CTD2 CTD2: Functional Cancer Genomics cancer genomics cancer therapeutics 1

Characterization of cancer genomes is essential but not sufficient Hundreds to thousands of candidates in each tumor Distinguishing Driver vs. Passenger mutations Drivers: Tumor initiation or maintenance Context-specific actions of particular genetic elements Prioritization must be based on both genomic and biological weight of evidence 2

Functional interrogation of cancer genomes Gain-of-function: ORFs Connect genotype to function Cancer Genome Annotation Experimental cancer models Loss-of-function: RNAi Identify potential Achilles’ Heels

Integrated functional cancer genomics

DFCI Cancer Target Discovery & Development Center William Hahn Lynda Chin Ron DePinho

Genome scale barcoded shRNA screens Pooled shRNA plasmid library Packaged into lentivirus 45,000 distinct shRNA plasmids 4-week culture Infect … We have developed a genome scale barcoded shRNA screening method. This method takes advantage of lentiviral stable integration into genomic DNA and our ability to use custom microarray to measure the hairpin abundance. Cancer cells were infected with pooled viruses and propagated for 4 weeks and harvested for Biao Luo Tony Cheung Aravind Subramanian David Root 6

Identification of genes essential in ovarian cancer I applied this method to screen two ovarian cancer cell lines for essential genes. We present our data in heatmap. We performed multiple replicate infections for each cell line. The blue color in these two columns indicate shRNAs that drop out in ovarian cancer cell lines these shRNAs appear to have less effect to a panel of twelve non ovarian cancer cell lines we screened previously. I identified a gene named ID4, which ranked top in our list. Tony Cheung, Glenn Cowley, Barbara Weir, Biao Luo Jesse Boehm, Dave Root 7

Integrating functional and structural genomics in ovarian cancer

Transformation of immortalized ovarian surface epithelial cells Transformed SV40 LT/ST, hTERT Cell line # tumors/# injection sites Next, we tested the ability of ID4 expression to transform ovarian surface epithelial cell which were immortalized by SV40 large t and small t and htert. Overexpression of ID4 together with active MEK allele, significantly promoted tumor formation after subcutaneous injection into immunocompromised mice. such effect can not be observed with non-functional ID4 mutant. Vector ID4 MEKDD + lacZ MEKDD + ID4 MEKDD + ID4_DM 0/9 4/21 21/27 2/18

Identification of ID4 as an ovarian cancer oncogene Loss-of-Function Genes essential for ovarian cancer proliferation Cancer Genome Annotation Control Anti-ID4 ID4 Cross reference with genes in amplified regions in OvCa (TCGA) Gain-of-Function Genes that induce ovarian tumor formation Yin Ren, Sangeeta Bhatia Tony Cheung, Jesse Boehm, Glenn Cowley 10 10

Context Specific Functional Genomic Screening Platform ORF Library of GEOI Genetically defined Context specific TARGET cell Orthotopic injection Phenotype = tumorigenicity Control shRNA Biological Validation GEOI “HIT” Responder ID Context specific Tumor cell GEOI shRNA

Integrated genomic pipeline Genome-scale LOF Screens Targeted GOF in vitro Screens Genome-scale LOF screens Identify genes essential to ovarian cancer and GBM viability. Context-specific GOF screens Define cell- and genetic contexts in which GEOI is functionally relevant Identify GEOIs with in vivo activity in specific context Context Specific in vivo Screens Novel validated cancer drivers that merit consideration for drug discovery efforts 12

Scott Powers, Scott Lowe Cold Spring Harbor Laboratory Cancer Target Discovery & Development Center Scott Powers, Scott Lowe Integrate cancer genome computational analysis, mouse models, and in vivo screening to identify and validate new cancer genes, pathways, and tumor dependencies / therapeutic targets

Computational Analysis of Cancer Genomes Construction of oncogenomically focused shRNA and cDNA libraries Screen for oncogenicity with a transplantable mouse model Test for tumor dependency with mouse models and human cancer cell lines Focused libraries Under construction p53-/-hepatoblasts

Cold Spring Harbor Laboratory Cancer Target Discovery & Development Center Summary of findings Discovery and validation of 20 novel TSGs and oncogenes Unexpected number of identified tumor suppressors encode secreted proteins Discovered FGF19 oncogene dependency in human HCC cell lines containing the FGF19 amplicon This pinpoints for the first time a candidate cancer drug that selectively targets a genetic abnormality in HCC.

UTSW Cancer Target Discovery & Development Center A CONCERTED ATTACK ON PATIENT SPECIFIC ONCOGENIC VULNERABILITIES IN LUNG CANCER Objective : to employ parallel phenotypic screening of genome-wide siRNA libraries and a diverse chemical compound file to return authentic drug lead/target relationships Mike Roth Michael White John Minna

mRNA Expression Profiles Identify 6 Major Subtypes (Clades) of Non-Small Cell Lung Cancer HBEC C7 C6 6 HBECs 56 NSCLCs C4 C1 C2 C5 C3 17 17

mRNA Defined Clades from the NSCLC Lines Are also Found in Primary NSCLCs Probability (using PAM, prediction analysis of microarray method) of each primary tumor sample belonging to a particular NSCLC Line Defined Clade Low probability of belonging to a Clade High probability of belonging to a Clade (NSCLC Data from Bild Nature 2006 (439), 353-357) 18

mRNA Defined Clades Identify Different NSCLC Drug Response Phenotypes Drug Sensitivity Frequency in Clades C5 C4 C3 C2 C1 High probability of sensitivity High probability of Resistance 19

Genome Wide siRNA Library Screens Reveal Clade-Selective Vulnerabilities No Kill Kill C1 C4 C6 C5 C7 Genome-wide siRNA screens on 4 tumor derived NSCLC cell lines revealed tumor specific siRNAs. At the same time, compound screens on cancer cell lines harboring different oncogene mutations revealed a reasonable rate of tumor cell line specific toxic compounds. This suggests that different tumors have different vulnerabilities, supporting a hypothesis that the price of uncontrolled growth is a set of metabolic vulnerabilities specific for the mutational background of the cancer. 20

Network interactions and synergy State of the art technological platforms Data sharing Model sharing Development of new informatics Deliverables to cancer research community Reagents and Informatic tools Large scale functional datasets (in vitro and in vivo) Experimental models Integrative data to inform investigator initiated research