Allele-Specific Effects Of U2AF1 Mutations On Alternative Splicing

Four independent groups, including ours, recently discovered recurrent mutations in core components of the pre-mRNA splicing complex (the “spliceosome”) in myelodysplastic syndrome (MDS) patient samples using next-generation sequencing approaches. We previously identified mutations in U2AF1 affecting codons S34 (S34F and S34Y) or Q157 (Q157R and Q157P) in 11% of patients with de novo MDS. Although the role of U2AF1 as an accessory factor in the U2 snRNP is well established, it is not yet clear how these mutations affect splicing or contribute to MDS. To determine the effects of S34 and Q157 mutations on U2AF1 splicing activity, we utilized GH1 and FMR1 minigene assays, which we have previously used to assess alternative splicing. Here, we report that recurrent mutations in U2AF1 have allele-specific effects on alternative splicing for both GH1 and FMR1 minigenes when transiently transfected in 293T cells. The U2AF1 S34F mutant allele yielded the most significant increase in alternative splicing activity for both GH1 and FMR1 (GH1: p<0.01; FMR1: p<0.001; n>3 biological replicates), compared to wildtype (WT). The S34Y allele modestly enhanced expression of the alternative isoforms for both minigenes (GH1: p<0.05; FMR1: p<0.01). Conversely, we saw a reduction in expression of the alternative isoforms in cells expressing the Q157R and Q157P alleles compared with WT (p<0.01) or the S34 alleles (p<0.001). In cells expressing the S34F/Q157R mutant (in which both the S34F and Q157R mutations occur on one allele, discovered in one patient with MDS), GH1 and FMR1 splicing was indistinguishable from WT. We then assessed the effect of the S34F mutation on U2AF1 sub-cellular localization and protein interaction within the spliceosome. Using fluorescence immunocytochemistry in transfected 293T cells, we found that S34F U2AF1 localized normally within the nuclear speckles and co-localized with U2AF2 and SRSF2. However, immunoprecipitation of epitope-tagged constructs transfected in K562 cells demonstrated that S34F U2AF1 had reduced direct interaction with U2AF2, compared to WT U2AF1. We next performed RNA-seq to comprehensively determine the effects of the S34F mutation on pre-mRNA splicing and gene expression. We transfected primary human CD34+ cells (derived from cord blood) with constructs expressing either WT or S34F U2AF1 (3 biological replicates for each condition) and sorted for transfected (GFP positive) cells after 24 hours. cDNA libraries depleted of ribosomal RNA were sequenced on the Illumina platform. Transient overexpression of WT vs. S34F U2AF1 did not significantly alter global gene expression profiles (by Cufflnks FPKM) in an unsupervised analysis. Analyses of splice junctions using ALEXA-Seq and EdgeR revealed significant differences in the abundance of known splice junctions (S34F>WT: 104, S34F<WT: 188; FDR<0.05, fold change >4). We also detected expression differences at novel splice junctions resulting from use of either known or novel alternative splice acceptor and/or donor sites (S34F>WT: 135, S34F<WT: 229). Both the known and novel junctions reflect alterations in canonical splicing, exon skipping or retention, intron retention and cryptic site usage. We selected junctions with higher expression in S34F samples for validation in CD34+ cells using RT-PCR and gel electrophoresis. 16/27 genes with known junctions and 10/10 with novel junctions validated (p<0.05; n=3 biological replicates). 9 of the validated changes were tested in primary clinical samples by RT-PCR (n=6 MDS bone marrow with S34F U2AF1 vs. n=6 MDS controls with no U2AF1 splicing mutations) and 5 were confirmed (p<0.01), including alternative splicing involving known (DEK, SERPIN8B, KIAA1033, IFI44) and novel (ABI1) junctions. We conclude that the S34F mutation affects U2AF1 function, leading to aberrant alternative splicing of target genes. Whether alternative splicing perturbs hematopoiesis and contributes to MDS pathogenesis is not yet known.