A Unified Post-Transcriptional Mechanism Regulates Intron Retention in Splicing Factor-Mutant MDS

Introduction: Splicing factors (SFs)- including SF3B1, SRSF2, and U2AF1- are the most commonly mutated gene group in Myelodysplastic Syndromes (MDS). SF mutations cause distinct alternative splicing patterns (AS), with minimal overlap. Only AS events with significant convergence are those that change intron retention (IR) (Pellagatti et al., Blood 2018, Hershberger et al., Leukemia 2020), suggesting a unified post-transcriptional mechanism underlying IR alterations. Mechanisms by which SF mutations change IR and how such changes impact MDS disease physiology remain unexplored.

Methods: We analyzed RNA-seq datasets from CD34-sorted bone marrow cells of 228 SF3B1-mutant, 106 SRSF2-mutant, 47 U2AF1-mutant, and 59 wild-type (WT) controls (Hershberger et al., Leukemia 2020). IR ratios were quantified using IRFinder, and filtered for introns with a minimum junction-span read count of 10 and no overlapping annotated exonic features. Differential IR (dIR) between SF-mutants and WT, identified as significant (FDR < 0.05) were selected for downstream analysis.

Results: 3650 significant loss of IR events (loIR) were noted across the three mutation groups of which 2302 (63%) were common to all mutations, and 2965 (81%) were present in at least two groups. While previous reports focused on loIR in SF mutations, we also found 3329 gain of IR events (gIR) with 1216 (36%) common to all three and 2005 (60%) in at least two groups. Notably, none of the 6979 IR events showed discordance in IR direction, indicating a remarkably high concordance and likely shared biology. To further characterize these IR events, we analyzed their distribution across nuclear sub-compartments in HEK293T cells (Barutcu et al., Mol Cell 2022). loIR events had higher median IRs in each sub-compartment (median IRs: 0.05, 0.08, 0.06 in nucleoplasm, speckles, and lamina) compared to those without IR changes (median IRs: 0.02, 0.04, 0.03; p<0.001), suggesting enrichment for nuclear-retained introns (nucRIs). This aligns with the paradigm that loIR manifests typically in introns retained in normal states but that undergo more splicing in SF mutant cells. Surprisingly, gIR events also showed significant enrichment for nucRIs (median IRs: 0.1, 0.17, 0.13; p<0.001). This suggests that introns, undergoing a change in IR (irrespective of loIR or gIR) in the presence of SF mutations, have high baseline retention levels in the nucleus. These findings also support our hypothesis that post-transcriptional, rather than co-transcriptional mechanisms, drive the IR perturbations in SF mutants. Consistently, nascent long-read sequencing of SF3B1 mutant and wild-type K562 cells confirmed that co-transcriptional splicing was not responsible for change in IR. RNA processing, including post-transcriptional splicing and export to the cytoplasm, is regulated by RNA binding proteins (RBPs). To identify candidate RBPs regulating IR in these introns, we used Encode Data (RNA-seq of K562 shRNA knockdown of 175 RBPs and CRISPR deletion of 180 RBPs) to determine the effects of loss of these RBPs on IR. Hierarchical clustering of 4181 introns (with dIR >|0.1|) revealed that the IR changes correlated closely with SRSF1 loss and anti-correlated with HNRNPAB and HNRNPA0 loss. This suggests a change in balance of active SR proteins and HNRNPs to underlie the observed changes to IR. Reduction in SRSF1 phosphorylation (pSRSF1) was confirmed in SF-mutant cell lines. pSRSF1 levels are regulated by various kinase families, notably SRPK1/Clk1 and MAPKs. These kinases are affected in turn by DNA damage and replicative stress-features of mutant SF expression (Chen et al., Mol Cell 2020). Gene expression analysis revealed a statistically significant reduction in Clk1 expression in all three mutation states across two independent large patient sample datasets (Pellagatti et al., Blood 2018, Hershberger et al., Leukemia 2020).

Conclusion: Our findings highlight a shared post-transcriptional mechanism underlying broad changes to IR in SF mutations. We attribute this to kinase signaling from replicative stress, common in all SF-mutant MDS, which reduces pSRSF1, disrupting the balance of active SR proteins and HNRNPs. Ongoing studies will validate pSRSF1 using phospho-proteomics in patient samples, explore the functional impact of IR alterations in SF-MDS, and determine ways to target pSRSF1 regulatory pathways and associated stress mechanisms therapeutically.

No relevant conflicts of interest to declare.