic targets for potential clinical use in CCC, a kinase inhibitor screen was performed on these nearly isogenic cell lines, plus 18 additional ovarian cancer cell lines, including ten clear cell carcinoma lines with known ARID1A mutation status.  The kinase inhibitor panel consisted of over 340 compounds, with many that were successfully used in at least Phase 1 clinical trials, as well as a number of target specific kinase inhibitor tool compounds.   From the kinase inhibitor screen, several novel and promising compounds with Phase 1 clinical trial data emerged, including GSK461364 (targeting PLK1), as well as Pelitinib and Tovok (both inhibitors of EGFR).  To investigate the therapeutic potential of the shortlisted compounds,  98 ten-point dose response experiments were performed, and IC50 values were determined.  The compounds with the lowest IC50 values (and therefore the potential for the greatest target specificity with the lowest toxicity) were taken forward for further characterization of functional effects on the panel of CCC lines. 5.2 Methods 5.2.1 Isogenic cell line model creation strategies Gateway® vector creation The full-length human ARID1A cDNA ORF clone NM_006015.4 in pCMV6-XL4 was obtained from OriGene (OriGene Technologies Inc., Rockville, MD).   The shuttling of the full length ARID1A ORF from pCMV6-XL4 into the Gateway® entry vector pDONR221 was performed by Genscript (GenScript, Piscataway, NJ).  The correct orientation of the ARID1A clones in pDONR221 clone was confirmed by Sanger sequencing the full length of the ARID1A gene, as previously described (83). Sequencing primer sequences located in Appendix B (for Exon 1), and Appendix D.  Two ARID1A mutant constructs [C1570T (Q524*) and C1680A (Y560*)] were created by designing primers with adapter specific for the Gateway BP reaction to allow for shuttling of the PCR product into pDONR221 (primer sequences are located in Appendix D).  PCR products were generated for the ARID1A mutant constructs using TaKaRa LA Taq® Hotstart with the manufacturer’s recommended protocols (TaKaRa), and were shuttled into pDONR221 with a standard BP reaction (Invitrogen), followed by cloning into One-shot® Top-10 E. coli (Invitrogen).  ARID1A knockdown with shRNAmir Lentiviral constructs expressing ARID1A (NM_006015) shRNAmir were obtained from the  99 OpenBiosystems pGIPZ vector libraries (clone V2LHS_71866) and (clone V2LHS_71862). The hairpin sequence for V2LHS_71866 was: TGCTGTTGACAGTGAGCGCCCGCAGGAGCTATCTCAAGATTAGTGAAGCCACAG ATGTAATCTTGAGATAGCTCCTGCGGTTGCCTACTGCCTCGGA with the target sequence CGCAGGAGCTATCTCAAGA, and the hairpin sequence for V2LHS_71862 was TGCTGTTGACAGTGAGCGAGCATGTCCTATGAGCCAAATATAGTGAAGCCACAG ATGTATATTTGGCTCATAGGACATGCGTGCCTACTGCCTCGGA, with the target sequence CATGTCCTATGAGCCAAAT. A negative control vector, pGIPZ-EFGP was included in parallel. A day before infection 250,000-300,000 cells (RMG1 and HCT116) were plated to provide 50-60% confluence the following day.  Viral titre was estimated, and a multiplicity of infection of 1 was calculated.  Following infection with lentiviral constructs for 4-6 hours, the media was changed, and cells were allowed to recover for 2-3 days under puromycin selection.  Fluorescence microscopy was used to assess experimental success, and cells were FACS sorted to obtain a pure population of ARID1A-shRNAmir expressing cells. Protein from FACS sorted cells was harvested for western blot analysis, and RNA from FACS sorted cells was collected for RT-PCR.  Re-expression of ARID1A using the inducible Tet-On® Advanced system ARID1A was transferred into the doxycycline inducible vector pLVX-V5-mcs2GW through a Gateway® LR Clonase® reaction (InvitrogenTM, Life Technologies Corporation, Burlington, ON). EGFP (enhanced GFP) was used as a control gene in the empty vector.  Plasmids were cloned into One Shot® TOP10 Chemically Competent E. coli, as per the manufacturer’s instructions (Invitrogen™), prepared with QIAprep® (QIAGEN, Hilden, Germany), and end-  100 sequenced by Sanger sequencing to check for correct orientation of the inserts with Forward primer (5’-AACGTATGTCGAGGTAGGCGTGTA-3’), and the reverse primer (5’- ACTTCCATTTGTCACGTCCTGCAC-3’), as previously described (83). Lentivirus particles containing pLVX-V5-mcs2GW-ARID1A and pLVX-V5-mcs2GW-EGFP were generated for the infection of the ARID1A/BAF250a null cell lines TOV21G and JHOC9 that had been previously infected with lentiviral particles containing the rtTA-expressing Tet-On® Advanced plasmid (Clontech, Mountain View, CA), and selected with G418 (129). Following infection and recovery, as described above, cells were induced for 48-hours with doxycycline, and re-expression of ARID1A was assessed using western blotting.  Since extremely limited re-expression was obtained, and only in TOV21G, fluorescence-activated cell sorting (FACS) was used to try and isolate specific clones with successful re-expression of ARID1A/BAF250a.  Briefly, cells were stained with propidium iodide to allow for detection of cell viability, and sorted at a density of one cell per well into five 96-well plates (BD-FACS).  Following a two-week recovery and selection with G418 and puromycin, twenty-seven wells with viable clones were expanded, induced with doxycycline for 48- hours and re-assayed for ARID1A/BAF250a expression by western.  Transient transfections of ARID1A The pDONR221 vectors containing ARID1A or the ARID1A mutants C1570T (Q524*) and C1680A (Y560*) described above, were used to transfer ARID1A through Gateway® LR- clonase reactions (InvitrogenTM, Life Technologies Corporation, Burlington, ON) into the two vectors pSG5-IRES-EGFP and pCDNA-TM6-2-N-EMGFP-DEST (InvitrogenTM). Transient transfections of the ARID1A-null cell lines TOV21G and OVTOKO, plus the  101 ARID1A expressing cell line JHOC5 were performed using FuGENE® 6, as per the manufacturer’s protocol (Promega, Madison, WI).  Following transfection, plates were incubated in the IncuCyte™ live cell kinetic imaging system with four images collected for each well at 10X objective for 96 hours.  Expression of GFP was plotted using the IncuCyte™ software (Essen BioScience Inc., Ann Arbor, MI).  siRNA pre-treatment Knockdown of ARID1A was achieved using siRNA Smartpool from Dharmacon (Thermo Fisher Scientific Inc., Waltham, MA) with RNAiMAX™ (Invitrogen™) as described and shown in Chapter 4, Figure 4.4.  Real-time PCR for ARID1A Total RNA (1µg) was converted to cDNA using random hexamer priming and Superscript III reverse transcriptase, as per the manufacturer’s instructions (Invitrogen™).  Primers were designed using the Roche Universal Probe Library software, designed to span an intron-exon boundary, and ordered from IDT (Integrated DNA Techologies, Inc., Coralville, IA).  The forward primer sequence was 5’-CCAACAAAGGAGCCACCA-3’, and the reverse primer sequence was 5’-TTGCCCATCTGATCCATTG-3’, resulting in an amplicon length of 108nt. RT-PCR using SYBR® Green PCR Master Mix (Applied Biosystems, Life Technologies Corporation, Burlington, ON) on the ABI 7300 (Applied Biosystems).   Relative quantification was utilized with normalization against β-actin, and a melting curve analysis was performed to confirm amplification of a single product.   102 Western blotting ARID1A expression was confirmed by western blotting with an anti-BAF250a (ARID1A) antibody (Sigma) at a 1:1000 dilution, with actin was as a loading control (Sigma).  The primary antibodies were incubated for 2 hours at room temperature, and the secondary antibody (anti-rabbit or anti-mouse HRP conjugate) for 1 hour with agitation.  IHC for EGFR, HER2, and BAF250a IHC was performed on 4µm thick paraffin sections of a cell line array. EGFR and HER2 were stained on automated Ventana Discovery XT using the DAB Map kit. The EGFR antibody was purchased from Epitomics (Catalog number AC-0025, Clone EP22), and was used at a dilution of 1:50.  The HER2 antibody was purchased from Thermo Scientific (Catalog number RM-9103, Clone SP3) and used at a dilution of 1:100.   IHC for BAF250a was performed as previously described.  5.2.2 Methods for screening Kinase inhibitor screening The kinase inhibitor library of 340 kinase inhibitors was obtained through the OICR (Ontario Institute for Cancer Research). A complete list of all compounds used in the screen is provided in Appendix D.  Information on the 19 cell lines used in the screen is summarized in Table 5.1.  Additional lines included the screen were the RMG1 and HCT116 isogenic lines with shRNAmir knockdown of ARID1A (RMG1-71862) and RMG1-71866). Cells were plated in 96 well plate format at 10,000 cells/well and compounds were added to a concentration of 1µM per well. Following treatment, cells were incubated at 37°C for 48hr  103 for all lines except RMG1 and RMG2, which were incubated for 72hr as they grew much more slowly than the other cell lines. Following incubation, cells were fixed with 10% neutral buffered formalin and stained with 0.05% Crystal Violet (CV) and OD590 values were read on the Tecan Safire2 (Tecan Group Ltd., Männedorf, Switzerland).  The percent of cell viability was calculated from the OD590 ratio between treated wells with the untreated control wells present in the first and last column of each plate.  Secondary validation of kinase inhibitor screen Confirmatory tests were performed in triplicate with candidate compounds of most interest identified in the primary kinase screen. For secondary tests, compounds were cherry-picked from the kinase library master plates, and re-tested in triplicate at 1µM. For compounds that revalidated, stock compounds were ordered and IC50 values were determined by plating at 10,000 cells per well in 96 well plate format and the next morning treating with each compound at nine different concentrations, (1 nM, 3 nM, 10 nM, 30 nM, 100 nM, 300 nM, 1 µM, 3 µM, and 10 µM), plus untreated control wells, in triplicate, followed by 37°C incubation with 5% CO2 for 48 hours (or 72 hours for RMG1 and RMG2, the slowest growing lines) followed by formalin fixation and Crystal Violet staining (as described above).  IC50 values were determined using GraphPad Prism5 (GraphPad Software Inc., Playa La Jolla, CA).  Compounds for follow-up testing after secondary validation The twelve candidate compounds of the highest interest after secondary validations were purchased in larger quantities:  BIBW-2992 (Tovok) (Active Biochem), EKB-569 (Pelitinib)  104 (Selleckchem), AS601245 (Enzo), PHA-739358 (Danusertib) (Selleckchem), GSK461364 (Selleckchem), GW843682X (Sigma), BMS-3 (Symansis), PD17395-Analogue1 (Symansis), 6-bromoindirubin-3'-oxime, BIO (Tocris), XRP44X (Tocris), NH125 (Tocris), AP-24534 (Tocris).  Compounds were reconstituted to 1 mM or 10 mM stock concentration in DMSO as per manufacturer’s instructions, aliquots were stored at -20C, and working stocks of the compounds were made immediately before use.  Data analysis for kinase inhibitor screen Candidate compounds with the greatest growth inhibition of Clear Cell Carcinomas (CCCs) were identified as compounds that caused a decrease in cell viability of ≥50% compared with control untreated wells (present in the first and last column of each plate in the kinase inhibitor library), and which did not cause inhibition of growth in normal primary culture mouse embryonic fibroblasts (MEFs).  Hierarchical clustering of kinase inhibitor hits was performed with the ward dendrogram method.  Kinetic live-cell imaging assays: For both cytoxicity and apoptosis assays, cells were plated at 5,000 or 10,000 cell per well (depending on growth rate of the cell line), and were treated in triplicate with the test compounds at differing concentrations (Figure 5.5 and 5.6).  For monitoring of cellular cytotoxicity, Yoyo-1® dye from Essen Scientific was added to the cell culture media at a concentration of 1:10,000 (Essen BioScience Inc., Ann Arbor, MI). For apoptosis assessment, CellPlayer™ Caspase 3/7 reagent from Essen Scientific was added at a concentration 1:1000 to the culture media.  Following treatment, plates were incubated at  105 37°C in the IncuCyte™ for 72hr, and images were captured at 2-hour intervals from 4 different regions of each well using the 10X objective.  The IncuCyte™ software was utilized to quantify fluorescent objects over time from all four regions of each plate well, and values were pooled and averaged across the three replicates, and graphed.  Plates were then removed from the instrument, formalin fixed, and stained with crystal violet to allow for the assessment of non-cytotoxic (i.e. cytostatic) growth inhibition.  Real-time PCR for PLK and EGFR family members RNA was harvested from cells using QIAGEN RNeasy and first-strand cDNA synthesis was performed using Superscript III (Invitrogen).  Real-Time PCR was performed for all PLK family members (1-4) and EGFR family members (1-4) using primers designed using the Roche Universal Probe library system and SybrFast Green from Biorad on the ABI7300. Data was quantified against a standard curve run in parallel, and normalized to the housekeeping gene Lamin A/C.  Data from this experiment is summarized in Appendix D.  5.3 Results Cell line model creation RMG1 and HCT116 both showed lentiviral- mediated knockdown of greater than 80% for BAF250a by western blot (Figure 5.1a).  For both RMG1 and HCT116 cell lines, at the protein level the sh71866 construct appeared to provide better knockdown of BAF250a/ARID1A.  RT-PCR showed at least 70% knockdown of BAF250a transcript compared to the control lines with the pGIPZ (Figure 5.1b).   106 A  B    Figure 5.1 Lentiviral knockdown of ARID1A in RMG1 and HCT116. (A) Western blot for ARID1A/BAF250a (250kDa) in HCT116 and RMG1 with PGIPZ Control, sh71862 and sh71866, with Actin (42kDa) as a loading control. TOV21G and VOA782 are negative controls for BAF250a expression, and IOSE398 is a positive control.  (B) Real-Time PCR for ARID1A to assess lentiviral knockdown in RMG1 and HCT116 with control (PGIPZ), sh71862 and sh71866, normalized to β-actin.   BAF250a Actin  107 The use of inducible ARID1A containing lentiviral vectors to restore ARID1A expression back into the well-characterized clear cell carcinoma line TOV21G and JHOC9 was of little success, possibly due to the limitation of lentiviral constructs to package inserts greater than 8-10kb (130, 131). An extremely weak band was visible at 250kDa (the expected size for BAF250a) after selection and induction with doxycycline.  To try and locate a clone with expression of BAF250a, FACS sorting was utilized.  Out of the 580 wells seeded by FACS sorting (five 96 well plates) 27 clones recovered over the two weeks under G418 and puromycin selection.  These 27 clones were expanded and induced with doxycyline to check for re-expression of ARID1A/BAF250a.   Out of the 27 clones obtained, no clones appeared to express the 250kDa protein.  To investigate how cell lines tolerated ARID1A or ARID1A mutants by transient transfection, full length ARID1A, as well as two truncated versions of ARID1A (C1680A and C1570T), were shuttled into the vectors pSG5-IRES-GFP and pCDNA6.2-N-GFP.  Three cell lines (TOV21G, OVTOKO, and JHOC5) were transiently transfected with the ARID1A wild type or mutant constructs, and monitored on the IncuCyte™ (Figure 5.2 A-C); in all three cell lines, wild type ARID1A transfection was not well tolerated.  Figure 5.2 A-C (pages 124-126) Transient transfection of (A) TOV21G, (B) Ovtoko, and (C) JHOC5 with ARID1A or ARID1A mutants C1570T (Q524*) and C1680A (Y560*) in pSG5-IRES-EGFP or pCDNA-TM6-2-N-EMGFP-DEST on IncuCyte™ Zoom live cell kinetic imaging system, with images captured at 10X every two hours.  For each set, phase contrast and green fluorescence images are shown in the inset panels A-H, green fluorescent object plots are shown in I, and confluence plots in J.   108 Figure 5.2  (A) TOV21G 48hr transfection with full length ARID1A or ARID1A truncated mutants C1570T (Q524*) and C1680A (Y560*)  109   Figure 5.2  (B) OVTOKO 48hr transfection with full length ARID1A or ARID1A mutants C1570T (Q524*) and C1680A (Y560*)   110 Figure 5.2  (C) JHOC5 48hr transfection with full length ARID1A or ARID1A mutants C1570T (Q524*) and C1680A (Y560*)   111   Figure 5.3 (A) Overview of kinase inhibitor screening on 19 ovarian cancer cell lines including ten clear cell carcinoma (CCC) lines with known mutation or protein status of ARID1A/BAF250a. (B,C) TOV21G and ES2, respectively- two clear cell carcinoma (CCC) cell lines, each treated with the panel of ~340 OICR kinase inhibitors for 48 hours, then stained with crystal violet (CV) to assess cell viability.  Wells with inhibition stain less intensely and can be read at OD590, allowing for the calculation of the percent inhibition against untreated wells (present in the first and last columns of each plate) A  112 Table 5.1 Cell lines used in the screening.  Subtypes as reported in the literature, and also classified by immuno- and mutational profiling (a version of this table is used by permission and in preparation for publication elsewhere- please see preface).   p1 6 (C D K N 2A ) M D M 2 T FF 3 p5 3 (T P5 3) V IM E N T IN W T 1 H N F1 B PR D K K 1 C C O C E N O C a Se ro us M U C A R ID 1A ¶ (B A F2 50 A ) E G FR H E R 2 JHOC-5 CCC 1 0 0 1 1 0 1 0 0 85 13 2 0 1 3 0 none detected CCC JHOC-7 CCC 1 1 0 1 1 0 1 0 0 99 1 0 0 0 3 0 PIK3CA CCC JHOC-9 CCC 1 0 0 1 1 0 1 0 0 85 13 2 0 0# 3 0 PIK3CA/ARID1A CCC RMG-2 CCC 0 1 0 1 1 0 1 0 1 97 3 0 0 0 3 0 PPP2R1A/ARID1A CCC TOV21G CCC 0 0 0 1 1 0 1 0 1 55 41 4 0 0#* 1 0 KRAS/PTEN/PIK3CA/ARID1A CCC OVTOKO CCC 1 1 0 1 1 0 1 0 0 99 1 0 0 0 n/a n/a none detected CCC OVMANA CCC 1 1 0 1 1 0 1 0 0 99 1 0 0 0 n/a n/a PIK3CA/ARID1A CCC OVISE§ CCC 0 0 0 1 0 0 1 0 1 20 54 26 0 0 n/a n/a ARID1A EC VOA782_XL EC 0 1 0 1 1 0 1 0 0 100 0 0 0 0# n/a n/a PIK3CA/ARID1A CCC A2780 Adenocarcinoma 0 0 0 1 1 0 0 0 1 0 94 6 0 0* 1 0 PTEN/ARID1A EC ES-2 CCC 1 0 0 1 1 0 0 0 1 0 100 0 0 1 3 0 BRAF EC TOV112D EC 0 0 0 2 1 0 0 0 1 0 38 62 0 1 0 0 CTNNB1 EC*** OVSAYO CCC 0 0 0 2 0 1 0 0 1 0 0 100 0 1 n/a n/a (n/a) HGSC OVCAR-4 Serous Adenocarc. 0 0 0 0 1 0 0 0 1 0 0 100 0 1 3 0 none detected HGSC SKOV3 adenocarcinoma 0 0 0 0 0 0 1 1 1 0 0 100 0 1* n/a n/a PIK3CA/ARID1A Inconclusive (EC) **** RMG-1 CCC 0 0 0 0 0 0 1 0 1 22 0 76 3 1 2 0 none detected HGSC KLE-1 Endometrial - - - - - - - - - - - - - 0* n/a n/a ARID1A - OVSAHO HGS - - - - - - - - - - - - - 1 n/a n/a (n/a) - HCT116 Colon - - - - - - - - - - - - - 1 n/a n/a - - Purple highlighting indicates cell lines profiled for COSP markers scored as IHC negative (0) or positive (1), except for p53.  P53 null mutation (0), wildtype (1), mutated (2). EGFR scoring is (1) normal, (2) positive, or (3) overexpressed Green highlighting indicates subtype most highly predicted by COSP ¶ARID1A nonsense or frameshift mutation as detected via sequencing (#) or reported in CCLE (*), (0) is BAF250a negative by IHC, and (1) is BAF250a positive by IHC § DKK1 IHC staining for Ovise is questionable (if negative would be CCC) **In the case of VOA1056_CL the mutation data, primary histology and/or primary expression data is influencing histo-type prediction. *** Beta-Catenin (CTNNB1) mutation and genome-wide expression profiling of the primary tumour consistent with EC (Madore et al) supports classification as EC. **** A truncating non-sense ARID1A mutation from sequencing suggests this cell line is not HGSC. Mutational profiles determined by targeted amplicon sequencing of BRAF, KRAS, ERBB2, NRAS, CTNNB1, EGFR, PTEN, PIK3CA, PPP2R1A, DICER1. Table 5.1  Validation of the histotype of commonly used ovarian cancer cell lines using immuno- and mutational profiling Cell Line Reported Histotype in Literature COSP Markers COSP Prediction (Clinical) Mutational Profile Validated Histotype based on immuno- and mutational profiles Non-COSP Markers  113 Kinase inhibitor screening results An outline of the workflow for screening of the 19 cell lines plus the RMG1 and HCT116 isogenic cell lines with the 340 candidate kinase inhibitors is outlined in Figure 5.3. Based on the known ARID1A defects in the CCC lines, no compounds emerged that were specifically targeted to ARID1A defective cell lines. However, from the initial screen 26 candidates were identified that appeared to be effective at inhibiting growth of ≥50% of the clear cell carcinoma (CCC) cell lines included in the screen, without causing cell death of normal Mouse Embryonic Fibroblasts (MEFs).  Interestingly, based on the literature classifications of the cell line subtypes, there was no one compound that uniformly affected a particular subtype (Figure 5.4), however, there was evidence for the emergence of treatable subgroups.  For example, in Figure 5.4 the bottom 4 cases of the cluster with 2 HGS and 2 CCC cell lines.  From the primary screen, the 26 shortlisted compounds were cherry-picked from the library for secondary validation.  Each of the 26 compounds was tested again in triplicate on the CCC cell lines using the same conditions as in the primary screen.  The results from the secondary screening are summarized in Table 5.1.  Following the secondary validation of the compounds, the twelve compounds showing the greatest inhibitory effects on the CCC cell lines were selected and purchased from individual vendors. Five of these compounds, marked in Table 5.2, were found in clinicaltrials.org to have passed at least Phase 1 clinical trials.  One compound included in the shortlist (AV-412) had a clinical trial terminated, and  114  Figure 5.4 Hierarchical, two-way clustering utilizing the Ward algorithm was used to examine the impact on the growth of 19 cell lines against 340 compounds. Along the Y-axis are 30 clusters which were derived using the cubic clustering criteria (Sarle, The Cubic Clustering Criteria,1983, SAS Institute, Cary NC U.S.A.).  These clusters contain compounds which have similar in-vitro efficacy profiles across groups of cell lines.  The X-axis represents the similarity of the cell lines with respect to their sensitivity to the 340 compounds used in this analysis.       115 this was taken out of the candidate list.  Since AV-412 had been replaced by a more advanced compound (EKB-569), this compound was purchased instead and carried forward into additional testing.   For each of the twelve compounds obtained, ten point dose response curves were generated, and IC50 values for each compound were determined.  The results for the compounds with the lowest IC50 are summarized in Table 5.3. Lower IC50 values indicate greater target specificity and lower potential toxicities to patients. The three top candidates, as indicated in Table 5.3, were the PLK1 inhibitor GSK461364, BIBW-2992 (also known as Afatinib or Tovok), and EKB-569 (Pelitinib), both inhibitors of EFGR1 and EGFR2.  From the primary screen of the RMG1 and HCT116 isogenic cell lines with pGIPZ control screened in parallel with the panel of kinase inhibitors, 3 candidate compounds NH125 (an inhibitor of EEF2K), PD407824 (a CHK1, Wee1 inhibitor), and 547757-23-3 (an IKK inhibitor) appeared to be specific for the HCT116 line with ARID1A knockdown, however these did not hold up on secondary validation.  116 Table 5.2 Secondary screening results: compounds most effective at inhibiting CCC cell line growth       117 Table 5.3 IC50 (nM) values of top four candidate compounds on ten CCC cell lines  Ovsayo Ovtoko Ovise Ovmana TOV21G RMG1 ES2 JHOC5 JHOC9 RMG2 Average  ( IC50(nM ) GSK461364 14.3  8.8 13.6 14.2 24.4 19.9 31.7 68.9 9.1 >1µM 22.8 Pelitinib 56.8 97.5 72.1 79.3 333.4 79.4 494.0 95.4 22.1 45.2 137.5 BMS-3 337.7 808.9 546.6 422.2 623.3 >1µM >1µM 578.5 182.2 855.9 544.4 Tovok 9.9 325.9 13.1 12.8 832.5 50.8 >1µM 34.1 858.9 226.4 262.7  118 Table 5.4 Summary of cytostatic and cytotoxic effects of GSK461364, Pelitinib and Tovok on 10 CCC Cell Lines and MEFs (normal cells)   Cell Line GSK-461364 Pelitinib Tovok RMG1 Cytostatic Cytostatic Cytostatic RMG2 Cytotoxic Cytostatic Cytostatic JHOC5 No effect No effect No effect JHOC9 Cytotoxic No effect No effect Ovmana Cytostatic Cytostatic Cytostatic Ovise Cytostatic No effect No effect Ovsayo Inhibition- odd Cytostatic Cytostatic Ovtoko Cytotoxic Cytostatic Cytostatic TOV21G Cytotoxic No effect No effect ES2 Cytotoxic No effect No effect MEFs (normal) No effect No effect No effect      Cytostatic  Cytostatic 3/10 (30%) 5/10 (50%) 5/10 (50%) Cytotoxic 5/10 (50%) 0 0 Undetermined 1/10 (10%) 0 0            119 Cytotoxicity and apoptosis assays with kinetic live cell imaging To determine whether the growth inhibition observed in the clear cell carcinoma cell lines by the kinase inhibitors GSK461364, EKB-569, and BIBW-2992 was cytotoxic or cytostatic, the cell lines were treated with the compounds in media containing Yoyo-1®, a normally cell- impermeable DNA stain, and assayed for 72 hours in the IncuCyte Zoom kinetic live cell imaging system.  Results from the cytotoxicity assay are summarized in Table 5.4, with representative graphs and images of the responses shown in Figure 5.5 and 5.7. Strong cytotoxic response to the PLK1 inhibitor GSK461364 was observed in half of the CCC lines including TOV21G, ES2, RMG2, OVTOKO, and JHOC9, as shown by increasing intracellular green fluorescence due to the cellular uptake of Yoyo-1®, indicating the loss of membrane integrity and decreasing cellular confluence over the 72 hour treatment period, while viable cells remain unstained and reach cellular confluence. In three out of five remaining cell lines treated with PLK1 inhibitor that did not show cytotoxic response, cytostatic response was achieved (RMG1, Ovmana, and Ovise). Cytostatic response was defined as reduced cellular confluence over time and lack of increased Yoyo-1® staining, indicating growth inhibition without cytotoxic response.  Cytostatic effects of Pelitinib and Tovok, the inhibitors of EGFR, were evident consistently on five cell lines (Table 5.4), including RMG1 and RMG2, plus three out of the four Japanese Clear Cell Lines (Ovmana, Ovsayo, and Ovtoko).      120 Apoptosis assays on the IncuCyte™ Zoom In an effort to determine the mechanism of the response obtained in the CCC lines, three cell lines showing strong cytotoxic response with the PLK1 inhibitor GSK461364 were treated with inhibitor and Essen’s CellPlayer Caspase 3/7 activation kit was used on the IncuCyte Zoom to obtain kinetic measurements of the number of Caspase-3/7 positive cells. Phase contrast with fluorescence images show fluorescent activation of Caspase-3/7 using Essen’s CellPlayer Caspase 3/7 activation kit on the IncuCyte Zoom with typical morphologies observed in apoptotic cells (Figure 5.6 A), with the results plotted on a graph in Figure 5.6 B.  121  A                                                                                          B      Figure 5.5  (A) Dose response titration of GSK461364 (Plk1 Inhibitor), with and without retinoic acid (RA), plus Pelitinib and Tovok (both EGFR inhibitors) on TOV21G and ES2 cell lines with the IncuCyte Zoom YOYO-1® cytotoxicity assay. (B) Increased cellular cytotoxicity occurs over 72 hour treatment of CCC lines TOV21G and ES2 with the Plk1 inhibitor as shown by increasing green fluorescent YOYO-1® object counts over a 72 hour period.  Nuclear staining by the normally cell-impermeable DNA stain YOYO-1® indicates loss of membrane integrity, a hallmark of cell death. Of note, the inhibitory growth effects exhibited by the EGFR inhibitors Pelitinib and Tovok are not detectable using the cytotoxic assay on the TOV21G and ES2 clear cell carcinoma lines, indicating that the mechanism of inhibition by these compounds is cytostatic.  122 A   B  Figure 5.6 GSK461364 (PLK1 Inhibitor) induces Caspase-3/7 activity in Clear Cell Carcinoma (CCC) cell lines TOV21G, JHOC9 and OVTOKO. (A) Representative phase contrast with fluorescence images show fluorescent activation of Caspase-3/7. (B) Kinetic measurement of the number of Caspase-3/7 positive cells is recorded over time and plotted as fluorescent objects detected (three replicates per data point shown).  123 Figure 5.7 GSK461364, Pelitinib and Tovok 48hr treatment of (A) ES2, JHOC9, and Ovtoko (B) RMG2, TOV21G and MEFs (normal)     A    124 Figure 5.7 B    125 Investigating the possible ARID1A dependency of GSK461364, Pelitinib, and Tovok JHOC5 is a clear cell carcinoma cell line with no known ARID1A mutations, normal BAF250a protein expression, and no response to the inhibitors GSK461364, Pelitinib, or Tovok.  ARID1A siRNA pre-treatment of the cell line JHOC5, and also JHOC5 with shRNAmir knockdown of ARID1A were tested to determine if the inhibition of ARID1A affected the response to the three kinase inhibitors GSK461364, Pelitinib, and Tovok.  Cells were either pre-treated with ARID1A specific siRNA to induce ARID1A knockdown, then treated with kinase inhibitor, or, in the case of the shRNAmir JHOC5 line, plated and treated with inhibitor the following day. Neither the pre-treatment of JHOC5 with the ARID1A siRNA nor shRNAmir sensitized the JHOC5 cell lines to any of the three inhibitors (data not shown).  5.4 Discussion and conclusions  Isogenic cell line model creation strategies To create an appropriate set of isogenic cell lines for functional studies of ARID1A, several approaches were taken.  First, lentiviral vectors expressing shRNAmir were used to try and knock out ARID1A from two cell lines containing normal levels of the gene- RMG1 an ovarian clear cell carcinoma line (according to the literature) with no known ARID1A mutations and normal BAF250a expression, plus HCT116- a well established colon cancer line, used in many cancer studies. Lentiviral vectors expressing shRNAmir constructs allow for stable integration and expression of constructs and should allow for the creation of stable isogenic cell line pairs. Lentiviral vectors containing shRNA and shRNAmir utilize infection-  126 based delivery, which has the advantage of being effective in most cell lines, even those that are normally difficult to transfect (including cells that are not actively dividing or cells in primary culture). An additional advantage of utilizing shRNA and shRNAmir knockdown is that the off-target effects sometimes seen using synthetic siRNAs are not present. Lentiviral vectors with second-generation shRNAmir design offer an advantage over traditional shRNA, as the mir element gives greater processing efficiency by Drosha and Dicer compared to shRNA that are not based on microRNA.  This enhanced processing ability results in greater production of siRNA/miRNA and should result in increased knockdown efficiency.  The OpenBiosystems pGIPZ shRNAmir lentiviral vector used in this study incorporated green fluorescent protein (GFP) and antibiotic (puromycin) selection markers, which allowed for FACS sorting and stable expression of the constructs.  Using this approach, with both FACS sorting and post-infection selection of the cells on puromycin, the result was the generation of cell lines with substantial knockdown of the ARID1A transcript or BAF250a protein (nearly isogenic cell line pairs), as the majority of BAF250a protein was abolished after knockdown and selection in RMG1 and HCT116 (Figure 5.1).  Next, a lentiviral approach was used in a converse strategy to create an isogenic cell line pair with restored ARID1A function using the inducible RTTA system and the prototypic CCC cell line TOV21G, known to have mutated ARID1A and complete loss of the BAF250a protein by IHC and western blot. Inducible shRNAs are frequently used to investigate the function of essential genes.  The inducible vector chosen (PLVX-V5-puro) is unable to express the shRNA unless doxycycline is added (and providing the RTTA element has already been stably introduced into the cell line).  This approach proved to be problematic, as  127 the lentiviral reintroduction resulted in very little re-expression of the protein. FACS (fluorescence activated cell sorting) of the TOV21G-PLVX-V5-mcs2GW-ARID1A cell line was employed to try and isolate clones with stable re-expression of ARID1A, however this strategy also proved to be unfruitful. The difficulties observed could possibly be due to the limitation of lentiviral constructs to package inserts greater than 8-10kb (130, 131).  A second factor that may have limited the success of the approach is the RTTA system itself, as the cell line must first have the RTTA expressing plasmid successfully in place for the system to function, and this should be the case as the RTTA element contains a G418 selection marker, however the selection may not result in a perfect population of cells that all express the RTTA element.  These two confounding factors may have had a negative effect in trying to establish an ARID1A-inducible cell line, since in fact what you are trying to achieve is a double-stable cell line with both the RTTA element and lentiviral construct with the gene of interest both in place.  To determine whether the problems encountered in trying to re-express ARID1A were perhaps in fact due to the functional role of the BAF250a protein, transient transfection of ARID1A with either full length ARID1A or truncated mutant versions of the ARID1A gene [C1570T (Q524*) or C1680A (Y560*)] was performed in the CCC cell lines TOV21G and OVTOKO (BAF250a-null- Table 5.1), and JHOC5 (BAF250a positive- Table 5.1) using two vectors pSG5-IRES-EGFP and pCDNA-TM6-2-N-EMGFP-DEST (Figure 5.2 A-C). As suspected and shown in Figure 5.2, the mutant ARID1A constructs were more easily tolerated by the cell lines than full length ARID1A, with the N-terminally tagged mutants being more highly expressed than the IRES versions in all three cell lines.   It has been  128 reported BAF250a functions to repress several genes involved in the cell cycle, including Cyclin A, Cyclin E, and c-myc (72).  Therefore, reintroduction of ARID1A could effectively put the breaks on cell growth, which would make it exceedingly difficult to select for cells expressing the gene, as they would effectively stop growing.  Kinase inhibitor screening of clear cell carcinoma cell lines To identify novel therapeutic targets for potential clinical use in clear cell carcinoma, a kinase inhibitor screen was performed using a panel of over 340 compounds, many of which have been successfully used in at least Phase 1 clinical trials.  Nineteen different ovarian cancer cell lines were included in the screen, including ten clear cell carcinoma lines with known BAF250a/ARID1A status (Table 5.1), plus the isogenic RMG1 and HCT116 cell lines with lentiviral mediated knockdown of ARID1A.  The rationale for this screening strategy was that cell lines harboring the ARID1A defect may have an Achilles heel rendering them preferentially sensitive to specific compounds in a similar synthetic-lethal way that cell lines and tumours with BRCA mutations are defective in the homologous recombination pathway are sensitive to PARP inhibitors. Based on the known ARID1A defects in the CCC lines (Table 5.1), no compounds emerged that were specifically targeted to ARID1A defective cell lines, potentially because CCC cell lines without ARID1A mutations or BAF250a loss may have some equivalent but as of yet undetermined genetic or epigenetic abnormality. GSK461364 was found to have the lowest average IC50 value of 22.7nM on the CCC lines. From the cytotoxicity data obtained, the mechanism of inhibition by GSK461364 was strongly cytotoxic at low concentrations in half of the CCC lines (Table 5.4). The observed cellular cytotoxicity was due to apoptosis through Caspase 3/7 activation (Figure 5.6), which  129 is in agreement with the mechanism of response observed in other cell line systems (132). Conversely, the two EGFR inhibitors were found to inhibit growth through a cytostatic mechanism, as growth inhibition was clearly visible from compound treatment followed by crystal violet staining, however no cytotoxic response was generated when the CCC lines were treated with the EGFR inhibitors and monitored for cellular staining with the YOYO- 1®, which would be indicative of compromised membrane integrity (a characteristic feature of cell death) (Table 5.4). Combination treatment of GSK461364 with either Pelitinib or Tovok was tried, however no additional inhibition was observed (data not shown).  PLK1 inhibitor GSK461364 GSK461364, an inhibitor of PLK1 caused cytotoxic inhibition in 50% the CCC cell lines and was cytostatic in another 30% of the CCC lines, without having any effect on normal cells (in this case, Mouse Embryonic Fibroblasts (MEFs)) (Table 5.4, Figures 5.5 and Figure 5.7). PLK1 (polo-like kinase 1), one of four members of the polo-like kinase family, is an essential serine/threonine kinase required for the cell cycle and cell division, playing diverse roles in mitotic entry, centrosome maturation, and cytokinesis (133).  PLK1 has been found to be overexpressed in ovarian cancer and many other cancer types, and is becoming an increasingly attractive target for cancer therapeutics (133-135).  Studies have shown targeting PLK1 with the ATP-competitive inhibitor GSK461364 can significantly inhibit the growth of many different cancer cell types without adverse effects on non-dividing human cells (136). GSK461364 has been used in one phase 1 study of patients with advanced disease, which established a recommended dose and schedule for phase II trials (137).   130 To investigate whether or not the inhibition by PLK1 might be due to defective ARID1A in the cell lines, GSK461364 was tested on the RMG1 lentiviral knockdown of ARID1A, and additionally, GSK461364 was tested on JHOC5 with ARID1A knock down by siRNA.  In these instances, no increased inhibition of growth was observed with the knockdown of ARID1A, suggesting other molecular features of the cell lines may be responsible for the effectiveness of GSK461364 inhibition in the CCC lines.  However, by immunoprofiling RMG1 may actually more closely resemble high grade serous carcinoma (Table 5.1), so this may not have been the most appropriate model to choose- unfortunately the immunotyping data was not available until very recently- after completion of the study.  JHOC5 on the other hand expresses BAF250a/ARID1A and does not have a detectable ARID1A mutation, however by immunoprofiling it looks to be a “true” clear cell carcinoma line, suggesting perhaps it has a functionally equivalent mutation to ARID1A.  Other model systems, besides RMG1 and JHOC5, may have been better choices for determining ARID1A dependence.  Liu-Sullivan et al determined from an shRNA library screen that two shRNAs targeting the nuclear retinoic acid receptor silenced RARA expression and conferred resistance to GSK461364 (138). This led them to test whether activation of RARA receptor with retinoids could sensitize cells to GSK461364. They found that retinoids increased the sensitivity to GSK461364 and enhanced the ability of the compound to induce mitotic arrest and apoptosis in lung cancer cell lines (138). To test this in ovarian cancer models, GSK461364 was tried in combination with 1µM all-trans retinoic acid (ATRA) (Figure 5.5 A and B).  No enhancement in GSK461364 was observed with the addition of the ATRA to the treatment. One possibility for the observed difference in sensitivity could be that the Liu-Sullivan group  131 utilized a 96-hour pre-treatment with retinoids, followed with a 72-hour treatment with PLK1 inhibitor GSK461364, instead of utilizing the treatments simultaneously.  Reports are conflicting whether or not the effects of PLK1 are due to defects in p53 (139).  It has been reported that CCCs have mutually exclusive mutations with either ARID1A or p53 (115), however in the clear cell carcinoma cell lines we have studied we have not found this to be the case, as few of the lines harbor mutations in p53 (Table 5.1). It has been reported that cells with defective DNA damage response (i.e., depletion of ATM) may be more sensitive to inhibitors of PLK1, and undergo apoptosis (132).  It is possible that the subset of clear cell carcinomas more sensitive to the GSK461364 inhibitor may have undetermined DNA damage response defects or other abnormalities that render half of the clear cell carcinoma lines more sensitive to inhibition of PLK1.  Further investigation will be required to determine what the underlying defect (or defects) are that cause specific clear cell carcinoma cell lines to be more sensitive to inhibition of PLK1 or EGFR.  To determine whether the PLK1 inhibitors and EGFR inhibitors might prove to be synergistic in the killing of CCC lines, a combination treatment strategy was tried with GSK- 641364 and each EGFR inhibitor, however no synergistic effect was achieved by the addition of both treatments at the same time (data not shown).  An alternate approach, which might be worth investigating, could be a pre-treatment strategy, with the addition of the EGFR inhibitors prior to GSK461364, as suggested by Liu-Sullivan for the retinoic acid treatments as their paper lists EGFR inhibition as one of the top cell surface receptors that sensitized to inhibition by PLK1 (138).  132 EGFR inhibitors Pelitinib and Tovok In this study, half of the clear cell lines exhibited a cytostatic response to Tovok (BIBW2992) and Pelitinib (EKB-569).  EGFR (also known as ErbB-1), is a member of the ErbB family of receptor tyrosine kinases, activated by ligands such as EGF (epidermal growth factor) and TGF-β (transforming growth factor-beta) (140).  EGFR activation leads to activation of signaling pathways such as MAPK and PI3K, which are involved in cellular processes such as proliferation, differentiation and survival (140). It has been reported that EFGR is mutated in 15% of ovarian clear cell carcinomas (141).  To determine whether or not the EGFR response obtained was directly related to levels of receptor, IHC was performed on six of the CCC cell lines involved in the study (Table 5.1).  Interestingly, five of the six cell lines tested appeared to have overexpression of the EGFR receptor, however, only two of the lines overexpressing the receptor showed cytostatic response to the inhibitor.  This may indicate downstream signaling targets of the EGFR receptors are the true molecular targets of EGFR inhibition, and signaling mechanisms are blocked regardless of the EGFR levels expressed by the cell.  To investigate whether or not the inhibitory effects of Pelitinib and Tovok might be due to defective ARID1A in the cell lines, the same approach was taken as with the GSK461364 inhibitor, with tests of both inhibitors performed on the JHOC5 cell lines with siRNA and on the RMG1 lines with lentiviral knockdown of ARID1A.  As was observed with the GSK461364 tests, with both of these compounds no increased inhibition of growth was observed with the knockdown of ARID1A, once again suggesting other molecular features of  133 the cell lines are responsible for the effects of Pelitinib and Tovok in the CCC lines, or different model systems should be tested for this purpose.  It will be of importance to determine which genetic characteristics of the clear cell carcinoma cell lines are responsible for the inhibitory effects observed with treatment by the PLK1 inhibitor GSK461364 and the EGFR inhibitors Pelitinib and Tovok.  With respect to the EGFR pathway inhibition, one possibility emerging from the NanoPro phosphoprotein profiling experiments is that the ERK pathway is dramatically activated with EGF stimulation in a subset of the CCC cell lines, therefore ERK may be the true target inhibited by Pelitinib and Tovok.  However, there are other possibilities for the effects observed, and further studies will be required to determine this.  Overall, there were several limitations to this study. First of all, the CCC cell lines without ARID1A mutations may have other equivalent mutations or epigenetic alterations that have yet to be discovered, which may help to explain why there were no compounds that specifically targeted the ARID1A mutant lines.  Another limitation of this study and with the analysis of the data is that the cell line models may not be exactly what has been reported in the literature.  By using immunomarkers to determine the cellular phenotype some of the cell lines used as typical clear cell carcinoma lines may actually more closely resemble other subtypes (Table 5.1).  The COSP (142) profiles were established after the screening had been completed, so unfortunately this was not taken into account with the present analysis, but it would definitely be worth revisiting the dataset utilizing the new COSP-defined immunophenotypes.  On that note, as mentioned, one of the cell lines chosen for the  134 shRNAmir work, RMG1, by COSP analysis looks like it more closely resembles the HGS subtype- therefore this may have been a poor choice for development into an ARID1A isogenic line.  Another potential weakness of this study was that the initial screen was conducted at a single point per cell line per compound, which may have led to generation of false positives initially.  Also, the 1µM concentration chosen for the initial and secondary screen is quite high for a kinase inhibitor, therefore off target effects may have been observed for some of the compounds with the initial screens.  As the combination treatment of the PLK inhibitors with EGFR inhibitors identified in this study did not provide synergistic effects, combinations with other candidate inhibitors would warrant investigation- potentially by targeting PIK3CA, pAKT, or ERK (as per the findings in Chapter 4), although combining kinase inhibitors brings the risk for increased adverse patient reactions.  Other avenues of combination treatments in CCC may also be worth evaluating, perhaps kinase inhibitors with radiotherapy, as that has recently shown some promise in CCC (143).  In conclusion, the mechanisms underlying the responsiveness of clear cell carcinomas to PLK1 and EGFR warrant further investigation to determine which tumours will be successfully treated with these inhibitors, but the Achilles heel of tumours carrying ARID1A remains elusive.     135 Chapter 6:  Overall summary, conclusions, and future directions  6.1 Overall summary and conclusions The studies presented in this dissertation focused on determining the molecular events critical in the development of clear cell carcinoma, a subtype of ovarian cancer with a poor prognosis at advanced stages that does not respond well to traditional platinum/taxane chemotherapy. In Chapter 2, frequent mutations in the SWI/SNF gene ARID1A were described in clear cell and endometrioid carcinomas. ARID1A is located at 1p36.11 (96), a region long recognized as being commonly deleted in cancers (68), and suspected to contain tumour suppressor genes. Prior to this study, there were some suggestions that ARID1A might be important in human malignancies (65, 70, 71).  The study presented in Chapter 2 provided the first evidence for a potential tumour suppressor role of ARID1A in endometriosis associated ovarian carcinomas, with 46% of CCC and 30% of EC demonstrating somatic truncating or missense mutations in ARID1A. The ARID1A mutations identified were mostly truncating mutations evenly distributed across the gene, consistent with the mutation patterns commonly seen in classic tumour suppressor genes. Unlike BRCA or TP53 mutations, which can be found in the germline DNA, all of the truncating ARID1A mutations tested in germline DNA were found to be somatic.  This is not surprising as haploinsufficiency of ARID1A results in embryonic lethality in mice (76). A simultaneous publication describing frequent ARID1A mutations in clear cell carcinomas confirmed the results that we found (84), and since the publication of these manuscripts many reports have been published describing the mutation of ARID1A in other gynecological malignancies as well as completely different types of cancer (recently reviewed by Wu and Roberts) (144).  136 We were initially a bit surprised that additional mutations were identified when the initial CCC cases submitted for RNA-seq were analyzed by targeted genomic whole exon resequencing.  However, the additional mutations were likely not initially detected in RNA- seq data either due to transcripts being rapidly targeted for nonsense mediated decay (NMD), or due to a decreased sensitivity to mutations at the 5’ end of transcripts- a limitation inherent to the RNA-seq methodology.  Although RNA-seq proved to be an extremely useful discovery tool for this project, targeted exon resequencing or whole genome resequencing would likely be more appropriate for determining the true mutation frequency.  It was extremely fortunate that we identified early on an antibody that worked extraordinarily well for detecting nuclear staining of the BAF250a protein with IHC.  This allowed for the interrogation of BAF250a/ARID1A loss utilizing tissue microarrays. This IHC analysis showed loss of nuclear BAF250a was visible in 36% of CCCs and ECs but only 1% of HGS carcinomas. The presence of mutations was strongly correlated with loss of BAF250a protein, with 73% and 50% of mutation positive CCCs and ECs showing loss of BAF250a, but only 11% and 9% of mutation negative cases showing loss, respectively. This may be due to immunohistochemical detection of truncated but non-functional BAF250a protein. The antibody used targets a 111 amino acid region in the middle of the protein (amino acids 1216-1326) and 7 of the 15 IHC positive cases had mutations that would result in truncations C-terminal to this epitope. Work is ongoing to try and develop antibodies that detect the C- terminal portion of the BAF250a protein.  For our studies to date, we have scored BAF250a IHC using a binary positive/negative system, and further refinement of this may lead to better  137 correlations with ARID1A mutant cases that are still expressing some level of BAF250a protein.  In Chapter 2, by comparing the mutations present in CCCs to their contiguous atypical endometriotic lesions, we demonstrated that the same mutations are present in the putative precursor lesions as the tumours. In contrast, the distant endometriotic lesions did not have ARID1A mutations. In one specific example, the mutation (G6139T (E2047*)) was shown to be present before the atypical endometriosis has developed the immunophenoptype associated with the cancer (ER negative, HNF-1β positive (20)) suggesting that the mutation of ARID1A is a very early event in neoplastic transformation. In a continuation of this work, in Chapter 3 we studied BAF250a expression in complex atypical hyperplasia of the endometrium (considered to be the precursor of endometrial carcinoma), and additional cases of atypical endometriosis, with the hope of gaining further insight into the timing of ARID1A mutations during tumour development. In this small follow-up study we did not identify BAF250a loss in any of the nine cases of atypical endometrial hyperplasia. One of the ten cases of atypical endometriosis had loss of BAF250a expression. That particular patient returned two years later with an endometrioid carcinoma at the location of the atypical endometriosis. Stemming from this work, a large initiative is now underway to collect samples of endometriosis to screen for mutations in ARID1A.  The study in Chapter 3 reported on the large-scale interrogation of over 3000 cases of other malignancies for ARID1A/BAF250a loss by IHC.   Overall, loss of BAF250a expression measured by IHC was a less common event in non-gynecological malignancies, with loss of  138 BAF250a in greater than 10% of cases of a given tumour type only seen in gastric cancer (14%) and anaplastic thyroid carcinoma (14%).   Otherwise, cancers of endometrial origin showed the highest frequency of BAF250a loss, with 29% of Grade 1 or 2 endometrioid, 39% of Grade 3 endometrioid, 26% of clear cell, and 18% of high-grade serous cancers of the endometrium showing BAF250a expression loss, while 14% of uterine carcinosarcomas showed BAF250a loss.  Following the publication of this study, other reports investigating the loss of ARID1A in gastric cancer reported ARID1A in to be mutated or lost in between 8- 27% of gastric cancer (100-102), with one study reporting loss of 51% (103). In addition, a controversy over the prognostic significance of ARID1A loss in gastric cancer emerged, with one report showing ARID1A mutation to be associated with better prognosis (101), and two reports of ARID1A being a negative prognostic marker (102, 103).  We conducted a follow- up study on an additional 262 cases of gastric carcinomas from Vancouver and Toronto, presented in the second part of Chapter 3.  We observed BAF250a loss in one cohort (Vancouver) at 22.5% and the other (Toronto) at 19.1%, for an overall frequency of BAF250a loss of 21% in the combined set of 262 cases.  This was higher than we initially would have suspected from our preliminary screening of diverse malignancies for ARID1A deficiency, which only reported 14% loss in gastric cancer for that small series.  To clarify whether ARID1A is indeed of prognostic and potentially therapeutic significance in gastric cancer, in the analysis of the two separate cohorts from Vancouver and Toronto, BAF250a expression was significantly associated with poor overall survival in one cohort (Toronto) whereas no significant correlation with overall survival was observed in the Vancouver group.  This finding may have indicated an underlying selection bias between the two cohorts, a possible explanation for which could be the fact that the Toronto cohort contained  139 an increased frequency of earlier stage disease. Our findings confirm that ARID1A appears to have an important prognostic value within gastric cancer, particularly in patients with lower stage.  The study presented in Chapter 4 aimed to identify functional consequences of ARID1A loss. The proteomic patterns of expression of the three major ovarian cancer subtypes (HGSC, CCC and EC) were examined using the reverse phase protein array (RPPA) platform. Whole tumour lysates from 127 ovarian carcinomas, including 34 CCC, 28 EC and 65 HGSC were profiled by RPPA, with the ARID1A mutation and IHC status previously defined for 96 of the ovarian cancers in the RPPA series. The RPPA proteomic assessment of CCCs and ECs identified a number of pathways that may be important as potential targets. This study provided evidence that BAF250a expression identifies a subgroup of CCC and ECs with higher pAKT-Thr308 phosphorylation that could contribute to changes in treatment responsiveness or provide a novel approach to target tumours with ARID1A/BAF250a aberrations.  Tumours lacking BAF250a expression showed higher levels of pAKT-Thr308 independent of PIK3CA mutation and PTEN loss suggesting a novel mechanism for activation of the PI3K/AKT pathway.  Similar findings were published recently in a large cohort of endometrial cancers (122). Unlike the findings in endometrial cancers, we were unable to convincingly demonstrate the mechanism by which this occurs in cell lines. It is possible the ovarian cancer cell line models utilized in our study may not accurately reflect the signaling changes observed in the tumour samples.  One possible explanation for this could be that the signaling changes observed in the tumours are effected by tumour/stromal interactions; however further work will be required to evaluate this theory.  140 Chapter 5 addressed the challenge of translating the basic finding of mutations in ARID1A, into the development of new therapeutic approaches for clear cell carcinoma. At 8,577 bp, ARID1A is a huge gene, and as such presented a rather formidable challenge to work with in the lab for the creation of appropriate models to carry into these functional studies. The closest cell lines resembling isogenic models were the RMG1 and HCT116 cell lines with lentiviral knockdown of approximately 70% of the BAF250a protein.  From the initial screen of this set, a few candidate compounds looked to have potential synthetic lethality with ARID1A, however these did not pass secondary validation.  In lieu of having the perfect isogenic set, nineteen ovarian cancer cell lines were taken through the screening of 340 kinase inhibitors.  Ten of these lines were, according to published literature, clear cell carcinoma derived cell lines.  These ten lines were assessed for mutation and protein status of ARID1A/BAF250a. Analysis of the screening data did not reveal compounds that appeared to be specific for ARID1A mutation status, possibly because the cell lines without detected ARID1A/BAF250a deficiencies could harbor some as of yet unidentified but equivalent genetic alteration.  For this reason, compounds that were effective at inhibiting growth in more than half of the clear cell lines without perturbing the growth of normal mouse embryonic fibroblasts were chosen for further studies. These included GSK461364 (an inhibitor of PLK1 kinase), plus Pelitinib and Tovok (both inhibitors of EGFR).  6.2 Future directions Although the development of model systems to study ARID1A proved to be rather challenging, the system developed to reintroduce ARID1A back into the cells will hopefully provide a good model for further analysis of the functional significance of ARID1A  141 mutations. As the data from sequencing ARID1A in different malignancies becomes available, the most frequently mutated regions of the gene (i.e. hotspot mutation regions) should become apparent; in particular, pCDNA-TM6-2-N-EMGFP-DEST vector system will potentially allow for the identification of the functional relevance of the common missense mutations in ARID1A- if any given mutation is well tolerated by the cell, it would likely be pathogenic, as re-expression of full length functional ARID1A is not well tolerated by the cells.  A possible direction of future work on this project is to address the question of what is altered in clear cell carcinoma cell lines or tumours without ARID1A mutations.  Perhaps mutations will be discovered in other members of the SWI/SNF family, or other chromatin remodeling genes, however in our data sets such mutations have not been identified to date. It is possible we have not identified equivalent mutations because we have focused our initial sequencing efforts on RNA-seq rather than the sequencing of the full genome; mutated genes may not generate transcripts, or these altered transcripts could be rapidly degraded, which would mean they would not be detected with RNA-seq. It is also possible that an equivalent functional change to an ARID1A mutation could occur in another as of yet unexplored area- perhaps in the epigenomic space (such as the methylome) or non-coding genomic space.  Another potential direction of study is the investigation of other genes responsible for the transformation of endometriosis-associated carcinomas, and the determination of which genetic alterations occur in “normal” endometriosis compared to endometrium. Since ARID1A mutations have been detected in other cancer types, it will be of interest to  142 determine whether or not mutations or loss of ARID1A will be associated with the known (or suspected) precursor lesions of other diseases.  With the evolution of the new sequencing technologies, studies of this nature are becoming more cost effective, and less genomic material is being required, which will be helpful in the study of precursor lesions which typically yield and extremely limited quantity of genomic material for study.  We still know extremely little about the role of ARID1A in cancer development. Another logical line of investigation would be to determine if the loss of ARID1A in different malignancies plays similar or different roles in each type of disease.  It is still not known whether ARID1A is a common initiation event, which then remains as a “driving” event during neoplastic progression, or whether ARID1A is an initiating event which after starting the neoplastic process then turns into more of a back seat driver. It is also not entirely known which regions of the genome ARID1A binds to; it should be possible to determine this utilizing ChIP-seq technology.  In Chapter 2, we started to address the relationship between ARID1A/BAF250a, P53, and MSI in gastric cancer- further studies in this are may help to resolve what the actual relationship is between these genetic aberrations, and whether or not certain mutation spectrums or hot-spot regions can be defined in the cancers with MSI.  Endometrial carcinomas and clear cell carcinomas are also known to be associated with MSI, and it remains to be seen whether potential hot-spot regions of ARID1A mutation in one MSI associated cancer type are also present in other types of cancers with MSI association.   143 Another intriguing area of exploration arising from this work is the interplay of the signaling mechanisms between the tumour and the stroma.  From Chapter 4, we saw that the cell line models did not seem to replicate what happened in the tumour- it is possible and rather likely that there are levels of interaction between tumour/stroma that have not been explored yet. A logical extension of this work would be to establish co-culture systems that might more closely resemble the biology observed in the tumour samples.  Along these lines, it would also be worthwhile investigating whether the re-introduction of ARID1A as detailed in Chapter 5, which appeared to halt cellular proliferation, could be rescued by stromal-derived (or other) factors introduced either through co-culture with fibroblasts, by culturing with fibroblast conditioned media.  It might also be possible to screen the lines with reintroduced ARID1A against a library of growth factors and compounds to see if compounds or factors which permit rescue could be identified.  The next stages of preclinical work for assessing the value of the compounds identified in the kinase inhibitor screen would likely involve in vivo evaluation.  A new preclinical model is being developed presently in-house, utilizing zebra fish.  The zebra fish can be microinjected with the clear cell carcinoma cell lines which have been labeled with a red nuclear stain (NucLight Red from Essen Scientific).  The compound of therapeutic interest can be added to the water to test for inhibition of cell line growth within the zebra fish.  This model has several advantages.  Zebra fish are an extremely cost effective model system compared to working with rodents (they require less space, less food, and less maintenance).  Another positive feature of the zebra fish is they can be rendered transparent, so it is easy to visualize processes occurring within the fish.  An additional advantage of the zebra fish is that they  144 have a short gestation period, and produce embryos in abundance.  However, zebra fish are evolutionally further away from humans as a model system than rodents, so eventually any compounds of interest would still require testing in another pre-clinical model system. Mouse xenograft models for CCC should be available locally in the near future, so any compounds that remain promising after the zebra fish trials should advance into testing with the mouse xenografts. Ultimately, any compounds that remain promising after the preclinical model tests would hopefully advance into clinical trials in patients with clear cell carcinoma.  Another area of further investigation arising from the kinase inhibitor screen is to try and identify the underlying reasons for the sensitivity of the CCC lines to the PLK1 and EGFR inhibitors.  Sequencing the genomes or exomes of CCC cell lines may allow for the identification of the underlying mutations responsible for susceptibility to specific therapeutic compounds including the PLK1 and EGFR inhibitors identified in the screen. Identification of the molecular features which predict the potential response to a particular compound is essential for the development of personalized cancer therapeutics.  It is rather unfortunate that several of the cell lines classified by the immunophenotyping project do not match the historical literature classifications (Anglesio et al., manuscript in preparation).  This work was completed after the kinase inhibitor screening was completed. With this in mind, an important future direction for the screening work would be to reanalyze the data generated from the kinase inhibitor screen to see if the results for compounds targeting clear cell carcinoma lines could be refined.   145 Another possible follow-up to the kinase inhibitor screen (Chapter 5) and the RPPA project (Chapter 4) could be to compare the two data sets to see whether any of the up-regulated or ARID1A-specific candidates are implicated as potential hits in the compound screen.  Since the kinase inhibitor screen has several limitations, as discussed in Chapter 5, it could also be worthwhile to evaluate therapeutic compounds against the most prominent candidates that emerged from the RPPA work; even though the cell lines have limitations this still could prove to be worthwhile.  For the generation of better isogenic models there are other strategies that could be employed, which might be more appropriate given the size of the ARID1A gene.   One possibility would be targeted knock-out by zinc-finger nucleases, however these are prohibitively expensive for the average lab, time consuming, and present different technical challenges.  If true isogenic model systems could be established, it may be worthwhile undertaking a full genomic siRNA screen to identify candidate genes targets with true synthetic-lethality with ARID1A defects.  With true isogenic cell lines in hand, it might also be useful to undertake larger screens of therapeutic compounds to try and identify novel candidates with synthetic lethal activity for ARID1A defective tumours.  Although much work remains to determine its functional and therapeutic significance, the loss of ARID1A in endometriotic epithelium appears to be of fundamental importance in malignant transformation in endometriosis associated ovarian cancers, as well as other diverse malignancies.  The studies presented in this dissertation have provided the first evidence that ARID1A is a novel epigenetic tumour suppressor in endometriosis associated  146 ovarian cancers. Intense worldwide investigation into the role of ARID1A in ovarian cancers and other types of malignancies is now well underway. It will be interesting to see if other abnormalities impacting the ARIDIA locus or dysregulation of other chromatin remodeling genes will be found in the ARID1A mutation negative CCCs and ECs.  This idea is supported by the clinical similarity between ARID1A mutation positive and mutation negative CCCs. It is possible that defects in other epigenetic genes, such as ARID1A, that alter the accessibility of transcription factors to chromatin, along with previously described WNT and PI3 kinase pathway mutations (26) will define CCCs and ECs, and work is ongoing in the Huntsman lab and labs across the world to determine if this is indeed the case.          147 Bibliography   1. Jemal A, Siegel R, Xu J, Ward E. Cancer Statistics, 2010. CA: A Cancer Journal for Clinicians. 2010(Journal Article). 2. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA Cancer J Clin. 2011 Mar-Apr;61(2):69-90. 3. Gilks CB, Prat J. Ovarian carcinoma pathology and genetics: recent advances. Human pathology. 2009;40(9):1213-23. 4. Kurman RJ, Shih Ie M. Pathogenesis of ovarian cancer: lessons from morphology and molecular biology and their clinical implications. Int J Gynecol Pathol. 2008 Apr;27(2):151- 60. 5. LaGrenade A, Silverberg SG. Ovarian tumors associated with atypical endometriosis. Human pathology. 1988;19(9):1080-4. 6. Moll UM, Chumas JC, Chalas E, Mann WJ. Ovarian carcinoma arising in atypical endometriosis. Obstetrics and gynecology. 1990;75(3 Pt 2):537-9. 7. Prat J. Ovarian carcinomas: five distinct diseases with different origins, genetic alterations, and clinicopathological features. Virchows Arch. 2012 Mar;460(3):237-49. 8. McAlpine JN, Wiegand KC, Vang R, Ronnett BM, Adamiak A, Kobel M, et al. HER2 overexpression and amplification is present in a subset of ovarian mucinous carcinomas and can be targeted with trastuzumab therapy. BMC Cancer. 2009;9:433. 9. Anglesio MS, Kommoss S, Tolcher MC, Clarke B, Galletta L, Porter H, et al. Molecular characterization of mucinous ovarian tumours supports a stratified treatment approach with HER2 targeting in 19% of carcinomas. J Pathol. 2013 Jan;229(1):111-20. 10. Kobel M, Kalloger SE, Huntsman DG, Santos JL, Swenerton KD, Seidman JD, et al. Differences in tumor type in low-stage versus high-stage ovarian carcinomas. International journal of gynecological pathology : official journal of the International Society of Gynecological Pathologists. 2010;29(3):203-11. 11. Integrated genomic analyses of ovarian carcinoma. Nature. 2011 Jun 30;474(7353):609-15.  148 12. Willner J, Wurz K, Allison KH, Galic V, Garcia RL, Goff BA, et al. Alternate molecular genetic pathways in ovarian carcinomas of common histological types. Hum Pathol. 2007 Apr;38(4):607-13. 13. Crum CP, Drapkin R, Miron A, Ince TA, Muto M, Kindelberger DW, et al. The distal fallopian tube: a new model for pelvic serous carcinogenesis. Curr Opin Obstet Gynecol. 2007 Feb;19(1):3-9. 14. Auersperg N. The origin of ovarian cancers--hypotheses and controversies. Front Biosci (Schol Ed). 2013;5:709-19. 15. Sugiyama T, Kamura T, Kigawa J, Terakawa N, Kikuchi Y, Kita T, et al. Clinical characteristics of clear cell carcinoma of the ovary: a distinct histologic type with poor prognosis and resistance to platinum-based chemotherapy. Cancer. 2000;88(11):2584-9. 16. Ushijima K. Current status of gynecologic cancer in Japan. J Gynecol Oncol. 2009 Jun;20(2):67-71. 17. Tavassoli FA, Devilee P, International Agency for Research on C, World Health O. Pathology and genetics of tumours of the breast and female genital organs. Lyon: IARC Press; 2003. 18. Kato N, Sasou S, Motoyama T. Expression of hepatocyte nuclear factor-1beta (HNF- 1beta) in clear cell tumors and endometriosis of the ovary. Mod Pathol. 2006 Jan;19(1):83-9. 19. Kobel M, Kalloger SE, Boyd N, McKinney S, Mehl E, Palmer C, et al. Ovarian carcinoma subtypes are different diseases: implications for biomarker studies. PLoS medicine. 2008;5(12):e232. 20. Kobel M, Kalloger SE, Carrick J, Huntsman D, Asad H, Oliva E, et al. A limited panel of immunomarkers can reliably distinguish between clear cell and high-grade serous carcinoma of the ovary. The American Journal of Surgical Pathology. 2009;33(1):14-21. 21. Press JZ, De Luca A, Boyd N, Young S, Troussard A, Ridge Y, et al. Ovarian carcinomas with genetic and epigenetic BRCA1 loss have distinct molecular abnormalities. BMC cancer. 2008;8(Journal Article):17. 22. Gilks CB. Molecular abnormalities in ovarian cancer subtypes other than high-grade serous carcinoma. Journal of oncology. 2010;2010(Journal Article):740968. 23. Campbell IG, Russell SE, Choong DY, Montgomery KG, Ciavarella ML, Hooi CS, et al. Mutation of the PIK3CA gene in ovarian and breast cancer. Cancer Res. 2004 Nov 1;64(21):7678-81.  149 24. Wang Y, Helland A, Holm R, Kristensen GB, Borresen-Dale AL. PIK3CA mutations in advanced ovarian carcinomas. Hum Mutat. 2005 Mar;25(3):322. 25. Kolasa IK, Rembiszewska A, Felisiak A, Ziolkowska-Seta I, Murawska M, Moes J, et al. PIK3CA amplification associates with resistance to chemotherapy in ovarian cancer patients. Cancer Biol Ther. 2009 Jan;8(1):21-6. 26. Kuo KT, Mao TL, Jones S, Veras E, Ayhan A, Wang TL, et al. Frequent activating mutations of PIK3CA in ovarian clear cell carcinoma. Am J Pathol. 2009 May;174(5):1597- 601. 27. Oliva E, Sarrio D, Brachtel EF, Sanchez-Estevez C, Soslow RA, Moreno-Bueno G, et al. High frequency of beta-catenin mutations in borderline endometrioid tumours of the ovary. J Pathol. 2006 Apr;208(5):708-13. 28. Cho KR, Shih Ie M. Ovarian cancer. Annu Rev Pathol. 2009;4:287-313. 29. Leitao MM, Jr., Boyd J, Hummer A, Olvera N, Arroyo CD, Venkatraman E, et al. Clinicopathologic analysis of early-stage sporadic ovarian carcinoma. The American Journal of Surgical Pathology. 2004;28(2):147-59. 30. Gilks CB, Ionescu DN, Kalloger SE, Kobel M, Irving J, Clarke B, et al. Tumor cell type can be reproducibly diagnosed and is of independent prognostic significance in patients with maximally debulked ovarian carcinoma. Human pathology. 2008;39(8):1239-51. 31. Pectasides D, Pectasides E, Psyrri A, Economopoulos T. Treatment issues in clear cell carcinoma of the ovary: a different entity? The oncologist. 2006;11(10):1089-94. 32. Goff BA, Sainz de la Cuesta R, Muntz HG, Fleischhacker D, Ek M, Rice LW, et al. Clear cell carcinoma of the ovary: a distinct histologic type with poor prognosis and resistance to platinum-based chemotherapy in stage III disease. Gynecologic oncology. 1996;60(3):412-7. 33. Crotzer DR, Sun CC, Coleman RL, Wolf JK, Levenback CF, Gershenson DM. Lack of effective systemic therapy for recurrent clear cell carcinoma of the ovary. Gynecologic oncology. 2007;105(2):404-8. 34. Fukunaga M, Nomura K, Ishikawa E, Ushigome S. Ovarian atypical endometriosis: its close association with malignant epithelial tumours. Histopathology. 1997;30(3):249-55. 35. Ogawa S, Kaku T, Amada S, Kobayashi H, Hirakawa T, Ariyoshi K, et al. Ovarian endometriosis associated with ovarian carcinoma: a clinicopathological and immunohistochemical study. Gynecologic oncology. 2000;77(2):298-304.  150 36. Varma R, Rollason T, Gupta JK, Maher ER. Endometriosis and the neoplastic process. Reproduction (Cambridge, England). 2004;127(3):293-304. 37. Giudice LC. Clinical practice. Endometriosis. The New England journal of medicine. 2010;362(25):2389-98. 38. Lapp T. ACOG issues recommendations for the management of endometriosis. American College of Obstetricians and Gynecologists. American Family Physician. 2000;62(6):1431, 4. 39. Bulun SE. Endometriosis. The New England journal of medicine. 2009;360(3):268- 79. 40. Holoch KJ, Lessey BA. Endometriosis and infertility. Clinical obstetrics and gynecology. 2010;53(2):429-38. 41. Sampson JA. Endometrial carcinoma of the ovary, arising in endometrial tissue in that organ. Archives of Surgery. 1925;10(1):1-72. 42. Osuga Y. Current concepts of the pathogenesis of endometriosis. Reprod Med Biol. 2010;9(1):1-7. 43. Sampson JA. Metastatic or Embolic Endometriosis, due to the Menstrual Dissemination of Endometrial Tissue into the Venous Circulation. The American journal of pathology. 1927;3(2):93-110.43. 44. Kumar V, Abbas AK, Fausto N, Robbins SL, Cotran RS. Robbins and Cotran pathologic basis of disease. Philadelphia: Elsevier Saunders; 2005. 45. Sato N, Tsunoda H, Nishida M, Morishita Y, Takimoto Y, Kubo T, et al. Loss of heterozygosity on 10q23.3 and mutation of the tumor suppressor gene PTEN in benign endometrial cyst of the ovary: possible sequence progression from benign endometrial cyst to endometrioid carcinoma and clear cell carcinoma of the ovary. Cancer research. 2000;60(24):7052-6. 46. Dinulescu DM, Ince TA, Quade BJ, Shafer SA, Crowley D, Jacks T. Role of K-ras and Pten in the development of mouse models of endometriosis and endometrioid ovarian cancer. Nature medicine. 2005;11(1):63-70. 47. Prowse AH, Manek S, Varma R, Liu J, Godwin AK, Maher ER, et al. Molecular genetic evidence that endometriosis is a precursor of ovarian cancer. International journal of cancerJournal international du cancer. 2006;119(3):556-62.  151 48. Roberts CW, Orkin SH. The SWI/SNF complex--chromatin and cancer. Nat Rev Cancer. 2004 Feb;4(2):133-42. 49. Reisman D, Glaros S, Thompson EA. The SWI/SNF complex and cancer. Oncogene. 2009;28(14):1653-68. 50. Weissman B, Knudsen KE. Hijacking the chromatin remodeling machinery: impact of SWI/SNF perturbations in cancer. Cancer Res. 2009 Nov 1;69(21):8223-30. 51. Versteege I, Sevenet N, Lange J, Rousseau-Merck MF, Ambros P, Handgretinger R, et al. Truncating mutations of hSNF5/INI1 in aggressive paediatric cancer. Nature. 1998 Jul 9;394(6689):203-6. 52. Fukuoka J, Fujii T, Shih JH, Dracheva T, Meerzaman D, Player A, et al. Chromatin remodeling factors and BRM/BRG1 expression as prognostic indicators in non-small cell lung cancer. Clin Cancer Res. 2004 Jul 1;10(13):4314-24. 53. Lee D, Kim JW, Seo T, Hwang SG, Choi EJ, Choe J. SWI/SNF complex interacts with tumor suppressor p53 and is necessary for the activation of p53-mediated transcription. J Biol Chem. 2002 Jun 21;277(25):22330-7. 54. Bochar DA, Wang L, Beniya H, Kinev A, Xue Y, Lane WS, et al. BRCA1 is associated with a human SWI/SNF-related complex: linking chromatin remodeling to breast cancer. Cell. 2000 Jul 21;102(2):257-65. 55. Trouche D, Le Chalony C, Muchardt C, Yaniv M, Kouzarides T. RB and hbrm cooperate to repress the activation functions of E2F1. Proc Natl Acad Sci U S A. 1997 Oct 14;94(21):11268-73. 56. Wang W, Xue Y, Zhou S, Kuo A, Cairns BR, Crabtree GR. Diversity and specialization of mammalian SWI/SNF complexes. Genes & development. 1996;10(17):2117-30. 57. Sif S, Saurin AJ, Imbalzano AN, Kingston RE. Purification and characterization of mSin3A-containing Brg1 and hBrm chromatin remodeling complexes. Genes & development. 2001;15(5):603-18. 58. Peterson CL, Tamkun JW. The SWI-SNF complex: a chromatin remodeling machine? Trends Biochem Sci. 1995 Apr;20(4):143-6. 59. Kingston RE, Bunker CA, Imbalzano AN. Repression and activation by multiprotein complexes that alter chromatin structure. Genes Dev. 1996 Apr 15;10(8):905-20.  152 60. Dallas PB, Cheney IW, Liao DW, Bowrin V, Byam W, Pacchione S, et al. p300/CREB binding protein-related protein p270 is a component of mammalian SWI/SNF complexes. Mol Cell Biol. 1998 Jun;18(6):3596-603. 61. Dallas PB, Pacchione S, Wilsker D, Bowrin V, Kobayashi R, Moran E. The human SWI-SNF complex protein p270 is an ARID family member with non-sequence-specific DNA binding activity. Mol Cell Biol. 2000 May;20(9):3137-46. 62. Nie Z, Xue Y, Yang D, Zhou S, Deroo BJ, Archer TK, et al. A specificity and targeting subunit of a human SWI/SNF family-related chromatin-remodeling complex. Mol Cell Biol. 2000 Dec;20(23):8879-88. 63. Van Rechem C, Boulay G, Leprince D. HIC1 interacts with a specific subunit of SWI/SNF complexes, ARID1A/BAF250A. Biochem Biophys Res Commun. 2009 Aug 7;385(4):586-90. 64. Fleuriel C, Touka M, Boulay G, Guerardel C, Rood BR, Leprince D. HIC1 (Hypermethylated in Cancer 1) epigenetic silencing in tumors. Int J Biochem Cell Biol. 2009 Jan;41(1):26-33. 65. Wang X, Nagl NG, Jr., Flowers S, Zweitzig D, Dallas PB, Moran E. Expression of p270 (ARID1A), a component of human SWI/SNF complexes, in human tumors. International journal of cancerJournal international du cancer. 2004;112(4):636. 66. Inoue H, Furukawa T, Giannakopoulos S, Zhou S, King DS, Tanese N. Largest subunits of the human SWI/SNF chromatin-remodeling complex promote transcriptional activation by steroid hormone receptors. J Biol Chem. 2002 Nov 1;277(44):41674-85. 67. Trotter KW, Fan HY, Ivey ML, Kingston RE, Archer TK. The HSA domain of BRG1 mediates critical interactions required for glucocorticoid receptor-dependent transcriptional activation in vivo. Mol Cell Biol. 2008 Feb;28(4):1413-26. 68. Mitelman F, Mertens F, Johansson B. A breakpoint map of recurrent chromosomal rearrangements in human neoplasia. Nature genetics. 1997;15 Spec No(Journal Article):417- 74. 69. Mertens F, Johansson B, Hoglund M, Mitelman F. Chromosomal imbalance maps of malignant solid tumors: a cytogenetic survey of 3185 neoplasms. Cancer research. 1997;57(13):2765-80. 70. Huang J, Zhao YL, Li Y, Fletcher JA, Xiao S. Genomic and functional evidence for an ARID1A tumor suppressor role. Genes, chromosomes & cancer. 2007;46(8):745-50.  153 71. Decristofaro MF, Betz BL, Rorie CJ, Reisman DN, Wang W, Weissman BE. Characterization of SWI/SNF protein expression in human breast cancer cell lines and other malignancies. Journal of cellular physiology. 2001;186(1):136-45. 72. Nagl NG, Jr., Wang X, Patsialou A, Van Scoy M, Moran E. Distinct mammalian SWI/SNF chromatin remodeling complexes with opposing roles in cell-cycle control. The EMBO journal. 2007;26(3):752-63. 73. Nagl NG, Jr., Patsialou A, Haines DS, Dallas PB, Beck GR, Jr., Moran E. The p270 (ARID1A/SMARCF1) subunit of mammalian SWI/SNF-related complexes is essential for normal cell cycle arrest. Cancer Res. 2005 Oct 15;65(20):9236-44. 74. Nagl NG, Jr., Zweitzig DR, Thimmapaya B, Beck GR, Jr., Moran E. The c-myc gene is a direct target of mammalian SWI/SNF-related complexes during differentiation- associated cell cycle arrest. Cancer Res. 2006 Feb 1;66(3):1289-93. 75. Boyer LA, Lee TI, Cole MF, Johnstone SE, Levine SS, Zucker JP, et al. Core transcriptional regulatory circuitry in human embryonic stem cells. Cell. 2005 Sep 23;122(6):947-56. 76. Gao X, Tate P, Hu P, Tjian R, Skarnes WC, Wang Z. ES cell pluripotency and germ- layer formation require the SWI/SNF chromatin remodeling component BAF250a. Proc Natl Acad Sci U S A. 2008 May 6;105(18):6656-61. 77. Shah SP, Kobel M, Senz J, Morin RD, Clarke BA, Wiegand KC, et al. Mutation of FOXL2 in granulosa-cell tumors of the ovary. The New England journal of medicine. 2009;360(26):2719-29. 78. Itamochi H, Kigawa J, Terakawa N. Mechanisms of chemoresistance and poor prognosis in ovarian clear cell carcinoma. Cancer science. 2008;99(4):653-8. 79. Dent J, Hall GD, Wilkinson N, Perren TJ, Richmond I, Markham AF, et al. Cytogenetic alterations in ovarian clear cell carcinoma detected by comparative genomic hybridisation. Br J Cancer. 2003 May 19;88(10):1578-83. 80. Suehiro Y, Sakamoto M, Umayahara K, Iwabuchi H, Sakamoto H, Tanaka N, et al. Genetic aberrations detected by comparative genomic hybridization in ovarian clear cell adenocarcinomas. Oncology. 2000 Jun;59(1):50-6. 81. Ness RB. Endometriosis and ovarian cancer: thoughts on shared pathophysiology. American Journal of Obstetrics and Gynecology. 2003;189(1):280-94.  154 82. Vigano P, Somigliana E, Chiodo I, Abbiati A, Vercellini P. Molecular mechanisms and biological plausibility underlying the malignant transformation of endometriosis: a critical analysis. Human reproduction update. 2006;12(1):77-89. 83. Wiegand KC, Shah SP, Al-Agha OM, Zhao Y, Tse K, Zeng T, et al. ARID1A mutations in endometriosis-associated ovarian carcinomas. The New England journal of medicine. 2010;363(16):1532-43. 84. Jones S, Wang TL, Shih I, Mao TL, Nakayama K, Roden R, et al. Frequent mutations of chromatin remodeling gene ARID1A in ovarian clear cell carcinoma. Science (New York, NY). 2010;330(6001):228-31. 85. Provencher DM, Lounis H, Champoux L, Tetrault M, Manderson EN, Wang JC, et al. Characterization of four novel epithelial ovarian cancer cell lines. In Vitro Cell Dev Biol Anim. 2000 Jun;36(6):357-61. 86. Lau DH, Lewis AD, Ehsan MN, Sikic BI. Multifactorial mechanisms associated with broad cross-resistance of ovarian carcinoma cells selected by cyanomorpholino doxorubicin. Cancer Res. 1991 Oct 1;51(19):5181-7. 87. Morin R, Bainbridge M, Fejes A, Hirst M, Krzywinski M, Pugh T, et al. Profiling the HeLa S3 transcriptome using randomly primed cDNA and massively parallel short-read sequencing. BioTechniques. 2008;45(1):81-94. 88. Li H, Ruan J, Durbin R. Mapping short DNA sequencing reads and calling variants using mapping quality scores. Genome research. 2008;18(11):1851-8. 89. Goya R, Sun MG, Morin RD, Leung G, Ha G, Wiegand KC, et al. SNVMix: predicting single nucleotide variants from next-generation sequencing of tumors. Bioinformatics (Oxford, England). 2010;26(6):730-6. 90. Rozen S, Skaletsky H. Primer3 on the WWW for general users and for biologist programmers. Methods Mol Biol. 2000;132:365-86. 91. Hochberg Y, Benjamini Y. More powerful procedures for multiple significance testing. Stat Med. 1990 Jul;9(7):811-8. 92. McPherson A, Hormozdiari F, Zayed A, Giuliany R, Ha G, Sun MG, et al. deFuse: an algorithm for gene fusion discovery in tumor RNA-Seq data. PLoS Comput Biol. 2011 May;7(5):e1001138.  155 93. Bengtsson H, Ray A, Spellman P, Speed TP. A single-sample method for normalizing and combining full-resolution copy numbers from multiple platforms, labs and analysis methods. Bioinformatics. 2009 Apr 1;25(7):861-7. 94. Conrad DF, Pinto D, Redon R, Feuk L, Gokcumen O, Zhang Y, et al. Origins and functional impact of copy number variation in the human genome. Nature. 2010 Apr 1;464(7289):704-12. 95. Chang YF, Imam JS, Wilkinson MF. The nonsense-mediated decay RNA surveillance pathway. Annual Review of Biochemistry. 2007;76(Journal Article):51-74. 96. Kozmik Z, Machon O, Kralova J, Kreslova J, Paces J, Vlcek C. Characterization of mammalian orthologues of the Drosophila osa gene: cDNA cloning, expression, chromosomal localization, and direct physical interaction with Brahma chromatin- remodeling complex. Genomics. 2001;73(2):140-8. 97. Ahmed AA, Etemadmoghadam D, Temple J, Lynch AG, Riad M, Sharma R, et al. Driver mutations in TP53 are ubiquitous in high grade serous carcinoma of the ovary. J Pathol. 2010 May;221(1):49-56. 98. Wiegand KC, Lee AF, Al-Agha OM, Chow C, Kalloger SE, Scott DW, et al. Loss of BAF250a (ARID1A) is frequent in high-grade endometrial carcinomas. J Pathol. 2011 Jul;224(3):328-33. 99. Jones S, Li M, Parsons DW, Zhang X, Wesseling J, Kristel P, et al. Somatic mutations in the chromatin remodeling gene ARID1A occur in several tumor types. Hum Mutat. 2012 Jan;33(1):100-3. 100. Zang ZJ, Cutcutache I, Poon SL, Zhang SL, McPherson JR, Tao J, et al. Exome sequencing of gastric adenocarcinoma identifies recurrent somatic mutations in cell adhesion and chromatin remodeling genes. Nat Genet. 2012 May;44(5):570-4. 101. Wang K, Kan J, Yuen ST, Shi ST, Chu KM, Law S, et al. Exome sequencing identifies frequent mutation of ARID1A in molecular subtypes of gastric cancer. Nat Genet. 2011 Dec;43(12):1219-23. 102. Abe H, Maeda D, Hino R, Otake Y, Isogai M, Ushiku AS, et al. ARID1A expression loss in gastric cancer: pathway-dependent roles with and without Epstein-Barr virus infection and microsatellite instability. Virchows Arch. 2012 Oct;461(4):367-77. 103. Wang DD, Chen YB, Pan K, Wang W, Chen SP, Chen JG, et al. Decreased expression of the ARID1A gene is associated with poor prognosis in primary gastric cancer. PLoS One. 2012;7(7):e40364.  156 104. Alkushi A, Clarke BA, Akbari M, Makretsov N, Lim P, Miller D, et al. Identification of prognostically relevant and reproducible subsets of endometrial adenocarcinoma based on clustering analysis of immunostaining data. Mod Pathol. 2007 Nov;20(11):1156-65. 105. Edge SB, American Joint Committee on Cancer. AJCC cancer staging manual. 7th ed. New York: Springer; 2010. 106. Ferlay J, Shin HR, Bray F, Forman D, Mathers C, Parkin DM. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int J Cancer. 2010 Dec 15;127(12):2893-917. 107. Jorgensen JT, Hersom M. HER2 as a Prognostic Marker in Gastric Cancer - A Systematic Analysis of Data from the Literature. J Cancer. 2012;3:137-44. 108. Arai T, Sakurai U, Sawabe M, Honma N, Aida J, Ushio Y, et al. Frequent microsatellite instability in papillary and solid-type, poorly differentiated adenocarcinomas of the stomach. Gastric Cancer. 2012 Dec 29. 109. Lee HS, Choi SI, Lee HK, Kim HS, Yang HK, Kang GH, et al. Distinct clinical features and outcomes of gastric cancers with microsatellite instability. Mod Pathol. 2002 Jun;15(6):632-40. 110. Iizasa H, Nanbo A, Nishikawa J, Jinushi M, Yoshiyama H. Epstein-Barr Virus (EBV)-associated Gastric Carcinoma. Viruses. 2012 Dec;4(12):3420-39. 111. Gonzalez-Angulo AM, Ferrer-Lozano J, Stemke-Hale K, Sahin A, Liu S, Barrera JA, et al. PI3K pathway mutations and PTEN levels in primary and metastatic breast cancer. Mol Cancer Ther. 2011 Jun;10(6):1093-101. 112. McConechy MK, Ding J, Cheang MC, Wiegand KC, Senz J, Tone AA, et al. Use of mutation profiles to refine the classification of endometrial carcinomas. J Pathol. 2012 Sep;228(1):20-30. 113. Agarwal R, Carey M, Hennessy B, Mills GB. PI3K pathway-directed therapeutic strategies in cancer. Curr Opin Investig Drugs. 2010 Jun;11(6):615-28. 114. Hennessy BT, Smith DL, Ram PT, Lu Y, Mills GB. Exploiting the PI3K/AKT pathway for cancer drug discovery. Nat Rev Drug Discov. 2005 Dec;4(12):988-1004. 115. Guan B, Wang TL, Shih Ie M. ARID1A, a factor that promotes formation of SWI/SNF-mediated chromatin remodeling, is a tumor suppressor in gynecologic cancers. Cancer Res. 2011 Nov 1;71(21):6718-27.  157 116. Lowery WJ, Schildkraut JM, Akushevich L, Bentley R, Marks JR, Huntsman D, et al. Loss of ARID1A-associated protein expression is a frequent event in clear cell and endometrioid ovarian cancers. Int J Gynecol Cancer. 2012 Jan;22(1):9-14. 117. Tibes R, Qiu Y, Lu Y, Hennessy B, Andreeff M, Mills GB, et al. Reverse phase protein array: validation of a novel proteomic technology and utility for analysis of primary leukemia specimens and hematopoietic stem cells. Mol Cancer Ther. 2006 Oct;5(10):2512- 21. 118. Stemke-Hale K, Gonzalez-Angulo AM, Lluch A, Neve RM, Kuo WL, Davies M, et al. An integrative genomic and proteomic analysis of PIK3CA, PTEN, and AKT mutations in breast cancer. Cancer Res. 2008 Aug 1;68(15):6084-91. 119. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A. 2001 Apr 24;98(9):5116-21. 120. Ho CM, Lin MC, Huang SH, Huang CJ, Lai HC, Chien TY, et al. PTEN promoter methylation and LOH of 10q22-23 locus in PTEN expression of ovarian clear cell adenocarcinomas. Gynecol Oncol. 2009 Feb;112(2):307-13. 121. Carey MS, Agarwal R, Gilks B, Swenerton K, Kalloger S, Santos J, et al. Functional proteomic analysis of advanced serous ovarian cancer using reverse phase protein array: TGF-beta pathway signaling indicates response to primary chemotherapy. Clin Cancer Res. 2010 May 15;16(10):2852-60. 122. Liang H, Cheung LW, Li J, Ju Z, Yu S, Stemke-Hale K, et al. Whole-exome sequencing combined with functional genomics reveals novel candidate driver cancer genes in endometrial cancer. Genome Res. 2012 Nov;22(11):2120-9. 123. Cheung LW, Hennessy BT, Li J, Yu S, Myers AP, Djordjevic B, et al. High frequency of PIK3R1 and PIK3R2 mutations in endometrial cancer elucidates a novel mechanism for regulation of PTEN protein stability. Cancer Discov. 2011 Jul;1(2):170-85. 124. Vasudevan KM, Barbie DA, Davies MA, Rabinovsky R, McNear CJ, Kim JJ, et al. AKT-independent signaling downstream of oncogenic PIK3CA mutations in human cancer. Cancer Cell. 2009 Jul 7;16(1):21-32. 125. Takakura S, Takano M, Takahashi F, Saito T, Aoki D, Inaba N, et al. Randomized phase II trial of paclitaxel plus carboplatin therapy versus irinotecan plus cisplatin therapy as first-line chemotherapy for clear cell adenocarcinoma of the ovary: a JGOG study. Int J Gynecol Cancer. 2010 Feb;20(2):240-7.  158 126. Cizkova M, Susini A, Vacher S, Cizeron-Clairac G, Andrieu C, Driouch K, et al. PIK3CA mutation impact on survival in breast cancer patients and in ERalpha, PR and ERBB2-based subgroups. Breast Cancer Res. 2012 Feb 13;14(1):R28. 127. Rahman M, Nakayama K, Rahman MT, Nakayama N, Ishikawa M, Katagiri A, et al. Clinicopathologic and biological analysis of PIK3CA mutation in ovarian clear cell carcinoma. Hum Pathol. 2012 Jun 15. 128. Annunziata CM, O'Shaughnessy J. Poly (ADP-ribose) polymerase as a novel therapeutic target in cancer. Clin Cancer Res. 2010 Sep 15;16(18):4517-26. 129. Welm BE, Dijkgraaf GJ, Bledau AS, Welm AL, Werb Z. Lentiviral transduction of mammary stem cells for analysis of gene function during development and cancer. Cell Stem Cell. 2008 Jan 10;2(1):90-102. 130. Thomas CE, Ehrhardt A, Kay MA. Progress and problems with the use of viral vectors for gene therapy. Nat Rev Genet. 2003 May;4(5):346-58. 131. Dissen GA, Lomniczi A, Neff TL, Hobbs TR, Kohama SG, Kroenke CD, et al. In vivo manipulation of gene expression in non-human primates using lentiviral vectors as delivery vehicles. Methods. 2009 Sep;49(1):70-7. 132. Liu X, Erikson RL. Polo-like kinase (Plk)1 depletion induces apoptosis in cancer cells. Proc Natl Acad Sci U S A. 2003 May 13;100(10):5789-94. 133. Strebhardt K. Multifaceted polo-like kinases: drug targets and antitargets for cancer therapy. Nat Rev Drug Discov. 2010 Aug;9(8):643-60. 134. Weichert W, Denkert C, Schmidt M, Gekeler V, Wolf G, Kobel M, et al. Polo-like kinase isoform expression is a prognostic factor in ovarian carcinoma. Br J Cancer. 2004 Feb 23;90(4):815-21. 135. Schmit TL, Ledesma MC, Ahmad N. Modulating polo-like kinase 1 as a means for cancer chemoprevention. Pharm Res. 2010 Jun;27(6):989-98. 136. Gilmartin AG, Bleam MR, Richter MC, Erskine SG, Kruger RG, Madden L, et al. Distinct concentration-dependent effects of the polo-like kinase 1-specific inhibitor GSK461364A, including differential effect on apoptosis. Cancer Res. 2009 Sep 1;69(17):6969-77. 137. Olmos D, Barker D, Sharma R, Brunetto AT, Yap TA, Taegtmeyer AB, et al. Phase I study of GSK461364, a specific and competitive Polo-like kinase 1 inhibitor, in patients with advanced solid malignancies. Clin Cancer Res. 2011 May 15;17(10):3420-30.  159 138. Liu-Sullivan N, Zhang J, Bakleh A, Marchica J, Li J, Siolas D, et al. Pooled shRNA screen for sensitizers to inhibition of the mitotic regulator polo-like kinase (PLK1). Oncotarget. 2011 Dec;2(12):1254-64. 139. Degenhardt Y, Greshock J, Laquerre S, Gilmartin AG, Jing J, Richter M, et al. Sensitivity of cancer cells to Plk1 inhibitor GSK461364A is associated with loss of p53 function and chromosome instability. Mol Cancer Ther. 2010 Jul;9(7):2079-89. 140. Idbaih A, Ducray F, Sierra Del Rio M, Hoang-Xuan K, Delattre JY. Therapeutic application of noncytotoxic molecular targeted therapy in gliomas: growth factor receptors and angiogenesis inhibitors. Oncologist. 2008 Sep;13(9):978-92. 141. Tanaka Y, Terai Y, Tanabe A, Sasaki H, Sekijima T, Fujiwara S, et al. Prognostic effect of epidermal growth factor receptor gene mutations and the aberrant phosphorylation of Akt and ERK in ovarian cancer. Cancer Biol Ther. 2011 Jan 1;11(1):50-7. 142. Kalloger SE, Kobel M, Leung S, Mehl E, Gao D, Marcon KM, et al. Calculator for ovarian carcinoma subtype prediction. Mod Pathol. 2011 Apr;24(4):512-21. 143. Hoskins PJ, Le N, Gilks B, Tinker A, Santos J, Wong F, et al. Low-stage ovarian clear cell carcinoma: population-based outcomes in British Columbia, Canada, with evidence for a survival benefit as a result of irradiation. J Clin Oncol. 2012 May 10;30(14):1656-62. 144. Wu JN, Roberts CW. ARID1A Mutations in Cancer: Another Epigenetic Tumor Suppressor? Cancer Discov. 2013 Jan;3(1):35-43.   ! 160! Appendix A:  Supplemental Material for Chapter 1 The Study of HER2 amplification in mucinous carcinoma                 The work presented in Appendix A has been published: McAlpine JN, Wiegand KC, Vang R, Ronnett BM, Adamiak A, Köbel M, Kalloger SE, Swenerton KD, Huntsman DG, Gilks CB, Miller DM.  HER2 overexpression and amplification is present in a subset of ovarian mucinous carcinomas and can be targeted with trastuzumab therapy.  BMC Cancer.  2009; 9:433. Copyright © 2009, Biomed Central. Reprinted with permission.     ! 161!   Abstract: Background: The response rate of ovarian mucinous carcinomas to paclitaxel/carboplatin is low, prompting interest in targeted molecular therapies. We investigated HER2 expression and amplification, and the potential for trastuzumab therapy in this histologic subtype of ovarian cancer. Methods: HER2 status was tested in 33 mucinous carcinomas and 16 mucinous borderline ovarian tumors (BOT)). Five cases with documented recurrence and with tissue from the recurrence available for testing were analyzed to determine whether HER2 amplification status changed over time. Three prospectively identified recurrent mucinous ovarian carcinomas were assessed for HER2 amplification and patients received trastuzumab therapy with conventional chemotherapy. Results: Amplification of HER2 was observed in 6/33 (18.2%) mucinous carcinomas and 3/16 (18.8%) BOT. HER2 amplification in primary mucinous carcinomas was not associated with an increased likelihood of recurrence. The prospectively identified recurrent mucinous carcinomas showed overexpression and amplification of HER2; one patient's tumor responded dramatically to trastuzumab in combination with conventional chemotherapy, while another patient experienced an isolated central nervous system recurrence after trastuzumab therapy. Conclusions: HER2 amplification is relatively common in ovarian mucinous carcinomas (6/33, 18.2%), although not of prognostic significance. Trastuzumab therapy is a treatment option for patients with mucinous carcinoma when the tumor has HER2 amplification and overexpression. ! 162!  Background: The majority of ovarian mucinous tumors are borderline tumors or stage I carcinomas, and the prognosis, overall, for patients with early stage mucinous carcinoma is excellent. The prognosis in patients with spread beyond the ovaries, however, is extremely poor. Chemotherapy with paclitaxel and carboplatin is recommended for patients with metastatic mucinous carcinoma, but response rates are considerably lower than are observed in other subtypes of epithelial ovarian cancer (EOC) (1-6). At present no superior alternative treatment options exist.  HER2 is a member of the epidermal growth factor family of tyrosine kinase receptors. Activation of HER2 triggers a cascade of cellular responses, impacting cellular proliferation, angiogenesis and metastasis (7-9). Amplification and overexpression of HER2 is seen in approximately 15% of breast carcinomas and is associated with a poor prognosis (10-14). Adjuvant therapy using a monoclonal antibody against HER2 protein (trastuzumab) is effective alone and in combination with conventional cytotoxic chemotherapy in patients whose breast carcinomas have amplification of HER2 (15-18). In contrast, the significance of HER2 overexpression and amplification in EOC is less well understood. The reported frequency of HER2 overexpression in EOC ranges from 5-66% (119-23), although more recent studies using validated techniques for detection of HER2 overexpression or amplification have consistently shown results at the low end of this range (21,22). Clinical response to single agent trastuzumab in EOC has been disappointing. In a series of 41 patients with HER2 overexpressing EOC, identified from a series of 837 EOC tested for HER2 expression, there was only one complete responder and two partial responders for an overall response rate of 7.3% and a median progression-free interval of two ! 163! months (19). In this series, HER2 expression was determined by immunohistochemistry (IHC) only, and none of the patients in this series had carcinomas of mucinous subtype. There has been an increasing appreciation of the molecular differences between the different histologic subtypes of EOC (24-26). Differences in initial presentation, metastasis, response to therapy, and overall prognosis have been described and there has been criticism of the conventional approach of treating EOC as one entity (27). Most series analyzing HER2 expression in EOC have not performed subtype analysis based on histology and often have poor or absent representation of mucinous carcinoma (19, 20, 22, 23, 28).  Given the absence of data on mucinous ovarian tumors and HER2 expression, inference may be permitted based on histological and immunohistochemical similarities between mucinous ovarian tumors and tumors of the upper gastrointestinal tract (29-31). Activity of trastuzumab has been demonstrated in preclinical models of gastric and esophageal cancers (32-35); approximately 7- 15% of gastroesophageal adenocarcinomas show amplification of HER2. This prompted our investigation of HER2 expression in patients with recurrent mucinous EOC. Our objectives in this study were 1) to look for HER2 protein overexpression (IHC) and gene amplification (FISH) in our current and a historical patient population of patients with mucinous EOC and mucinous borderline ovarian tumors (BOT), 2) examine the correlation between HER2 immunostaining and amplification, 3) determine if HER2 expression or amplification status changed from the time of initial presentation to recurrence, 4) treat patients with recurrent mucinous ovarian carcinoma with trastuzumab, when the tumor has HER2 amplification and overexpression, and monitor for response to treatment.   ! 164! Methods:  Case Selection Following Institutional Review Board approval the following cases were identified: 1) a cohort of 34 cases of mucinous carcinoma from 1984-2000 in British Columbia (BC); these were identified as part of a population-based review of cases of ovarian carcinoma who had no microscopic residual disease after primary surgery. This cohort has been described previously (36) and the 34 mucinous carcinomas are part of a tissue microarray consisting of 541 cases of ovarian cancer, 2) three mucinous carcinomas and seven mucinous BOT collected as part of our Ovarian Tumor Bank in BC, since 2000, 3) three archival cases from a previously published series on mucinous ovarian carcinomas from our institution (37), and 4) 15 mucinous BOT cases, gastrointestinal type, from the pathology archives of Johns Hopkins University School of Medicine collected from their institution (n=12) or as consults from other institutions (n=3) during the period between 1994-2005. Three patients with recurrent mucinous carcinoma and HER2 amplification were treated with a combination of HER2 targeted therapy (trastuzumab) and platin-based chemotherapy and followed prospectively. Response was based on serial examinations, tumor markers, and CT imaging (RECIST criteria).  Immunohistochemistry IHC was performed on either whole sections, or for the retrospective, population based series, on tissue microarray slides in which duplicate 0.6 mm cores from each case were present in the array. Four-micron thick sections were immunostained on a Ventana ! 165! Benchmark XT staining system (Ventana Medical Systems, Tucson, AZ, USA). Sections were deparaffinized in xylene, dehydrated through three alcohol changes and transferred to Ventana Wash solution. Heat antigen retrieval was used. Endogenous peroxidase activity was blocked in 3% hydrogen peroxide. Slides were then incubated with rabbit monoclonal anti- HER2 Ab (clone SP3) at a dilution of 1:50, at 37oC for 32 min, and developed with a proprietary Ventana amplification reagent kit followed by DAB chromogen. Finally, sections were counterstained with hematoxylin and mounted. HER2 was scored visually according to the ASCO/CAP guidelines (38): 0 or 1+ (negative): no staining or incomplete membrane staining in >30% of tumor cells; 2+ (weakly positive, equivocal): strong, complete membranous staining in < 30% cells or weak to moderate heterogeneous staining in > 10% cells); 3+ (strongly positive: strong complete membrane staining in > 30% of tumor cells). All cases were reviewed and scored by one pathologist (CBG). Tissue cores that were missing, or were otherwise uninterpretable were not included in the analysis.  Fluorescence in situ hybridization Six-micron sections of the TMA slides were hybridized with probes to LSI® Her-2/neu and CEP® 17 with the PathVysionTM HER-2 DNA Probe Kit using a modified protocol. Briefly, slides were baked overnight at 60°C, deparaffinized and dehydrated. Pre-treatment washes included 10mM citric acid buffer (pH 6.0) (80°C, 45 minutes), 2X SSC twice (5 minutes each), distilled water (1 minute). Slides were protease treated at 37°C for 12 minutes, washed with 2XSSC twice (5 minutes each), dehydrated and air-dried, then counterstained with DAPI and visualized on a Zeiss Axioplan epifluorescent microscope. Analysis of FISH signals was performed using MetasystemsTM automated image acquisition and analysis ! 166! system, Metafer (Metasystems, Altlussheim, Germany). This FDA-approved, automated system scores FISH signals by employing specific measurement algorithms to detect and quantify clustered signals. A high correlation between manual and automated scoring of FISH signals has been previously reported (39). Average copy number for each probe was calculated and the amplification ratio (ratio between the average copy per cell for HER2 and the average copy for centromere 17) determined. Amplification ratios > 2.2 are considered positive (38). Tumors that failed to hybridize were not included in the analysis.  Statistical analysis Tests for heterogeneity were performed for the parameters: age, stage, grade, residual disease, exposure to previous chemotherapy, and the mean follow-up time with regard to both progression and overall survival for the HER2 positive and HER2 negative cohorts utilizing the Welch ANOVA, or the Pearson χ2 statistic as appropriate. Fisher’s Exact Test was used to test the prognostic significance of HER2 with respect to progression free (PF) and overall survival (S) time. Progression free survival (PF) interval (PFSI) or time is defined as the time from surgery to the first clinical evidence of recurrence (“chemical” recurrences i.e., tumor marker elevations not included). Overall survival (OS) time is defined as the time from surgery until death from any cause, or until the last date of follow-up. Kaplan-Meier survival analyses and the log-rank test were used to assess the impact of various clinicopathologic parameters and HER2 amplification on PFS and OS time. Results: Immunostaining and FISH data were available for 33 cases of mucinous carcinoma and 16 cases of mucinous BOT (Figure 1). Loss of cases from the original pool of 40 carcinomas ! 167! was primarily due to use of tissue microarrays where small sample size with few tumor cells or loss of tissue after digestion result in inability to assess amplification (39) Demographic and clinicopathologic data for the mucinous carcinomas with and without HER2 amplification are shown in Table 1. Tumor grade was the only parameter shown to be associated with progression free (PFS) (p<0.001) and overall survival (OS) time (p<0.001). HER2 overexpression by IHC was seen in five of the carcinomas (3+ in four cases, 2+ in one case). There was high-level HER2 amplification (HER2/CEP17 ratios >5) in six cancer cases (6/33=18.2%), including the five cases with HER2 overexpression. However, one case had discordant IHC and FISH results (IHC score of 0, FISH HER2/CEP ratio of 6.7) (Table 2). Of sixteen mucinous borderline tumors, three (3/16=18.8%) demonstrated HER2 amplification (HER2/CEP17 ratios of 3.1-3.3). One case of BOT showed discordant results with an IHC score of 0 and HER2/CEP ratio of 2.4 (Table 2).  We then looked at any recurrences with tissue available for FISH and IHC, to determine if HER2 expression levels/copy number changed in the recurrence. Seven patients whose tumor was represented on the TMA and all three of the cases from the previously published case series (33) developed recurrent disease. . None of these 10 cases that recurred had initial HER2 amplification. Tissue specimens were available for testing in 2/10 recurrences, neither of which demonstrated HER2 immunoreactivity or amplification (HER2 amplification ratio of 1.2 and 0.77, respectively). Among the cases represented on the TMA, there were no recurrences in the six patients with HER2 amplification, and seven recurrences in the 27 cases where HER2 was not amplified. There was no significant difference in prognosis ! 168! associated with HER2 amplification at the time of diagnosis (pLog-Rank p=0.0920; Fisher’s Exact two-tailed p=0.1445) (Figure 2).  Of the three cases of recurrent mucinous carcinoma identified prospectively, all showed strong HER2 expression and amplification at the time of recurrence and were treated with a combination of conventional chemotherapy and trastuzumab. These cases are described in detail below.  Case 1: The first patient with recurrent mucinous carcinoma initially presented at age 19 with irregular periods, pelvic pain, increased abdominal girth, and an elevated CA125 of 110 kU/L (other markers normal). Imaging revealed a 15 cm mass and ascites. She underwent surgical staging including unilateral salpingoophorectomy (USO), appendectomy, omentectomy, peritoneal biopsies, and washings. Pathology reported a 20 x 15 x 14 cm mucinous BOT, intestinal type with focal inatraepithelial carcinoma of the ovary, all other specimens negative, stage Ia. She was observed and did well until 15 months later when she was noted to have an elevation in her CA125 to 81 kU/L. A CT scan revealed ascites and a mass in the contralateral ovary. She underwent USO and multiple biopsies. Pathology showed a mucinous borderline tumor of the ovary with intraepithelial carcinoma, but there were now implants of invasive mucinous carcinoma on the peritoneal surfaces. She received carboplatin and paclitaxel (CP) for six cycles, with normal CA125 throughout but again recurred four months after completion of therapy, based on reaccumulation of ascites, omental disease, elevated CA125 (130 kU/L), and abdominal symptoms. Pathology review of her first recurrence was performed to assess for molecular markers. This revealed the ! 169! overexpression and amplification of HER2 (IHC 3+, HER2/CEP ratio 7.2) (Figure 3) and trastuzumab (6mg/kg) was given in addition to single agent monthly carboplatin (600mg/m2). A dramatic response, based on imaging and tumor markers, was noted after three cycles (Figure 4) and she completed a total of six cycles of this combination. She then received trastuzumab alone for three cycles with stable disease after which her markers began to rise and ascites and omental disease were seen on CT scan. Carboplatin was reintroduced but her markers continued to increase and she was changed to gemcitabine in combination with trastuzumab. Her CA125 level dropped from 1800 to 180 kU/L after the first cycle but she developed signs and symptoms of large bowel obstruction. She was taken to surgery for necrotic tumor in her cecum and splenic flexure and underwent a hemicolectomy and debulking without complications. She continued on gemcitabine and trastuzamab for six cycles with stable markers (CA125 range 50-210 kU/L). She progressed and failed three other traditional chemotherapy agents (capecitabine, liposomal doxorubicin, and etoposide) before ultimately succumbing to her disease. She died 50 months from time of diagnosis secondary to respiratory distress with massive intractable pleural effusions and pulmonary emboli.  Case 2: The second patient was a 33yo taken to the operating room for a 10 cm mass suspected to be benign. There was intra operative rupture of thick mucus within the abdominal cavity. RSO and washings were performed. Final pathology revealed a FIGO grade 2 invasive mucinous carcinoma with destructive stromal invasion, and normal fallopian tube. She was fully staged (USO, appendectomy, biopsies, washings) at a second procedure one month later, with all specimens negative, Stage Ic. She received three cycles of CP ! 170! followed by pelvic and whole abdominal radiation. She recurred 40 months later with a large pulmonary metastasis and subcarinal lymphadenopathy. She was initially deemed unresectable and received CP for four cycles and achieved a partial remission. The CA 19-9 had also decreased from a high of 1000 kU/L to 80 kU/L pre-thoracotomy. She underwent right middle and lower lobectomy. Immunohistochemistry of her tumor at this time revealed 3+ positivity for HER2 protein and a HER2/CEP17 ratio of 7.5 (Figure 3). She was then changed to trastuzumab monotherapy, which she took for a total of 5 cycles (6mg/kg for three weeks) and remained without clinical evidence of disease and with normal tumor markers. One month after the discontinuation of trastuzumab therapy she began experiencing severe headaches, neck spasms and vomiting. A CT scan of the head revealed multiple bilateral brain metastases (prior CT’s of the head negative, within six months), predominantly in her frontal lobes with possible interventricular extension. She was given whole brain radiation, 2000 cGy prescribed to the midplanes in five fractions. Despite radiation the patient developed progressive intracranial tumor without evidence of disease elsewhere. She died less than three months after discovery of her brain metastases, 56 months from initial diagnosis. In the third case, evaluation of her response to trastuzumab alone and in combination with platin- based chemotherapy was not possible by RECIST criteria (not imaged pre/post therapy and inconsistent tumor marker assessment). Interestingly, evaluation of her primary presentation, first recurrence and second recurrence showed an apparent change in HER2 amplification status. Careful re-analysis of the primary tumor identified an area of tumor heterogeneity. The primary tumor was predominantly HER2 negative with only focal HER2 expression (Figure 5). The areas showing overexpression also showed HER2 amplification ! 171! (data not shown). In the recurrent specimens (28 and 57 months from initial diagnosis) there was diffuse HER2 overexpression and amplification.  Discussion: Development of treatments for rare tumors is challenging. The NCI State of the Science meeting on ovarian cancer in 2005 recognized the need for separate trials for ovarian mucinous carcinoma, a rare subtype of EOC that responds poorly to conventional chemotherapy (23,27). An increased understanding of the importance of histologic subtype in EOC has resulted in an increased emphasis on understanding the molecular changes leading to the development of tumor subtypes with the goal of targeted therapy specific to each subtype. Success with this strategy is evident in breast cancer and there is increasing evidence from preclinical models of gastroesophageal cancers that HER2 can be targeted in this disease (28-30,32-34). Mucinous EOC resembles adenocarcinoma of the gastroesophageal region and molecular targeted therapy may also be indicated in appropriately selected cases of mucinous EOC.  Previously reported series investigating the prognostic implications of HER2 overexpression or HER2 targeted therapy in EOC included few or no cases of mucinous histology (19-23, 28). Varying techniques have been used to determine HER2 overexpression, often with less specific IHC assays, no FISH correlation, and inconsistent scoring/classification systems. Our series suggest that immunohistochemistry, FISH, and a scoring system similar to that used for breast cancer can be used for mucinous EOC and that there is good correlation between IHC and FISH results (2/49 or 4% discrepant, all with negative IHC and positive ! 172! FISH). The correct interpretation for these cases with discordant IHC and FISH is not clear. The frequency of HER2 overexpression/amplification is higher than previously reported (18%) and these cases are candidates for molecular therapy.  A dramatic response was observed in a patient with recurrent mucinous EOC showing amplification of HER2, treated with trastuzumab in combination with traditional chemotherapy after conventional therapy had ceased to work. The second prospectively identified case with HER2 overexpression and amplification received trastuzumab treatment, however she developed isolated intracranial tumor metastasis, a rare site of metastatic tumor in EOC. HER2 overexpression may provide tumor cells with increased metastatic aggressiveness thereby increasing the spread to sites such as the lungs and the central nervous system (40). In breast cancer patients, isolated central nervous system metastases have been observed in 9-10% of patients receiving trastuzumab-based therapy (41). The development of central nervous system metastases in these patients may occur due to increased patient survival times (i.e., brain metastases may become symptomatic as a result of an extended life span), and the inability of trastuzumab to penetrate the blood-brain barrier (40). We postulate that this limitation in trastuzumab therapy explains the isolated brain recurrence in this patient who had complete resolution of her disease process in all other locations. The prognostic implications of HER2 amplification in mucinous EOC or BOT have not been studied previously. None of the cases with HER2 overexpression or amplification identified in the retrospective case series experienced a recurrence. Determination of HER2 status at the time of diagnosis is unlikely to be a clinically relevant prognostic indicator. We believe, however, that assessment of HER2 status can provide valuable information in patients with advanced stage or recurrent mucinous EOC. For those patients whose tumors demonstrate ! 173! overexpression and amplification of HER2, targeted therapy with trastuzumab (+/- conventional chemotherapy) can be considered. As seen in other cancers, HER2 heterogeneity was demonstrated in one of our mucinous ovarian carcinomas and repeat analysis of tumors of interest may be warranted.  Conclusions: Prior investigations suggest HER2 amplification does not seem to be a significant event in epithelial ovarian cancers when analyzed across all histologic subtypes. However, we have demonstrated that in ovarian mucinous carcinomas HER2 amplification is relatively common (6/33, 18.2%), although not necessarily of prognostic significance. Response to conventional therapy is limited in this rare histologic subtype of EOC and trastuzumab therapy provides a treatment option for patients with mucinous carcinoma when the tumor has HER2 amplification and overexpression.    Competing interests: Martin Kobel was supported through a non-directed educational grant from Eli Lilly Canada. The authors declare that they have no competing interests. Author’s contributions: JM, DM and KS provided clinical care and pertinent clinical information on the involved patients. KW interpreted the FISH results. BG, DH, and MK reviewed the pathology, scored the IHC, and BG and DH confirmed discrepant FISH results. AA performed the IHC. BR and RV provided cases and clinical histories from their ! 174! institution, and shared their expertise in mucinous ovarian tumors. JM, DH, DM, KW and BG participated in the design, and coordination of the manuscript with JM, KW, and BG principally involved in its draft. All authors read and approved the final version of the manuscript.  Acknowledgements: We would like to acknowledge the technical assistance of Lindsay Brown and Melinda Miller in performing the HER2 FISH assays. This work was supported by a unit grant from the Michael Smith Foundation for Health Research to OvCaRe. CBG was supported by the National Cancer Institute of Canada (#017051) and an unrestricted educational grant from Sanofi-Aventis. Patient outcome data and support in data analysis was provided by the Cheryl Brown Ovarian Cancer Outcomes Unit of the British Columbia Cancer Agency. MK is affiliated with the Institute of Pathology, Charité Hospital, Berlin, Germany, and has received fellowship support from Eli Lilly Canada.  References: 1. Winter WE, 3rd, Maxwell GL, Tian C, Carlson JW, Ozols RF, Rose PG, et al: Prognostic factors for stage III epithelial ovarian cancer: a Gynecologic Oncology Group Study. J Clin Oncol 2007, 25:3621-7.  2. Hess V, A’Hern R, Nasiri N, et al: Mucinous epithelial ovarian cancer: a separate entity requiring specific treatment. J Clin Oncol 2004,22:1040-4.  3. Pectasides D, Fonutrilas G, Aravantinos G, et al : Advanced stage mucinous epithelial ovarian cancer ; the Hellenic Cooperative Oncology Group experience. Gynecol Oncol 2005,99:7988-90. 15 ! 175! 4. Tabrizi AD, Kalloger SE, Kobel M, Cipollone J, Roskelley CD, Mehl E, Gilks CB. Primary ovarian mucinous carcinoma of intestinal type. Analysis of histologic invasion patterns, survival and immunohistochemical expression profile in a series of 31 cases. Int J Gynecol Path 2009, In press.  5. Pignata S, Ferrandina G, Scarfone G, Scollo P, Odicino F, Cormio G, Katsaros D, Villa A, Mereu L et al. Activity of chemotherapy in mucinous ovarian cancer with a recurrence free interval of more than 6 months: results from the SOCRATES restrospective study. BMC Cancer 2008, 8: 252.  6. Alexandre J, Ray-Coquard I, Selle F, Floquet A, Cottu P, Weber B, Falandry C, Lebrun D, Pujade-Lauraine E. Clinical presentation and sensitivity to platinum-based chemotherapy (CT) of mucinous advanced epithelial ovarian carcinoma (M-AEOC): The GINECO-Group experience. J Clin Oncol 27:15s, 2009 (suppl; abstr 5572).  7. Salomon DS, Brandt R, Ciardiello F, Normanno N: Epidermal growth factor-related peptides and their receptors in human malignancies. Crit Rev Oncol Hematol 1995,19:183- 232.  8. Slamon DJ, Godolphin W, Jones LA, Holt JA, Wong SG, Keith DE, et al: Studies of the HER3772/neu proto-oncogene in human breast and ovarian cancer. Science 1989,244:707- 12.  9. Alimandi M, Romano A, Curia MC, Muraro R, Fedi P, Aaronson SA, et al : Cooperative signaling of ErbB3 and ErbB2 in neoplastic transformation and human mammary carcinomas. Oncogene 1995,10:1813-21.  10. Cheang MC, Chia SK, Voduc D, Gao D, Leung S, Snider J, Mark Watson, Davies S, Bernard PS, Parker JS, Perou CM, Ellis MJ, Nielsen TO. Ki67 Index, HER2 Status, and Prognosis of Patients With Luminal B Breast Cancer. J Natl Cancer Inst 2009, 101(10): 736– 750. ! 176! 11. Slamon DJ, Clark GM, Wong SG, Levin WJ, Ullrich A, McGuire WL: Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene. Science 1987,235:177-82.  12. Pauletti G, Dandekar S, Rong H, Ramos L, Peng H, Seshadri R, et al: Assessment of methods for tissue-based detection of the HER-2/neu alteration in human breast cancer: a direct comparison of fluorescence in situ hybridization and immunohistochemistry. J Clin Oncol 2000,18:3651-64. 16  13. Sauer T, Wiedswang G, Boudjema G, Christensen H, Karesen R: Assessment of HER- 2/neu overexpression and/or gene amplification in breast carcinomas: should in situ hybridization be the method of choice? APMIS 2003,111:444-50.  14. Tapia C, Glatz K, Novotny H, Lugli A, Horcic M, Seemayer CA, et al: Close association between HER-2 amplification and overexpression in human tumors of non-breast origin. Mod Pathol 2007,20:192-8.  15. Goldenberg MM: Trastuzumab, a recombinant DNA-derived humanized monoclonal antibody, a novel agent for the treatment of metastatic breast cancer. Clin Ther 1999,21:309- 18.  16. Shak S: Overview of the trastuzumab (Herceptin) anti-HER2 monoclonal antibody clinical program in HER2-overexpressing metastatic breast cancer. Herceptin Multinational Investigator Study Group. Semin Oncol 1999,26:71-7.  17. Piccart-Gebhart MJ, Procter M, Leyland-Jones B, Goldhirsch A, Untch M, Smith I, et al: Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N Engl J Med 2005,353:1659399 72.   ! 177!  18. Romond EH, Perez EA, Bryant J, Suman VJ, Geyer CE, Jr., Davidson NE, et al : Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer. N Engl J Med 2005, 353:1673-84.  19. Bookman MA, Darcy KM, Clarke-Pearson D, Boothby RA, Horowitz IR: Evaluation of monoclonal humanized anti-HER2 antibody, trastuzumab, in patients with recurrent or refractory ovarian or primary peritoneal carcinoma with overexpression of HER2: a phase II trial of the Gynecologic Oncology Group. J Clin Oncol 2003,21:283-90.  20. Camilleri-Broet S, Hardy-Bessard AC, Le Tourneau A, Paraiso D, Levrel O, Leduc B, et al: HER 2 overexpression is an independent marker of poor prognosis of advanced primary ovarian carcinoma: a multicenter study of the GINECO group. Ann Oncol 2004,15:104-12.  21. Mano MS, Awada A, Di Leo A, Durbecq V, Paesmans M, Cardoso F, et al: Rates of topoisomerase II-alpha and HER-2 gene amplification and expression in epithelial ovarian carcinoma. Gynecol Oncol 2004,92:887-95.  22. Lee CH, Huntsman DG, Cheang MC, Parker RL, Brown L, Hoskins P, et al: Assessment of Her-1, Her-2, and Her-3 expression and Her-2 amplification in advanced stage ovarian carcinoma. Int J Gynecol Pathol 2005,24:147-52.  23. Mayr D, Kanitz V, Anderegg B, Luthardt B, Engel J, Lohrs U, et al: Analysis of gene amplification and prognostic markers in ovarian cancer using comparative genomic hybridization for microarrays and immunohistochemical analysis for tissue microarrays. Am J Clin Pathol 2006,126:101-9.  24. Soslow RA: Histologic subtypes of ovarian carcinoma: an overview. Int J Gynecol Pathol 2008,27:161-74. ! 178!  25. Gilks CB: Subclassification of ovarian surface epithelial tumors based on correlation of histologic and molecular pathologic data. Int J Gynecol Pathol 2004,23:200-5.  26. Kobel M, Huntsman D, Gilks CB: Critical molecular abnormalities in high-grade serous carcinoma of the ovary. Expert Rev Mol Med 2008,10:e22.  27. Fountain J, Trimble E, Birrer MJ: Summary and discussion of session recommendations. Gynecol Oncol 2006,103:S23-5.  28. Wu Y, Soslow RA, Marshall DS, Leitao M, Chen B: Her-2/neu expression and amplification in early stage ovarian surface epithelial neoplasms. Gynecol Oncol 2004,95:570-5.  29. Vang R, Gown AM, Wu LS, Barry TS, Wheeler DT, Yemelyanova A, et al: Immunohistochemical expression of CDX2 in primary ovarian mucinous tumors and metastatic mucinous carcinomas involving the ovary: comparison with CK20 and correlation with coordinate expression of CK7. Mod Pathol 2006,19:1421-8.  30. Vang R, Gown AM, Barry TS, Wheeler DT, Yemelyanova A, Seidman JD, et al: Cytokeratins 7 and 20 in primary and secondary mucinous tumors of the ovary: analysis of coordinate immunohistochemical expression profiles and staining distribution in 179 cases. Am J Surg Pathol 2006,30:1130-9.  31. Ji H, Isacson C, Seidman JD, Kurman RJ, Ronnett BM: Cytokeratins 7 and 20, Dpc4, and MUC5AC in the distinction of metastatic mucinous carcinomas in the ovary from primary ovarian mucinous tumors: Dpc4 assists in identifying metastatic pancreatic carcinomas. Int J Gynecol Pathol 2002,21:391-400.  ! 179! 32. Matsui Y, Inomata M, Tojigamori M, Sonoda K, Shiraishi N, Kitano S: Suppression of tumor growth in human gastric cancer with HER2 overexpression by an anti-HER2 antibody in a murine model. Int J Oncol 2005,27:681-5.  33. Fujimoto-Ouchi K, Sekiguchi F, Yasuno H, Moriya Y, Mori K, Tanaka Y: Antitumor activity of trastuzumab in combination with chemotherapy in human gastric cancer xenograft models. Cancer Chemother Pharmacol 2007,59:795-805.  34. Kim SY, Kim HP, Kim YJ, Oh do Y, Im SA, Lee D, et al: Trastuzumab inhibits the growth of human gastric cancer cell lines with HER2 amplification synergistically with cisplatin: Int J Oncol 2008,32:89-95.  35. Reichelt U, Duesedau P, Tsourlakis M, Quaas A, Link BC, Schurr PG, et al: Frequent homogeneous HER-2 amplification in primary and metastatic adenocarcinoma of the esophagus. Mod Pathol 2007,20:120-9.  36. Gilks CB, Ionescu DN, Kalloger SE, Kobel M, Irving J, Clarke B, et al: Tumor cell type can be reproducibly diagnosed and is of independent prognostic significance in patients with maximally debulked ovarian carcinoma. Hum Pathol 2008, 39(8):1239-51.  37. Ludwick C, Gilks CB, Miller D, Yaziji H, Clement PB: Aggressive behavior of stage I ovarian mucinous tumors lacking extensive infiltrative invasion: a report of four cases and review of the literature. Int J Gynecol Pathol 2005,24:205-17.  38. Wolff AC, Hammond ME, Schwartz JN, Hagerty KL, Allred DC, Cote RJ, et al: American Society of Clinical Oncology/College of American Pathologists guideline recommendations for human epidermal growth factor receptor 2 testing in breast cancer. J Clin Oncol 2007,25:118- 45.  39. Brown LA, Huntsman D: Fluorescent in situ hybridization on tissue microarrays: challenges and solutions. J Mol Histol 2007, 38:151-7. ! 180!  40. Weil RJ, Palmieri DC, Bronder JL, Stark AM, Steeg PS: Breast cancer metastasis to the central nervous system. Am J Pathol 2005 ,167:913-920.  41. Brustein HJ, Lieberman G, Slamon DJ, Winer EP, Klein P: Isolated central nervous system metastases in patients with HER2-overexpressing advanced breast cancer treated with first- line trastuzumab-based therapy. Ann Oncol 2005,16:1772-1777.                   ! 181!      Figure 1. Flowchart outlining the process of case identification for our retrospective series of mucinous ovarian cancers and mucinous borderline ovarian tumors.   ! 182!            Figure 2. Kaplan-Meier survival curves demonstrating that the presence of HER2 amplification in primary mucinous carcinomas is not of prognostic significance with respect to disease recurrence.     ! 183!                 Figure 3. HER2 immunostaining and FISH of tumors from cases 1 and 2 (Case 2-sample from lung), who subsequently received trastuzumab either alone or in combination with conventional chemotherapy. Each tumor shows strong immunoreactivity for HER2 and amplification by FISH (HER2 probe – red, CEP17 probe – green).            ! 184!     Figure 4. Computed tomography images of Case 1. The first image (a) was taken four months after completion of (surgery and) chemotherapy treatment for her first recurrence. Imaging had been ordered for increased gastrointestinal symptoms and an elevation in her tumor markers. Ascites and omental disease are noted. Carboplatin and trastuzumab were commenced with a dramatic response (b) resolution of ascites and omental nodules) seen after only three cycles. Graphic representation of CA125 levels (c) also demonstrates a drop in CA125 levels after the initiation of carboplatin and trastuzamab therapy and stable CA125 levels during trastuzamab monotherapy for at least three cycles. ! 185!            Figure 5. Prospectively identified Case 3: Tumor from initial presentation classified as a mucinous borderline ovarian tumor (BOT) and shows a discrete area of HER2 positivity in what is predominantly a HER2 negative tumor.         ! 186!             Table 1 Summary: Demographics and clinicopathologic parameters for the 33 mucinous ovarian carcinoma cases identified retrospectively. Progression free survival (PF) interval (PFSI) or time is defined as the time from surgery to the first clinical evidence of recurrence. Overall Survival (OS) time is defined as the time from surgery until death from any cause, or until the date of last follow-up. All times are given in years.       ! 187!   Table 2        ! 188!  Table 2, Continued:   Table 2 Summary: Immunohistochemistry (IHC) and fluorescence in-situ hybridization (FISH) results for HER2 protein expression and gene amplification respectively. Amplification was observed in a) 6/33 (18.2%) mucinous carcinomas and b) 3/16 (18.8%) mucinous borderline tumors of the ovary. a) Ovarian mucinous carcinomas (n=33 scorable)   ! 189! Appendix B:  Supplementary Material for Chapter 2              The work presented in Appendix B has been published: Wiegand KC, Shah SP, Al-Agha OM,  Zaho Y, Tse K, Zeng T, Senz J, McConechy MK, Anglesio MS, Kalloger SE, Yang W, Heravi-Moussavi A, Giuliany R, Chow C, Fee J, Zayed A, Prentice L, Melnyk N, Turashvili G, Delaney AD, Madore J, Yip S, McPherson AW, Ha G, Bell L, Fereday S, Tam A, Galletta L, Tonin PN, Provencher D, Miller D, Jones SJM, Moore RA, Morin GB, Oloumi A, Boyd N, Aparicio SA, Shih IeM, Mes-Masson AM, Bowtell DD, Hirst, M, Gilks B, Marra MA, Huntsman DG.  ARID1A Mutations in Endometriosis-Associated Ovarian Carcinomas. New England Journal of  Medicine. 2010; Oct 14; 363(16):1532-43. Epub 2010 Sep 8. Copyright © 2010, Massachusetts Medical Society.  Reprinted with permission.   ! 190! B.1 Supplemental Methods and Results Sanger sequencing of ARID1A exon1 The Illumina based targeted exon sequencing of ARID1A did not provide coverage of exon1. To obtain sequence information for exon 1, four overlapping PCR primer sets were designed, priming sites for M13 forward and M13 reverse added to their 5′ ends to allow direct Sanger sequencing of amplicons. For the PCR, after denaturation at 94oC for 1 min, DNA was amplified over 35 cycles (94oC 30 sec, 58-60oC 30sec, 72oC 30 sec) using an MJ Research Tetrad (Ramsey, MN). Final extension was at 72oC for 5 min. PCR products were purified using ExoSAP-IT® (USB® Products Affymetrix, Inc., Cleveland, OH) and sequenced using an ABI BigDye terminator v3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA) and an ABI Prism 3130xl Genetic Analyzer (Applied Biosystems, Foster City, CA). All capillary traces were visually inspected to confirm their presence in tumor and absence from germline traces or analyzed using Mutation Surveyor.  Sequencing of two mutations in CCC13 CCC13 has two somatic mutations (5541insG and T5953C (S1985P)), which were close enough to be sequenced from a single PCR fragment. PCR products were cloned and then resequenced (see Supplemental Figure 2.2).      ! 191!  Gene Expression Analysis For gene expression analysis, the RNA-sequencing reads initially were mapped to the genome (NCBI36/hg 18) using MAQ (0.7.1). We used the Sequence Alignment/Map (SAMtools 0.1.7) for downstream processing. Up to five mismatches was allowed. Raw expression values (read counts) were obtained by summing the number of reads that mapped to human genes based on the Ensembl database (Release 51). The initial gene expression values were normalized using a quantile normalization procedure using aroma.light (1.16.0.) package in R (2.11.1). Results for the 50 genes with the greatest differential expression with respect to ARID1A mutation status are shown in Supplemental Table 2.4.  Fluorescent In-Situ Hybridization Tissue samples from CCC13 and CCC23 were assayed for deletion of ARID1A using fluorescent in-situ hybridization (FISH). Six micrometer-thick sections were pre-treated as described previously.1  Three-color FISH assays were performed using BACs specific to the regions flanking ARID1A (RP11-35M8 (chr1:26,609,021-26,767,926) and RP11-285H13 (chr1:27,033,759-27,216,771)) and fosmids specific to the ARID1A locus (G248P86703G10 (chr1:26,976,949-27,017,636), G248P89619A2 (chr1:26,954,143- 26,991,761), and G248P88415D8 (chr1:26,914,023-26,954,284)). BAC and fosmid probes were obtained from British Columbia Genome Sciences Centre, and were directly labeled with Spectrum Red, Spectrum Blue, or Spectrum Green using a Nick ! 192! Translation Kit (Abbott Molecular Laboratories, Abbott Park, Il). Analysis was done on a Zeiss Axioplan epifluorescent microscope. Images were captured using Metasystems Isis FISH imaging software (MetaSystems Group, Inc. Belmont MA). Loss of heterozygosity was confirmed in CCC23 and the results were inconclusive for CCC13.  B.2  Reference for supplemental methods and results 1. Makretsov N, He M, Hayes M, et al. A fluorescence in situ hybridization study of ETV6- NTRK3 fusion gene in secretory breast carcinoma. Genes Chromosomes Cancer 2004;40:152-7                       ! 193!  B.3 Supplemental Figures and Tables for Chapter 2  Figure 2.1 A                                            ! !  ! 194!  Supplemental Figure 2.1 Specimen CCC13 and adjacent atypical endometriosis. A. H&E stained sections from clear cell carcinoma (*) arising in an endometriotic cyst (†) at low power showing adjacent histologies (a), and at higher power showing regions of the clear cell carcinoma (b) and adjacent atypical endometriosis (c). A distant region of endometriosis from the same individual is shown at low power (d). B. BAF250a immunoreactivity is lost in the epithelial portion of both the clear cell carcinoma and adjacent atypical endometriosis, however is maintained in the distant endometriosis. HNF-1β can be seen in both the tumor and the adjacent atypical endometriosis, however is largely negative with only occasionally positive cells in the distant endometriosis. ER is highly expressed only in the distant endometriosis and is lost in both the tumor sample and adjacent atypical endometriosis. C. Sequencing chromatograms from the clear cell carcinoma and a PCR clone from contiguous atypical endometriosis clearly show the nucleotide variation corresponding to T5953C (S1985P). This mutation was present in 20/51 clones from the contiguous atypical endometriosis. In contrast, all cloned PCR products (from 58 clones) from distant endometriosis, which maintained BAF250a expression, show only wild type sequence. A heterozygous peak is seen in the DNA from the tumor. Micro-dissected material from both endometriosis samples was used to extract DNA, amplify by PCR, clone and sequence. None of the PCR clones from the distant endometriosis showed variation from the wild-type sequence. D. As in “C” Sequencing chromatograms from the clear cell carcinoma and a PCR clone from contiguous atypical endometriosis show an insertion of an additional G (5541InsG). This mutation was present in 3/54 clones from the contiguous atypical endometriosis. In contrast, all cloned PCR products (from 59 clones) from the distant endometriosis, which maintained BAF250a expression, show only wild type sequence. Sequencing read from the tumor sample shows characteristic overlapping reads corresponding to the in frame and out of frame alleles after the insertion point. As in “C” sequence from PCR clones are shown for both adjacent atypical endometriosis and distant endometriosis.               ! 195! Supplemental Figure 2.2       Supplemental Figure 2.2 Sequencing and cloning of two somatic mutations in CCC13. The two somatic mutations (5541insG and T5953C (S1985P)) were sequenced from a single PCR fragment. PCR products were cloned and then resequenced. In total, sequences from 45 clones were analyzed. We found 15/45 (33%) wildtype sequence, 9/45 (20%) sequences with the T5953C (S1985P) mutation, 9/45 (20%) sequences with the 5541insG mutation, and 12/45 (27%) sequences with both mutations in a single Sanger sequence trace. This reveals the complex relationship between the mutations,which occur both in trans (on independent alleles) and also in cis (on the same allele). This finding along with the presence of wildtype alleles, suggest that this tumor is aneuploid and a gene conversion or other rearrangement at the ARID1A locus has occurred and is present in a subset of cells.       ! 196! Supplemental Table 2.1 Summary of data generated from whole-transcriptome paired-end RNA sequencing. *     *Paired-End (PE) reads: the number of two short sequences with an unsequenced insertion between the sequences resulting from sequencing of fragmented and size-selected complementary DNA from both ends. Mapped PE reads: the number of resulting PE reads which were mapped back to the reference human genome sequence Genes expressed: the total number of human genes expressed in each case based on the reads that mapped to each gene RNA Sequencing: Illumina based RNA sequencing Exon Resequencing: Illumina based targeted exon sequencing. ! 197! Supplemental Table 2.2 ARID1A targeted exon sequence Validation Primers    ! 198! Supplemental Table 2.3 Summary of all ARID1A mutations in discovery and validation cohorts  CASE SOURCE MUTATION TYPE GENOMIC LOCATION VARIANT NO. OF READS CONTAINING MUTANT SEQUENCE/TOTAL NO. OF READS OVERLAPPING VALIDATED IN TUMOUR VALIDATED IN GERMLINE MUTATION STATUS BAF250a EXPRESSION ES2 (cell line) VGH missense 1:26974775 A5114G (N1705S) 1183/2652 (44.61%) yes not applicable not applicable positive TOV21G* (cell line) VGH indel 1:26930523 1645insC 484/1821 (26.58%) yes not applicable predicted somatic§ negative indel 1:26961245 2268delC 806/2022 (39.86%) no not applicable not applicable CCC01*± VGH indel 1:26978993 6018-6020delGCT 223//1529 (14.58%) yes no (FFPE) somatic positive CCC02* VGH indel 1:26895885 404delC exon 1 yes no (FFPE) somatic negative CCC03*± VGH indel 1:26978493 5518delG 395/1725 (22.90%) yes no (FFPE) somatic positive CCC04* VGH indel 1:26898389 deletion/fusion with ZDHHC18 † not applicable yes (by SNP 6.0) not applicable predicted somatic negative missense 1:26928901 G1310T (R437L) 31/244 (12.70%) no# not applicable inconclusive missense 1:26930260 G1381T (G461W) 105/648 (16.20%) no# not applicable inconclusive missense 1:26930385 G1506T (Q502H) 73/518 (14.09%) not done# not applicable inconclusive missense 1:26930555 C1676A (P559H) 97/536 (18.10%) no# not applicable inconclusive CCC05*± VGH missense 1:26965352 C2786A (P929H) 61/329 (18.54%) no# not applicable inconclusive positive missense 1:26979094 G6118T (G2040W) 159/1261 (12.61%) no# not applicable inconclusive missense 1:26979127 G6151T (D2051Y) 140/1392 (10.06%) no# not applicable inconclusive CCC06*± VGH nonsense 1:26973506 C4201T (Q1401*) 100/914 (10.94%) yes no (FFPE) somatic positive CCC07 VGH indel 1:26972761 3971insA 71/528 (13.45%) yes no (blood) somatic positive CCC08 VGH missense 1:26970384 A3386C (K1129T) 762/2536 (30.05%) yes no (FFPE) somatic positive CCC09* VGH nonsense 1:26978140 C5164T (R1722*) 1132/1513 (74.82%) yes no (FFPE) somatic negative CCC10* VGH indel 1:26972738 3948delG 166/758 (21.90%) yes no (blood) somatic negative CCC11± VGH nonsense 1:26930320 C1441T (Q481*) 204/813 (25.09%) yes no (FFPE) somatic positive CCC12± VGH nonsense 1:26978137 C5161T (R1721*) 514/1698 (30.27%) yes no (blood) somatic positive ‡± CCC13* VGH indel 1:26978517 5541insG¥ 395/1518 (26.02%) yes no (blood) somatic negative missense 1:26978929 T5953C (S1985P) 540/1712 (31.54%) yes no (blood) somatic CCC14*± VGH nonsense 1:26930559 C1680A (Y560*) 1411/2651 (53.23%) yes no (blood) somatic negative CCC15± VGH nonsense 1:26978542 C5566T (Q1856*) 447/1299 (34.41%) yes no (blood) somatic positive CCC16± VGH indel 1:26895767 286-296delGCGGAGCCGGA exon 1 yes no (blood) somatic positive ‡± CCC17 VGH nonsense 1:26928776 T1185A (Y395*) 392/787 (49.81%) yes no (blood) somatic negative indel 1:26974013 4709-4712delCTAA 426/2303 (18.50%) yes no (blood) somatic CCC18± VGH nonsense 1:26928768 C1177T (Q393*) 51/387 (13.18%) yes no (blood) somatic negative CCC19± VGH missense 1:26978470 G5494T (G1832W) 83/826 (10.05%) no# not applicable inconclusive positive ‡± CCC20 VGH nonsense 1:26930245 C1366T (Q456*) 512/2238 (22.88%) yes no (blood) somatic negative indel 1:26978280 5305delC 950/2225 (42.70%) yes no (blood) somatic CCC21 VGH missense 1:26978591 C5615T (A1872V) 838/1078 (77.74%) yes yes (FFPE) germline positive CCC22± VGH indel 1:26974657 4997delC 695/3171 (21.92%) yes no (FFPE) somatic negative CCC23*± VGH nonsense 1:26979115 G6139T (E2047*) 952/1374 (69.29%) yes no (blood) somatic negative CCC24 Montreal nonsense 1:26961284 C2306G (S769*) 505/1386 (36.44%) yes germline DNA not available somatic negative nonsense 1:26972770 C3979T (Q1327*) 147/751 (19.57%) no# germline DNA not available inconclusive missense 1:26978525 A5549G (D1850G) 92/739 (12.45%) no# germline DNA not available inconclusive CCC25 Montreal indel 1:26978269 5294delA 670/1823 (36.75%) yes no (blood) somatic negative CCC26 Montreal nonsense 1:26972779 C3988T (Q1330*) 265/1570 (16.88%) no# not applicable inconclusive positive CCC27 Montreal nonsense 1:26930449 C1570T (Q524*)¥ 310/786 (39.44%) yes no (blood) somatic positive missense 1:26965379 C2813T (A938V) 97/612 (15.85%) not done# no (blood) inconclusive indel 1:26979390 6415insC 779/6650 (11.71%) no not applicable did not validate CCC28 Montreal missense 1:26973770 G4465C (A1489P) 354/3084 (11.48%) no# not applicable inconclusive positive missense 1:26974082 C4777T (R1593W) 230/1782 (12.91%) no# not applicable inconclusive CCC29 Montreal indel 1:26978271 5296insG 613/1797 (34.11%) yes no (blood) somatic negative indel 1:26978517 5541insG¥ 141/1231 (11.45%) no not applicable did not validate CCC30 Montreal indel 1:26962348 2718-2719insCT 2445/4619 (52.93%) yes no (blood) somatic negative ! 199! EXPRESSIONCASE SOURCE MUTATION TYPE GENOMIC LOCATION VARIANT NO. OF READS CONTAINING MUTANT SEQUENCE/TOTAL NO. OF READS OVERLAPPING MUTATION VALIDATED IN TUMOUR VALIDATED IN GERMLINE MUTATION STATUS BAF250a CCC32 Montreal indel 1:26896551 1070delG exon 1 yes not done¶ predicted somatic negative indel 1:26979722 6747-6748delAG 341/3401 (10.03%) no not applicable did not validate CCC33 Montreal indel 1:26978976 6001-6002insCA 477/2309 (20.66%) yes no (blood) somatic negative ‡ CCC34 Montreal indel 1:26973944 4640delC 583/2274 (25.64%) yes no (blood) somatic negative indel 1:26979639 6664-6665delCC 380/2192 (17.34%) yes no (blood) somatic ‡ CCC35 Montreal indel 1:26930654 1776delT 470/2074 (22.66%) yes no (blood) somatic negative indel 1:26961268 2291insC 527/1569 (33.59%) yes no (blood) somatic CCC36 Montreal missense 1:26974298 G4993A (G1665R) 1177/3894 (30.23%) PCR failure not applicable inconclusive negative indel 1:26895496 14- 41insGGTCGCCCCCCGCCTCCAGCAC exon 1 yes no (blood) somatic ‡ CCC37 Montreal nonsense 1:26930506 C1627T (Q543*)¥ 422/1262 (33.44%) yes no (blood) somatic negative missense 1:26979172 A6196G (N2066D) 759/2003 (37.89%) yes no (blood) somatic CCC38 Montreal indel 1:26978517 5541insG¥ 460/1104 (41.67%) yes no (blood) somatic negative CCC39‡ Australia indel 1:26930276 1398delC 359/1101 (32.61%) yes no (blood) somatic not available indel 1:26960119 2107delC 229/1147 (19.97%) yes no (blood) somatic missense 1:26965568 G2912C (G971A) 64/569 (11.25%) not done# not applicable inconclusive CCC40‡ Australia indel 1:26895632 150insG exon 1 yes no (blood) somatic not available indel 1:26896589 1108insG exon 1 yes not done¶ predicted somatic indel 1:26978517 5541insG¥ 136/1173 (11.59%) no not applicable did not validate ± CCC41 Australia missense 1:26979278 A6302G (D2101G) 379/1727 (21.95%) no not applicable did not validate not available nonsense 1:26979631 C6655T (Q2219*) 175/1496 (11.70%) no# not applicable inconclusive CCC42± Australia indel 1:26965566 2911insG 158/722 (21.88%) yes no (blood) somatic not available CCC43 Australia nonsense 1:26895695 C214T (E72*) exon 1 yes not done¶ predicted somatic not available CCC44± Australia indel 1:26895601 120- 145delGGCGGCGGCAGCGGCCGAGC exon 1 yes not done¶ predicted somatic not available CCC45± Australia nonsense 1:26930506 C1627T (Q543*)¥ 659/1442 (45.70%) yes no (blood) somatic not available ‡ CCC46 Australia nonsense 1:26930521 C1642T (Q548*) 278/651 (42.70%) yes not done¶ predicted somatic not available nonsense 1:26973887 C4582T (R1528*) 574/1408 (40.77%) yes not done¶ predicted somatic CCC47± Australia indel 1:26971689 3519delC 787/2196 (35.84%) yes no (blood) somatic not available CCC48 Australia indel 1:26930676 1798insC 306/994 (30.78%) yes no (blood) somatic not available CCC49 Australia nonsense 1:26930449 C1570T (Q524*)¥ 515/1006 (51.19%) yes no (blood) somatic not available ‡± CCC50 Australia indel 1:26974155 4851delT 454/1810 (25.08%) yes not done¶ predicted somatic not available indel 1:26895538 57-62insGCCGCC (looks homozygous!) exon 1 yes not done¶ predicted somatic CCC51 Australia indel 1:26928787 1197delC 578/2043 (28.29%) yes no (blood) somatic not available missense 1:26928788 C1197A (N399K) 55/292 (18.84%) no# not applicable inconclusive indel 1:26928790 1200delA 2020/2024 (99.80%) no not applicable did not validate CCC52 Australia missense 1:26978351 G5375A (G1792D) 427/3073 (13.90%) no# not applicable inconclusive not available CCC53‡± Australia indel 1:26895951 470-486delGCCCGTCTGCCGTCGCC exon 1 yes no (blood) somatic not available nonsense 1:26973929 G4624T (E1542*) 1547/3013 (51.34%) yes not done¶ predicted somatic missense 1:26978534 A5558G (E1853G) 176/1232 (14.29%) no# not applicable inconclusive CCC54± Australia nonsense 1:26974145 C4840T (Q1614*) 1697/2306 (73.59%) yes no (blood) somatic not available CCC55 Australia indel 1:26973691 4387insA 841/1685 (49.91%) yes no (blood) somatic not available ± CCC56 Australia indel 1:26973434 4130insA 347/1365 (25.42%) yes no (blood) somatic not available indel 1:26973435 4131insC 1352/1355 (99.78%) no not applicable did not validate CCC57 JHU indel 1:26896215 734-747delCGGCTGCCGGCTCC exon 1 yes no (FFPE) somatic not available CCC58± JHU indel 1:26896488 1007delG exon 1 yes not done¶ predicted somatic not available CCC59 JHU missense 1:26974775 A5114G (N1705S) 852/2166 (39.34%) yes yes (cultured fibroblasts) germline positive CCC60 JHU indel 1:26972006 3660- 3684delGATGGGGCGCATGTCCTATGA 285/759 (37.55%) yes no (cultured fibroblasts) somatic negative ! 200! CASE SOURCE MUTATION TYPE GENOMIC LOCATION VARIANT NO. OF READS CONTAINING MUTANT SEQUENCE/TOTAL NO. OF READS OVERLAPPING VALIDATED IN TUMOUR VALIDATED IN GERMLINE MUTATION STATUS BAF250a EXPRESSION identical change is a germline variant in CCC59, thus not assumed to be somatic change in ES2 cells § cell line variant is assumed to occur as somatic mutation in CCCs† for details of deletion/fusion see Supplemental Figure 1 ¥ recurrent somatic/predicted somatic mutation# mutant allele frequency less than 20% ¶ presumed somatic in absence of germline DNA analysis because mutation is truncating (see Results) * cases analysed by RNA sequencing ‡ cases with multiple (more than one) somatic or predicted somatic mutations± cases with associated endometriosis no not applicable did not validate LEGEND HGS05 VGH missense 1:26960136 1:26895710 1:26973470 A2123C (Q708P) 1145/1439 (79.57%) yes yes (FFPE) germline positive HGS06 VGH missense G229A (A77T) exon 1 yes yes (blood) germline positive HGS07 VGH missense T4165C (Y1389H) 170/907 (18.74%) no# not applicable inconclusive positive nonsense 1:26978941 C5965T (R1989*)¥ 428/1569 (27.28%) HGS03 VGH missense 1:26973501 A4196C (Q1399P) 655/1154 (56.76%) yes yes (FFPE) germline positive HGS04 VGH missense 1:26896062 C581T (P194L) exon 1 yes yes (FFPE) germline positive HGS01 VGH missense 1:26978591 C5615T (A1872V) 310/1918 (16.16%) yes# yes (FFPE) germline positive HGS02 VGH missense 1:26960136 A2123C (Q708P) 667/1500 (44.47%) yes yes (FFPE) germline positive negative nonsense 1:26978479 C5503T (Q1835*) 533/1740 (30.63%) yes no (FFPE) somatic ‡± EC13 VGH missense 1:26973470 T4165C (Y1389H) 529/1363 (38.81%) yes no (blood) somatic positive nonsense 1:26978941 C5965T (R1989*)¥ 397/958 (41.44%) yes no (blood) somatic not applicable inconclusive missense 1:26978753 G5777T (G1926V) 137/1087 (12.60%) no# not applicable inconclusive missense 1:26979031 C6055A (H2019N) 131/1100 (11.91%) no# not applicable inconclusive EC12‡± VGH indel 1:26962057 2427delC 381/2381 (16.00%) yes no (FFPE) somatic missense 1:26965612 G2956T (D986Y) 19/170 (11.18%) no# not applicable inconclusive EC11 VGH missense 1:26965622 C2966A (P989H) 17/163 (10.43%) no# not applicable inconclusive positive missense 1:26973942 C4637A (P1546H) 143/1397 (10.24%) no# not applicable inconclusive missense 1:26978107 G5131T (G1711W) 91/829 (10.98%) no# EC09 VGH nonsense 1:26896126 C645A (Y215*) exon 1 yes not done¶ predicted somatic positive EC10± VGH indel 1:26978517 5539delG 151/873 (17.30%) yes no (FFPE) ti positive EC08 VGH nonsense 1:26979448 C6472T (R2158*) 339/1093 (31.02%) yes no (blood) somatic positive missense 1:26895837 A356G (E119G) exon 1 yes yes (blood) germline ‡± EC07 VGH indel 1:26962055 2425insT 85/639 (13.30%) yes no (blood) somatic negative nonsense 1:26896286 C805T (Q269*) exon 1 yes no (blood) somatic EC05 VGH nonsense 1:26962317 G2686T (E896*) 358/2075 (17.25%) no# not applicable inconclusive negative ‡ EC06 VGH indel 1:26961268 2291delC 432/2160 (20.00%) yes no (FFPE) somatic negative indel 1:26979390 6415delC 1820/6232 (29.20%) yes no (FFPE) somatic EC03± VGH nonsense 1:26978941 C5965T (R1989*)¥ 554/1474 (37.58%) yes no (blood) somatic negative ‡ EC04 VGH nonsense 1:26930566 C1687T (Q563*) 790/2002 (39.46%) yes no (blood) somatic negative indel 1:26970208 3211delA 1100/2999 (36.68%) yes no (blood) somatic EC01 VGH indel 1:26896510 1029-1043delAGCTGCGGCGGCGGC exon 1 yes not done¶ predicted somatic positive ± EC02 VGH missense 1:26973924 A4619G (N1540S) 285/1985 (14.36%) no# not applicable inconclusive positive missense 1:26973927 A4622C (H1541P) 194/1775 (10.93%) no# not applicable inconclusive ‡± CCC65 JHU indel 1:26978875 5900-5901insTG 359/1075 (33.40%) yes no (cultured fibroblasts) somatic negative nonsense 1:26896034 C553T (Q185*) exon 1 yes no (cultured fibroblasts) somatic CCC63± JHU indel 1:26974232 4928-4931insCTGG 551/2698 (20.42%) no not applicable did not validate not available CCC64± JHU indel 1:26972561 3852insA 463/1589 (29.14%) yes no (FFPE) somatic negative CCC61 JHU indel 1:26972885 4011-4012delTT 1158/6957 (16.65%) yes no (cultured fibroblasts) somatic negative CCC62± JHU indel 1:26974232 4928-4931insCTGG 537/3019 (17.79%) no not applicable did not validate negative ! 201! Supplemental Table 2.4 Differential expression of genes based on ARID1A mutation status   ! 202! Supplemental Table 2.5 ARID1A Mutation Status and Presence of Endometriosis               ! 203! Appendix C:  Supplementary Tables and Figures for Chapter 4 Protein Phosphorylation Site Company Cat # Concentration Status Notes 4EBP1 Cell Signaling 9452 1:250 V 4EBP1 T37 T46 Cell Signaling 9459 1:500 V 4EBP1 S65 Cell Signaling 9456 1:250 V acetyl Co-A carboxylase alpha (1) Epitomics 1768-1 1:300 C acetyl Co-A carboxylase alpha (1) S79 Cell Signaling 3661 1:500 V AKT Cell Signaling 9272 1:250 V AKT S473 Cell Signaling 9271 1:250 V AKT T308 Cell Signaling 9275 1:250 V Alpha-catenin EMD Millipore 1030 1:3000 V Antibody no longer available AMPK Cell Signaling 2532 1:250 C AMPK T172 Cell Signaling 2535 1:200 V Androgen Receptor Epitomics 1852 1:100 V ataxia telangiectasia mutated Abcam 32420 1:1000 C ATR interacting protein Cell Signaling 2737 1:100 C BAD S112 Cell Signaling 9291 1:200 V Bcl-2 Epitomics 1017-1 1:500 V BCL2L1 Cell Signaling 2762 1:250 V β-Catenin Cell Signaling 9562 1:250 V BIM Epitomics 1036 1:500 V B-raf Santa Cruz 5284 1:250 C Caspase 7 (cleaved) Cell Signaling 9491 1:250 C Caveolin 1 Cell Signaling 3238 1:250 V Collagen VI Santa Cruz 20649 1:250 V Cyclin B1 Epitomics 1495 1:500 V Cyclin D1 Santa Cruz 718 1:1000 V Cyclin E1 Santa Cruz 247 1:250 V Cyclin E2 Epitomics 1142 1:200 C CD31 Dako M0823 1:750 V E-Cadherin Cell Signaling 4065 1:200 V checkpoint kinase 1 Cell Signaling 2345 1:100 C checkpoint kinase 2 Cell Signaling 3440 1:100 C checkpoint kinase 1 S345 Cell Signaling 2348 1:100 C c-Jun Cell Signaling 9165 1:150 Validation  pending c-Jun S73 Cell Signaling 9164 1:100 V C-KIT Epitomics 1522-1 1:1000 V Appendix C:  Supplemental Material for Chapter 4 Supplemental Table 4.1:  RPPA Antibody List Protein Phosphorylation Site Company Cat # Concentration Status Notes C-MYC Cell Signaling 9402 1:100 C Cofilin-1 (non-muscle) S3 Cell Signaling 3313-S 1:500 Validation pending cytochrome c oxidase subunit II Epitomics 2969 1:500 C E-Cadherin Cell Signaling 3195 1;250 V EGFR Cell Signaling 2232 1:200 V EGFR Epitomics 1124 1:100 V Eukaryotic translation initiation factor 4E Cell Signaling 9742 1:200 V Estrogen induced gene Provided by collaborator 1:500 C E74-like factor 2 S51 Cell Signaling 9721S 1:100 Validation pending member of ETS oncogene family S383 Cell Signaling 9181 1:250 C Estrogen Receptor Alpha S167 Cell Signaling 2514 1:200 Validation  pending Estrogen Receptor Alpha S118 Epitomics 1091-1 1:300 V Estrogen Receptor Alpha Neomarkers RM-9101-S1:200 V Fibronectin Epitomics 1574-1 1:500 V FORTILIN Provided by R. Chambers 1:5000 C FOXO3A Cell Signaling 9467 1:500 Validation  pending FOXO3A S318 S321 Cell Signaling 2402 1:250 C mTOR Cell Signaling 2983 1:800 V GRB2-associated binding protein2 Cell Signaling 3239 1:500 V GATA-2 Sigma 1:100 Validation  pending GATA-3 BD Biosciences 558686 1:100 V GSK-3β S9 Cell Signaling 9336 1:500 V GSK-3α/β S21 S9 Cell Signaling 9331 1:250 V Heat Shock Protein 27 Cell Signaling 2402 1:250 C Heat Shock Protein 70 Cell Signaling 4872 1:250 C IGFBP2 Cell Signaling 3922 1:100 V IGFRβ Cell Signaling 3027 1:250 C Insuln receptor substrate 1 UBI 06-248 1:1000 V lymphocyte-specific protein tyrosine kinase Cell Signaling 2752 1:500 C ERK1/ERK2 T202 Y204 Cell Signaling 4377 1:800 V MEK Epitomics 1235-1 1:5000 V MEK S217 S221 Cell Signaling 9154 1:1000 V MIG-6 (mitogen-inducible gene 6 protein) Sigma 1:200 V mTOR Cell Signaling 2983 1:250 V mTOR S2448 Cell Signaling 2971 1:500 C Protein Phosphorylation Site Company Cat # Concentration Status Notes NOTCH3 Cell Signaling 3268 1:100 C P21 Santa Cruz 397 1:100 V P27 Epitomics 1591 1:200 V P38 MAPK (14) Cell Signaling 9212 1:100 V p38 MAPK (14) T180 Y182 Cell Signaling 9211 1:200 V P53 Cell Signaling 9282 1:3000 C Regulatory subunit PI3-Kinase UBI 06-195 1:4000 V P70S6K Epitomics 1494 1:500 V P70S6K T389 Epitomics 1175-1 1:250 V P90RSK T359 S363 Cell Signaling 9344 1:500 C PDK1 Cell Signaling 3062 1:250 V PDK1 S241 Cell Signaling 3061 1;100 V PI3K P110 ALPHA Epitomics 1683 1:500 C PROGESTERONE RECEPTOR Epitomics 1483-1 1:300 V RB S807 S811 Cell Signaling 9308 1:100 V PCNA Abcam 29100 1:1000 C PROGESTERONE RECEPTOR Epitomics 1483 1:250 V PROTEIN KINASE C ALPHA Upstate 5-154 1:2000 V PROTEIN KINASE C ALPHA S657 Upstate(UBI) 06-822 1:2000 C PTEN Cell Signaling 9552 1:300 V RAD51 Calbiochem 71 1:250 C S6 S236 S236 Cell Signaling 2211 1:5000 V S6 S240 S244 Cell Signaling 2215 1:2000 V SMAD3 Epitomics 1735-1 1:200 V SRC Upstate 05-184 1:200 V SRC Y416 Cell Signaling 2101 1:100 C SRC Y527 Cell Signaling 2105 1:400 V STAT3 Upstate 6-596 1:500 Validation pending STAT3 S727 Cell Signaling 9134 1:100 Validation pending STAT3 Y705 Cell Signaling 9131 1:500 V STAT5 Epitomics 1289 1:250 V STAT5 Y694 Epitomics 1208 1:250 Validation pending STAT6 Y641 Cell Signaling 9361 1:150 Validation pending STATHMIN Epitomics Jan-72 1:500 V Spleen Tyrosine kinase Santa Cruz 1240 1:500 V Protein Phosphorylation Site Company Cat # Concentration Status Notes Telomerase SDI 1706 1:250 C Antibody no longer available TAZ Abcam 3961 1:250 V TRANSGLUTAMINASE Neomarker MS224 1:750 V TUBERIN Epitomics 1613-1 1:500 V TUBERIN T1462 Cell Signaling 3617 1:200 V VEGFR2 Cell Signaling 2479 1:700 V DNA repair protein XRCC1 Cell Signaling 2735 1:100 C YES-ASSOCIATED PROTEIN Santa Cruz 15407 1:500 V 14-3-3 beta Santa Cruz 628 V 14-3-3 epsilon Santa Cruz 23957 C 14-3-3 zeta Santa Cruz 1019 V V=Validated C=Caution. For further evalutation Gene$ID Gene$Name Score(d) Numerator(r) Denominator(s+s0) Fold$Change q=value(%) Erα 67 &6.059 &1.631 0.269 0.323 0 SYK 54 &5.124 &1.07 0.209 0.476 0 GAB2 34 &4.763 &0.913 0.192 0.531 0 YAP 2 &3.895 &0.425 0.109 0.745 0 FKHRL1pS318<(c) 71 &3.519 &0.678 0.193 0.625 0 IGFBP2 17 &3.519 &0.67 0.19 0.629 0 CofilinpS3<(vp) 33 &3.404 &0.718 0.211 0.608 0 AR 96 &3.318 &0.652 0.197 0.636 0 STATHMIN 11 &2.923 &0.549 0.188 0.683 0 βcatenin 45 &2.898 &0.789 0.272 0.579 0 PKCAα 30 &2.854 &0.474 0.166 0.72 0 cMYC 74 &2.703 &0.396 0.147 0.76 0 ATM<(c) 69 &2.537 &0.492 0.194 0.711 0 STAT5 62 &2.026 &0.391 0.193 0.762 0 FibronecPn 73 &1.968 &0.499 0.253 0.708 0 AKT 21 &1.903 &0.367 0.193 0.776 0 MIG&6 81 &1.797 &0.218 0.121 0.86 0 MAPKpT202<Y204 15 &1.656 &0.39 0.235 0.763 1.116 PTEN 58 &1.656 &0.276 0.167 0.826 1.116 CAV1 43 &1.629 &0.538 0.33 0.689 1.116 CHK2<<(c) 83 &1.462 &0.239 0.164 0.847 2.828 Telomerase<(c) 10 &1.445 &0.145 0.1 0.905 2.828 4EBP1 95 &1.443 &0.189 0.131 0.877 2.828 PKCαpS657<(c) 88 &1.406 &0.146 0.104 0.904 2.828 SRCpY527 7 &1.331 &0.141 0.106 0.907 2.828 4EBP1pT37 24 &1.295 &0.213 0.164 0.863 2.828 CCNB1 50 &1.266 &0.396 0.313 0.76 4.079 PR 1 &1.189 &0.325 0.273 0.798 4.079 EIG121<(c) 53 &1.166 &0.158 0.136 0.896 4.079 ATRIP<(c) 99 &1.162 &0.14 0.12 0.908 4.079 p85 70 &1.161 &0.141 0.122 0.907 4.079 LCK<(c) 35 &0.972 &0.146 0.15 0.904 10.443 XRCC1 78 &0.93 &0.095 0.102 0.936 10.443 ClCasp7<(c) 44 &0.825 &0.19 0.23 0.877 11.57 CollagenVI 42 &0.768 &0.134 0.175 0.911 11.57 p70S6K 37 &0.73 &0.116 0.158 0.923 15.536 PCNA<(c) 29 &0.699 &0.085 0.122 0.943 15.536 PDK1pS241 4 &0.676 &0.077 0.114 0.948 17.435 TSC2 6 &0.563 &0.066 0.116 0.956 20.929 Note: All<anPbodies<validated<with<the<excepPon of<the<following<notaPons: (c<)<=<CauPon<&<Further<tesPng<pending (vp)<=<validaPon<sPll<pending Supplemental Table 4.2a:  Complete SAM List, CCC vs. HGS Down-regulated genes Gene$ID Gene$Name Score(d) Numerator(r) Denominator(s+s0) Fold$Change q=value(%) HSP27&(c) 28 5.413 0.866 0.16 1.822 0 4EBP11pS65 25 5.354 0.562 0.105 1.476 0 CCNE1 91 4.671 1.039 0.222 2.055 0 p38pT180 38 4.078 0.717 0.176 1.644 0 αCatenin 85 3.182 0.431 0.136 1.349 0 ACC1&(c) 22 3.054 0.485 0.159 1.4 0 HSP70&(c) 97 3.044 0.611 0.201 1.527 0 SRCpY416&(c) 86 3.019 0.494 0.164 1.409 0 EBCadherin 49 2.974 0.7 0.235 1.625 0 RAD51&(c) 93 2.967 0.382 0.129 1.303 0 ACCpS79 23 2.612 0.443 0.17 1.359 0 p38 12 2.466 0.312 0.126 1.241 0 STAT5pY694&(vp) 63 2.401 0.363 0.151 1.286 0 SMAD3 52 2.394 0.354 0.148 1.278 0 p21 13 2.185 0.222 0.102 1.167 0 AKTpT308 19 2.168 0.364 0.168 1.287 0 STAT3pS727&(vp) 60 1.853 0.295 0.159 1.227 0 STAT6pY641&(vp) 64 1.819 0.232 0.128 1.175 0 GSKpS9 76 1.507 0.171 0.114 1.126 2.828 GSK3pS21 66 1.457 0.194 0.133 1.144 2.828 GATA2&(vp) 77 1.446 0.219 0.151 1.164 2.828 p90RSKpT359&(c) 59 1.293 0.137 0.106 1.1 5.12 S6pS235 56 1.267 0.267 0.21 1.203 5.12 p27 68 1.25 0.197 0.158 1.146 5.12 EIF4E 46 1.248 0.148 0.118 1.108 5.12 AKTpS473 20 1.2 0.256 0.214 1.194 5.12 CD31 90 1.162 0.114 0.098 1.082 5.12 S6pS240 57 1.106 0.187 0.169 1.139 6.259 EGFRpY1173 47 1.102 0.145 0.131 1.105 6.259 Transglut 27 1.059 0.23 0.218 1.173 6.259 BCL2 26 0.982 0.151 0.154 1.11 10.443 EGFR 48 0.934 0.118 0.127 1.086 10.443 14B3B33&Epsilon&(c) 92 0.821 0.082 0.1 1.058 15.536 CCND1 40 0.793 0.104 0.131 1.075 15.536 14B3B3&β 84 0.679 0.08 0.117 1.057 17.435 STAT3&(vp) 61 0.622 0.092 0.148 1.066 20.929 Note: All&anTbodies&validated&with&the&excepTon of&the&following&notaTons: (c&)&=&CauTon&B&Further&tesTng&pending (vp)&=&validaTon&sTll&pending Supplemental Table 4.2b:  Complete SAM List, CCC vs. HGS Up-regulated genes Gene$ID Gene$Name Score(d) Numerator(r) Denominator(s+s0) Fold$Change q=value(%) CofilinpS3*(vp) 33 .3.525 .0.765 0.217 0.588 0 STAT5 62 .3.244 .0.662 0.204 0.632 0 GAB2 34 .3.198 .0.649 0.203 0.638 0 cMYC 74 .2.966 .0.44 0.148 0.737 0 YAP 2 .2.78 .0.308 0.111 0.808 0 S6pS235 56 .2.713 .0.545 0.201 0.685 0 CCNE1 91 .2.667 .0.52 0.195 0.697 0 S6pS240 57 .2.644 .0.408 0.154 0.754 0 Transglutaminase 27 .2.635 .0.54 0.205 0.688 0 4EBP1pT37 24 .2.549 .0.466 0.183 0.724 0 FKHRL1pS318*(c) 71 .2.498 .0.503 0.201 0.706 0 CCNB1 50 .2.141 .0.796 0.372 0.576 0 MAPKpT202*Y204 15 .2.113 .0.539 0.255 0.688 0 STAT3pS727*(vp) 60 .1.863 .0.296 0.159 0.814 0 14.3.3*Zeta 75 .1.797 .0.245 0.136 0.844 0 cJUNpS73 18 .1.765 .0.212 0.12 0.864 0 p70S6KpT389 55 .1.618 .0.222 0.137 0.857 1.721 BIM 41 .1.617 .0.315 0.195 0.804 1.721 MEK1_2pS217*S221 3 .1.561 .0.208 0.133 0.866 1.721 STAT3*(vp) 61 .1.429 .0.2 0.14 0.87 2.816 PKCAα 30 .1.362 .0.235 0.172 0.85 2.816 SRCpY527 7 .1.291 .0.133 0.103 0.912 4.891 Telomerase*(c) 10 .1.231 .0.126 0.102 0.917 6.864 p90RSKpT359*(c) 59 .1.176 .0.123 0.104 0.919 6.864 MIG.6 81 .1.035 .0.113 0.109 0.925 11.616 CAV1 43 .1.001 .0.362 0.362 0.778 11.616 PCNA*(c) 29 .0.996 .0.126 0.126 0.917 11.616 MEK1 89 .0.86 .0.137 0.16 0.909 17.156 p85 70 .0.808 .0.098 0.121 0.934 17.156 ATRIP*(c) 99 .0.765 .0.095 0.124 0.936 20.287 CollagenVI 42 .0.748 .0.142 0.189 0.907 20.287 Note: All*anWbodies*validated*with*the*excepWon of*the*following*notaWons: (c*)*=*CauWon*.*Further*tesWng*pending (vp)*=*validaWon*sWll*pending Supplemental Table 4.2c:  Complete SAM List, EC vs. HGS Down-regulated genes Gene$ID Gene$Name Score(d) Numerator(r) Denominator(s+s0) Fold$Change q=value(%) PR 1 7.17 2.548 0.355 5.848 0 CCND1 40 3.854 0.545 0.142 1.46 0 AMPKpT172 36 3.569 0.626 0.175 1.543 0 IRS1 72 3.313 0.546 0.165 1.46 0 BCL2 26 2.346 0.393 0.168 1.313 0 COX2<(c) 82 2.223 0.368 0.166 1.291 0 AKTpT308 19 2.194 0.395 0.18 1.315 0 p21 13 1.901 0.204 0.107 1.152 1.721 ERαpS118 39 1.807 0.302 0.167 1.233 1.721 EIG121<(c) 53 1.707 0.261 0.153 1.198 2.816 SYK 54 1.686 0.392 0.233 1.312 2.816 Erα 67 1.652 0.499 0.302 1.413 2.816 HSP70<(c) 97 1.602 0.356 0.222 1.28 2.816 CHK1<(c) 80 1.507 0.182 0.12 1.134 4.891 4EBP11pS65 25 1.49 0.166 0.111 1.122 4.891 14F3F3<Epsilon(c) 92 1.441 0.152 0.106 1.111 4.891 EGFR 48 1.345 0.197 0.147 1.147 4.891 AKTpS473 20 1.287 0.296 0.23 1.228 6.864 SRCpY416<(c) 86 1.275 0.221 0.173 1.165 6.864 α<Catenin 85 1.23 0.166 0.135 1.122 6.864 EGFRpY1173 47 1.15 0.162 0.141 1.119 7.228 GSK3pS21 66 1.081 0.142 0.132 1.104 7.228 ACC1<(c) 22 1.064 0.164 0.154 1.12 7.228 SMAD3 52 1.004 0.138 0.138 1.1 11.616 RAD51<(c) 93 0.975 0.116 0.119 1.084 11.616 GATA3 31 0.885 0.116 0.131 1.083 11.616 GATA2<(vp) 77 0.878 0.148 0.168 1.108 11.616 IGFRβ<(c) 16 0.878 0.155 0.176 1.113 11.616 EIF4E 46 0.851 0.108 0.127 1.078 11.616 p38pT180 38 0.787 0.147 0.187 1.107 17.156 ELF2ApS51<(vp) 87 0.766 0.108 0.141 1.077 17.156 HSP27<© 28 0.765 0.164 0.215 1.121 17.156 ACCpS79 23 0.753 0.126 0.168 1.092 17.156 PDK1pS241 4 0.721 0.087 0.121 1.062 17.156 STAT6pY641<(vp) 64 0.683 0.096 0.14 1.068 17.156 CD31 90 0.674 0.067 0.1 1.048 17.156 14F3F3<β 84 0.671 0.08 0.12 1.057 17.156 ATM<(c) 69 0.616 0.114 0.186 1.082 20.287 SRC 8 0.609 0.122 0.201 1.088 20.287 BRAF<(c) 32 0.59 0.089 0.151 1.064 20.287 Note: All<anSbodies<validated<with<the<excepSon of<the<following<notaSons: (c<)<=<CauSon<F<Further<tesSng<pending (vp)<=<validaSon<sSll<pending Supplemental Table 4.2d:  Complete SAM List, EC vs. HGS Up-regulated genes Gene$ID Gene$Name Score(d) Numerator(r) Denominator(s+s0) Fold$Change q=value(%) GATA2%(vp) 77 +4.034 +0.708 0.176 0.612 0 HSP70%(c) 97 +3.773 +0.913 0.242 0.531 0 CollagenVI 42 +3.423 +0.754 0.22 0.593 0 FibronecFn 73 +3.116 +1 0.321 0.5 0 CAV1 43 +3.107 +1.414 0.455 0.375 0 Transglutaminase 27 +2.162 +0.54 0.25 0.688 0 CD31 90 +2.146 +0.253 0.118 0.839 0 14+3+3%Epsilon%%(c) 92 +2.012 +0.256 0.127 0.838 0 CCND1 40 +1.895 +0.321 0.17 0.8 0 COX2%(c) 82 +1.801 +0.308 0.171 0.808 2.085 p85 70 +1.732 +0.251 0.145 0.84 2.085 CHK1%(c) 80 +1.7 +0.225 0.132 0.856 2.085 Bcl+xL 98 +1.604 +0.229 0.143 0.853 2.552 LCK%(c) 35 +1.477 +0.268 0.181 0.83 3.759 ClCasp7%(c) 44 +1.441 +0.445 0.309 0.734 3.759 FKHRL1pS318%(c) 71 +1.352 +0.355 0.263 0.782 4.814 BCL2 26 +1.311 +0.228 0.174 0.854 4.814 GATA3 31 +1.282 +0.189 0.147 0.877 4.814 AKTpT308 19 +1.144 +0.222 0.194 0.857 6.021 AKTpS473 20 +1.111 +0.299 0.269 0.813 6.021 Notch3 51 +1.106 +0.178 0.161 0.884 6.021 p21 13 +1.106 +0.151 0.136 0.901 6.021 RAD51%(c) 93 +1.054 +0.146 0.139 0.904 8.38 YAP 2 +1.01 +0.141 0.14 0.907 8.38 cJUNpS73 18 +0.964 +0.146 0.151 0.904 10.207 MEK1 89 +0.938 +0.182 0.194 0.882 10.207 MIG+6 81 +0.883 +0.125 0.141 0.917 11.991 PCNA%(c) 29 +0.835 +0.14 0.167 0.908 11.991 p27 68 +0.833 +0.164 0.197 0.893 11.991 ATRIP%(c) 99 +0.784 +0.129 0.165 0.914 11.991 ELK11pS383%(c) 94 +0.732 +0.123 0.168 0.919 11.991 14+3+3%Zeta 75 +0.713 +0.126 0.177 0.916 11.991 ELF2ApS51%(vp) 87 +0.712 +0.115 0.162 0.923 11.991 XRCC1 78 +0.698 +0.091 0.131 0.939 14.612 STAT6pY641%(vp) 64 +0.671 +0.109 0.163 0.927 14.612 HSP27%(c) 28 +0.637 +0.133 0.209 0.912 14.612 STAT5pY694%(vp) 63 +0.613 +0.12 0.196 0.92 14.612 CHK1pS345%(c) 79 +0.599 +0.075 0.126 0.949 14.612 ERpS167%(vp) 65 +0.583 +0.073 0.125 0.951 14.612 EIG121%(c) 53 +0.556 +0.102 0.184 0.931 15.957 14+3+3%β 84 +0.495 +0.077 0.155 0.948 17.914 EGFR 48 +0.402 +0.068 0.168 0.954 19.467 p38pT180 38 +0.395 +0.09 0.227 0.94 19.467 SRCpY527 7 +0.392 +0.05 0.126 0.966 19.467 4EBP1 95 +0.38 +0.063 0.167 0.957 19.467 MAPKpT202%Y204 15 +0.291 +0.09 0.31 0.94 21.065 Note: All%anFbodies%validated%with%the%excepFon of%the%following%notaFons: (c%)%=%CauFon%+%Further%tesFng%pending (vp)%=%validaFon%sFll%pending DN%=%group%of%18%serous%cases%ER%negaFve Supplemental Table 4.2e:  Complete SAM List, HGS vs. DN Down-regulated genes Gene$ID Gene$Name Score(d) Numerator(r) Denominator(s+s0) Fold$Change q=value(%) Erα 67 4.26 1.373 0.322 2.591 0 βcatenin 45 3.885 1.237 0.318 2.357 0 BRAF:(c) 32 3.769 0.612 0.162 1.528 0 mTOR 14 3.351 0.519 0.155 1.433 0 ERαpS118 39 3.141 0.596 0.19 1.512 0 p70S6K 37 2.642 0.514 0.194 1.428 0 AR 96 2.6 0.696 0.268 1.62 0 αCatenin 85 2.566 0.397 0.155 1.317 0 PKCAα 30 2.378 0.494 0.208 1.408 0 BIM 41 2.344 0.471 0.201 1.386 0 EGcadherin 49 2.208 0.677 0.307 1.599 0 AKT 21 2.113 0.456 0.216 1.372 0 AMPKpT172 36 2.057 0.413 0.201 1.332 0 CCNB1 50 1.938 0.815 0.42 1.759 0 CHK2:(c) 83 1.833 0.383 0.209 1.304 0 PR 1 1.788 0.662 0.37 1.582 0 IGFRβ:(c) 16 1.582 0.345 0.218 1.27 1.847 SYK 54 1.569 0.406 0.259 1.325 1.847 RbpS807 9 1.519 0.357 0.235 1.281 1.847 ATM:(c) 69 1.509 0.332 0.22 1.258 1.847 EIF4E 46 1.47 0.229 0.155 1.172 1.847 CCNE1 91 1.439 0.357 0.248 1.281 1.847 PDK1pS241 4 1.385 0.202 0.146 1.151 1.847 TSC2 6 1.297 0.193 0.148 1.143 2.552 Telomerase:(c) 10 1.267 0.159 0.126 1.117 2.552 GAB2 34 1.22 0.31 0.254 1.24 2.552 PTEN 58 1.2 0.25 0.209 1.19 3.759 STATHMIN 11 1.144 0.296 0.259 1.228 3.759 cMYC 74 1.103 0.216 0.196 1.161 4.408 GSKpS9 76 0.922 0.132 0.143 1.096 8.38 PKCαpS657:(c) 88 0.842 0.114 0.135 1.082 10.207 p70S6KpT389 55 0.759 0.132 0.174 1.096 11.991 TSC2pT1462 5 0.741 0.107 0.145 1.077 11.991 4EBP1pT37 24 0.697 0.158 0.227 1.116 14.612 4EBP11pS65 25 0.649 0.081 0.125 1.058 14.612 S6pS235 56 0.606 0.162 0.267 1.119 15.957 IRS1 72 0.605 0.106 0.176 1.077 15.957 MEK1_2pS217:S221 3 0.601 0.104 0.172 1.075 15.957 CofilinpS3:(vp) 33 0.56 0.144 0.257 1.105 15.957 SRCpY416:(c) 86 0.497 0.107 0.216 1.077 17.914 S6pS240 57 0.486 0.102 0.209 1.073 17.914 GSK3pS21 66 0.477 0.077 0.161 1.055 17.914 SRC 8 0.436 0.115 0.263 1.083 19.467 Note: All:anVbodies:validated:with:the:excepVon of:the:following:notaVons: (c:):=:CauVon:G:Further:tesVng:pending (vp):=:validaVon:sVll:pending DN:=:group:of:18:serous:cases:ER:negaVve Supplemental Table 4.2f:  Complete SAM List, HGS vs. DN Up-regulated genes Supplemental Figure 4.1 A pAKT-Thr308 and B pAKT-Ser473  compared to PIK3CA mutation type ! 215! Appendix D:  Supplementary Tables and Figures for Chapter 5 ! 216! !!   Supplemental Table 5.2 Sequencing primers for ARID1A ! NAME* SEQUENCE 5'-3' COMMENT Length (nt) Tm ARID1A_FSP_1288 GTACCCGATGACCATGCAG Forward Sequencing primer for ARID1A Align to cDNA pos 1288 19 59.5 ARID1A_FSP_1599 CATACCCCTCCCAGCAGTC Forward Sequencing primer for ARID1A Align to cDNA pos 1599 19 61.6 ARID1A_FSP_1989 ACATCAGGGATTTCCAGCAG Forward Sequencing primer for ARID1A Align to cDNA pos 1989 20 58.4 ARID1A_FSP_207 GGGAAAGGAGCTGCAGGA Forward Sequencing primer for ARID1A Align to cDNA pos 207 18 58.4 ARID1A_FSP_2273 CCAGATGCCCCAGTACAGTT Forward Sequencing primer for ARID1A Align to cDNA pos 2273 20 60.5 ARID1A_FSP_2674 ACCGGAAAACCCAAGAAACT Forward Sequencing primer for ARID1A Align to cDNA pos 2674 20 56.4 ARID1A_FSP_3046 GTGGTGAGCCTGAGAGGAAG Forward Sequencing primer for ARID1A Align to cDNA pos 3046 20 62.5 ARID1A_FSP_3372 AAGAAGTCCCAGCCCAAGAT Forward Sequencing primer for ARID1A Align to cDNA pos 3372 20 58.4 ARID1A_FSP_3714 AAAGCTCCAGGGAGTGATCC Forward Sequencing primer for ARID1A Align to cDNA pos 3714 20 60.5 ARID1A_FSP_4101 CAGAATTACAAGCGGCCAAT Forward Sequencing primer for ARID1A Align to cDNA pos 4101 20 56.4 ARID1A_FSP_4424 AAACATGCCACCACAAATGA Forward Sequencing primer for ARID1A Align to cDNA pos 4424 20 54.3 ARID1A_FSP_4794 CGCACCTCTCCTAGCAAGTC Forward Sequencing primer for ARID1A Align to cDNA pos 4794 20 62.5 ARID1A_FSP_5099 CAGCATCATGACCTTCAACC Forward Sequencing primer for ARID1A Align to cDNA pos 5099 20 58.4 ARID1A_FSP_5469 TTTGTGGTGGACTGCTCAGA Forward Sequencing primer for ARID1A Align to cDNA pos 5469 20 58.4 ARID1A_FSP_5812 GCAAGTTTCCATTTGGCATT Forward Sequencing primer for ARID1A Align to cDNA pos 5812 20 54.3 ARID1A_FSP_597 CAGCAGAACTCTCACGACCA Forward Sequencing primer for ARID1A Align to cDNA pos 597 20 60.5 ARID1A_FSP_6123 GTGAGCTGCAACAAAGTGGA Forward Sequencing primer for ARID1A Align to cDNA pos 6123 20 58.4 ARID1A_FSP_6493 GGGAGATGGCTGTGGTACTG Forward Sequencing primer for ARID1A Align to cDNA pos 6493 20 62.5 ARID1A_FSP_882 ACCCTCAACCAACTGCTCAC Forward Sequencing primer for ARID1A Align to cDNA pos 882 20 60.5 ARID1A_RSP_7081 AGGAACAGAAAGGCGTGAGGTGAT Reverse Sequencing primer for ARID1A Align to cDNA pos 7081 24 60.1 ARID1A_RSP_6697 TAGTTGGCTCAAAGGGTGGGTTCT Reverse Sequencing primer for ARID1A Align to cDNA pos 6697 24 60 ARID1A_RSP_6110 TAAGTTAGTGGTGCCTGCTTCCGT Reverse Sequencing primer for ARID1A Align to cDNA pos 6110 24 60 ARID1A_RSP_5879 ATCTTGATGTTCCGGTGGCTCTGT Reverse Sequencing primer for ARID1A Align to cDNA pos 5879 24 60.1 ARID1A_RSP_5504 TCTGAGCAGTCCACCACAAATGGA Reverse Sequencing primer for ARID1A Align to cDNA pos 5504 24 60.1 ARID1A_RSP_5048 TTGAGGGACATCATTACCCGCCAT Reverse Sequencing primer for ARID1A Align to cDNA pos 5048 24 60.1 ARID1A_RSP_4494 AGCAACCTCAGCTGATGCCTGTAT Reverse Sequencing primer for ARID1A Align to cDNA pos 4494 24 60.2 ARID1A_RSP_4139 CCATTGGCCGCTTGTAATTCTGC Reverse Sequencing primer for ARID1A Align to cDNA pos 4139 24 59.9 ARID1A_RSP_3626 TTCCGCTTCTGGAATGTGGAGTCA Reverse Sequencing primer for ARID1A Align to cDNA pos 3626 24 60 ARID1A_RSP_3261 ATTGAGGTTGGTTGCAAGTTCCCG Reverse Sequencing primer for ARID1A Align to cDNA pos 3261 24 59.9 ARID1A_RSP_2849 TGAGGGTTGATCATGCCAGCCATA Reverse Sequencing primer for ARID1A Align to cDNA pos 2849 24 60.1 ARID1A_RSP_2661 TGACCCAACCTGAGGTGGCATATT Reverse Sequencing primer for ARID1A Align to cDNA pos 2661 24 60.1 ARID1A_RSP_2043 ATTACTCTGCTCTCCTTGGCTGCT Reverse Sequencing primer for ARID1A Align to cDNA pos 2043 24 59.8 ARID1A_RSP_1860 TGAGGATGCCTGAGACCCAAATGA Reverse Sequencing primer for ARID1A Align to cDNA pos 1860 24 59.8 ARID1A_RSP_1539 TTGAGACTGTGGCTGCTGCTGATA Reverse Sequencing primer for ARID1A Align to cDNA pos 1539 24 60 ARID1A_RSP_1175 TGATCCATTGGACTGGATGGCTGA Reverse Sequencing primer for ARID1A Align to cDNA pos 1175 24 59.9 *FSP= Forward seqencing primer, RSP= Reverse Sequencing Primer ! 217! Supplemental Figure 5.1  RT-PCR data for PLK and EGFR family !! ! 218! Supplemental Table 5.3 BP-Adapted primers for ARID1A mutant creation  PrimerName GW-BP-primer add-on Full Primer Sequence ARID1A-5BPgw GGGGACAAGTTTGTACAAAAAAGCAGGCTTC GGGGACAAGTTTGTACAAAAAAGCAGGCTTCATGGCCGCGCAGGTCGCCCCCGCCG ARID1A-3BPgw_C1680A GGGGACAAGTTTGTACAAAAAAGCAGGCTTC GGGGACAAGTTTGTACAAAAAAGCAGGCTTCTTAAGGAGACTGAGCCTGTGGC ARID1A-3BPgw_C1570T GGGGACAAGTTTGTACAAAAAAGCAGGCTTC GGGGACAAGTTTGTACAAAAAAGCAGGCTTCGGCTAGGATGGCTGCTGGGAG  

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.24.1-0073992/manifest

Comment

Related Items