Abstract Mutations in the RNA splicing factor genes SRSF2, SF3B1, U2AF1 and ZRSR2 are present in 50% of patients with myelodysplastic syndromes (MDS), chronic myelomonocytic leukemia (CMML) and secondary acute myeloid leukemia (AML). Although recent studies have suggested that leukemia cells with mutations in U2AF1 are sensitive to genotoxic stress such as with PARP and ATR inhibition, the precise mechanism for this association remains to be elucidated as these cells do not have defects in homologous recombination (HRD). We recently reported that under the replication stress, loss of Microhomology-mediated End-joining (MMEJ), one of the major DNA double strand break repair pathways, sensitizes cells to PARP and ATR inhibitors (Abeykoon and Asada et al. J Clin Invest. 2025). Therefore, we predict that a potential MMEJ defect in the U2AF1 S34-mutated myeloid malignancies offers an opportunity to target other DNA repair pathways via synthetic lethality. First, analysis of RNA-seq data from an erythroleukemia cell line HEL (GSE195620) revealed that an exon 7 skipping of the POLQ gene is one of the most robust alternative splicing events in U2AF1 S34F-mutant samples. POLQ encodes DNA polymerase theta, the primary DNA polymerase in the MMEJ pathway. The POLQ exon7 skipping is significantly observed in patients with the U2AF1 S34F mutation from the Leucegene cohort as well as in endogenous U2AF1 S34F knock-in cells identified by long-read RNA sequencing. Indeed, forced expression of U2AF1 S34F or depletion of U2AF1 markedly increased POLQ exon 7 skipping, resulting in a profound POLQ protein reduction and decrease of the MMEJ activity. We have recently shown that U2AF1, along with another splicing factor, CCAR1, is also required for the proper splicing of the FANCA pre-mRNA and promoting the Fanconi anemia (FA)/BRCA pathway (Harada and Asada et al. Mol Cell. 2024). The FANCA gene has a poison exon in intron 14, and the U2AF1/2 heterodimer is required for the removal of the poison exon from the FANCA mRNA. Indeed, depletion of U2AF1 caused inclusion of the poison exon and loss of the FANCA protein. Interestingly, U2AF1 S34 mutants retain the activity of U2AF1 in the removal of the poison exon. On the contrary, the U2AF1 Q157P mutant did not affect the POLQ exon 7 splicing but failed to remove the FANCA poison exon, which is consistent with a recent finding that U2AF1 S34F and Q157P mutations exhibit a distinct phenotype. By analyzing a previous eCLIP dataset in primary hematopoietic cells (GSE16466), we found that U2AF1 S34F strongly binds to intron 6/7 of the POLQ pre-mRNA at non 3′ splice sites compared with wildtype U2AF1. Consensus motif analysis revealed that the motifs enriched in S34F-specific peaks are all based on a highly conserved “CAG” trinucleotide, suggesting that the S34F mutation drives the selection of cryptic non-3' splice sites with this consensus sequence. In contrast, motifs enriched in WT-specific peaks reveal classic U-rich polypyrimidine tracts (PPTs) upstream of the AG dinucleotide, consistent with canonical U2AF1/U2AF2-dependent splice-site recognition. The absence of PPT motifs among S34F-specific peaks suggests the U2AF1 S34F mutant may bypass the U2AF2 interaction for splicing, as U2AF2 depletion in WT cells had no effect on POLQ splicing or protein levels, pointing to a novel splicing complex or mechanism utilized by the U2AF1 S34F mutant. Finally, we exploited the therapeutic vulnerability of the MMEJ defect in the U2AF1 S34-mutated cells. We used isogenic control and U2AF1 S34F/Y mutant knock-in K562 cells as well as an NKM-1 cell line, which carries an U2AF1 S34F mutation. Since U2AF1 S34F mutations can be seen in lung adenocarcinoma, we also used U2AF1 S34F and S34S knock-in A549 cells for further validation. Genetic perturbation of FA (FANCA, FANCI, or FANCL) and HR (BRCA2) by CRISPR-Cas9 eliminated U2AF1 S34F-mutant A549 cells, which is a consistent phenotype with MMEJ deficient cells. In addition, K562 and A549 cells with the U2AF1 S34 mutation showed a higher sensitivity to a CDK12 inhibitor and RBM39 degrader, which inhibit the HR and FA pathways, respectively. Collectively, myeloid malignancies harboring the U2AF1 S34 mutations are functionally MMEJ-deficient and dependent on other DNA repair pathways including the FA and HR pathways, which can be targeted based on a synthetic-lethal mechanism.
Asada et al. (Mon,) studied this question.