Sten Linnarsson

Recent advances in single-cell biology have enabled very large-scale mapping of cell types and cell states. Emerging brain cell atlases now provide unprecedented access to the molecular composition of normal cell types, including gene expression and regulatory chromatin modifications. In parallel, efforts to map human brain tumors are underway in many labs, leading to an emerging conceptual framework where tumors are viewed as analogous to developmental processes. However, previous work in glioblastoma has largely neglected inter- and intra-tumor heterogeneity, the spatial dimension of tumor heterogeneity, and the dynamic aspect of interconversion between cellular states. Here, we propose to focus intensely on these aspects, and ask a number of related questions about tumor growth. For example, which subpopulations drive tumor growth? What is the rate of cell division in different subpopulations? Are there quiescent stem cells that can cause relapse? What is the lineage relationship between tumor subpopulations, and how do they relate to normal brain development? Can phenotypic subpopulations interconvert freely or do they form a hierarchy, or are they isolated independent clones? What is the relationship between tumor cell phenotype and local environment, e.g. proxmity to vessels or white matter? What is the phenotype of distant, disseminating tumor cells, versus those of the tumor mass? What is the variability between tumors from different patients?