Systematic analysis of A-to-I RNA editing upon release of ADAR from the nucleolus

Figures

• 

  Figure 1.. Nucleolar disassembly leads to gradual release of nucleolar ADAR into the nucleoplasm A) Rationale. The gradual dissociation of ADAR from the nucleolus was prompted by disassembly of the nucleolus triggered by treatment of cells with a low dose of actinomycin D (Act D, 0.1 µg/ml). The figure depicts the different distribution patterns of ADAR observed: marked nucleolar enrichment (0 min), progressive loss of nucleolar association, with partial redistribution to the nucleoplasm (30 min), and full redistribution of ADAR to the nucleoplasm and near complete exclusion from the nucleolus (60, 90, and 120 min). For reference, fibrillarin (FBL) and pescadillo 1 (PES1) were used, respectively, as markers of an internal and a peripheral layer of the nucleolus. As the nucleolus disassembles, FBL is redistributed to the periphery of the organelle, forming foci that coalesce in “caps”. Representative nucleolar caps are highlighted with green arrowheads. Immunostaining of A549 cells with antibodies specific to ADAR or to the nucleolar marker FBL or PES1. Insets, show a zoomed in cell. Scale bar, 20 µm. For colocalization and for the same analysis on SH-SY5Y cells, see Supplemental Figure 1.…

• 

  Figure 2.. Effect of nucleolar disassembly on editing frequency and the number of edited sites A) Comparison of editing frequencies in A549 and SH-SY5Y cells after 30 min (top left), 1 h (top right), 1.5 h (bottom left), and 2 h (bottom right). Black, green, and red dots respectively indicate, for each treatment time, the edited sites with no significant change in editing frequency, a significant increase, or a significant decrease as compared to untreated cells. Edited sites were selected with a ≥ 10x coverage and detected in each replicate for each condition. A Fisher test was performed on the mean proportion of adenosine converted to guanosine in order to assess the significance of differentially edited sites (q-value <0.05, q is a multiple-testing-corrected measure of statistical significance). B) Number of edited sites (left) and edited sites per base (right) in A549 and SH-SY5Y cells for each treatment time and for untreated cells. For changes in the number of edited sites, see main text. For number of edited sites per base: After 30 minutes, we observed an average increase of 16% in A549 and of 12% in SH-SY5Y compared to untreated cells. In 2 h-treated cells, it was a 1.84 and…

• 

  Figure 3.. Substrate distribution of ADAR editing upon ADAR release from the nucleolus A) Distribution of editing among transcribed gene regions (left) and transcribed repetitive regions (right) in A549 and SH-SY5Y. Edited sites detected with coverage ≥ 10x and in at least two replicates for each condition were included in this tally. Editing occurs mostly in intronic and SINE regions and both proportions increase upon actinomycin treatment. B) Ratio of the number of edited sites per base in intronic versus 3’ UTR regions (left) and in intronic versus CDS regions (right) in A549 and SH-SY5Y. Editing ratios increase with treatment time, suggesting intronic editing as the major contribution to editing. C) Edited sites in ‘unspliced reads’ (left) and ‘spliced reads’ (right). A significant increase in editing is observed in unspliced reads during Act D treatment, but not in spliced reads (p-values = 0.88, 0.63). D) The scatter plots depict the editing frequency of edited sites shared between spliced and unspliced reads. Most sites (~97%) show no difference in editing frequency between the spliced and the unspliced read (q-values > 0.05), which indicates that editing occurred before splicing.

• 

  Figure 4.. The effects of nucleolar disassembly on the transcriptome are monotonous A) Differentially expressed genes and isoforms in A549 and SH-SY5Y cell lines, respectively (red depicts |log2 FC| ≥ 1, q-value 0.01). Boxplots compare the absolute log2 fold changes in transcript levels. B) Variation of gene expression (transcripts per million, TPM) in A549 and SH-SY5Y cells in the course of Act D treatment. In each pair of panels, genes selected for |log2 FC| ≥ 1, q-value <0.01 are separated between left panel and right according to the direction of expression fold-change after two hours of Act D treatment. Expression either increases or decreases monotonously over the duration of the treatment. C) Same as panel B) for isoforms.

• 

  Figure 5.. Regulation of exon skipping and inclusion events is editing-dependent A) Differential alternative splicing between untreated cells and cells treated for 2 h with Act D, according to the type of alternative splicing event indicated. Results computed with rMATS (q-values < 0.05). B) Number of edited sites per base (i.e., normalized to gene or segment length) computed for exon-skipping and exon-inclusion events (events detected in at least two replicates at coverage ≥ 10x) in transcripts showing altered (‘aberrant’) exon skipping/inclusion or not (‘normal’) and in ‘segments’ defined as intronic sequences flanking or not an aberrantly skipped or included exon (median intron length = 2 kb). Segments flanking aberrantly skipped or included exons are significantly richer in editing events. Numbers of Alu elements per base (i.e., normalized to gene or segment length) are also higher in segments flanking aberrantly skipped or included exons (bottom). C) Numbers of edited sites per base (i.e., normalized to cassette or constitutive exon length, respectively) are compared between cassette exons (exons that may be spliced out) and constitutive exons (constitutively included exons) in control…

• 

  Figure 6.. Validation of exon skipping and intron retention events by RT-PCR in SH-SY5Y. Analysis of exon skipping and intron retention by deep sequencing and rMATS (panels A, C, and E) and by RT-PCR (B, D, and F). A-B) ZNF217, C-D) EIF4A2, E-F), MAT2A. rMATS analysis: the inclusion level (IncLevel) was computed as (I/LI) / (I/LI + S/LS), where I is the number of reads of the inclusion isoform, S the number of reads of the skipped isoform, LI the effective length of the inclusion isoform, and LS the effective length of the skipped isoform. RT-PCR analysis: samples loaded on a 2% agarose gel and stained with ethidium bromide. Negative control samples (-) consisted of PCR mix without cDNA. B), Size of PCR products: E4-E5 (product “c”), 207 bp; E3-E5 (product “b”), 270 bp; E3-E4-E5 (product “a”): 405 bp. D), Size of PCR products: E3-E4, (product “b”), 134 bp; E3-intron-E4 (product “a”), 400 bp; E3-E5 (product “d”), 152 bp; E3-E4-E5 (product “c”), 292 bp. EIF4A2 overexpression (right, log2 FC = 0.8, q-value = 1.5x10-85). F), Size of PCR products: E2-E3 (product “g”), 134 bp; E6-E7 (product “f”), 169 bp; E6-intron-E7 (product “e”), 360 bp; E7-E8 (product “d”), 131 bp; E7-intron-E8 (product “c”),…