Logo: to the web site of Uppsala University

The adult brain tumor glioblastoma is characterized by extensive heterogeneity and invasion, with intertumoral and intratumoral variation. This heterogeneity is partly driven by cellular plasticity, which permits glioblastoma cells to transition between cellular states. The plastic cells invade the brain, commonly via perivascular spaces, and diffuse invasion into the white matter, termed invasion routes. In this thesis, we investigate whether cellular states influence invasion routes using patient-derived models and develop new tools for real-time monitoring of glioblastoma across models. 

In Paper I, we connect cellular states to invasion routes by characterizing six patient-derived cell culture and xenograft models using single-cell RNA profiling and spatial proteomics. We connect bulk-forming capacity and perivascular invasion to MES-like and OPC-like cells, driven by ANXA1. NPC-like and AC-like cells, we link to diffuse invasion, driven by HOPX and RFX4. Perturbation of these genes shifts cellular state composition and invasion routes, suggesting that cellular state shapes invasion.

To directly monitor cellular states, paper II introduces the CRISPR-tag, which we use to fluorescently label genes representative of cellular states in patient-derived cells. In vitro, we observe differences and oscillations in protein levels. Ex vivo, we monitor CRISPR-tagged cells and detect spatially-dependent expression of cellular-state markers. During differentiation treatment, SOX2 expression remains high outside the tumor core, whereas it is lost in the central tumor regions. ANXA1-expressing cells display higher expression closer to a vessel. We performed a whole-genome knock-out screen to identify genetic dependencies that increase or decrease ANXA1 expression and identified several candidate genes that regulate ANXA1 expression.

Finally, in Paper III, we characterize eleven patient-derived cell cultures in a zebrafish model and monitor tumor initiation and growth using AI. The freely swimming fish are automatically imaged every four to six hours and display heterogeneity in growth and survival. Further characterization using light-sheet imaging revealed intratumoral variability in bulk-forming ability, tumor spread, and the presence of cells near vessels.

Together, our findings suggest an association between cellular states and invasion routes and introduce new tools to monitor cellular plasticity and tumor growth.