Nonsynonymous A-to-I RNA editing contributes to burden of deleterious missense variants in healthy individuals

Adenosine-to-inosine (A-to-I) RNA editing is a very common post-transcriptional modification that can lead to A-to-G changes at the RNA level and compensate for G-to-A genomic changes to a certain extent. It has been shown that each healthy individual can carry dozens of missense variants predicted to be severely deleterious. Why strongly detrimental variants are preserved in a population and not eliminated by negative natural selection remains mostly unclear. Here we ask if RNA editing correlates with the burden of deleterious A/G polymorphisms in a population. Integrating genome and transcriptome sequencing data from 447 human lymphoblastoid cell lines, we show that nonsynonymous editing activities (prevalence/level) are negatively correlated with the deleteriousness of A-to-G genomic changes and positively correlated with that of G-to-A genomic changes within the population. We find a significantly negative correlation between nonsynonymous editing activities and allele frequency of A within the population. This negative editing-allele frequency correlation is particularly strong when editing sites are located in highly important genes/loci. Examinations of deleterious missense variants from the 1000 genomes project further show a significantly higher mutational burden in G-to-A changes than in other types of changes. The level of the mutational burden in G-to-A changes increases with increasing deleterious effects of the changes. Moreover, the deleteriousness of G-to-A changes is significantly positively correlated with the percentage of binding motif of editing enzymes at the variants. Overall, we show that nonsynonymous editing contributes to the increased burden of G-to-A missense mutations in healthy individuals, expanding RNA editing in pathogenomics studies.