KMT2A::AF6 fusion protein localizes to PML nuclear bodies and undergoes ATO-induced degradation: A potential novel therapeutic approach for KMT2A::AF6-rearranged Acute Myeloid Leukemia

Abstract Chromosomal rearrangements of the human KMT2A gene are key drivers of both pediatric and adult acute leukemias. Over 100 known in-frame gene fusions involving KMT2A have been described, but the seven most frequent translocation partner genes account for over 90% of all KMT2A fusions. Notably, prognosis is closely linked to the specific KMT2A fusion partner. For instance, the KMT2A::AFDN (AF6) fusion is associated with poor prognosis in pediatric acute myeloid leukemia (AML), resulting in a 5-year event free survival rate of 20-25%. AF6 is normally a cytoplasmic protein involved in actin cytoskeleton organization, cell junctions, and the RAS pathway. However, we demonstrated that when fused to KMT2A, the fusion protein encoded by KMT2A::AF6 mis-localizes to the nucleus in a punctate pattern, suggesting localization to specific nuclear structures (Manara et al. Blood 2014), but the functional implications of this mis-localization remain unexplored. To elucidate the specific subnuclear localization of KMT2A::AF6, we conducted high-resolution immunofluorescence analysis. We detected a complete co-localization of KMT2A::AF6 and promyelocytic leukemia (PML), indicating that KMT2A::AF6 localizes to PML nuclear bodies (PML-NBs), which are membraneless nuclear structures mostly formed by PML, where client protein are recruited through SUMO-SUMO-interacting motifs (SIMs), that act as hubs for SUMOylation. We identified eight consensus SUMOylation sites and ten SIMs in the KMT2A::AF6 fusion protein, mostly within the AF6 domain. Accordingly, co-localization of AF6 with SUMO2/3 and PML was identified in both KMT2A::AF6 cell lines and primary AML samples (n=4), and confirmed in HEK293T cells transfected with a RFP:KMT2A::AF6 vector. Given that PML and its client proteins undergo proteasomal degradation upon arsenic trioxide (ATO) treatment via SUMO-dependent ubiquitination, we tested whether KMT2A::AF6 protein follows a similar fate. ATO treatment in KMT2A::AF6-rearranged AML cells induced a reduction in the number of nuclear foci positive for both, PML and AF6 staining, indicating a decrease of PML-NBs and their associated KMT2A::AF6 fusion proteins. Consistently, ubiquitin, as well as proteasome, also accumulated in these structures, indicating a ubiquitin-dependent proteasomal degradation. Blocking the proteasome with MG132 prevented this degradation, confirming that KMT2A::AF6, as well as PML, is degraded via the SUMO–ubiquitin–proteasome axis. Additionally, ATO treatment reduced proximity between AF6 and MENIN measured by proximity ligation assay, indicating that KMT2A::AF6 was no longer present in the MENIN–LEDGF complex. Transcriptomic analyses by RNAseq revealed downregulation of canonical KMT2A target genes following ATO treatment, including HOXA9, HOXA10 and HOXA11, and downregulation of KMT2A::AF6 gene signatures, supporting the loss of fusion-driven leukemic transcriptional programs. Functionally, ATO treatment significantly reduced cell viability in both AML cell lines and primary patient-derived xenograft (PDX) samples harboring KMT2A::AF6 to a greater extent compared with other AML genetic subgroups. This effect was partially reversed by N-acetylcysteine (NAC), suggesting a ROS-dependent mechanism, described to be the very first step by which ATO drives PML proteasomal-mediated degradation. Moreover, ATO treatment induced myeloid differentiation both in vitro and in ex vivo patient-derived cells (PDXs), supporting a transcriptional change at support of a targeted KMT2A:.AF6 fusion protein degradation mechanism. In vivo, ATO monotherapy (8 mg/kg) moderately delayed disease progression in KMT2A::AF6 PDX models, while its combination with low-dose cytarabine (20 mg/kg) significantly reduced leukemic burden and prolonged AML-PDXs survival (n=4, p<0.05). Interestingly, treatment with a Menin inhibitor (alone or in combination with ATO) had limited efficacy compared to other KMT2A fusion types, highlighting the unique biology of KMT2A::AF6 AML and underscoring the relevance of ATO as a specific and efficacious therapeutic strategy. Overall, we uncovered KMT2A::AF6 fusion as a client of PML-NBs that undergoes ATO-induced SUMO-dependent ubiquitin-mediated proteasomal degradation, similarly to PML::RARα in acute promyelocytic leukemia. These findings establish a novel vulnerability of KMT2A::AF6-rearranged AML and suggest that ATO may be repurposed as a targeted therapy for this aggressive and resistant leukemia subtype.