(P 78) NOVEL EPIGENETIC MECHANISMS DRIVE AGGRESSIVE OSTEOSARCOMA PATHOGENESIS AND DRUG RESISTANCE

Objective: Osteosarcoma (OS) is the most common mesenchymal neoplasm predominantly occuring in adolescents and young adults. After a decade of comprehensive characterization of the OS somatic genome, improvement in survival rates and clinical care remains stagnent. Conversely, the role of epigenome alterations and chromatin remodelling in OS is vastly limited eventhough evidence suggests that OS pathogenesis is linked to hijacked developmental programs. Given the pivotal role of epigenetics in orchestrating development, we hypothesized that epigenomic perturbations play a key role in driving OS with potential clinical implications.

Methods: knowing that alterations in epigenomic complexes are exceedingly rare in OS, we evaluated the expression of oncohistones and the H3K27M oncohistone-mimic, EZHIP, in 2 sizable OS patient cohorts by immunohistochemistry. We modeled oncohistone- and EZHIP- expression using CRISPR/Cas9 genetic knockout and lentiviral overexpression. Next, we examined resulting phenotypes in OS cell lines and human mesenchymal cells (hMSC) via live cell imaging, in-vitro differentiation assays and orthotopic and subcutaneous engraftment in immunocompromised mice. To dissect its molecular mechanims, we profiled the epigenome and transcriptome using CUT&RUN and RNAseq. Finally, drug assays were used in-vitro and in orthotopic and xenograft mouse models to assess druggable vulnerabilities.

Results: We discovered that EZHIP is ectopically expressed in ~22% of high-grade OS tumors leading to the loss of the repressive histone mark H3K27me3. Surprisingly, drastic loss of H3K27me3 significantly correlated with poor response to neoadjuvant chemotherapy and was enriched in chemoresistant tumors. EZHIP and H3K27M overexpression demonstrated that they are prominent oncogenes in xenograft models. EZHIP genetic knockout abrogated tumorigenic capacity of OS cells in-vivo. Tumorigenesis was only rescued by re-expressing wildtype EZHIP or the oncohistone H3K27M, but not other oncohistones nor EZHIP-R405E mutant. Converging on the same mode of action, this identified H3K27me3 loss as a novel mechanism of tumorigenesis. GSEA and deconvolution analyses of OS cells and xenografts revealed that EZHIP globally dysregulates the transcriptome to activate oncogenic signalling and stemness programs. Moreover, EZHIP expression permits the expansion of smooth muscle cells and newly characterized pre-osteoblastic progenitors such as Osteo-CAR cells at the expense of mature osteoblasts. In-vitro, myogenic features were validated in xenograft models and available patient tumors. Similarly, EZHIP blocked multilineage terminal differentiation of hMSCs in-vitro while enhancing myogenic differentiation. Although aggressive, EZHIP expressing cells demonstrated enhanced sensitivity to EZH1/2 inhibitors UNC1999, the FDA-approved drug Tazemetostat and EZH2 Proteolysis Chimeras (PROTACs).

Conclusion: EZHIP is a novel oncogene in a substantial subset of high-grade OS tumors. EZHIP stalls mesenchymal progenitors in a cycling state permissive of malignant transformation and promotes OS aggressiveness in different models of different genetic backgrounds. H3K27me3 loss represents a novel mechanism of pathogenesis which sheds light on uncharted epigenetic disease mechanisms in OS. These findings warrant assessing the use of epigenetic drugs as potential treatment options in combination with conventional polychemotherapy for select patients.