1. Master Regulators of Tumor Subtype and Associated Driver Mutations
Interrogating High-Grade Glioma Regulatory Networks to Identify :
Master Regulators of Tumor Subtype and Associated Driver Mutations

2. Systems Biology Cancer Genomics Drugs & Biomarkers Epigenomics
Prevention
Diagnosis
Treatment
Cancer
Drugs & Biomarkers
Epigenomics
Cell Regulatory Logic
Protein
Membrane
DNA
Transcriptomics
Other Omics:
Metabolomics
Proteomics
Glycomics …
Clinical Trials

3. Transcriptional Interactions POST-TRANSCRIPTIONAL INTERACTIONS
Developmental
Transcriptional Interactions
POST-TRANSCRIPTIONAL INTERACTIONS
Zhao X et al. (2009) Dev Cell. 17(2):
Mani KM et al. (2008) Mol Syst Biol. 4:169
Palomero T et al., Proc Natl Acad Sci U S A 103, (Nov 28, 2006).
Margolin AA et al., Nature Protocols; 1(2): (2006)
Margolin AA et al., BMC Bioinformatics 7 Suppl 1, S7 (2006).
Basso K et al. (2005), Nat Genet.;37(4): (Apr. 2005)
Basso et al. Immunity May;30(5):744-52
Klein et al, Cancer Cell, 2010 Jan 19;17(1):28-40.
Sumazin et al. 2011, in press
Post-translational Interactions
Master regulators and
mechanism of action
Wang K, Saito M, et al. (2009) Nat Biotechnol. 27(9):829-39
Zhao X et al. (2009) Dev Cell. 17(2):
Wang K et al. (2009) Pac Symp Biocomput. 2009:
Mani KM et al. (2008) Mol Syst Biol. 4:169
Wang K et al. (2006) RECOMB
The CTD2 Network (2010), Nat Biotechnol Sep;28(9):
Floratos A et al. Bioinformatics Jul 15;26(14):
Lefebvre C. et al (2010), Mol Syst. Biol, 2010 Jun 8;6:377
Carro MS et al. (2010) Nature Jan 21;463(7279):318-25
Mani K et al, (2008) Molecular Systems Biology, 4:169

4. MINDy: Reverse Engineering of Post Translational Modulators of Transcriptional Regulation
Mary
Joe
TERT
MYC
GSK3
Tony
Signal
GSK3
Degradation
MYC
TERT

5. Post-Translational Network Validation (MINDy)
STK38 (serine-threonine kinase 38, NDR1)
1) Protein-Protein interaction with MYC
2) STK38 silencing in ST486 decreases MYC stability
3) MYC mRNA is not affected
3) MYC targets are consistently affected
~400 Gene Expression Profiles for Normal and Tumor Related Human B Cells 1 3
20 ug of protein lysate/lane
Cells were harvested 96 hours post-infection.
NT shRNA is a non-targeting shRNA vector, which does not target any human or mouse gene.
B10 and B11 are two different shRNAs for STK38.
PARP antibody recognizes 85 kDa fragment of cleaved 116 kDa PARP protein, which can be used as a marker for detecting an apoptosis. 2
Wang K, Saito M, et al. (2009) Nat. Biotechnol. 27(9):829-39

6. Available Transcriptional Interactomes:
Cancer
B Cell interactome (BCi)
Breast Cancer Cell interactome (BCCi)
T-ALL interactome (TALLi)
AML
Prostate Cancer interactome (Pci mouse/human)
Glioblastoma Multiforme interactome (GBMi)
Ovarian
Non-small-cell Lung Cancer
Colon Cancer
Hepatocellular Carcinoma
Neuroblastoma
NET
Stem Cells
Mouse EpiSC and ESC
Human ESC
Germ Cell Tumors (Pluripotency, Lineage Differentiation)
Neurodegenerative Disease
Human and Mouse Motor Neuron (ALS)
Human and Mouse whole brain (Alzheimer’s)
Interactomes are generated from primary tissue profiles and thus reflect cell regulation in vivo, in the presence of all relevant paracrine, endocrine, and contact signals

7. Interrogating the Assembly Manual of the Cell
Disease Initiation & Progression
Biomarkers
Pluripotency and Lineage Differentiation
Cell Regulatory
Logic
Therapeutic Targets
Small-molecule Modulators
Drug MoA and Resistance
Mechanism of Action

8. MARINa: Master Regulator Inference algorithm
Phenotype 2 (Neoplastic)
Phenotype 1 (Normal)
MRx ?
A Master Regulator is a gene that is necessary and/or sufficient to induce a specific cellular transformation or differentiation event.
TF1:
Repressed: 5/7
Activated: 5/7
Coverage: 10/18 (55%)
TF2:
Repressed: 1/5
Activated: 1/6
Coverage: 2/18 (11%)
Under-expressed in Tumor
Over-expressed in Tumor
Tumor Signature
Carro, M. et al. (2010). "The transcriptional network for mesenchymal transformation of brain tumours." Nature 463(7279):
Lefebvre C. et al. (2009). "A Human B Cell Interactome Identifies MYB and FOXM1 as Regulators of Germinal Centers." Mol Syst Biol, in press
Lim, W. et al. (2009). "Master Regulators Used As Breast Cancer Metastasis Classifier." Pac Symp Biocomp 14:

9. Mesenchymal Subtype of High-Grade Gliomas
Unsupervised clustering of 176 high grade tumors by expression of 108 genes that are positively or negatively associated with survival reveals 3 tumors classes (Proneural (PN), Mesenchymal (Mes) and Proliferative (Prolif).
Phillips et al., Cancer Cell, 2006
PNGES
MGES
PROGES
Malignant gliomas belonging to the mesenchymal sub-class express genes linked to the most aggressive properties of glioblastoma (migration, invasion and angiogenesis).

10. Identification of a mesenchymal regulatory module
Master Regulators control >75% of the Mesenchymal Signature of High-Grade Glioma
Mes signature genes
Activator
Repressor
Biochemical Validation of ARACNe Inferred Targets of Stat3, C/EBPb, FosL2, and bHLH-B2
Hierarchical Regulatory Module

11. In Vitro Validation
We asked whether combined expression of Stat3 and C/EBPb in NSCs is sufficient to initiate mesenchymal gene expression and to trigger the mesenchymal properties that characterize high-grade gliomas. To do this, we used an early passage of the stable, clonal population of mouse NSCs known as C17.2 because its enhanced yet constitutively self-regulated expression of stemness genes permits its cells to be efficiently grown as undifferentiated monolayers in sufficiently large, homogeneous and viable quantities to ensure reproducible patterns of self-renewal and differentiation without ever behaving in a tumorigenic fashion in vitro or in vivo{Lee, 2007 #1332; Park, 2006 #1331; Parker, 2005 #1281}. Following ectopic expression of C/EBPb and a constitutively active form of Stat3 (Stat3C, Supplementary Fig. 1){Bromberg, 1999 #1312} in NSCs, we observed dramatic morphologic changes, consistent with loss of ability to differentiate along the neuronal lineage (Fig. 4a). Parental and vector-transfected NSCs have the classical spindle-shaped morphology that is associated with the neural stem/progenitor cell phenotype. When grown in the absence of mitogens, these cells display efficient neuronal differentiation characterized by extensive formation of a neuritic network (Fig. 4a, top-right panel). Conversely, expression of C/EBPb and Stat3C leads to cellular flattening and manifestation of a fibroblast-like morphology. Remarkably, depletion of mitogens resulted in additional flattening with complete loss of every neuronal trait

12. In Vivo Validation Mouse Survival Data Human Survival Data
Control Vector
Stat3-
C/EBPb-
Stat3-/C/EBPb-
Human Survival Data
Mouse immunohistochemistry
Carro, M. et al. (2010). Nature 463(7279):

13. TCGA dataset on different networks
Phillips (2)
Sun (3)
TCGA (1) 22 10 5 6 8 4
1.9x10-7
1Cancer Genome Atlas Research Network, Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature Oct 23;455(7216):1061-8
2 Phillips, H.S., et al., Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. Cancer Cell, (3): p
3 Sun, L., et al., Neuronal and glioma-derived stem cell factor induces angiogenesis within the brain. Cancer Cell, (4): p `
Example of MaRINA results for inference of MR of the mesenchymal phenotype on the TCGA dataset by using three different interactomes: (1) TCGA, (2) Phillips, and (3) Sun. The Venn diagram shows a very significant overlap between the MR inferred from each interactome, showing that the three interactomes are functionally similar. The heatmap on the right shows the differential expression (first column) and the differential activity (NES, columns labelled as 1-3) inferred from each interactome for the MRs commonly identified on the three interactomes (top red and blue blocks), or identified only by two interactomes (bottom red and blue blocks).
Distinct Programs with significant overlap across distinct datasets

14. The Bottleneck Hypothesis
Glioblastoma:
Carro MS et al. The transcriptional network for mesenchymal transformation of brain tumours. Nature Jan 21;463(7279):
Master Regulators: C/EBP + Stat3
Diffuse Large B Cell Lymphoma:
Compagno M et al. Mutations of multiple genes cause deregulation of NF-kappaB in diffuse large B-cell lymphoma. Nature Jun 4;459(7247):717-21
Master Regulator: Nf-kB pathway
GC-Resistance in T-ALL:
Real PJ et al. Gamma-secretase inhibitors reverse glucocorticoid resistance in T cell acute lymphoblastic leukemia. Nat Med Jan;15(1):50-8.
Master Regulator: NOTCH1 pathway G1 G2 G3 G4 X Y Z W V
Master
Regulator
Module(s) G5 G6 G7 G8
Therapeutic
Targets G9
G10
G11
G12
Patient X
= EGFR
= PDGFRA
= p16
= p53
= PTEN
= MDM2
= MDM4
= MYC
= NF1
= ERBB2
= RB1
= CDK4 G1 G2 G3 G4 G5 G6 G7 G8 G9
G10
G11
G12
Molecular
Phenotype
Disease
Stratification
Biomarkers …
E.g. GBM subtypes

15. Inhibitors of C/EBP Activity
(c) MINDy Analysis
(a) Protein Binding Assays M2 M1 Mn M2 M1 Mn
C/EBP
Comp1
Comp1
STAT3
Compn
MGES
Compn
Collaboration with:
S. Schreiber (a)
B. Stockwell (b)
A. Iavarone and A. Lasorella (a, b, c)
(b) High Throughput Screening

16. MINDy Modulators of miRNA activity
mature
miRNA
miR
target
Mod
Sumazin et al. Cell, 2011 Oct 14;147(2):

17. Novel miR-program mediated regulatory network
Analysis of TCGA data for GBM and Ovarian Cancer
including matched gene and miRNA expression profiles for 422 and 587 samples
Modulation of miRNA activity on targets
7,000 Sponge modulators, participating in 248,000 miR-mediated mRNA-mRNA interactions
148 Non-sponge modulators affecting more than 100 miRs (using only experimentally validated miRs targets)
17/430 are RNA-binding proteins or a component of the spliceosome

18. 13 miR-mediated regulators of PTEN
G14 – G 563 G1 G2 G3
G13
564 node, 111 core sub-graph
PTEN G4
G11 G5
G10 G9 G6 G8 G7
p < 5 x 10-23
p < 2 x 10-10

19. 13 miR-mediated regulators of PTEN

20. A B C D Days Days Days Days
PTEN over expression and silencing effects on SNB19 cell growth rate
Silencing PTEN mPR regulators affects SNB19 cell growth rate
Proliferation fold change
Proliferation fold change
Days
Days C D
PTEN over expression and silencing effects on SF188 cell growth rate
Silencing PTEN mPR regulators affects SF188 cell growth rate
Proliferation fold change
Proliferation fold change
Days
Days

21. Targeted Drug Development
Molecular Target(s)
Hit compound
CTD2
Network
Broad Inst.
S.Schreiber
CSHL S.Power
Columbia
A.Califano
Dana Farber
B.Hahn
UT South-western
M.Reich
Clinical Trials
Biomarkers:
Response/Efficacy
Compound Mechanism
of Action

22. Conclusions and Reflections
Current emphasis on genes harboring genetic and epigenetic alterations may not be sufficient
We should also focus on Master Regulator and Master Integrator genes
Current approach to biomarker discovery should be re-evaluated in a molecular interaction network context.
It is not the genes/proteins that change the most but rather those that change most consistently. (mRNA is not informative)
From GWAS (Genome-Wide Association Studies) to NBAS (Network-Based Association Studies)
Califano A, Butte A, Friend S, Ideker T, and Schadt EE, Integrative Network-based Association Studies: Leveraging cell regulatory models in the post-GWAS era, Nat. Genetics, in press. Accessible in Nature Preceedings:
One disease – One target – One drug Multi-target combinations
Optimal combination of drugs selected from a repertoire of safe, target-specific compounds using predictive tools.
Identification of genetic dependencies (addictions) from Ex Vivo Models
Identification of candidate therapeutic agents from In Vitro mechanistic models.

23. Acknowledgements Funding Sources: NCI, NIAID, NIH Roadmap
Califano Lab (Computational)
Mukesh Bansal, Ph.D.
Archana Iyer, Ph.D.
Celine Lefebvre, Ph.D.
Yishai Shimoni, Ph.D.
Maria Rodriguez-Martinez, Ph.D.
Antonina Mitrofanova, Ph.D.
Jose’ Morales, Ph.D.
Paola Nicoletti, Ph.D.
Pavel Sumazin, Ph.D.
Gonzalo Lopez, Ph.D.
James Chen, (GRA)
Hua-Sheng Chu (GRA)
Wei-Jen Chung (GRA)
In Sock Jang (GRA)
William Shin (GRA)
Jiyang Yu (GRA)
Alex Lachman (GRA)
Pradeep Bandaru M.A.
Manjunath Kustagi (Programmer)
Software Development
Aris Floratos, Ph.D. (Exec. Director)
Ken Smith, Ph.D.
Min Yu, (Programmer)
Califano Lab Experimental
Gabrielle Rieckhof, Ph.D. (Exec Director)
Mariano Alvarez, Ph.D.
Brygida Bisikirska, Ph.D.
Xuerui Yang, Ph.D.
Yao Shen, Ph.D.
Presha Rajbhandari, M.A. (Sr. Res. Worker)
Jorida Coku, M.A. (Staff Associate)
Hesed Kim, (Staff Associate)
Sergey Pampou, Ph.D.
A. Iavarone & A. Lasorella (CU)
Maria Stella Carro
K. Aldape (MD Anderson)
R. Dalla Favera (CUMC)
Katia Basso
Ulf Klein
R. Chaganti (MSKCC)
M. White & J. Minna (UTSW)
J. Silva (CU)
C. Abate-Shen & M. Shen (CU)
D. Felsher (Stanford)
Funding Sources: NCI, NIAID, NIH Roadmap