ESTRO 2024 - Abstract Book

S5213

Radiobiology - Microenvironment

ESTRO 2024

1468

Proffered Paper

Hypoxia drives spatially distinct epigenetic and transcriptional cell states in Glioblastoma

Bradly G Wouters, Phoebe Lombard, Ronald Wu, Mark Zaidi, Gelareh Zadeh, Sheila Mansouri, Anna Dvorkin-Gheva

Princess Margaret Cancer Centre, University Health Network, Toronto, Canada

Purpose/Objective:

Glioblastoma (GBM) is a highly aggressive disease with poor overall survival rates that have not changed substantially in decades. One of the key features that contributes to treatment resistance is the morphological and phenotypic diversity in this disease. Phenotypic diversity was first recognized across patients in analyses from The Cancer Genome Atlas (TCGA) project, which identified transcriptional tumour subtypes using bulk RNA profiling techniques. More recently, the advent of single-cell sequencing has demonstrated that these bulk subtypes approximate a much greater intratumour diversity made up of various cellular states coexisting at the single-cell level. In this study we have investigated the role of hypoxia in driving cellular phenotype and cell state in GBM patients.

Material/Methods:

We have assessed the influence of hypoxia on cellular state in GBM patients administered pimonidazole prior to surgery as part of a clinical trial at Princess Margaret. We carried out spatial transcriptomics on patient samples and constructed a pimonidazole-based hypoxia signature for GBM to define the spatial localization of hypoxia in the spatial transcriptomic data sets. We then compared hypoxic spatial regions with previously defined cell states in GBM. Spatial transcriptomic data from Princess Margaret Patients was combined with publicly available data to create the largest global dataset in GBM and this set was used to discover novel phenotypic cell states and their association with hypoxic regions across all samples. Finally, we investigated the underlying molecular mechanisms through which hypoxia may influence cell state using patient derived glioma stem cells (GSCs) cultured in vitro under normoxic and hypoxic conditions. Using global epigenetic profiling technologies, we identified specific chromatin regions and genes regulated by hypoxia and mapped gene expression back to spatial transcriptomic data. Our data demonstrate a remarkable spatial overlap between hypoxia and cell phenotypic state across patient samples. Hypoxia is strongly associated with the mesenchymal cell state and is absent from more differentiated cell states. Epigenetic profiling of chromatin state in GBM stem cells, as assessed by ATAC-seq, and CHIP-seq for activating (H3K4me3) and inactivating (H3K27me3) marks identified specific regions of the genome that are epigenetically modified by hypoxia. Chromatin areas that open during hypoxia and show a corresponding increase in gene expression are enriched for known HIF transcription factors. In contrast, chromatin regions that close and show a corresponding decrease in gene expression are enriched in binding sites for neurodevelopmental transcription factors and genes involved in stem cell maintenance and differentiation. A GSC-derived hypoxia gene signature obtained from integrating chromatin and transcriptional changes was created and used to explore the spatial localization of these genes in GBM patients. We confirm that these same genes are epigenetically regulated in patient samples and demonstrate that they are spatially associated with defined cell states in the microenvironment. Furthermore, this signature is prognostic in data from TCGA. Results: