Myeloid disorders arise in Dnmt3a-null marrow

In this issue of Blood, Celik et al¹  and Mayle et al²  report in 2 separate studies that loss of Dnmt3a in hematopoietic stem cells (HSCs) using an Mx-cre inducible knockout model causes disturbed blood cell formation.

Both studies report that irradiated mice transplanted with Dnmt3a-null HCSs died within 1 year of hematologic malignancies. The malignancies of which the mice died represented myeloid abnormalities that are mostly consistent with myelodysplastic syndromes. A low percentage of the animals died with acute leukemia. Sequencing experiments revealed that the leukemia cells in these animals, in contrast to the mice that showed preleukemic disease, had acquired additional mutations, frequently in genes encoding signaling molecules involved in proliferation. Both teams propose that the Dnmt3a-null mouse model and the myriad of hematologic malignancies observed are representative of hematopoietic disorders found in humans carrying DNMT3A mutations.^3-5  The hypothesis that mutant DNMT3A acts as a dominant negative of wild-type DNMT3A causing the same effect as Dnmt3a knockout maybe underscored by a set of experiments by Celik et al, showing that the DNMT3A^R882H variant is able to drive the development of myeloproliferative disorders in vivo as well.¹ 

The outcome of the 2 studies differs from the data presented in a previous study by Challen et al, whose authors included the 2 senior authors of these papers.⁶  In that study, mice transplanted with Dnmt3a-null HSCs did not develop malignancies on transplantation. The major difference between that previous report and the ones that are published here is that, in the Nature Genetics study, the effects of the Dnmt3a-null HSCs were investigated in a competitive transplantation setting. Dnmt3a-null HSCs in those experiments appeared less proliferative than normal HSCs. However, this difference in proliferative activity was not sufficient to develop hematologic disorders. The fact that, in the competitive transplant studies, no bone marrow failures were observed is not surprising because sufficient normal bone marrow HSCs were present to outcompete the Dnmt3a-null HSCs in the transplanted irradiated animals and regenerate the bone marrow. Leukemias, however, were not found in the competitive transplant setting. Both current studies show that the Dnmt3a-null cells do have a moderate leukemogenic potential. There is no reason to believe that the cells had lost this capacity in the competitive transplantation setting. Suppression of malignant transformation of the Dnmt3a-null HSCs by healthy bone marrow cells is a likely explanation for this finding. Acute myeloid leukemia (AML) in humans is a clonal disease, and the early appearance of DNMT3A mutations in hematopoietic progenitor or stem cells in humans can be viewed as a competitive situation as well. The primitive cells that are initially transformed by mutant DNMT3A mutations reside in a marrow that consists of sufficient healthy HSCs that do not carry the mutation. However, with time, leukemias can arise in those individuals. The investigators propose that the Dnmt3a-null HSCs resemble the DNMT3A mutation as found in primary AMLs. Therefore, one could also argue that in the Dnmt3a-null competitive transplantation setting, leukemias may arise, but that the in vivo experiments did not take long enough and additional serial transplantations may be required to obtain leukemias in those mice, notwithstanding the presence of healthy HSCs in the marrow. No matter the outcome of such studies, the 2 reports published here, together with the previous Nature Genetics study, emphasize the importance of applying complementary in vivo approaches to obtain a more complete understanding of mechanisms of transformation.

Importantly, the studies as carried out in these 2 papers allowed the investigators to study the in vivo effects of Dnmt3a loss of function as a single event and in combination with additional mutations in the Dnmt3-null HSCs. In fact, the investigators suggest that a major role for Dnmt3a is “to balance proliferation and differentiation to maintain the hematopoietic progenitor populations.”¹  Moreover, knockout of Dnmt3a in HSCs is the onset of preleukemia in those animals, which, as the result of additionally acquired mutations, may transform into AML or acute lymphoblastic leukemia. These cooperating mutations seem to occur frequently in genes that play a role in the signaling cascades that regulate proliferation, eg, tyrosine kinase or Ras genes. Leukemias appeared with high incidence as the result of cooperation between Dnmt3a loss of function and mutations in either NRas or c-Kit. The fact that comparable mutations were found in patients with myeloid malignancies with DNMT3A mutations is another reason for the investigators to believe that Dnmt3a-null HSCs may resemble human DNMT3A mutant transformed cells and that the model can be applied to unravel the mechanism of transformation in humans.

Because Dnmt3a encodes a DNA methyltransferase, altered DNA methylation patterns are predicted in the genome of the transformed HSCs of Dnmt3a knockout mice or in human disease with DNMT3A mutations. The presence of disturbed methylation patterns, as reported by Mayle et al,²  points to such effects. No common methylation profiles were found among the distinct diseases, but rather specific methylation patterns were apparent that discriminated the different malignancy types. The absence of the DNA methyltransferase activity encoded by Dnmt3a most likely plays an essential role in the generation of these defective epigenomic patterns in these malignancies. It is unclear, however, whether these different patterns are relevant for the development of the malignancies and involve genes that play a role in transformation. We are only at the beginning of understanding the function of DNMT3A on DNA methylation and how defects in this enzyme are involved in epigenomic alterations and the development of hematological malignancies. Mouse models as reported here are important to increase our understanding of these underlying mechanisms. These models may also prove valuable in our search to treat malignancies with DNMT3A defects, ie, either to find cures for the leukemia or prevent leukemias to arise from preleukemic disorders.