Summary In acute myeloid leukemia (AML), the cell of origin, nature and biological consequences of initiating lesions and order of subsequent mutations remain poorly understood, as AML is typically diagnosed without observation of a pre-leukemic phase. Here, highly purified hematopoietic stem cells (HSC), progenitor and mature cell fractions from the blood of AML patients were found to contain recurrent DNMT3a mutations (DNMT3amut) at high allele frequency, but without coincident NPM1 mutations (NPM1c) present in AML blasts. DNMT3amut-bearing HSC exhibited multilineage repopulation advantage over non-mutated HSC in xenografts, establishing their identity as pre-leukemic-HSC (preL-HSC). preL-HSC were found in remission samples indicating that they survive chemotherapy. Thus DNMT3amut arises early in AML evolution, likely in HSC, leading to a clonally expanded pool of preL-HSC from which AML evolves. Our findings provide a paradigm for the detection and treatment of pre-leukemic clones before the acquisition of additional genetic lesions engenders greater therapeutic resistance.

Acute promyelocytic leukemia (APL), a cytogenetically distinct subtype of acute myeloid leukemia (AML), characterized by the t(15;17)-associated PML-RARA fusion, has been successfully treated with therapy utilizing all-trans-retinoic acid (ATRA) to differentiate leukemic blasts. However, among patients with non- APL AML, ATRA-based treatment has not been effective. Here we show that, through epigenetic reprogramming, inhibitors of lysine- specific demethylase 1 (LSD1, also called KDM1A), including tranylcypromine (TCP), unlocked the ATRA-driven therapeutic response in non-APL AML. LSD1 inhibition did not lead to a large-scale increase in histone 3 Lys4 dimethylation (H3K4me2) across the genome, but it did increase H3K4me2 and expression of myeloid-differentiation–associated genes. Notably, treatment with ATRA plus TCP markedly diminished the engraftment of primary human AML cells in vivo in nonobese diabetic (NOD)- severe combined immunodeficient (SCID) mice, suggesting that ATRA in combination with TCP may target leukemia-initiating cells. Furthermore, initiation of ATRA plus TCP treatment 15 d after engraftment of human AML cells in NOD-SCID γ (with interleukin-2 (IL-2) receptor γ chain deficiency) mice also revealed the ATRA plus TCP drug combination to have a potent anti-leukemic effect that was superior to treatment with either drug alone. These data identify LSD1 as a therapeutic target and strongly suggest that it may contribute to AML pathogenesis by inhibiting the normal pro-differentiative function of ATRA, paving the way for new combinatorial therapies for AML.