Dose-dependent effects of Dnmt3a in an inducible murine model of KrasG12D-driven leukemia

•Loss of Dnmt3a in mice with Kras-driven leukemias skews toward a myeloid phenotype.•Combined Dnmt3a knockout/Kras mutants quickly succumb to aggressive myeloid disease.•Dysregulation of cell cycle and MYC are associated with biallelic Dnmt3a deletion.•This is a robust model for investigating Ras/DNMT3A-mutant acute myeloid leukemia. DNMT3A mutations are frequently found in clonal hematopoiesis and a variety of hematologic malignancies, including acute myeloid leukemia. An assortment of mouse models have been engineered to explore the tumorigenic potential and malignant lineage bias due to loss of function of DNMT3A in consort with commonly comutated genes in myeloid malignancies, such as Flt3, Nras, Kras, and c-Kit. We employed several tamoxifen-inducible Cre-ERT2 murine model systems to study the effects of constitutively active KrasG12D-driven myeloid leukemia (Kras) development together with heterozygous (3aHet) or homozygous Dnmt3a deletion (3aKO). Due to the rapid generation of diverse nonhematologic tumors appearing after tamoxifen induction, we employed a transplantation model. With pretransplant tamoxifen induction, most Kras mice died quickly of T-cell malignancies regardless of Dnmt3a status. Using posttransplant induction, we observed a dose-dependent effect of DNMT3A depletion that skewed the leukemic phenotype toward a myeloid lineage. Specifically, 64% of 3aKO/Kras mice had exclusively myeloid disease compared with 36% of 3aHet/Kras and only 13% of Kras mice. Here, 3aKO combined with Kras led to increased disease burden, multiorgan infiltration, and faster disease progression. DOT1L inhibition exerted profound antileukemic effects in malignant 3aKO/Kras cells, but not malignant cells with Kras mutation alone, consistent with the known sensitivity of DNMT3A-mutant leukemia to DOT1L inhibition. RNAseq from malignant myeloid cells revealed that biallelic Dnmt3a deletion was associated with loss of cell-cycle regulation, MYC activation, and TNF⍺ signaling. Overall, we developed a robust model system for mechanistic and preclinical investigations of acute myeloid leukemia with DNMT3A and Ras-pathway lesions.