LSD1 Inhibition Exerts Its Anti-Leukemic Effect By Recommissioning PU.1- and C/Ebpα-Dependent Enhancers in AML

Epigenetic regulators are recurrently mutated and aberrantly expressed in acute myeloid leukemia (AML). Targeted therapies designed to inhibit these chromatin-modifying enzymes, such as the histone demethylase lysine specific demethylase 1 (LSD1) and the histone methyltransferase DOT1L, have been developed as novel treatment modalities for these often refractory diseases. We aimed to investigate the chromatin changes induced by an irreversible inhibitor of LSD1 (GSK-LSD1) on rMLL⁺ leukemias. For this, we treated rMLL⁺ murine AML and patient-derived xenografts with the GSK-LSD1 inhibitor in vivo and assessed chromatin accessibility changes using the Assay for Transposase-Accessible Chromatin with high throughput sequencing (ATAC-seq). Treated mice exhibited a lower proportion of MLL-AF9⁺ leukemia cells detected in the bone marrow (BM) and signs of myeloid differentiation. Treated mice also had markedly improved survival with a median survival of 78 days vs. control mice (39 days). After 6 weeks of treatment with GSK-LSD1, mice xenotransplanted with human rMLL⁺ AML samples showed a significant decrease in engraftment and myeloid differentiation. Serial transplantation of murine AML revealed eradication of LSC upon GSK-LSD1 treatment of donor mice. Most importantly, LSD1 inhibition resulted in gain in chromatin accessibility in both murine and PDX-AML cells in vitro as well as in vivo . We then wanted to compare the effect of another anti-leukemic epigenetic drug, also known to induce myeloid differentiation of MLL-AF9⁺ cells, such as the DOT1L inhibitor EPZ4777. In contrast to GSK-LSD1, EPZ4777 treatment caused a global decrease in chromatin accessibility. The intersection of sites with dynamic accessibility upon DOT1L and LSD1 inhibition showed minimal overlap between the 2 treatments. The distribution of genomic regions with dynamic accessibility upon these drugs indicated that DOT1L inhibition affects a greater number of sites in promoter regions, whereas LSD1 inhibition affected a higher proportion of loci occupied within the intergenic and intronic regions. Furthermore, we sought to identify TF motif signatures at GSK-LSD1-induced dynamic sites. The most highly enriched motifs captured were those for myeloid TFs PU.1 and C/EBPα-β. We then performed ChIP-seq to identify sites of C/EBPα and PU.1 occupancy in the setting of GSK-LSD1 treatment. Interestingly, we observed a redistribution of PU.1 and C/EBPα occupancy upon treatment, consistent with the role of LSD1 in regulating the binding and activity of TFs. While PU.1 ChIPseq signal increased at dynamic ATAC-seq peaks in response to LSD1 inhibition, no increase in C/EBPα occupancy was observed. These dynamic sites also demonstrates GSK-LSD1-dependent co-occupancy of PU.1 and C/EBPα. We next sought to delineate whether PU.1 and C/EBPα are critical for these dynamic changes in chromatin accessibility in response to LSD1 inhibition. We performed ATAC-seq on PU.1 URE KO MLL-AF9 leukemias and C/EBPα KO MLL-AF9 leukemias treated with vehicle or GSK-LSD1. The increase in chromatin accessibility in TF-WT MLL-AF9 leukemias upon GSK-LSD1 was abrogated in both PU.1 URE KO and C/EBPα KO MLL-AF9 leukemia cells. We next assess the impact of TF-depletion on sensitivity to LSD1 inhibition. We treated PU.1 URE KO leukemias cells with GSK-LSD1 for 48 hours in vitro . These cells were more resistant to LSD1 inhibition when compared to PU.1 URE heterozygous and WT AML cells. In vivo, we reported loss of survival benefit upon GSK-LSD1 treatment in PU.1 URE KO MLL-AF9⁺ mice vs. WT mice. Similarly, C/EBPα KO leukemia cells were more resistant to LSD1 inhibition when compared to WT leukemia cells. After 2 weeks of treatment in vivo, C/EBPα KO leukemic mice also showed no decrease in MLL-AF9 allele burden. Taken together, our comparative investigations into the chromatin-modifying effects of two distinct epigenetic therapies targeting either LSD1 or DOT1L is illustrative of the concept that multiple paths toward myeloid differentiation can potentially be engaged to achieve therapeutic efficacy in AML. Our findings reveal that pharmacologic inhibition of LSD1 depends on balanced expression of the TFs PU.1 and C/EBPα, which can be used as predictors of response in AML aberrantly expressing these factors.