Bi-Allelic DDX41 Mutations Exert Stage Specific Effects on Hematopoiesis to Promote Disease

Germline mutations in DEAD-box helicase 41 (DDX41) are the most common heritable predisposition to myeloid malignancies, accounting for ~4% of acute myeloid leukemia (AML) and the myelodysplastic syndromes (MDS). In 50% of cases that progress to AML/MDS, the other copy of DDX41 is somatically mutated. However, DDX41 germline/somatic (g/s-mutant) clones are always detected at very low levels, and it remains unclear how they cause disease. Due to the role of DDX41 in ribosome biogenesis, we hypothesized that complete loss of function promotes hematopoietic stem cell (HSC) expansion at the expense of effective differentiation, leading to lack of detection of g/s-mutant clones in sequencing of unfractionated bone marrow (BM). To test this, we performed error-corrected low-input targeted sequencing of purified HSCs (average number 2,191) and unfractionated BM from 11 patients with DDX41-mutant AML/MDS, 10 of whom had no known somatic DDX41 mutation. We discovered in 4/10 patients (40%) occult somatic DDX41 mutations, all of which were dramatically enriched in HSCs compared to unfractionated BM (average variant allele frequency 41% vs. 6.9%). On flow cytometry analysis, g/s-mutant cases exhibited a dramatic expansion of Lin-CD34+CD38- cells (74% of CD34+ cells) compared with germline mutation-only cases (27%, p = 0.0196), indicative of an arrest at an early stage of differentiation. To evaluate the functional potential of DDX41 g/s-mutant HSCs, and to understand the molecular pathways driving their expansion, we developed RDM-Seq, which leverages single-cell RNA-seq, single cell DNA-seq with cell surface protein measurements (Abseq), and Mitochondrial mutations identifiable across all datasets to perform genotype-aware differential gene expression analysis. We performed RDM-Seq on two DDX41 g/s-mutant AML cases and three healthy age-matched controls, totaling 12,222 cells sequenced with scDNAseq/Abseq, and 16,111 cells sequenced with scRNA-seq. One DDX41 g/s-mutant AML had known TET2 and SF3B1 mutations, while the other did not have known co-mutations. Evaluation of scDNA-seq/Abseq data allowed us to trace DDX41 g/s-mutant clones across hematopoietic differentiation, revealing that they comprised 87% of the HSC/multipotent progenitor pool, with their frequency dropping to 45% in committed myeloid progenitors, 14% in granulocyte macrophage progenitors, and 6% in megakaryocyte/erythroid progenitors. Strikingly, g/s-mutant cells completely failed to differentiate into any mature lineages, and this behavior was recapitulated upon engraftment of g/s-mutant HSCs into NSG mice. We identified driver mutations specific to the g/s-mutant clone, including TP53 K132R and EZH2 G625R, which were previously undetected due to their restriction to HSCs/MPPs. In contrast, the TET2 and SF3B1 mutations were present as two separate clones harboring only the germline DDX41 mutation and with intact differentiation, with enrichment in mature monocytes and CD34-CD71+ erythroid precursor cells. We next leveraged mitochondrial mutations identified in scDNA-seq as specific to g/s-mutant clones and also captured by scRNA-seq to perform differential gene expression analyses on g/s-mutant and healthy control HSCs. DDX41 g/s-mutant cells across both patients exhibited decreased expression of genes regulated by promotor H3K27me3. As noted above, we identified a g/s-mutant subclone with a previously undescribed EZH2 G625R mutation located at a highly conserved GWG motif within the SET domain. This suggests that DDX41 mutations may cooperate with perturbations in PRC2-mediated gene regulation to promote HSC expansion. Our findings suggest that germline DDX41-mutant AML and MDS have a distinct etiology compared with sporadic disease. They support a model where DDX41 germline and somatic co-mutations drive expansion of HSCs incapable of effective hematopoiesis, with cytopenias developing upon dropout of HSCs with functional DDX41, and with expanded DDX41 g/s-mutant HSCs detected as blasts. This may explain the clinical features of DDX41-mutant disease, such as severe cytopenias, a hypocellular BM, and the absence of proliferative AML. These studies also show how analyses of unfractionated BM may miss the DDX41 g/s-mutant clones that drive disease phenotypes. Further profiling of HSCs in these patients promises to improve our understanding of the biology of DDX41-associated disease.