Background: Diffuse midline gliomas (DMGs) are among the most aggressive pediatric brain tumors and remain one of the biggest therapeutic challenges in paediatric oncology, with a median survival of 9–11 months despite decades of clinical trials. In 2021, WHO reclassified these tumors according to their mutational status: among them, H3K27-altered DMGs account for ~80% of DMGs. K27M mutations in histone H3 leads to an overall decrease in H3 methylation and trigger distinct epigenetic reprogramming, contributing to patient heterogeneity and variable clinical outcomes. Therefore, understanding epigenetic dysregulation and identifying epigenetic driver genes (epidrivers) are essential to improving early risk stratification, prognosis, and prevention of treatment resistance in children.

Objectives: This project aims to decode epigenetic heterogeneity driving DMGs to further understand the etiology of the disease and characterize potential biomarkers of aggressiveness.

Methods: We performed an integrative analysis of publicly available single-cell omics datasets (scRNA-seq: 28 tumors; scATAC-seq: 7 tumors; scMultiome: 9 tumors). Following quality control, malignant cells were identified using Numbat-inferred copy number alterations integrated with graph-based clustering. Cell populations were annotated using SingleR with reference datasets and complemented by manual curation using established cell-type markers. Cluster-specific gene markers were identified through differential analysis with multiple-testing correction, enabling the detection of the expression profile of epigenetic regulator genes (ERGs) in each cell population. Based on this in silico approach, we used patient-derived H3K27-altered DMG cell lines and CRISPR-Cas9 genome editing approaches to target candidate ERGs for further validation of their role. More precisely, we are using 3D organoid models to reproduce tumor heterogeneity and microenvironmental interactions and investigate the phenotypic and functional impact of those ERG alterations (collaboration with Dr Cosset, CRCL, Lyon, France). Finally, expression of identified epidrivers will be analyzed in patient tumor samples (n=45) by immunostaining and spatial transcriptomics (Pr. Natacha Entz-Werle from the Pediatric Onco-Hematology Department of University Hospital of Strasbourg, France).

Results: Single-cell analyses revealed five major malignant cell populations (astrocytic-like, mesenchymal-like, oligodendrocytic-like, oligodendrocyte precursor cell–like (OPC-like), and proliferating OPC-like cells) each characterized by distinct ERGs expression displaying consistent up- or downregulation. These ERGs may serve as cell-type markers and their loss or gain could represent epidriver events underlying the emergence of aggressive subpopulations. Among them, we identified several histone modifiers regulating histone acetylation that are highly expressed during childhood, notably in the brain where they regulate metabolism to induce proliferation and differentiation of glial cells, therefore providing essential support for nervous system function. Using two patients-derived cell lines, we showed that CRISPR-Cas9-mediated knockdown of these ERGs leads to a decrease in cell proliferation and viability, supporting their relevance in tumor development. The further role of these promising targets will be assessed by phenotypical and functional analysis, focusing on their roles in epigenetic remodelling and in patient spatial transcriptomic data.

Conclusion: Overall, this integrative in silico/in vitro approach combining several partners and multi-omics analyses, will enhance our understanding of pediatric DMGs and enable the identification of prognostic biomarkers.