127. Characterization of alternative transcription start and termination sites in glioblastoma

Aly Abdelkareem

Varsha Thoppey Manoharan

Varsha Thoppey Manoharan is currently a PhD candidate specializing in bioinformatics at the University of Calgary studying under Dr. Sorana Morrissy. She holds a bachelor’s degree in Pharmaceutical Technology from Anna University, Chennai, India with emphasis on subjects like Pharmacology and Molecular Biology. Driven by her research interests in cancer biology and immunotherapy, she pursues her doctoral degree, with a thesis focused on investigating the landscape of alternative transcription processing as a broadened target discovery space in glioblastoma (GBM), a lethal brain tumor in adults. With a data-driven approach to delineate tumor heterogeneity, she aims to contribute novel insights on GBM transcriptome and explore its potential for clinical translation.


Varsha Thoppey Manoharan, Theodore Verhey, Sorana Morrissy

University of Calgary, Calgary, Alberta, CA

Glioblastoma (GBM) is a brain tumor characterized by therapeutic resistance and inevitable fatal relapse. GBM presents a major clinical challenge due to the co-existence of diverse and plastic cellular states and the role of non-neoplastic cells in promoting tumor growth and invasion (tumor microenvironment; TME). While gene expression has been widely applied to profile GBM, our understanding of alternative transcription control significantly lags behind. The role of alternative transcription start site (aTSS) and termination site (aTES) usage is now recognized as key to normal brain development and establishment of cell-type and cell-state diversity. This regulation is altered in cancer, affecting post-transcriptional regulation of mRNA transcripts and resulting proteins. We undertake an isoform-centric study focusing on aTSS/aTES usage patterns in GBM, aiming to identify tumor-specific isoform alterations. To achieve this we develop an unsupervised approach to classify genes based on their UTR profiles (WGCNA) in bulk RNAseq data and use these profiles to classify samples (NMF). Our approach reveals robust delineation between normal brain and GBM samples and further stratifies patients into subtypes distinguished by variable contribution of NMF metagenes. These represent (1) isoforms differentially expressed between different tumor cell states, and (2) between the tumor and the TME. We validate our results using single-cell analyses of aTSS/aTES from scATAC-seq and scRNA-seq data, and when available, matched proteomics data. Our approach enables unsupervised discovery of alternative UTR usage in cancer vs normal tissues, identifies isoform diversity within GBM and in the TME, and could lead to novel targets for therapy.