mRNA editing of the Alzheimer’s risk gene APOE

Abstract

Variants in the human APOE gene govern the risk of Alzheimer’s disease and other disorders. Three major APOE variants in humans reflect C→T replacements at two positions in a single exon: an upstream variant (AE4 site) that differs between the ancestral APOE ε4 allele (APOE4) (C) and human-specific APOE2/E3 (T), and a downstream variant (AE2 site) that differentiates APOE3/E4(C)from APOE2 (T). It has long been assumed that APOE allelotypes are genomically encoded, but here we report that multiple individuals express brain APOE C or U/T variant transcripts that differ from genomically templated versions. We demonstrate up to 10% C→U or U→C nucleotide replacement at AE4 and AE2, but not at other sites, and with no corresponding changes in genomic DNA. Single-cell transcriptomic datasets from brain microglia revealed sporadic (up to ∼8%) C→U replacement at AE2. We found 0.4–1.6% of brain transcripts in human APOE knock-in mice harbor selective C→U changes at either AE4 or AE2 sites. Transfection of HepG2 or Huh7 cells with either mouse or human APOBEC1 led to efficient (>90%) C→U editing of APOE4 mRNA at the AE4 but not AE2 site, with lower (<10%) C→T editing of genomic DNA at the AE4 site by mouse, but not human, APOBEC1. Furthermore, interrogation of proteomic datasets revealed up to 4% of non-genomically encoded APOE peptides in human plasma, indicating that the edited APOE transcripts are functional in vivo. These data suggest that APOE mRNA is subject to RNA editing that interconverts the different allelic forms of APOE. Author summary Human APOE gene variants govern the risk of Alzheimer’s disease (AD) and other disorders. Three alleles are widespread: ancestral E4 and human-specific E3 and E2. AD risk declines in the order E4 > E3 > E2. It has been assumed that the APOE allotype we inherit is laid down at birth, but we report that APOE mRNA is enzymatically edited to convert E4 to E3/E2, and/or E2 to E3/E4. Up to ∼10% conversion was seen in brain, and up to 100% in vitro driven by the RNA-editing enzyme APOBEC1. Proteomic analysis of human plasma argues that edited APOE transcripts are functional in vivo.