Slow-Cycling Cells in Glioblastoma: A Specific Population in the Cellular Mosaic of Cancer Stem Cells

SIMPLE SUMMARY: A major challenge in successfully managing glioblastoma is that we do not understand the types and dynamic behaviors of the cells that constitute these tumors. Combining bioinformatics and functional studies, we describe the presence of multiple independent lineages of cancer stem cells driving the heterogeneic nature of glioblastoma. Our results help us decode and map the transcriptional and functional diversity of glioblastoma cells. By revealing potential mechanisms underlying tumor resilience, the root of resistance to treatment, our study may inform novel strategies to develop precision and effective therapies to treat brain cancer. ABSTRACT: Glioblastoma (GBM) exhibits populations of cells that drive tumorigenesis, treatment resistance, and disease progression. Cells with such properties have been described to express specific surface and intracellular markers or exhibit specific functional states, including being slow-cycling or quiescent with the ability to generate proliferative progenies. In GBM, each of these cellular fractions was shown to harbor cardinal features of cancer stem cells (CSCs). In this study, we focus on the comparison of these cells and present evidence of great phenotypic and functional heterogeneity in brain cancer cell populations with stemness properties, especially between slow-cycling cells (SCCs) and cells phenotypically defined based on the expression of markers commonly used to enrich for CSCs. Here, we present an integrative analysis of the heterogeneity present in GBM cancer stem cell populations using a combination of approaches including flow cytometry, bulk RNA sequencing, and single cell transcriptomics completed with functional assays. We demonstrated that SCCs exhibit a diverse range of expression levels of canonical CSC markers. Importantly, the property of being slow-cycling and the expression of these markers were not mutually inclusive. We interrogated a single-cell RNA sequencing dataset and defined a group of cells as SCCs based on the highest score of a specific metabolic signature. Multiple CSC groups were determined based on the highest expression level of CD133, SOX2, PTPRZ1, ITGB8, or CD44. Each group, composed of 22 cells, showed limited cellular overlap, with SCCs representing a unique population with none of the 22 cells being included in the other groups. We also found transcriptomic distinctions between populations, which correlated with clinicopathological features of GBM. Patients with strong SCC signature score were associated with shorter survival and clustered within the mesenchymal molecular subtype. Cellular diversity amongst these populations was also demonstrated functionally, as illustrated by the heterogenous response to the chemotherapeutic agent temozolomide. In conclusion, our study supports the cancer stem cell mosaicism model, with slow-cycling cells representing critical elements harboring key features of disseminating cells.