Causal Mechanistic Regulatory Network for Glioblastoma Deciphered Using Systems Genetics Network Analysis

Christopher L Plaisier,Sofie O’Brien,Brady Bernard,Sheila Reynolds,Zac Simon,Chad M Toledo,Yu Ding,David J Reiss,Patrick J Paddison,Nitin S Baliga

doi:10.1016/j.cels.2016.06.006

Abstract

We developed the transcription factor (TF)-target gene database and the Systems Genetics Network Analysis (SYGNAL) pipeline to decipher transcriptional regulatory networks from multi-omic and clinical patient data, and we applied these tools to 422 patients with glioblastoma multiforme (GBM). The resulting gbmSYGNAL network predicted 112 somatically mutated genes or pathways that act through 74 TFs and 37 microRNAs (miRNAs) (67 not previously associated with GBM) to dysregulate 237 distinct co-regulated gene modules associated with patient survival or oncogenic processes. The regulatory predictions were associated to cancer phenotypes using CRISPR-Cas9 and small RNA perturbation studies and also demonstrated GBM specificity. Two pairwise combinations (ETV6-NFKB1 and romidepsin-miR-486-3p) predicted by the gbmSYGNAL network had synergistic anti-proliferative effects. Finally, the network revealed that mutations in NF1 and PIK3CA modulate IRF1-mediated regulation of MHC class I antigen processing and presentation genes to increase tumor lymphocyte infiltration and worsen prognosis. Importantly, SYGNAL is widely applicable for integrating genomic and transcriptomic measurements from other human cohorts.

Highlights

Glioblastoma multiforme (GBM) is the most common brain tumor and is nearly uniformly fatal
The transcription factor (TF)-to-target gene interactions were discovered by intersecting the locations of 2,331 unique DNA recognition motifs for 690 TFs across the human genome (Matys et al, 2006; Newburger and Bulyk, 2009; Jolma et al, 2013; Mathelier et al, 2014) and Encyclopedia of DNA Elements (ENCODE)-determined 8.4 million genomic sites with digital genomic footprints (DGFs) across 41 diverse cell and tissue types (Neph et al, 2012)
A DGF is experimental evidence that a DNA-binding protein was bound to a genomic location, and, when coincident with a motif instance, it suggests an interaction of a specific TF with that genomic location (Figure 1A)

Summary

Introduction

Glioblastoma multiforme (GBM) is the most common brain tumor and is nearly uniformly fatal. One possible strategy to achieve complete and durable remission is to tailor a combination of drugs that targets multiple vulnerabilities in a patient’s tumor. We hypothesized that knowledge of the detailed architecture of transcription factor (TF) and microRNA (miRNA) regulatory interactions in the form of a transcriptional regulatory network (TRN) would provide the mechanistic details required to prioritize combinatorial interventions. Both TFs (Cai et al, 1996) and, more recently, miRNAs (Bouchie, 2013) have been used as therapeutic targets. Therapies targeting TFs and miRNAs have the potential for a broader effect than those targeting a single gene, as these regulators control many genes associated with diverse oncogenic biological processes

Results

Discussion

Conclusion