Conversion of adenosine in RNA to inosine by ADAR enzymes, termed ‘RNA editing’, occurs at thousands of sites across the transcriptome, and is required for healthy development of the central nervous system. Editing can change the protein sequence, and dampen the innate immune response. This has been extensively documented in the brain and other body tissues, revealing tissue-specific and genetically-determined landscapes of modifications that contribute to multiple diseases. However despite the importance of RNA editing in the CNS, nothing is known about this process in the human retina.
We present the first ever analysis of ~127,000 high-confidence RNA-editing sites across the transcriptomes of >500 donor retinae. We found 80% of editing sites occurred within coding-RNA, with the majority occurring in Alu repeats, and 3’UTR sequence. Important editing-based amino acid substitutions reported in brain are highly prevalent in the retina. Enrichment analysis revealed the hypoxia factor HIF3a among 27 hyper-edited genes, and synaptic vesicle recycling genes to be protected from editing. We found that ~15% of sites are restricted to the retina, and within genes strongly enriched for photoreceptor functions. Differential editing between heathy (N=105) and AMD samples (346), revealed ~300 differentially edited sites that are potential risk factors for AMD pathogenesis. We identified editing QTLs (‘edQTLs’) in 391 donors, and found ~10,000 genetic loci impacting the editing frequency of 2,765 genes. There was little overlap (~15%) between edQTLs and eQTLs reported in the same cohort, however strong agreement was found between edQTLs in retina and 30 tissues, notably brain. To highlight the utility of our results to interpret GWAS, we performed colocalization of edQTLs with ocular GWAS disease risk loci. We identified six loci where RNA editing differences may drive AMD, and rarer diseases independent of eQTLs, suggesting a prominent role for RNA editing in the health of the retina.