Loss of Dnmt3b Accelerates MLL-AF9 Leukemia Progression

Acute myeloid leukemia (AML) is a heterogeneous hematopoietic disorder with poor prognosis. DNA methylation may play a vital role in the development of AML. Dnmt3a and Dnmt3b are de novo methyltransferases. Dnmt3a is frequently mutated in hematologic malignancies and in contrast, Dnmt3b is rarely mutated. While it was previously reported that Dnmt3b functions as a tumor suppressor in a mouse model of Myc-induced lymphomagenesis, its function in AML is yet to be determined. We genetically inactivated Dnmt3b in a mouse model of MLL-AF9 induced AML. c-Kit⁺ hematopoietic stem/progenitor cells (HSPC) of Dnmt3b ^fl/fl CreER mice were transduced with retroviral MLL-AF9 fusion gene and transplanted into lethally irradiated recipients. Tamoxifen (TAM) was administrated to delete Dnmt3b and we monitored AML progression of the mice. The deletion of Dnmt3b was confirmed in Dnmt3b ^fl/fl CreER AML cells at both mRNA and protein levels. We examined the survival of AML mice and found that Dnmt3b knockout (KO) accelerated MLL-AF9 leukemia development (p=0.0001, n=9). Morphological analysis showed that Dnmt3b KO AML cells were lack of segmented nuclei, suggesting that Dnmt3b deficiency might be associated with decreased differentiation. Colony forming assay showed that Dnmt3b KO AML cells formed colonies at a higher frequency than that of control AML cells (1.7 fold). Flow cytometry showed the frequency of leukemia stem cell (LSC) in KO AML cells was higher than in control AML cells (13% vs 3.3%). Cell cycle analysis indicated an increased cycling fraction in S/G2/M phases (30% vs 16%) of KO AML cells in comparison to control AML cells. The molecular basis was explored by qRT-PCR gene profiling analysis. LSC-associated genes such as Meis1 and Myb were up-regulated whereas the transcription factors involved in the differentiation of hematopoietic cells such as Runx1 and Gata1 were down-regulated in Dnmt3b KO AML cells compared with control. As a converse functional assessment, we employed a retroviral system to overexpress full-length cDNA of Dnmt3b in the MLL-AF9 cells. Compared with control AML cells, Dnmt3b overexpressed (OE) leukemia cells had reduced colony forming ability (0.7 fold), and the recipient mice of Dnmt3b OE leukemia showed longer survival time (p=0.0056, n=7). Dnmt3b OE leukemia cells had lower LSC frequency (1% vs 4.3%) and higher Gr-1 median intensity (1.4 fold), thereby suggesting that Dnmt3b overexpression induced leukemia cell differentiation. The qRT-PCR analysis showed up-regulation of p21, Gata1 and PU.1 and down-regulation of Meis1 and Myb. As Dnmt3a has been shown to be frequently mutated in multiple types of hematopoietic malignancies, we asked whether loss of both Dnmt3a and Dnmt3b would have synergistic effects in leukemogenesis. We generated Dnmt3a KO and Dnmt3a/3b double-KO (DKO) leukemia cells by the CRISPR-Cas9 technique. Compared with control AML cells, separate or double deletion of 2 genes all could accelerate leukemogenesis. Especially, Dnmt3a/3b DKO resulted in the shortest AML latency (p=0.0002 for Dnmt3a KO, n=7; p=0.0015 for Dnmt3b KO, n=7; and p=0.0001 for DKO, n=8). In addition, colony forming assay showed that DKO leukemia cells had the highest colony forming ability. In summary, these results demonstrated that loss of Dnmt3b accelerated MLL-AF9 leukemia progression by increasing the stemness and enhancing cell cycle movement. Loss of Dnmt3b was able to synergize with Dnmt3a deficiency in leukemia development. Therefore, our current study provides new insights into the roles of DNA methylation in leukemogenesis.