ASXL1 Mutations in AML Are Associated with a Distinct Epigenetic State Which Highlights Vulnerabilities to Specific Epigenetic-Targeted Agents

ASXL1 is frequently mutated in hematologic malignancies, with incidence ranging in AML from 5 to 17%. ASXL1 binds BAP1, part of the polycomb-repressive deubiquitinase (PR-DUB) complex. PR-DUB removes monoubiquitination of histone H2A at lysine 119, a repressive mark, catalyzed by the PRC1 (polycomb-repressive complex 1). Gain-of-function (GOF) mutations in ASXL1 result in stabilization of ASXL1 and the PR-DUB complex, causing a global reduction in H2AK119Ub and increased mRNA transcription. Poorer prognosis of AML with mutant (mt) ASXL1 creates an unmet need to identify and test novel therapies for this AML sub-type.

We generated two isogenic knock-in (KI) models of the clinically relevant mtASXL1 Y591* into the AML cell lines OCI-AML3 and UCSD-AML1, which endogenously express wildtype (wt) ASXL1. Both model cell lines exhibited elevated ASXL1, BAP1, and reduced H2AK119Ub histone mark. We utilized the isogenic AML OCIAML3 ASXL1 Y591* and UCSD-AML1 ASXL1 Y591* models for studying mutant (mt) ASXL1 biology and potential therapeutic interventions. Notably, ASXL1/BAP1 stabilization caused an increased expression of c-Myc at both the RNA and protein level. ChIP-Seq analysis was performed in the KI-OCI-AML3 cells to determine its specific effects on histone mark alterations. We observed reduced H2AK119Ub at H3K27Ac-marked chromatin on Wnt/β-catenin pathway genes that resulted in increased mRNA expression of FZD5, TCF7L2, and WNT5A, which are involved in a stem cell signaling pathway. This was also associated with a reduction in the cell surface CD14 expression, as determined by flow cytometry. KI of mtASXL1 in OCI-AML3 cells also increased cell-cycle G0/G1 with a concomitant reduction in S and G2/M phases of the cell-cycle. RNA-Seq analysis showed reduced expression of the REACTOME cell-cycle mitotic gene-set, associated with depletion of E2F1, AURKA/B, and PLK1 mRNA. Mass spectrometry analysis showed enriched gene-sets for translation initiation and elongation in the OCIAML3 ASXL1 Y591* cells, but concomitant reduction in the interleukin and interferon signaling pathways. Notably, compared to parental cells, OCIAML3 ASXL1 Y591* cells exhibited reduced sensitivity to standard anti-AML chemotherapeutic agents, including cytarabine, etoposide and daunorubicin. It has been reported that mtASXL1, but not wild-type ASXL1, binds BRD4 resulting in hypersensitivity to BET inhibitors (BETi). Our findings also demonstrate and confirm that mtASXL1-expressing AML cells exhibited increased sensitivity to BETi (pelabresib or birabresib). These cells were also more sensitive, compared to the parental cells, to other epigenetically targeted agents, including HATi (GNE-781 or CBP-IN-1), mediator kinase inhibitor (SEL120-34A/RVU120) and the dual HAT/BET inhibitor (NEO2734/EP31670). Treatment with NEO2734 reduced the permissive chromatin marks especially on c-Myc and SPI1 super enhancers, leading to a significant reduction in mRNA and protein expressions of c-Myc, c-Myb, and PU.1. RNA-Seq and mass-spectrometric analyses demonstrated that treatment with NEO2734 negatively enriched gene sets of c-Myc targets, rRNA processing, mRNA splicing, and translation. Treatment with SEL120-34A (the CDK8 and CDK19 mediator-kinases inhibitor) globally reduced active chromatin marks in the mtASXL1 expressing AML cells. This was associated with reduced protein expression of BAP1, pSTAT1 (S727), pSTAT5 (S726), MEIS1 and HOXA9 exclusively in mtASXL1 expressing cells. Co-treatment with NEO2734 or pelabresib and SEL120-34A synergistically induced apoptosis in mtASXL1-expressing cells (delta synergy scores > 20, by ZIP method). Importantly, in the NSG mice engrafted with luciferized OCIAML3 ASXL1 Y591*, monotherapy with NEO2734 (5 mg/kg QD), pelabresib (30 mg/kg QD) or SEL120-34A (40 mg/kg QD), compared to vehicle control, significantly reduced AML burden. Collectively these findings highlight previously uncharacterized biologic effects of the presence of mtASXL1 and support the rationale for further evaluating AML therapies incorporating BETi, HAT-BETi or inhibitor of mediator kinase.