DDX41-R525H mutation promotes dysplastic hematopoiesis via aberrant splicing of key regulators of HSC biology

Abstract Germline mutations in the RNA helicase DDX41 are the most common genetic predisposition to myelodysplastic syndromes (MDS) and acute myeloid leukemia (AML), representing 5-7% of AML. Over 70% of patients acquire a somatic mutation at hotspots in the helicase domain of DDX41 in trans, which is linked to disease progression. Hypomethylating agents and bone marrow transplant have shown promise in treatment of DDX41-associated disease, however age and donor-related germline disposition can complicate treatment. Thus, identifying novel therapeutic targets is crucial for DDX41-mutant patients. Recent studies have described the role of DDX41 in RNA splicing, ribosome biogenesis, R-loop resolution and inflammation. However, the causal mechanisms by which somatic DDX41 mutations drive MDS/AML pathogenesis are yet to be determined. To systematically evaluate the function of DDX41, we designed a series of variants including germline pathogenic missense mutations (M1I, G173R, E256K, Y259C, S363del), somatic mutations (R525H, G530D) and structural mutants lacking functional domains (N-terminus, zinc finger, HELICc) and examined their function in isogenic AML cell lines MOLM-13 and THP-1. Using an inducible shRNA targeting the 3'UTR of DDX41, and rescuing with DDX41-wildtype or mutant cDNAs, we found that i) somatic mutants R525H and G530D fail to rescue growth defects in DDX41-deficient cells, ii) germline missense mutants behave like wildtype DDX41 in rescuing growth defects, and iii) all structural mutants fail to rescue cell growth. Our data shows, for the first time, functional differences between DDX41-somatic and DDX41-germline mutations and provides a mechanistic basis to study the function of specific DDX41 domains. Recent structural data suggests that DDX41 associates with the catalytically active spliceosome and interacts with the DEAH-box ATPase DHX8. While DDX41 and DHX8 have been shown to coordinate 3' splice site (SS) selection in C. elegans, we found no evidence of direct interaction using purified human proteins in vitro, evenin the presence of RNA. Moreover, DDX41 shows 100-fold weaker RNA binding affinity than DHX8, suggesting that their interaction in the spliceosome may be indirect, potentially from DDX41 binding RNA and acting independently of DHX8. Based on these observations, we hypothesize that the R525H and G530D mutations likely abrogate the helicase activity of DDX41, causing mis-splicing of specific transcripts that, in turn, promote dysplastic hematopoiesis. To test our hypothesis, we performed differential splicing analysis on isogenic MOLM-13 cells and identified over 500 mis-spliced transcripts specific to DDX41-R525H, with intron retention and alternative 3' SS usage as predominant mis-splicing events. Target prioritization based on FDR, delta-PSI values and long-read RNA-sequencing revealed that DDX41-R525H mutation causes mis-splicing of a subset of genes (e.g. SEPTIN7, DVL1, PIKFYVE, CLK3, MED25), several of which are involved in hematopoietic stem cell (HSC) biology, such as SEPTIN7. RT-PCR validation of high-confidence targets in both MOLM-13 and THP-1 cells confirmed that these aberrantly spliced events are enriched only in R525H but not wildtype cells. Moreover, enhanced crosslinking and immunoprecipitation analysis of DDX41-R525H specific RNA targets showed enrichment for mis-spliced transcripts. Next, we validated this splicing phenotype in purified CD34+ cells from MDS and AML patients with DDX41 germline, combined germline and somatic mutations, and patients lacking DDX41 mutations. We confirmed alternative 3'SS usage of SEPTIN7, among others, only in patients with both germline and somatic mutations but not in samples with germline only or lacking DDX41 mutations. Of note, mis-splicing of SEPTIN7 introduces a premature stop codon resulting in its downregulation, seen by gene expression and western blot analysis in isogenic MOLM-13 cells. Studies have shown that SEPTIN7 regulates cytokinesis, maintains HSC polarity, and HSCs devoid of Septin7 show reduced engraftment potential. In summary, we have uncovered a novel mechanism by which the DDX41-R525H somatic mutation impairs hematopoiesis by aberrant splicing of key HSC regulators such as SEPTIN7. This is the first direct evidence of DDX41 regulating splicing in humans. Further interrogation of these targets in the context of DDX41 mutations could provide insights into developing targeted therapies for patients with DDX41-mutant MDS/AML.