Leveraging multi-omics data to infer regulators of mRNA 3-prime end processing in glioblastoma

Alterations in mRNA 3-prime end processing and polyadenylation are widely implicated in the biology of many cancer types, including glioblastoma (GBM), one the most aggressive tumor types. Although several RNA-binding proteins (RBPs) responsible for alternative polyadenylation (APA) were identified from functional studies in cell lines, their contribution to the APA landscape in tumors in vivo was not thoroughly addressed. In this study we analyzed a large RNA-seq data set of glioblastoma (GBM) patient samples from The Cancer Genome Atlas (TCGA) to identify APA patterns differentiating the main molecular subtypes of GBM. We superimposed these to RBP footprinting data and to APA events occurring upon depletion of individual RBPs from a large panel tested by the ENCODE Consortium. Our analysis revealed 32 highly concordant and statistically significant RBP-APA associations, whereby changes in RBP expression were accompanied by APA in both TCGA and ENCODE datasets. Among these, we found a previously unknown PTBP1-regulated APA event in the SC5D gene and an HNRNPU-regulated event in the PRRC2B gene. Both of these were further supported by RNA-sequencing data of paired tumor center-periphery GBM samples obtained at the University Hospital of Basel. The transcriptome analysis workflow that we present here enables the identification of concordant RBP-APA associations in cancers.