ID #675 Oncohistone H3.3K27M blocks neural differentiation by disabling the H3.3S31 phospho-switch and misdirecting KDM4B

Abstract

Diffuse midline gliomas (DMGs) driven by the H3.3K27M oncohistone are marked by a striking failure of neural differentiation, yet this phenotype is potentially reversible and therefore attractive for differentiation-based therapy. However, the molecular basis of the developmental block has remained poorly defined, limiting rational strategies to re-engage maturation programs. To pinpoint how K27M disrupts differentiation, we engineered mouse embryonic stem (ES) cells carrying a single endogenous H3.3K27M allele and followed neural differentiation, allowing us to define direct K27M consequences in isolation from additional tumour-associated alteration.

A single copy of H3.3K27M arrested neural differentiation at the progenitor stage and abolished H3K9me3-rich chromocenters, indicating collapse of pericentric heterochromatin. Chromatin profiling identified the H3K9me3 demethylase KDM4B as a central driver. In normal cells, KDM4B is enriched at H3K27me3-marked promoters and is kept away from pericentric H3K9me3, supporting ATRX-dependent heterochromatin establishment. In H3.3K27M cells, KDM4B mislocalises to pericentric regions, erases H3K9me3, and prevents chromocenter formation, likely undermining ATRX heterochromatin maintenance by depleting the H3K9me3 landscape that ATRX requires. Consistently, Kdm4b knockout restored chromocenters and reinstated neural differentiation, demonstrating that aberrant KDM4B activity enforces the developmental block.

Mechanistically, we found that H3.3K27M disrupts phosphorylation of H3.3 serine 31 (H3.3S31ph), a modification that normally restrains KDM4B activity and supports ATRX-dependent heterochromatin organisation during differentiation. Loss of H3.3S31ph in K27M cells permits a gain of KDM4B activity, leading to erosion of H3K9me3 and functional impairment of ATRX-driven heterochromatin establishment. Restoring this control with a phosphomimetic H3.3S31E mutant rescued chromocenter integrity and neuronal differentiation in both engineered ES models and patient-derived DMG lines.

Together, these data define a mechanistic and potentially reversible chromatin circuit that sustains the DMG differentiation block and nominate a rational differentiation-therapy strategy: restoring the H3.3S31 phosphorylation switch or inhibiting KDM4B to re-establish H3K9me3 heterochromatin, unlock neural maturation, and exploit a tumour-specific vulnerability.