PP14. NETWORK ANALYSIS REVEALS INSIGHTS INTO H3F3A BIOLOGY IN PEDIATRIC GLIOBLASTOMA

Abstract

AbstractBACKGROUND: H3 histone, family 3A (H3F3A) encodes H3.3 histone protein monomers, which plays a central role in chromatin structure and gene regulation. Recent studies have reported highly recurrent mutations affecting the N-terminal tail of H3.3 and causing amino acid substitution of lysine 27 to methionine (K27M) and of glycine 34 to arginine or valine (G34A or G34V) in pediatric glioblastoma (GBM), which in turns linked to two subgroups with distinct clinical and pathological features. However, effect of H3F3A mutations on other closely related genes is not yet fully explored. Thus, we performed a network analysis to gain insights into the biology of H3F3A in pediatric GBM. METHODS: We constructed H3F3A network by extracting gene products that directly interact with H3F3A from protein-protein interaction (PPI) databases. To examine the network wiring around H3F3A, we combined pediatric GBM gene expression data from three independent studies and computed the correlation between H3F3A and all interacting genes for each mutation status (K27M or G34) and H3F3A wild type (WT). We then identified network of differentially co-expressed gene pairs that either gained or lost correlation in mutation relative to the WT tumours. RESULTS: Through mining of PPI networks, 106 gene products were identified to interact with H3F3A. Many of these genes were known to be involved in chromatin modification (47 genes; FDR = 2.17E-38), DNA metabolic process (34 genes; FDR = 3.23E-16), DNA repair (22 genes; FDR = 8.18E-12), and histone methylation (13 genes; FDR = 2.75E-11). H3F3A network was largely perturbed by K27M mutation than G34 (44 interactions changed in K27M and 24 in G34 when compared with WT, FDR < 0.05). Of these interactions, only four were perturbed by both mutations: H3F3A - ASB9, H3F3A - CBX3, H3F3A - HIST2H2BE, and H3F3A - PDGFRA, suggesting distinct mechanisms by which H3F3A mutations might be inducing transcriptional programming in pediatric GBM. A relatively small number of H3F3A interactions were identified between IDH1 mutation and WT tumours from adult GBM (7 interactions from K27M and 1 from G34) and in pediatric medulloblastomas (2 interactions from K27M and 1 from G34), demonstrating the robustness and specificity of identified interactions in pediatric GBM. CONCLUSION: In conclusion, the network analysis represents a new approach for the prioritization of H3F3A interacting partners in pediatric GBM, providing new insights into network dependencies in pediatric GBM subgroups and potentially enabling the identification of new diagnostic and therapeutic targets.