OS01.1 Identification of the pre-malignant cell of origin in Human Glioblastoma Multiforme

Abstract

AbstractOne of the major difficulties in identifying the cell of origin for glioblastoma is the complex cellular composition of this disease. It is now hypothesized that the different subtypes of high-grade glioma may have different cell of origin and in mouse models it has been shown that the cell of origin resides within the progenitor populations. To identify the cell of origin in human GBM we decided to first dissect out the various stages of CNS stem cell differentiation in fetal and adult human brain. After a functional high throughput screen, we identified a panel of four antibodies (i.e CD15, Notch1, EGFR, and CD90) that in various combinations can separate long-term self-renewing multi-potent neural stem cells (NSCs: LinCD15Notch1EGFRCD90) from multi-potent progenitors with limited self-renewal capacity (NPC: LinCD15Notch1EGFRCD90) from non-neoplastic human fetal and adult Brain (sub-ventricular zone). FACS sorted cells were analyzed for single cell lineage multi-potentiality, lineage bias and self-renewal to trace and characterize their lineage relationships. In surgical human glioblastoma samples, however we observed in-vivo tumor initiating and in-vitro tumorsphere-forming frequency was predominant in the progenitor cells and not in the neural stem cells. Interestingly we did observe a small frequency of NSC cells forming distinct lesions in the brain, which on further analysis showed the presence of both the neural stem cells and Glioma stem cells, where as the tumor initiating enriched population (GSC: LinCD15Notch1EGFRCD90) did not show the presence of neural stem cells but only tumor initiating NPCs. limiting dilution transplantation and serial transplantation confirmed the LinCD15Notch1EGFRCD90 to have the highest frequency of self renewing cells compared to other populations. Taken together our data suggests the presence of “normal” or pre-malignant stem cells in glioblastoma, which suggests a possible mechanism of relapse. Using Transcriptome analysis and targeted sequencing of known mutations we identify the signal transduction and genomic events (PTEN point mutations vs Deletions) which distinguish pre-malignant neural stem cells from the tumor initiating glioma stem cells. We further show a significant enrichment of the protein translational machinery as cells transition from pre-malignant to tumor initiating cells. These pre-malignant cells comprise a cellular reservoir that needs to be targeted to prevent tumor relapse. SSM: Seibel Fellowship and Ludwig Cancer Center, ILW: National Institutes of Health, Ludwig Cancer Center, SHC: Stanford Child Health Research Institute, St Baldrick’s Foundation and Ludwig Cancer Center