A high resolution A-to-I editing map in the mouse identifies editing events controlled by pre-mRNA splicing

Abstract

Pre-mRNA-splicing and adenosine to inosine (A-to-I) RNA-editing occur mostly cotranscriptionally. During A-to-I editing, a genomically encoded adenosine is deaminated to inosine by adenosine deaminases acting on RNA (ADARs). Editing-competent stems are frequently formed between exons and introns. Consistently, studies using reporter assays have shown that splicing efficiency can affect editing levels. Here, we use Nascent-seq and identify ∼90,000 novel A-to-I editing events in the mouse brain transcriptome. Most novel sites are located in intronic regions. Unlike previously assumed, we show that both ADAR (ADAR1) and ADARB1 (ADAR2) can edit repeat elements and regular transcripts to the same extent. We find that inhibition of splicing primarily increases editing levels at hundreds of sites, suggesting that reduced splicing efficiency extends the exposure of intronic and exonic sequences to ADAR enzymes. Lack of splicing factors NOVA1 or NOVA2 changes global editing levels, demonstrating that alternative splicing factors can modulate RNA editing. Finally, we show that intron retention rates correlate with editing levels across different brain tissues. We therefore demonstrate that splicing efficiency is a major factor controlling tissue-specific differences in editing levels.

Figure 1.

Nascent-seq identifies almost 100,000 mouse editing sites. (A) Nascent-RNA was prepared from brains of wild-type mice and subjected to Illumina sequencing (n = 3). After mapping to the mouse genome (mm10), potential RNA-DNA differences (RDDs) were determined using RDDpred (Raw RDDpred). To improve the identification of true RDDs, we removed all RDDs that either could not be mapped unambiguously to one strand, only occurred in one replica, or were also detected in an editing-deficient mouse line (Adar^−/−, Adarb1^−/−; A^−/−, Ab1^−/−). Thereby, we enriched for A-to-G RDDs indicative of A-to-I editing (wt stranded). (B) For validation, we used Sanger sequencing. Twenty-two editing sites were tested, and representative cDNA plus corresponding gDNA sequencing traces are shown (all traces are shown in Supplemental Fig. S2).

Figure 2.

Both ADAR- and ADARB1-mediated editing is primarily associated with intronic regions and repeat elements. (A,B) Adarb1 deletion leads to an ∼60% reduction in editing activity. (A) The total number of editing sites is shown separately for wild type and Adarb1^−/− (sites edited by ADAR). (B) The average editing level of all identified editing sites is plotted for wild type and Adarb1^−/− (overall ADAR editing activity). Error bars = standard deviation. n = 3. (C) The genomic annotation for all sites (dark blue), ADAR sites (blue), and ADARB1 sites (light blue) is given (Exon, Intron, UTR, [n.a.] not annotated/intergenic). In addition, for the subgroups Exon, Intron, and UTR, the predicted effect of editing is given using Ensembl's variant effect predictor (VEP). VEP terms: intron variant (intron); noncoding transcript variant (noncoding); noncoding transcript exon variant (noncoding exon); NMD transcript variant (NMD); regulatory region (regulatory); 5′ or 3′ UTR variant (5′ or 3′ UTR); missense variant (missense); synonymous variant (synonymous). (D) The percentage of editing sites associated with a particular repeat as identified by RepeatMasker is shown. Colors as in C. (E,F) Sequences enriched close to editing sites not associated with repeat elements (upper panel) or associated with repeat elements (lower panel) depicted separately for (E) ADAR or (F) ADARB1. The height of the nucleotide indicates either the degree of overrepresentation (above the line) or the degree of underrepresentation (below the line).

Figure 3.

The persistence of the ECS increases editing levels. Box plot showing binned editing sites according to their log₁₀ ES/ECS coverage (red: log₁₀ ES/ECS coverage < 0 → ES saturated with ECS; blue: log₁₀ ES/ECS coverage > 0 → ES deprived of ECS) and the respective editing level (left side).

Figure 4.

Reduced splicing efficiency globally increases exonic and intronic editing. (A) Either bone marrow cells or primary neurons were treated with the splicing inhibitor meayamcyin (MEA) or vehicle control (DMSO). RNA was isolated after treatment, and poly(A)-selected RNA was subjected to RNA-seq. The relative intronic coverage over editing sites after treatment with MEA is shown. Bone marrow: n = 6, primary neurons: n = 5, error bars = SEM. (B) Bar plot displaying the overall number of significantly changed editing sites for bone marrow and primary neurons. (C) Box plots showing the log₂ fold change (log₂FC) for editing levels in untreated (DMSO) and treated (MEA) primary cells (bone marrow: left panel, primary neurons: right panel) separated into different genic locations (exonic, intronic, UTR, [n.a.] not annotated, i.e., intergenic). Dots represent single editing sites. Significantly changed sites are highlighted in red (P-value < 0.05). (D) Mean editing levels in bone marrow (left panel) or primary neurons (right panel) for grouped editing sites (intron, exon, UTR, n.a.) under DMSO conditions (blue dots) or meayamycin treatment (red dots). The blue (DMSO) or red (meayamycin) line is drawn at the mean editing level of all sites. Gray arrows indicate the shift in mean editing levels upon meayamycin treatment. (E) Validation of changed editing levels by Sanger sequencing. The editing site is marked by an arrow. The change in editing determined by RNA-seq is given below the chromatograms (+ or − indicates an increase or decrease upon treatment as determined by RNA-seq).

Figure 5.

The alternative splicing factors NOVA1 and NOVA2 modulate editing levels. (A) Re-analysis of publicly available RNA-seq data from the cortices of six wild-type and three Nova1^−/− and three Nova2^−/− mice (Saito et al. 2016). The reads have been mapped to the mouse genome (mm10), and the editing level of known editing sites was determined. The total number of significantly changed editing sites is given (P-value < 0.05). (B) Editing levels in cortices of wild-type and either Nova1^−/− (left panel) or Nova2^−/− (right panel) knockout mice were determined, and the change in editing levels was plotted. Dots represent single editing sites. Significantly changed sites are highlighted in red (P-value < 0.05). A separate box plot is given for different genic locations (exonic, intronic, UTR, [n.a.] not annotated/intergenic). (C) Box plot showing the log₂ fold change of editing levels for editing sites with up-regulated splicing efficiency (increase) or down-regulated splicing efficiency (decrease) for wild-type versus Nova1^−/− (green) or wild-type versus Nova2^−/− (blue) mice. Splicing efficiency is determined by a decrease (up-regulated splicing efficiency) or increase in intron-specific coverage (down-regulated splicing efficiency) as determined by DEXSeq (P-value < 0.1).

Figure 6.

Splicing controls tissue-specific A-to-I editing for intron-dependent sites. (A,B) Exonic editing sites either depend on an (A) intronic editing complementary site (intronic ECS) or (B) exonic ECS. Exons are shown as blue bars, introns as thin lines. The editing site (A → I) is depicted. (C,D) Representative graphs for the correlation between average editing level and average exon/intron coverage at the downstream intron (as a measure of splicing efficiency) are shown for sites with a (C) intronic ECS like FLNB and GRIK2 or (D) exonic ECS like GABRA3 across different brain regions (RNA-seq data from GTEx). (E,F) The correlation (r = Pearson correlation coefficient) between average editing levels and average exon/intron ratio for conserved editing sites (Coordinates: hg19; [Chr.] chromosome; [Str.] strand) is plotted when the read coverage allowed calculation of editing levels and exon/intron ratio for at least 10 tissues (n = number of tissues) with minimum of two samples. Color code ranges from blue = strong negative correlation over white = no correlation to red = positive correlation. The significance is indicated by one or two asterisks: (*) P-value < 0.05, (**) P-value < 0.01.