Investigating Polycomb group proteins as genetic dependencies in H3-K27M mutant paediatric glioma.

Abstract

Polycomb group proteins play an important role in the maintenance of cellular identity by contributing to transcriptional repression through the regulation of chromatin structure. These proteins form two main multimeric complexes, known as Polycomb Repressive Complex 1 and 2 (PRC1 and PRC2). PRC2 catalyses mono-, di- and tri- methylation of lysine 27 of Histone H3 (H3K27me1/2/3) and certain forms of PRC1 can interact with H3K27me3 through their chromobox (CBX) subunit. Diffuse midline gliomas (DMGs) including diffuse intrinsic pontine glioma (DIPG) are universally fatal childhood central nervous system tumours. Approximately 84% of DIPG tumours harbour a somatic heterozygous mutation encoding a lysine-to-methionine alteration of Histone H3 at the site of PRC2 catalytic activity (H3-K27M). In a dominant negative manner the presence of H3-K27M leads to a global reduction in PRC2-mediated H3K27me3 levels, although H3K27me3 is notably retained at a number of target sites. Reducing this residual H3K27me3 through the use of EZH2 inhibitors has shown some therapeutic promise in preclinical models. The aim of this PhD thesis was to identify what specific PRC1 and PRC2 members are genetic dependencies in cell line models of H3-K27M mutant DMG, which is currently an incurable cancer. Firstly, using patient- derived DMG cell lines, I performed proteomic analyses to understand the composition of both PRC1 and PRC2. These analyses detected a wide variety of PRC1 and PRC2 subunits, suggesting that a heterogenous mix of PRC1 and PRC2 are found in these cells. Secondly, having identified the PRC2 subunits present in DMG cells, I performed tiled pooled PRC2 CRISPR-Cas9 screens. These screens identified the PRC2 catalytic core subunits EZH2, SUZ12 and EED as genetic dependencies. Thirdly, I performed a genome-wide CRISPR-Cas9 screen to identify other essential genes in DMG. PRC1 member CBX4 was identified as a dependency but its paralogs (CBX2/6/7/8) were not. In subsequent experiments, PRC1 subunit PCGF4/BMI1 was also identified as an essential gene but its paralog PCGF2/MEL18 was not. Chromatin immunoprecipitation (ChIP) assays in isogenic models with or without H3-K27M suggest that the presence of H3-K27M disrupts PRC1 localisation. Taken together, the data generated improves our understanding of the composition of PRC1 and PRC2 in DMG and provides evidence to support further investigation of core PRC2 and specific PRC1 subunits as potential therapeutic targets in H3K27-altered DMG and potentially other cancers with disrupted Polycomb function.