Dnmt3a-Deletion Accelerates FLT3-ITD Malignancies In Mice By Hypomethylation Of Enhancer Sites and Activating Stem Cell Programs; Implications For Therapy

The de novo DNA methyltransferase (DNMT) 3A is mutated in 50% of patients with mixed phenotype acute leukemia, 20% with acute myeloid leukemia (AML) and 18% with T-cell acute lymphoblastic leukemia (T-ALL). The mechanisms through which mutant DNMT3A contributes to hematologic malignancy are poorly understood. In mice, deletion of Dnmt3a in hematopoietic stem cells (HSCs) leads to abnormal DNA methylation and inhibition of differentiation, but is insufficient for leukemic transformation. To study the role of Dnmt3a in leukemia, we combined Dnmt3a-deletion with the activated FLT3 proto-oncogene (FLT3-ITD), a frequent co-mutation with DNMT3A in AML patients, to establish a murine model of Dnmt3a-associated malignancy.

In mice transplanted with Dnmt3a-knockout (KO) or wild-type (WT) bone marrow cells transduced with a FLT3-ITD retrovirus, Dnmt3a-loss dramatically impacted the disease phenotype. Dnmt3aKO/ITD transplanted mice had significantly shortened survival (79 days vs. 116 days) and increased rate of acute leukemia compared to mice with ITD alone. The mice developed CD4+CD8+ Notch activation-associated T-ALL or myeloproliferative disease (MPD), or concurrently both, consistent with previous studies of FLT3-ITD in mice. To determine the leukemia-initiating population, we transplanted sorted HSC, myeloid, and lymphoid progenitors transduced with FLT3-ITD. All mice transplanted with HSC and myeloid progenitors succumbed to both malignancies.

To uncover the mechanisms by which Dnmt3a-deletion accelerated acute leukemia, we analyzed changes in DNA methylation in T-ALL blasts by whole genome bisulfite sequencing. Compared to Dnmt3aWT/ITD, Dnmt3aKO/ITD blasts exhibited global hypomethylation, particularly at distal enhancer sites. These hypomethylated enhancer sites were associated with genes in signaling pathways, transcription regulators, and metabolic pathways in cancer (KEGG and GO Analysis). Transcriptome analysis showed that relative to Dnmt3aWT/ITD, the Dnmt3aKO/ITD blasts had 1577 significantly differentially expressed genes positively related to cancer, cellular growth, and proliferation, and negatively to apoptosis by Ingenuity Pathway Analysis (IPA). Surprisingly, we observed increased expression of genes related to HSCs and myeloid function and decreased expression of genes related to lymphocyte function. Human AML signature genes (Oncomine) were also upregulated in our mouse model. Predicted activated pathways include Myc, Nfe2l2, Eif4e, E2f1, Csf2, Cebpb, Vegf, Rxra, Ezh2, and Brd4 and inhibited pathways include tumor suppressors Rb, let7, Cdkn2a, and Tob1 (IPA). We did not observe changes in genomic copy number variation by chromosomal comparative hybridization (cCGH).

To test whether Dnmt3a-deletion could functionally bestow stem cell properties on pre-leukemic cells, we examined self-renewal capabilities of malignant cells of Flt3^+/ITD knock-in mouse (an ITD mutation knocked in to the endogenous murine Flt3 allele causing MPD). Remarkably, when Dnmt3aKO; Flt3^+/ITD bone marrow cells were serially transplanted, MPD was seen in all recipients, compared to none in Dnmt3aWT; Flt3^+/ITD transplanted mice (n=7). Further, we transplanted sorted CLP, CMP, GMP, MPP, ST-HSC, LT-HSC populations and observed myeloproliferation in transplanted non-stem (CMP, GMP, ST-HSC) and stem cell (LT-HSC) populations. This strongly suggests that Dnmt3aKO synergized with Flt3-ITD to confer stem cell self-renewal abilities to transformed progenitor and stem cells.

Increasingly, decitabine is being used to treat patients with AML and MDS, but whether patients with DNMT3A mutations could benefit is unclear, so we examined the impact of decitabine treatment on the retroviral transduced Dnmt3aKO/ITD mice. Monthly treatment led to significantly increased survival of Dnmt3aKO/ITD mice from T-ALL and MPD and reduced presence of ITD-transduced KO cells.

Together, we demonstrate that Dnmt3aKO accelerated malignancies induced by FLT3-ITD in mouse and may shed light on how DNMT3A mutations contribute to lymphoid and myeloid disease in patients. Dnmt3a deletion ignited multilineage and stem cell programs at the expense of lymphoid programs to accelerate disease, but was extinguishable by decitabine therapy. The findings from our mouse model can be used for the development and testing of targeted epigenetic therapy for DNMT3A-associated malignancies.