A systematic analysis of intronic sequences downstream of 5' splice sites reveals a widespread role for U-rich motifs and TIA1/TIAL1 proteins in alternative splicing regulation

Abstract

To identify human intronic sequences associated with 5' splice site recognition, we performed a systematic search for motifs enriched in introns downstream of both constitutive and alternative cassette exons. Significant enrichment was observed for U-rich motifs within 100 nucleotides downstream of 5' splice sites of both classes of exons, with the highest enrichment between positions +6 and +30. Exons adjacent to U-rich intronic motifs contain lower frequencies of exonic splicing enhancers and higher frequencies of exonic splicing silencers, compared with exons not followed by U-rich intronic motifs. These findings motivated us to explore the possibility of a widespread role for U-rich motifs in promoting exon inclusion. Since cytotoxic granule-associated RNA binding protein (TIA1) and TIA1-like 1 (TIAL1; also known as TIAR) were previously shown in vitro to bind to U-rich motifs downstream of 5' splice sites, and to facilitate 5' splice site recognition in vitro and in vivo, we investigated whether these factors function more generally in the regulation of splicing of exons followed by U-rich intronic motifs. Simultaneous knockdown of TIA1 and TIAL1 resulted in increased skipping of 36/41 (88%) of alternatively spliced exons associated with U-rich motifs, but did not affect 32/33 (97%) alternatively spliced exons that are not associated with U-rich motifs. The increase in exon skipping correlated with the proximity of the first U-rich motif and the overall "U-richness" of the adjacent intronic region. The majority of the alternative splicing events regulated by TIA1/TIAL1 are conserved in mouse, and the corresponding genes are associated with diverse cellular functions. Based on our results, we estimate that approximately 15% of alternative cassette exons are regulated by TIA1/TIAL1 via U-rich intronic elements.

Figure 1.

Positional bias of intronic U-rich motifs. Distribution of U-rich (4/5 U’s) motifs in the first 100 nucleotides of intronic sequences downstream of constitutively spliced (CS) (A) and alternatively spliced (AS) (B) exons. U-rich motifs are enriched between nucleotides +6 and +30 of introns downstream of CS and AS exons. The difference in the nature of the plots between CS and AS events (i.e., more smooth for CS events) is due to different data set sizes (109,225 vs. 3872). U-rich motifs are not enriched in the corresponding intronic region downstream of pseudo exons (Supplemental Fig. 1). Other U-rich motifs (3/3, 3/4, 4/4, 5/5, 5/6, 6/6, 6/7, 7/7 Us) were also assessed and showed similar distributions in introns adjacent to constitutive exons, whereas only motifs that have at least one positionally flexible nucleotide showed similar distributions in introns downstream of alternative exons (Supplemental Fig. 1).

Figure 2.

Intron sequences with elevated U-content are associated with exons having reduced frequencies of ESEs and increased frequencies of ESSs. Distribution of constitutively spliced (CS) (A) and alternatively spliced (AS) (B) exons with different frequencies of ESEs that are associated with either “U-rich” downstream intronic sequences (blue lines) or non-U-rich downstream intronic sequences (red lines). Panels C and D show the corresponding plots when comparing numbers of exons with different frequencies of ESSs. A single cut-off for U-richness of 28% (>28% U-content defined as U-rich, and <28% U-content defined as non-U-rich) was used. ESE and ESS counts (X-axis) are binned into equiprobable bins over the entire corresponding event set (CS/AS), so that higher counts bins have a higher X-value. Panels E–H repeat this analysis for ESE (E,F) and ESS (G,H) elements for G-rich and non-G-rich intronic flanks. A single threshold of 22%–25% was used to separate events that contain G-rich or non-G-rich flanking intron sequences. (*)P-value too small to compute.

Figure 3.

Alternative exons associated with U-rich downstream intronic sequences are differentially spliced following TIA1/TIAL1 depletion. (A) Immunoblot analysis of HeLa cells transfected with pools of siRNAs targeted against TIA1 and TIAL1 transcripts (lane 1), or transfected with a pool of control, nontargeting siRNAs (lanes 2–4). The upper panel was probed with an anti-TIA1 antibody, the middle panel with an anti-TIAL1 antibody, and the bottom panel with an anti-GAPDH antibody. The anti-TIA1/TIAL1 antibodies each detect a protein doublet, presumably corresponding to variants of these proteins. Lanes 3 and 4 are dilutions corresponding to 30% and 10%, respectively, of the sample loaded in lane 2. (B) RT-PCR analysis of alternative spliced exons associated with U-rich intronic sequences in HeLa cells transfected with either a pool of control nontargeting siRNAs (−) or pools of siRNAs specific to TIA1/TIAL1 (+). (C) RT-PCR analysis of alternative spliced exons associated with non-U-rich introns in HeLa cells transfected with either a pool of control, nontargeting siRNAs (−) or a pool of siRNAs specific to TIA1/TIAL1 (+). (D) Densitometric analyses of the resulting PCR bands were carried out using ImageQuant software. The increase in exon skipping upon TIA1/TIAL1 depletion was calculated as the ratio of the band intensities of exon-included (−)/exon-excluded (−) over the band intensities of exon-included (+)/exon-excluded (+). A ratio equal to 1, as shown by the horizontal line, indicates no change in the level of exon skipping upon TIA1/TIAL1 depletion. The means and standard errors of the ratios from two independent RT-PCR analyses of two independent siRNA transfection experiments were calculated for each alternatively spliced cassette exon and plotted using GraphPad Prism 4. Light gray bars correspond to U-rich alternatively spliced cassettes; dark gray bars correspond to non-U-rich AS cassettes. T/T siRNA indicates TIA1/TIAL1 siRNA; N-T siRNA, nontargeting (control) siRNA.

Figure 4.

The increase in exon skipping correlates with the proportion of U’s and the proximity of the first U-rich motif to the exon–intron boundary. Spearman correlation tests showing significant correlations between the increase in exon skipping upon TIA1/TIAL1 silencing and proportion of U’s within the first 30 nt of the second intron of the splicing cassettes (A), and the proximity of the first U-rich motif to the exon–intron boundary (B). The correlations suggest that exons followed by extended TIA1/TIAL1 binding sites (longer stretches of U’s) and/or by binding sites that are located closer to the 5′ splice sites are more sensitive to the reduced levels of TIA1/TIAL1. EIB, exon–intron boundary; n = 27 alternative splicing cassettes.