U6 snRNA m6A modification is required for accurate and efficient splicing of C. elegans and human pre-mRNAs

Abstract

pre-mRNA splicing is a critical feature of eukaryotic gene expression. Both cis- and trans-splicing rely on accurately recognising splice site sequences by spliceosomal U snRNAs and associated proteins. Spliceosomal snRNAs carry multiple RNA modifications with the potential to affect different stages of pre-mRNA splicing. Here, we show that the conserved U6 snRNA m6A methyltransferase METT-10 is required for accurate and efficient cis- and trans-splicing of C. elegans pre-mRNAs. The absence of METT-10 in C. elegans and METTL16 in humans primarily leads to alternative splicing at 5' splice sites with an adenosine at +4 position. In addition, METT-10 is required for splicing of weak 3' cis- and trans-splice sites. We identified a significant overlap between METT-10 and the conserved splicing factor SNRNP27K in regulating 5' splice sites with +4A. Finally, we show that editing endogenous 5' splice site +4A positions to +4U restores splicing to wild-type positions in a mett-10 mutant background, supporting a direct role for U6 snRNA m6A modification in 5' splice site recognition. We conclude that the U6 snRNA m6A modification is important for accurate and efficient pre-mRNA splicing.

Graphical Abstract

Figure 1.

The absence of mett-10 causes cis- and trans-splicing defects. (A) Overview of pre-mRNA cis-splicing and cis-splicing defects detectable by RNA sequencing. The grey bars represent exons adjacent to alternative splicing events. Orange and blue bars represent sequences that are excluded or included depending on the alternative splicing event type. (B) Overview of pre-mRNA trans-splicing and trans-splicing defects detectable by RNA sequencing. (C) Volcano plot of all transcripts tested for cis-splicing changes between mett-10^−/− and wild-type animals. The X-axis shows PSI differences between mett-10 and wild-type animals, and the y-axis shows the significance in P-values. Transcripts that show increased splicing in mett-10^−/− (p ≤ 0.05 and ΔPSI ≤ −0.1) at a given splice site are coloured green, and transcripts that show reduced splicing in mett-10^−/− (P ≤ 0.05 and ΔPSI ≥ 0.1) at a given splice site are coloured purple. (D) Volcano plot of all transcripts tested for trans-splicing changes between mett-10^−/− and wild-type animals. Transcripts that show increased splicing in mett-10^−/− (P ≤ 0.05 and ΔPSI ≤ −3) at a given splice site are coloured green, and transcripts that show reduced splicing in mett-10^−/− (P ≤ 0.05 and ΔPSI ≥ 3) at a given splice site are coloured purple. −log₁₀P-value of 1 gene has been lowered to fit into the graph. (E) Classification of all significant (FDR < 0.05) cis-splicing defects in mett-10^−/− animals. (F) Classification of all significant (P-value < 0.05) trans-splicing defects in mett-10^−/− animals.

Figure 2.

mett-10 sensitive 5′SSs predominantly have +4A. (A) The sequence motif and frequency analysis of mett-10 sensitive 5′SSs (−3 to + 6) and the alternative 5′SSs that are used more often in mett-10^−/−. The sequence motif shows the probability of bases at each position around the 5′SS. U5 and U6 snRNA binding sequences are shown under the sequence motif logo. The frequency table is coloured based on the U5 snRNA interacting sequence frequency on the y-axis (−2 and −1) and U6 snRNA interacting sequence frequency on the x-axis (+3, +4 and + 5). (B) Normalised coverage of RNA-Seq reads for the pch-2 intron 1 boundary. Alt. 5′SS 1 is used more often in wild-type animals and alt. 5′SS 2 is used more often in mett-10^−/−. Barplot shows the fraction of reads supporting splicing at the Alt. 5′SS 2 over total reads in mett-10^−/− and wild-type animals. (C) Normalised ONT-DRS alignments for the pch-2 intron 1 boundary in wild-type, mett-10^−/−, and mett-10 germline rescued animals. (D) Heatmap showing the correlation between splice site usage in mett-10^−/− (y-axis) and the specific base at position +4 of the 5′SS. (E) Effect size plot for the 5′SS positions −3 to + 6. Negative values indicate bases at the specific position are associated with significantly more alternative splicing, and positive values indicate bases at the specific position are associated with significantly less alternative splicing events. The circles' size corresponds to the base frequency at a given position across all 5′SSs in the genome (e.g. position 1 is predominantly G, and position 2 is predominantly U across all 5′SSs). (F) Histogram for the distance between alternative splice site pairs. The Y-axis shows the number of alternative splice site pairs, and the X-axis shows the distance between the pairs, with negative values indicating the alternative splice site moves upstream and positive values indicating the alternative splice site moves downstream of the original splice site.

Figure 3.

mett-10^−/- animals have increased intron retention and exon-skipping. (A) 5′SS motif analysis of introns with increased (left) and decreased (right) retention in mett-10^−/−. The frequency of sequences corresponding to U5 and U6 binding are shown in the heatmap. (B) Normalized RNA-Seq coverage of Y18H1A.11 intron 3 in mett-10^−/− and wild-type animals. The bar plot shows the fraction of intronic reads over the retained intron. (C) 5′SS motif analysis of exons with increased skipping (upper panel) and increased retention (bottom panel) in mett-10^−/− animals. Sequence motifs are shown for the upstream exon 5′SSs (grey) and the retained/skipped exon (orange). (D) Normalised ONT-DRS alignments for Y18H1A.11 intron 3 in wild-type, mett-10^−/−, and mett-10 germline rescued animals. (E) Normalized RNA-Seq coverage of aqp-2 exon 5 in mett-10^−/− and wild-type animals. Bar plots show the fraction of exonic reads over the skipped exon. (F) Normalized ONT-DRS alignments for aqp-2 exon 5 in wild-type, mett-10^−/−, and mett-10 germline rescued animals.

Figure 4.

The absence of mett-10 affects 3′SS usage. (A) 5′ and 3′SS motif analysis of transcripts that are mett-10 sensitive (upper panel) and the corresponding alternative 3′SSs whose usage increases in mett-10^−/− (bottom panel). (B) Histogram for the distance between alternative splice site pairs. The Y-axis shows the number of alternative splice site pairs, and the X-axis shows the distance between the pairs. Negative values indicate the alternative splice site moving upstream, and positive values indicate the alternative splice site moving downstream of the original splice site. (C) Normalized RNA-Seq coverage of B0001.7 intron 6. The canonical splice position is Alt. 3′SS 1. In wild-type animals, weak upstream splice site Alt. 3′SS 2 is also utilized. The bar plot shows the fraction of reads covering the exon sequence to the right of the canonical splice site. (D) Normalized ONT-DRS alignments for B0001.7 intron 6 in wild-type, mett-10^−/− and mett-10 germline rescued animals. (E) Heat map (left) and bar plots (right) showing the frequency of 5′SS usage at specific splice sites in wild-type and mett-10^−/− animals. (F) Heat map (left) and bar plots (right) showing the frequency of 3′SS usage at specific splice sites in wild-type and mett-10^−/− animals.

Figure 5.

mett-10 is required for efficient trans-splicing. (A) The trans-splicing defects rescued by the germline expression of mett-10 are shown in orange over the total SL trans-splicing defects observed in mett-10^−/− animals as in Figure 1F (blue). (B–D) Examples for the (B) outron retention, (C) alternative 3′ trans-splice site usage, (D) cis-spliced outron retention events showing the normalised coverage of ONT-DRS alignments (left) and the fraction of reads covering the outron sequence (right) in wild-type, mett-10^−/− and mett-10 germline rescued animals. (E) 3′ trans-splice site motif of background transcripts that do not show the trans-splicing defect (top panel), transcripts that show the weak trans-splicing defect (middle panel) and transcripts that show the strong trans-splicing defect (bottom panel).

Figure 6.

snrp-27 is required for the efficient splicing of 5′SSs with +4A. (A) Classification of all significant (FDR < 0.05) cis-splicing defects in snrp-27(az26) M141T animals compared to CB936 (unc-73(e936)). (B) Sequence motif (upper panels) and the frequency of U5 and U6 interacting sequences (bottom panels) of the 5′SSs that are either sensitive to snrp-27(az26) (left panels) or used more often in snrp-27(az26) (right panels). (C) Effect size plot for the 5′SS positions −3 to + 6. Negative values indicate bases at the specific position are associated with significantly more alternative splicing, and positive values indicate bases at the specific position are associated with significantly less alternative splicing events. The circles' size corresponds to the base frequency at a given position across all 5′SSs in the genome. (D) Heat map showing the overlap of mett-10 and snrp-27 sensitive 5′SSs. Row and column names ‘true’ and ‘false’ refer to whether these rows and columns fulfil the condition in the axis names. The expected overlap between mett-10 and snrp-27 is 95, and the observed overlap is 169, P-value = 5.3e-20. (E) Sequence motif analysis of 3′SSs that are either sensitive to snrp-27(az26) (upper panel) or used more often in snrp-27(az26) (bottom panel). 5′SS motif of sensitive 3′SSs is shown on the left. (F) Histogram showing the distance between alternative 3′SS pairs (x-axis) and the number of 3′SS events (y-axis) (G) Heat-map showing the overlap of mett-10 and snrp-27 sensitive 3′SSs. Row and column names ‘true’ and ‘false’ refer to whether these rows and columns fulfil the condition in the axis names. The expected overlap between mett-10 and snrp-27 is 32, and the observed overlap is 11, P-value = 7.14 × 10^–5.

Figure 7.

Human METTL16 is required for efficient splicing of 5′SS with +4A, and in vivo, editing of 5′SS +4A to +4U can restore splicing in mett-10^−/− mutants. (A) Sequence motif analysis of METTL16 sensitive 5′SSs (left) and 5′SSs whose usage increases in METTL16 knock-down cells (right). The p-value for the comparison of motifs is 9.35 × 10⁻¹⁵⁶. (B) Human U6 snRNA showing the positions of m6A and m2G methylations in red. The UACAGA box is highlighted with a red dashed line. Grey boxes indicate 5′ exons and depict interactions between m6A43 and m2G72 with the 5′SSs. M1 depicts the catalytic metal ion. (C) Sequence motif analysis of THUMPD2 sensitive 5′SSs (left) and 5′SSs whose usage increases in THUMPD2 knock-out cells (right). The p-value for the comparison of motifs is 0.98. (D) Normalized read coverage of pipp-4P intron 4 5′SSs. In wild-type animals, splicing was observed most frequently at Alt. 5′SS 1 position followed by Alt. 5′SS 2 position. In mett-10 mutants, splicing was observed most frequently at Alt. 5′SS 3 position followed by Alt. 5′SS 2 position. Adenosine position edited to uracil is shown in red. (E) cDNA visualisation of three splice isoforms for pipp-4P exon 4 and 5. Orange depicts the upstream exon, and green depicts the downstream exon. The hpy166I restriction enzyme recognition sequence is shown by the red bar. (F) Gel electrophoresis analysis of pipp-4P RT-PCR products followed by Hpy166I digestion. The orange arrow shows the undigested product size, and the red arrow shows the digested product size. The table shows the expected product sizes for RT-PCR and digestion. (G) Model for the function of U6 snRNA m6A modification in 5′SS recognition, alternative splicing of 3′SS and trans-splicing. m6A methylated U6 snRNA facilitates accurate recognition of 5′SSs with a //GURAG motif through non-canonical base pairing with +4A (Top panel). SNRNP27K likely functions together with m6A methylated U6 snRNA to recognize 5′SSs with +4A. Efficient recognition of 5′SSs can facilitate alternative splicing at weak 3′SSs. Similarly, efficient recognition of SL RNA 5′SSs by m6A methylated U6 snRNA can facilitate trans-splicing at weak 3′ trans-splice sites.