Fig. 5

Effects of editing on mRNA abundance. a Scatterplot of coefficient estimate and statistical significance (log10-transformed adjusted p value) of editing level as a predictor of host mRNA expression in linear regression, accounting for potential confounding variables. For genes with multiple editing sites associated with expression, the most significantly associated site was used. Dashed line indicates significance threshold based on 10% false discovery rate (FDR). b Scatterplot of editing level coefficient estimate from multiple linear regression models used in A and log2-transformed fold change of the corresponding gene observed in ADAR1 KD cells. Red points indicate expression changes in the direction consistent with the sign of the editing association, in contrast to the gray points. c Editing sites associated with host expression (Expression-Correlated) are more often found in 3′ UTR regions, compared to all differential editing sites (Diff Edited, not including intergenic sites). d Validation of six editing sites affecting host mRNA abundance. For each site, a scatterplot of editing level and log2-transformed mRNA expression in the TCGA data is shown. On the right of each scatterplot is mCherry expression, normalized by eYFP expression, of minigenes with A or G, corresponding to nonedited or edited versions of the sites in the 3′ UTR of each gene. All minigenes were tested in Hela cells with five biological replicates. Normalized expression values (mean ± SD) were compared between edited and nonedited versions by two-sided t-test. *p < 0.05, **p < 0.01, ***p < 0.001. Note that RHOA and MRPS16 editing sites were identified as differential sites in the single-cell RNA-seq analysis (Fig. 3c)