U2AF1 Mutant Myeloid Neoplasms are Preferentially Sensitive to In Vivo Inhibition of Nonsense-Mediated RNA Decay

Expression of mutant spliceosome proteins (e.g., U2AF1S34F, SF3B1K700E, or SRSF2P95H) alters RNA splicing in myeloid neoplasms, leading to increased production of nonsense transcripts. We have previously shown that inhibiting the nonsense–mediated RNA decay (NMD) pathway, which is responsible for degradation of nonsense transcripts, preferentially kills cells expressing mutant spliceosome proteins. In this study, we used a novel inhibitor (SMG1i-63) of the kinase SMG1, a key regulator of NMD, to provide in vivo evidence that NMD is also a therapeutic vulnerability for splicing factor mutant myeloid neoplasms. We show that primary mouse acute myeloid leukemia cells and human K562 leukemia cell lines expressing splicing factor mutants were more sensitive than wild-type cells to in vivo inhibition of SMG1 (SMG1i). Disruption of NMD activity by SMG1i led to increased R-loops levels in spliceosome wild-type cells, which are further increased in U2AF1S34F treated cells. This R-loop accumulation was accompanied by an increase in DNA damage. Degradation of R-loops with RNase H1 rescued spliceosome mutant cells from NMD inhibition-induced cell death, indicating that R-loop formation is a primary mechanism of drug sensitivity. In U2AF1S34F cells, SMG1i led to increased detection of NMD transcript isoforms (with reduced but detectable protein levels) for DNA repair genes, including ATR and RAD51. Consequently, SMG1i-induced cell death in splicing factor mutant leukemias could be further enhanced by inhibition of the DNA damage response proteins ATR or RAD51. This study shows that in vivo targeting of NMD is a therapeutic strategy to treat myeloid neoplasms with aberrant splicing. Overall design: RNA was isolated from sorted GFP+ mouse MLL::AF9 BM cells after SMG1i-63 or vehicle treatment using the Quick-RNA MicroPrep Kit (Fisher-Zymo, No. R1050). Quality of isolated RNA was determined using Agilent RNA 6000 Nano chip using the Agilent 4200 TapeStation platform. Quantification was performed using Qubit Fluorometer (Thermo-Fisher Scientific). Total RNA samples with RNA integrity number (RIN) >9 were selected for Illumina sequencing library preparation. The RNA-Seq libraries were prepared using the KAPA RNA Hyper Prep Kit with RiboErase (Roche). After amplification, approximately 300 bp fragments were sequenced on Illumina NovaSeq 6000. The reads were then mapped using STAR (v2.7.0) against the GRCm38.95 version of the mouse genome from the Ensembl consortium. In parallel, gene level counts for mutant and control samples were processed using kallisto (v0.43.1). Differentially expressed genes were identified using DESeq2 (version 1.26.0). Additional filtering was applied to require genes to have 5 reads in at least half the samples, and an FDR < 0.1. TPM values were calculated using StringTie (version 1.3.3). Splicing event classification was per- formed using rMATS-turbo (version 4.1.0). Gene enrichment analysis was performed using clusterProfiler (version 3.10.1). Pathway enrichment analysis was performed using fgsea (version 1.17.0) against GO, Reactome, and MSIGDB gene sets (FDR < 0.1). For K562 same protocol was used for library preparation. The reads were then mapped using STAR (v2.7.0) against the GRCh38.95 version of the human genome from the Ensembl consortium. Pathway enrichment analysis for NMD predicted transcritps was performed using Enrichr.