Deciphering inosinome in glioblastoma versus normal cortex

Cancer is driven by alterations of the genomic information, which carries mutations in key genes providing selective advantage for clonal multiplication of cancer cells. However, mutations within DNA are not the only source for cell alteration. RNA molecules are targets of a series of post-transcriptional modifications, such as splicing and RNA editing, that can affect sequence, structure and stability. The most common type of RNA editing in humans converts Adenosine in RNA targets into Inosine (A-to-I) and is catalyzed by two adenosine deaminases that act on dsRNA (ADARs) family of enzymes (ADAR and ADARB1). Inosines are subsequently interpreted as guanosines by several cellular proteins and could ultimately lead to a genomic mutations (A-to-I/G). At present, it has been estimated that over 4 millions editing sites exist in our transcriptome involving coding and non-coding RNAs. These huge amounts of Inosine at RNA level are necessary for our survival and their levels is highly regulated in different tissues and during development. Considering the importance of ADAR activity in our cells we believe that if ADAR are not well regulated they may contribute to cancer on set and/or progression. The advent of high-throughput RNA sequencing has enabled identification of RNA editing sites and global analyses of cancer transcriptomes demonstrate that ADAR-mediated RNA editing dynamically contributes to genetic alterations in cancer, including high-grade gliomas. Glioblastoma (GBM) is one of the most common and aggressive primary brain tumor in humans and despite advances in understanding the molecular mechanisms underlying these tumors, current treatments are ineffective. In order to elucidate the glioma-specific RNA editing signature, we analyzed 146 RNA-Seq of primary glioblastomas from the TCGA dataset compared to 132 normal brain cortex RNA-Seq from the GTEx database and purified pools of normal cortex astrocytes. A-to-I editing events has been detected using a collection of more than 4 million annotated edited substrates and the REDItools suite of python scripts with stringent filters. By means of the Cuffquant/Cuffdiff tools, we have also compared global transcriptome profiles and ADAR genes expression patterns at gene level isoform level and using IHC at protein level. We found that a general down regulation of editing events at both recoding and non -recoding (Alus) sites that correlate with a down expression of ADARB1 enzyme. No differences were observed with ADAR expression. Overall, we observed a strong editing landscape perturbation in glioblastoma that could be important for identifying the most effective target genes for possible therapeutic intervention.