Intratumoral heterogeneity in glioblastoma: subtype transition and cell-to-cell communication

Abstract: Glioblastoma is a most aggressive brain tumor with a median overall survival of less than two years. The current treatment is surgery, chemotherapy, and radiotherapy. However, glioblastoma is difficult to treat and recurs some time after treatment. Several reports have shown that intratumoral heterogeneity is a hallmark of glioblastoma, and it is often said to be a factor that contributes to tumor progression and tumor relapse. Based on the single-cell gene expression profiling, glioblastoma is found to be composed of tumor cells with three different subtype signatures namely proneural, classical, and mesenchymal. These subtypes may transition between each other, which may be affected, and/or potentially controlled, by genetic and epigenetic modifications. Even microenvironmental events and communication between different subtype cells within the same tumor may play a role. Studies in this thesis focused on identifying regulators of subtype transitions and their involvement in cell-to-cell communication in individual glioblastomas. Study I used several analytical approaches including connectivity map analysis, overexpression screen, RNA-sequencing, and gene expression profiling. SOX2 was identified to have the capacity to transition mesenchymal gene expression subtype glioblastoma cells back to a non-mesenchymal signature. Subsequently, SFRP2 was identified as a SOX2 antagonist able to induce a mesenchymal transition. In line with this, SFRP2 was found expressed at a higher level in mesenchymal as compared to proneural and classical TCGA glioblastoma samples. In vitro, SFRP2 decreased tumor sphere formation and cell proliferation, and increased cell invasion capacity and sensitivity towards apoptotic stimuli. Spatial gene expression analysis found SFRP2- correlated genes to be expressed in a higher level in vascularized areas in glioblastoma, as opposed to SOX2-correlated genes that were highly expressed in cellular tumor regions. Moreover, conditioned media from SFRP2 transitioned cells induced more MRC1 (encodes CD206) expression in human monocytes in vitro. Collectively, these experiments identified SFPR2 as a SOX2 antagonist and inducer of mesenchymal transition in glioblastoma. In Study II PROX1 was found expressed at a lower level in glioblastoma as compared to lowgrade gliomas. Tumors with lower PROX1 correlated with a mesenchymal subtype and patients exhibited shorter survival. Suppression of PROX1 in glioblastoma cell lines with high PROX1 levels induced transition to a mesenchymal glioblastoma subtype signature. Conversely, overexpression of PROX1 transitioned cells to a non-mesenchymal subtype signature. In co-occurrence with these transitions, PROX1 functionally impacted on cell proliferation and levels of several cell cycle proteins. SOX2 was identified as an upstream signaling component of PROX1, and thus SOX2 and PROX1 levels decreased upon treatment with a CDK inhibitor. By co-immunoprecipitation experiments PROX1 was found to interact with THRAP3 in the nucleus. Depletion of THRAP3 increased PROX1 expression and protein stability. Based on previous reports on THRAP3, these findings suggest THRAP3 to be involved in the transcriptional regulation of PROX1. These results underscore a functional role of PROX1 in glioblastoma development and with implications for survival outcome. In study III the well-known U-343 cell line system was used as a model system to study intratumoral heterogeneity and cell-to-cell communication. The four cell lines U-343 MG, U343 MGa, U-343 MGa 31L and U-343 MGa Cl2:6 are derived from the same glioblastoma, and display different phenotypes including marker expression, cells shape, and proliferation. Gene expression and DNA copy number analyses suggested their shared derivations from a common ancestor in a tumor evolutionary relationship. Temozolomide (TMZ) sensitivity of the individual cell lines was determined. In a cell culture composed of a mixture of all four cell lines, the one with the lowest sensitivity outlived the others during TMZ treatment, which modeled the appearance of drug resistance. Finally, co-culture and conditioned media experiments revealed complex interactions between the U-343 cell lines through signaling by cell-to-cell contact or secreted proteins. These findings provide a model for research on intratumoral heterogeneity, both with regard to appearance of drug resistance and to interclonal interactions. Collectively, these studies provide insights into the complexity of intratumoral heterogeneity, including subtype transitions during tumor progression and interclonal communication. Intratumoral heterogeneity is thus important to consider when designing new therapeutic strategies and may also in itself pose novel therapeutic targets.