Single cell RNA sequencing analysis of GFAP alternative splice isoforms in glioblastoma

Abstract

Glioblastoma multiforme is a very aggressive brain tumor with extremely poor patient prognosis. Clinical treatment fails to fully treat glioblastoma due to its intratumoral heterogeneity and population of quiescent stem cells propagating tumor recurrence. The intermediate filament protein GFAP is highly expressed in brain matter. The ratio of two prominent splice isoforms of GFAP, α and δ, are implicated in glioma tumor grade and migration. In this study, the ratio of these two isoforms was investigated in glioblastoma using published single cell RNA SMART-seq data. Two very large datasets were investigated in this study, one with grade III and IV gliomas consisting of 14,517 cells and 41 tumors. The second dataset investigated was on pediatric ependymoma with 5,900 cells. To analyze isoform level information, the datasets were mapped with the pseudoalignment program Kallisto and then analyzed with the Seurat single cell analysis package. GFAP expression was found to be highly heterogenous between tumors and cell clusters, however on average there was no difference between ratio of GFAPδ/α between glioma grade. This led to the suggestion that cells are characterized based on GFAP isoform expression profile. When comparing dividing and nondividing cells, GFAP was found to be highly expressed in neural quiescent stem cells compared to actively dividing cells. However, invasive score was not found to be correlated with GFAPα and δ expression or ratio. Through the use of a bioinformatics pipeline a large glioma dataset was analyzed for intermediate filament expression and scored for invasion. This dataset can be further investigated for other intermediate filaments and any isoform switches between grade III to IV glioma. This analysis is highly beneficial to discern the relationship between isoform switches and the heterogeneity of glioblastoma multiforme at the single cell level.