Among genetic aberrations responsible for ineffective hematopoiesis in myelodysplastic syndromes (MDS) and acute myeloid leukemia, somatic mutations in splicing factors such as U2AF1 are of significant interest as they are recurrent, mutually exclusive and early occurring. U2AF1 participates in mRNA splicing through the recognition of the intronic 3' splice site, forming the U2AF complex as a heterodimer with U2AF2. Heterozygous hotspot mutations at S34 or Q157, in the two U2AF1 zinc fingers respectively, result in sequence dependent aberrant splicing, suggestive of altered RNA binding. The mechanism by which these mutations alter U2AF1-U2AF2-RNA interactions has to date not been elucidated, yet understanding the structure-function relationship is critical to devise novel therapeutic strategies that either aim to correct or exploit RNA binding and splicing defects.
To address this issue, we profiled the transcriptome of HEL erythroleukemic cell lines expressing wild-type (WT) and mutant U2AF1. U2AF1 S34F and Q157R mutants induced widespread alterations in splicing patterns of 3250 and 1791 genes respectively, with an overlap of 23.8% genes. On the other hand, we observed only minor alterations in gene expression levels. Meta-analysis and comparison with published RNA sequencing datasets on U2AF1 mutants revealed both conserved and unique splicing changes, with a strong enrichment for genes involved in cell cycle (P=6.7E-15) and DNA repair (P=2.6E-5). Confirming previous literature, U2AF1 S34F preferentially leads to the exclusion of exons preceded by 3' splice sites bearing an intronic UAG motif, while U2AF1 Q157R preferentially excludes exons starting with the AGA motif (Figure 1A). Collectively, the S34F mutation has a stronger effect on splicing, ultimately decreasing the global translation state of cells.
To understand how mutations eventually result in the observed splicing alterations, we also profiled with unprecedented resolution the RNA interactome of the physiological and pathological U2AF heterodimer. We first performed enhanced crosslinking immunoprecipitation (eCLIP) on U2AF1 WT, U2AF1 mutants and U2AF2. Comparison of U2AF1 and U2AF2 binding profiles revealed a high degree of similarity, suggesting that they mostly bind to RNA as a tight dimer. Only by performing fractionated eCLIP on U2AF1 we were able to isolate, at the molecular level, the individual contributions of the U2AF components in the recognition of the 3' splice site. In particular, we deconvolved the U2AF2 signal, insisting on the polypyrimidine region, and the U2AF1 signal, peaking on the AG dinucleotide at the intronic end (Figure 1B). Importantly, the S34F mutant displays an aberrant binding profile, with a specific peak on the nucleotide in position -3, matching the sequence specificity previously observed in aberrant splicing events (Figure 1A-B). Systematic analysis of bound junctions suggests a complex model where the S34F mutation does not simply abrogate the ability of U2AF1 to bind splicing junctions ending with the UAG sequence, but rather alters the conformation of the U2AF complex bound to RNA, resulting in a differential ability to effectively recruit the U2 complex. To confirm this model, we identified and validated a set of gain-of-function splice junctions in genes contributing to hemopoiesis and cell cycle, characterized by increased binding of U2AF1 S34F mutant and parallel decreased binding of U2AF2.
In summary, we identified novel RNA sequence and structure determinants of U2AF complex conformation, uncovered by the binding alterations induced by the U2AF1 S34F mutation. Our data further dissect the complexities of post-transcriptional regulation and provide the basis for development of U2AF directed cancer therapies.
Hunck:B**hringer-Ingelheim Foundation.: Other: During my stay in the Halene Lab I was founded by an MD fellowship.
Asterisk with author names denotes non-ASH members.