Abstract
Introduction
Splicing factor (SF) mutations represent a novel class of driver mutations highly prevalent in myelodysplastic syndromes (MDS), where four genes, including SF3B1, SRSF2, U2AF1, and ZRSR2, are most frequently affected. SF3B1 and SRSF2 mutations show prominent specificity to RARS/RCMD-RS and CMML subtypes, respectively. Although abnormal RNA splicing is thought to play a central role in the pathogenic mechanism of mutated SFs, little is known about exact gene targets, whose abnormal splicing is implicated in the pathogenesis of MDS or about the molecular mechanism that explains the unique subtype specificity of SF mutations, especially to those subtypes characterized by increased ring sideroblasts.
Methods
To address these issues, comprehensive analysis of abnormal RNA splicing was performed for a total of 336 MDS patients with different SF mutations. High-quality RNA was extracted from bone marrow mononuclear cells (BM/MNCs) and/or CD34+ cells and subjected to high-throughput sequencing, followed by exhaustive detection of splicing junctions for all relevant reads. Aberrant splicing events associated with different SF mutations were explored by comparing observed splicing junctions between samples with and without mutations. To specifically determine the role of SF3B1 mutations in ring sideroblast formation, CD34+ bone marrow cells from 13 patients with or without SF3B1 mutations were differentiated in vitro into erythroid cells. RNA sequencing was performed on cells recovered on day 7 and day 14 and differentially spliced genes in erythroid cells between SF3B1-mutated and unmutated samples were investigated.
Results
SF3B1, SRSF2, U2AF1, and ZRSR2 were mutated in 28%, 18%, 5%, and 7% of the patients, respectively. First, we compared SF3B1-mutated samples with those without known SF mutations. RNA sequencing of CD34+ cells revealed 230 splicing events significantly enriched in SF3B1-mutated cases, of which 90% (n = 206) were caused by misrecognition of 3' splice sites. A similar result was obtained in the experiment for BM/MNCs, where 177 (83%) out of 206 splicing events significantly enriched in SF3B1-mutated samples were caused by misrecognition of 3' splice sites. These observations were in accordance with the known function of SF3B1 in branch point recognition in the U2 snRNP complex. In both BM/MNCs and CD34+ cells, approximately 70% of the unusual 3' splice sites were located from 5 to 25 bases downstream from the authentic junctions. The bases immediately upstream to these 3' splice sites were more often pyrimidines, which was not accordance with the general rule: the bases next to 3' splice sites are purines, especially guanines. About 50% of these altered 3' splice sites resulted in frameshift, indicating that SF3B1 mutations cause deleterious effects in many genes simultaneously. Next, to explore the genes whose abnormal splicing is responsible for increased ring sideroblast formation, RNA sequencing was carried out for erythroid progenitor cells differentiated in vitro from CD34+ cells from MDS patients with or without SF3B1 mutations. We found that a total of 146 altered 3' splice sites were significantly associated with SF3B1 mutations, of which 87 were overlapped to the aberrant splice sites shown to be enriched in SF3B1 mutated primary MDS specimens. These splice sites were found in genes involved in heme biosynthesis, cell cycle progression, and DNA repair and their consequence was mostly deleterious due to aberrant frameshifts. Abnormal splicing events associated with other SF mutations were also identified. Among these, the most common abnormalities associated with mutated SRSF2 and U2AF1 were alternative exon usage. Misrecognition of 3' splice sites was also common in U2AF1-mutated cases. ZRSR2 mutations were associated with retentions of U12 introns, which is consistent with the known role of ZRSR2 as an essential component of the minor (U12-type) spliceosome.
Conclusion
SF mutations were associated with characteristic abnormal splicing changes in primary MDS samples as well as in vitro cultured cells. Our results provide insights into the pathogenic role of SF mutations in MDS.
No relevant conflicts of interest to declare.
Author notes
Asterisk with author names denotes non-ASH members.