Impact of spliceosome mutations on RNA splicing in myelodysplasia: dysregulated genes/pathways and clinical associations

Abstract

SF3B1, SRSF2, and U2AF1 are the most frequently mutated splicing factor genes in the myelodysplastic syndromes (MDS). We have performed a comprehensive and systematic analysis to determine the effect of these commonly mutated splicing factors on pre-mRNA splicing in the bone marrow stem/progenitor cells and in the erythroid and myeloid precursors in splicing factor mutant MDS. Using RNA-seq, we determined the aberrantly spliced genes and dysregulated pathways in CD34⁺ cells of 84 patients with MDS. Splicing factor mutations result in different alterations in splicing and largely affect different genes, but these converge in common dysregulated pathways and cellular processes, focused on RNA splicing, protein synthesis, and mitochondrial dysfunction, suggesting common mechanisms of action in MDS. Many of these dysregulated pathways and cellular processes can be linked to the known disease pathophysiology associated with splicing factor mutations in MDS, whereas several others have not been previously associated with MDS, such as sirtuin signaling. We identified aberrantly spliced events associated with clinical variables, and isoforms that independently predict survival in MDS and implicate dysregulation of focal adhesion and extracellular exosomes as drivers of poor survival. Aberrantly spliced genes and dysregulated pathways were identified in the MDS-affected lineages in splicing factor mutant MDS. Functional studies demonstrated that knockdown of the mitosis regulators SEPT2 and AKAP8, aberrantly spliced target genes of SF3B1 and SRSF2 mutations, respectively, led to impaired erythroid cell growth and differentiation. This study illuminates the effect of the common spliceosome mutations on the MDS phenotype and provides novel insights into disease pathophysiology.

Graphical abstract

Figure 1.

Aberrant splicing events in CD34⁺cells of patients with SFmut MDS. (A-C) Venn diagrams showing the aberrant splicing events identified in SF3B1 (A), SRSF2 (B), and U2AF1 (C) mutant MDS cases vs healthy control individuals and patients with SFwt MDS. (D-F) Doughnut charts showing the distribution of the aberrant splicing events identified in SF3B1 (D), SRSF2 (E), and U2AF1 (F) mutant MDS cases by event type. For each category, the number of significant aberrant splicing events was normalized to the total number of events identified by the rMATS pipeline. (G-I) Hierarchical clustering of SF3B1 (G), SRSF2 (H), and U2AF1 (I) mutant MDS samples, with wild-type MDS and healthy control samples using the rMATS-calculated inclusion levels of the 245, 236, and 287 aberrant splicing events identified.

Figure 2.

Gene ontology and Ingenuity pathway analysis of aberrantly spliced genes in SFmut MDS. (A) Venn diagram showing the overlap of significant GOs identified in SF3B1, SRSF2, and U2AF1 mutant MDS, and visualization of the significant BP GO terms common to all splicing factor mutant MDS using a REVIGO treemap. REVIGO panel sizes are inversely proportional to enrichment P values. (B-D) Ranked heat maps, as determined by collective significance across all splicing factor mutation groups, showing the significant dysregulated pathways (B), top 10 transcriptional regulators (C), and top 6 drug/chemical gene sets (D) in SF3B1, SRSF2, and U2AF1 mutant MDS. Only heat map tiles with a -log10pvalue > 1.3 (P value < .05) are shown. Within each heat map, dysregulated pathways, transcriptional regulators and drug/chemical names are ranked by the lowest P-value identified in the SFmut group.

Figure 3.

Associations between aberrant splicing and clinical variables or patient survival. (A-C) Scatterplots of aberrant splicing values in AP1G2 (A), DOM3Z (B), and ERCC3 (C) and neutrophil counts (ANC) in patients with MDS. (D) Scatterplot of aberrant splicing values in NICN1 and platelet counts in patients with MDS. (E-H) Kaplan-Meier survival plots for individual isoforms of PTPRC and IFI44L in our MDS cohort (E and G, respectively), and the Cancer Genome Atlas AML cohort (F and H, respectively).

Figure 4.

Aberrant splicing in BM cell populations of SFmut MDS. (A-B) UpSet plots showing the overlap of aberrant splicing events identified in monocyte (MON), granulocyte (GRA), and erythroid (ERY) precursor cell populations isolated from SF3B1 (A) and SRSF2 (B) mutant MDS patient samples. (C-H) Ranked heat maps showing the top 15 dysregulated pathways (C,F), top 15 transcriptional regulators (D,G), and top 6 drug/chemical gene sets (E,H) in MON, GRA, and ERY populations of SF3B1 mutant and SRSF2 mutant patients with MDS. Only heat map tiles with a -log10pvalue > 1.3 (P value < .05) are shown. Within each heat map, dysregulated pathways, transcriptional regulators and drug/chemical names are ranked by the IPA ranking score.

Figure 5.

Functional effects of AKAP8 and SEPT2 knockdown on erythroid differentiation. (A,H) Real-time quantitative PCR showing the mRNA knockdown of AKAP8 (A) and SEPT2 (H) in erythroid cells. (B,I) Growth curves for erythroid cells with knockdown of AKAP8 (B) and SEPT2 (I). (C,J) Cell cycle analysis of erythroid cells with knockdown of AKAP8 (C) and SEPT2 (J) on day 11 of culture. (D-F and K-M) Flow cytometry quantification of erythroid differentiation. (D and K) Percentage of CD71⁺CD235a⁺ cells in erythroid cultures with knockdown of AKAP8 (D) and SEPT2 (K) on day 11. (E and L) Percentage of CD36⁺CD235a⁺ cells in erythroid cultures with knockdown of AKAP8 (E) and SEPT2 (L) on day 11. (F and M) Percentage of CD71⁻CD235a⁺ cells in erythroid cultures with knockdown AKAP8 (F) and SEPT2 (M) on day 14. (G and N) Number of BFU-E and CFU-E obtained from CD34⁺ progenitors with knockdown of AKAP8 (G) and SEPT2 (N) after 14 days in methylcellulose (colony-forming cell assays). Results shown in A-G were obtained from 5 independent experiments, except for C (3 replicates). Results shown in H-N were obtained from 4 independent experiments. Data represent the mean ± SEM. All P-values were obtained by 1-way ANOVA with Bonferroni’s posttest with the exception of G and N, in which 2-way ANOVA with Bonferroni’s posttest was used. *P < .05; **P < .01; ***P < .001.