Parsing the roles of DExD-box proteins DDX39A and DDX39B in alternative RNA splicing

Abstract

DExD-box RNA proteins DDX39A and DDX39B are highly homologous paralogs that are conserved in vertebrates. They are required for energy-driven reactions involved in RNA processing. Although we have some understanding of how their functions overlap in RNA nuclear export, our knowledge of whether or not these proteins have specific or redundant functions in RNA splicing is limited. Our previous work has shown that DDX39B is responsible for regulating the splicing of important immune transcripts IL7R and FOXP3. In this study, we aimed to investigate whether DDX39A, a highly homologous paralog of DDX39B, plays a similar role in regulating alternative RNA splicing. We find that DDX39A and DDX39B have significant redundancy in their gene targets, but there are targets that uniquely require one or the other paralog. For instance, DDX39A is incapable of complementing defective splicing of IL7R exon 6 when DDX39B is depleted. This exon and other cassette exons that specifically depend on DDX39B have U-poor/C-rich polypyrimidine tracts in the upstream intron and this variant polypyrimidine tract is required for DDX39B dependency. This study provides evidence that despite a high degree of functional redundancy, DDX39A and DDX39B are selectively required for the splicing of specific pre-mRNAs.

Graphical Abstract

Figure 1.

Genes differentially regulated upon DDX39A and DDX39B depletion: Volcano plots of genes differentially expressed upon (A) DDX39A knockdown or (B) DDX39B knockdown. Upregulated genes (P_adj value < 0.05 and log₂fold change > 0.36) are represented in red, and downregulated genes (P_adj value < 0.05 and log₂fold change <−0.36) are represented in blue.

Figure 2.

Overlap between genes regulated by DDX39A or DDX39B expression: (A) GSEA results of enrichment of DDX39A downregulated genes in control (NSC) over DDX39A-knockdown (left panel) and DDX39B-knockdown (right panel) in HeLa cells. (B) GSEA results of enrichment of DDX39B downregulated genes in control (NSC) over DDX39B-knockdown (left panel) and DDX39A-knockdown (right panel) in HeLa cells.

Figure 3.

Changes in RNA alternative splicing upon DDX39A and DDX39B depletion: (A) number of alternative splicing events that are significantly different (P< 0.05 and |inclusion level difference| > 0.1) between DDX39A or DDX39B knockdown as compared to control (NSC) (CE: cassette exon splicing; IR: intron retention; Alt3’SS: alternative 3′splice site; Alt5’SS: alternative 5′SS; MXE: mutually exclusive exons). (B) Inclusion level differences of exons that are differentially spliced upon DDX39A and DDX39B knockdown. A positive inclusion level difference indicates exons that are included more upon knockdown, and a negative inclusion level difference indicates exons that are skipped more (included less) upon knockdown. (C) Inclusion level differences of introns differentially spliced upon DDX39A and DDX39B knockdown. A positive inclusion level difference indicates introns that are retained (included more) upon knockdown, and a negative inclusion level difference indicates introns that are spliced (included less) more upon knockdown. The red line indicates the median inclusion level difference. (D) GSEA results of enrichment of exons skipped more upon DDX39A knockdown in NSC control compared to DDX39A knockdown (left panel) and DDX39B knockdown (right panel). (E) GSEA results for enrichment of exons skipped more upon DDX39B knockdown in controls compared to DDX39B knockdown (left panel) and DDX39A knockdown (right panel).

Figure 4.

DDX39A does not regulate IL7R exon 6 splicing. (A) RT-qPCR analysis of the transcript levels of DDX39A and DDX39B upon DDX39A knockdown (siD09) and DDX39B knockdown (siD13) compared to control (NSC) in THP1 cells. Statistical significance was assessed using Student's t test (***P ≤ 0.001; **P ≤ 0.01; ns = not significant). (B) The top panel provides a diagrammatic representation of IL7R exon 6 splicing in the presence and absence of DDX39B. The bottom panel shows the RT-PCR analysis of IL7R exon 6 splicing in transcripts from endogenous IL7R. (C, D) Rescue experiments of IL7R exon 6 splicing in HeLa cells stably expressing siRNA-resistant DDX39A transgene under the control of doxycycline promoter. (C) Immunoblot depicting the protein abundance of DDX39A and DDX39B with respect to actin. The numbers indicating protein abundance relative to actin are listed below each immunoblot. (D) RT-PCR analysis of IL7R exon 6 splicing from minigene reporter. In all panels, the data are shown as mean ± s.d., and statistical significance was assessed using one-way ANOVA (****P ≤ ***P ≤ 0.001; **P ≤ 0.01; ns = not significant).

Figure 5.

Exons sensitive to DDX39B have U-poor/C-rich py tracts in their upstream introns. (A) rMAPS analysis showing the enrichment of U2AF2 binding motif in introns upstream of cassette exons that are differentially regulated upon DDX39B depletion. (B) Violin plots depicting the nucleotide frequencies in py tracts of introns upstream of cassette exons skipped more (included less) upon DDX39A (n = 160) and DDX39B (n = 396) depletion compared to exons that are unchanged (n = 150) upon either knockdown. Statistical significance is calculated using Kruskal–Wallis test (****P ≤ 0.0001; ***P ≤ 0.001; *P ≤ 0.05; ns = not significant). (C) Sequence comparisons of py tracts of introns upstream of cassette exons skipped more upon DDX39A and DDX39B depletion.

Figure 6.

U-poor py tract of GOLGA2 exon 8 dictates its dependency on DDX39B expression. (A) Top panel shows the change in alternative splicing of GOLGA2 exon 8 and CELF1 exon 2 upon DDX39B and DDX39A depletion, respectively. (Bottom panel) Construction of GOLGA2 exon 8 reporters with wild-type GOLGA2 intron 7 py tract and CELF1 intron 1 py tract. RT-PCR analysis of GOLGA2 exon 8 reporters splicing changes with (B) wild-type GOLGA2 intron 7 and (C) CELF1intron 1 py tracts in DDX39A (siD09) and DDX39B (siD13) knockdown and control cells. The top panel shows the RT-PCR amplicons on 6% TBE gels and the bottom panel shows the bar graphs depicting the percent exon 8 skipping based on quantification of these gels. Statistical significance was calculated using the Student's t test (**P ≤ 0.01)

Figure 7.

Introns dependent on DDX39B expression have U-poor/C-rich py tracts. (A) Violin plots depicting the nucleotide frequencies in py tracts of introns retained more upon DDX39A (n = 34) and DDX39B (n = 89) depletion compared to introns unchanged (n = 100) upon either knockdown. Statistical significance is calculated using Kruskal–Wallis test (****P ≤ 0.0001; ns = not significant). (B) Sequence comparisons of py tracts of introns retained more upon DDX39A or DDX39B depletion and unchanged introns.

Figure 8.

DDX39A and DDX39B regulate the splicing of their respective introns. Increase in DDX39A and DDX39B intron 6 retention upon DDX39A and DDX39B depletion, respectively. (A) Bed graphs depicting the read coverage for DDX39A intron 6 (left panel) and DDX39B intron 6 (right panel) in DDX39A-depleted (siD06 and siD09), DDX39B-depleted (siD11 and siD13) and control-treated (NSC-5nM and NSC-25nM) samples. (B) RT-qPCR analysis of the expression of DDX39A intron 6-retained transcripts relative to total DDX39A transcripts and DDX39B intron 6-retained transcripts relative to total DDX39B transcripts in DDX39A knockdown (siD09), DDX39B knockdown (siD13) and control HeLa cells. (C) Rescue experiments to measure the levels of DDX39A intron 6 and DDX39B intron 6 retained transcripts using RT-qPCR in HeLa cells stably expressing siRNA-resistant DDX39A or DDX39B transgene under the control of the doxycycline operator. (D) RT-qPCR analysis of DDX39A intron 6-retained and DDX39B-intron 6-retained transcripts in subcellular compartments. MALAT1 and 18S ribosomal RNA (rRNA) were used as normalization controls for measuring transcript levels in the nuclear and cytoplasmic fractions, respectively. In all panels, the data are shown as mean ± s.d. and statistical significance was calculated using one-way ANOVA (****P ≤ 0.0001; ***P ≤ 0.001; **P ≤ 0.01; *P ≤ 0.05; ns = not significant).