1. CTD2 CTD2: Functional Cancer Genomics cancer genomics
cancer therapeutics 1

2. 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

3. 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

4. Integrated functional cancer genomics

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

6. 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

7. 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

8. Integrating functional and structural genomics
in ovarian cancer

9. 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

10. 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

11. 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

12. 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

13. 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

14. 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

15. 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.

16. 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

17. 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

18. 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), ) 18

19. 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

20. 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

21. 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