Mutations affecting the gene for the de novo DNA Methyltransferase 3A (DNMT3A) are the most common drivers of clonal hematopoiesis (CH) and amongst the most common somatic events in acute myeloid leukaemia (AML). Approximately two thirds of AML-associated DNMT3A mutations are heterozygous substitutions affecting codon R882, located within the methytransferase domain and are correlated with global hypomethylation. DNMT3A mutations are initiating events in leukemogenesis and leukemic progression relies on the acquisition of additional mutations in genes such as NPM1, TET2, IDH1/2, FLT3 andNRAS. AMLs harboring DNMT3A-R882 mutations display increased resistance to chemotherapy and carry a worse prognosis for patients, but the molecular basis of this is not well understood.

To improve our understanding of the molecular effects of mutant DNMT3A, we developed a conditional model of DNMT3A-R882H. Our conditional Dnmt3a-flox-R882H allele retains the native locus intact, but generates the Dnmt3a-R882H mutant upon Cre-loxP recombination (Mx1-Cre). We observed that Dnmt3aR882H/+ mutant bone marrow (BM) cells had markedly enhanced self-renewal potential in serial re-plating assays and increased BM repopulation ability in competitive transplants. Approximately 30% of Dnmt3aR882H/+ mice developed leukaemia after a long latency (median of 532 days). Introduction of Flt3-ITD mutation significantly accelerated spontaneous disease, in agreement with other published reports, with majority of mice succumbing to AML after a median of 192 days.

Due to the frequent co-occurrence and strong cooperation between Dnmt3aR882H/+ with Flt3ITD/+, we sought to identify potential genetic vulnerabilities of this combination, to facilitate patient treatment. We therefore generated Dnmt3aR882H/+/Flt3ITD/+/Mx1-Cre+/Rosa-Cas9+ mice, induced Dnmt3aR882H mutation with pIpC and isolated hematopoietic stem and progenitor cells (HSPCs). We cultured HSPCs in vitro and performed whole-genome CRISPR dropout screens to identify genetic vulnerabilities. This identified 526 dropout genes at a genome-wide FDR of 20%, including pan-essential genes such as components of the ribosome, proteasome or spliceosome. To identify essential genes specific to the Dnmt3aR882H/+/Flt3ITD/+ genotype, we performed analogous screens in the non-leukemic mouse hematopoietic precursor cell-7 (HPC-7) and used published data indicating essential genes for normal human CD34 cells. Overlap of these screens revealed 196 dropouts present only in Dnmt3aR882H/+/Flt3ITD/+/Mx1-Cre+/Rosa-Cas9+ cells. Among these were genes coding for "druggable" candidates such as kinases and histone modifying enzymes, including SET Domain Containing 1B (Setd1b), one of the six mammalian proteins (also Setd1a and Mll1-4) that can catalyze methylation of H3K4. Each of these six proteins can interact with obligate proteins as a part of the methyltransferase complex, but also form distinct complexes with non-redundant function. We firstly validated that the knockout of Setd1b induced by gRNA significantly reduced the growth of pre-leukemic and leukemic HSPCs from both Dnmt3aR882H/+/Flt3ITD/+/Mx1-Cre+/Rosa-Cas9+ and Dnmt3aR882H/+/Mx1-Cre+/Rosa-Cas9+ mice. Importantly, Setd1b deletion in normal HSPC had no major effect on cell growth. To investigate the molecular role of Setd1b in Dnmt3aR882H context, we asked if Setd1b catalytic activity and linked transcriptional programs were responsible for the observed phenotypes. For this, we performed histone profiling and RNA-sequencing in Dnmt3aR882H/+/Flt3ITD/+/Mx1-Cre+/Rosa-Cas9+HSPC upon Setd1b KO vs control and observed a striking reduction of H3K4me3 marks at particular loci in association with loss of expression of corresponding genes. These included several known to be required for cellular proliferation and transformation, some of which were themselves dropouts in our original screen. Importantly the majority of genes were also significantly upregulated when gene expression from Dnmt3aR882H/+/Flt3ITD/+ HSPC were compared with Flt3ITD/+ HSPC.

Our data propose that the catalytic activity of Setd1b plays a mechanistic role in driving the growth/fitness-promoting effects of Dnmt3aR882H/+. Blocking the catalytic activity of SETD1B may therefore represent a plausible therapeutic option for the treatment of AML of even CH patients harboring mutations inDNMT3A.


Mazan:Selvita S.A.: Employment. Vassiliou:Kymab Ltd: Consultancy, Other: Minor Stockholder; Oxstem Ltd: Consultancy; Celgene: Research Funding.

Author notes


Asterisk with author names denotes non-ASH members.