Accelerating myelofibrosis through loss of Dnmt3a

In this issue of Blood, Jacquelin et al report that loss of DNA methyltransferase 3A (Dnmt3a) in hematopoietic cells expressing mutant Jak2V617F promotes progression from polycythemia vera to myelofibrosis leading to bone marrow failure and premature mortality.¹ 

DNMT3A is an enzyme that adds methyl groups in CpG-rich regions of the chromosomes, which are involved in control of transcription. Mutations in DNMT3A are frequently found in normal individuals with clonal hematopoiesis of indeterminate potential,^2,3  consistent with effectively increasing the self-renewal potential of hematopoietic stem cells (HSCs).⁴  Monoallelic or biallelic mutations in DNMT3A are found in acute myeloid leukemias and myelodysplastic syndromes and are also detectable in 7% to 10% of patients with myeloproliferative neoplasms (MPN), most frequently in combination with JAK2V617F.^5,6  Although the loss of DNMT3A leads to demethylation of DNA, the exact mechanism of how DNMT3A mutations contribute to the multistep genesis of cancer is still largely unknown.⁴  The report by Jacquelin et al provides new insights into how the Jak2V617F driver mutation and the loss of Dnmt3a collaborate to accelerate MPN disease progression.

Jacquelin and colleagues used clustered regularly interspaced short palindromic repeats (CRISPR) with CRISPR-associated protein 9 (Cas9) technology to introduce deletions in the Dnmt3a gene into HSCs from a previously characterized Jak2V617F knockin mouse model of MPN (see figure). These targeted HSCs were then transplanted into lethally irradiated wild-type recipients. The Jak2V617F;Dnmt3a double-mutant HSCs exhibited a competitive advantage over the Jak2V617F single-mutant HSCs and over time dominated hematopoiesis in the reconstituted recipients. These double-mutant mice displayed progression from an initial polycythemia phenotype to advanced myelofibrosis accompanied by premature lethality. Hematopoiesis was oligoclonal in respect to the Dnmt3a mutations (ie, each recipient mouse had hematopoietic cells carrying a number of different Dnmt3a deletions induced by CRISPR-Cas9). Importantly, the Dnmt3a gene in double-mutant cells was inactivated on both chromosomes, resulting in loss of expression of the Dnmt3a protein in western blots (supplemental Figure 1B in Jacquelin et al).

To understand the mechanisms of action, the authors performed a detailed gene expression, DNA methylation, and chromatin accessibility analysis. Although the global DNA methylation changes did not correlate well with alterations in gene expression, the authors found that loss of Dnmt3a in conjunction with Jak2V617F upregulated self-renewal programs mainly through regulation of chromatin topology and methylation of specific chromatin marks at enhancer sites that have been shown to regulate HSC gene expression. Thus, the loss of Dnmt3a resulted in an “enhanceropathy” driving reproducible transcriptional programs that alter stem cell properties. Surprisingly, despite upregulating self-renewal gene expression programs, the Jak2V617F;Dnmt3a double-mutant HSCs exhausted earlier leading to bone marrow failure and inability to engraft into secondary recipients than Jak2V617F single-mutant HSCs. These data suggest that double-mutant HSCs might be sensitive to drugs, such as bromodomain inhibitors, that exert activity in the context of enhancer activation.

The accelerated transition to myelofibrosis is reminiscent of the phenotypes observed in Jak2V617F;Ezh2 double-mutant mice.^7-9  Indeed, Jacquelin et al found that Jak2V617F;Dnmt3a double-mutant cells show changes in chromatin accessibility consistent with decreased repressive H3K27me3 marks that are normally deposited by the Ezh2 containing polycomb repressive complex 2 (PRC2) on histone proteins in key enhancer regions and are instead replaced by activating H3K27Ac marks histone placed by the acetyltransferases p300 and CBP. Thus, the loss of Dnmt3a in part mimics the loss of PRC2 activity, although at present it remains unclear how these DNA methylation and histone modification pathways interact at the molecular level. The transition to myelofibrosis in Jak2V617F;Dnmt3a double-mutant mice is further promoted by increased chromatin accessibility and enhancer activation in genes that orchestrate a proinflammatory program with strong upregulation of tumor necrosis factor α (TNFα) via NF-κB pathways. Increased proinflammatory cytokines as a consequence of Dnmt3a loss were also observed in conjunction with other activated oncogenes such as NRas or FLT3-ITD.

Some of the key gene expression alterations leading to activated TNFα signaling and proinflammatory gene expression observed in the Jak2V617F;Dnmt3a double-mutant mice were also confirmed in primary cells from MPN patients with vs without mutated DNMT3A. In addition to advancing our understanding of the role of Dnmt3a loss in MPN disease progression, this study raises additional questions. In the current prognostic scores for MPN patients, loss of DNMT3A is not considered a “high-risk” mutation, whereas in the mouse model a strong negative effect on survival was seen. The order of acquisition of the Jak2V617F and Dnmt3a mutations may influence the severity of the phenotype. In the majority of MPN patients, the Dnmt3a mutation precedes the acquisition of the Jak2V617F mutation, whereas, as in the mouse model studied by Jacquelin et al, the Dnmt3a mutation is introduced on top of a preexisting Jak2V617F mutation. Indeed, differences in phenotypes were reported for “JAK2-first” vs “DNMT3A-first” patients with MPN.¹⁰  Finally, this work provides a strong rationale for testing therapeutic approaches that target the activity of gene enhancers for MPN patients selected for specific mutational profiles.