Dissecting the basis for differential substrate specificity of ADAR1 and ADAR2

Abstract

Millions of adenosines are deaminated throughout the transcriptome by ADAR1 and ADAR2, modulating double-stranded RNA (dsRNA) immunogenicity and recoding mRNA. The high variability in the susceptibility of different adenosines to editing begs the question of what are the determinants of substrate specificity. Here, we systematically monitor how secondary structure modulates ADAR2 vs ADAR1 substrate selectivity, on the basis of systematic probing of thousands of synthetic sequences transfected into ADAR1-deleted cell lines exogenously expressing either ADAR2 or ADAR1. In both cases, structural disruptions gave rise to symmetric, strand-specific induced editing at a fixed offset, but of varying length: -26 nt for ADAR2, and -35 nt for ADAR1. We dissect the basis for the differences in offset between ADAR1 and ADAR2 via diverse mutants, domain-swaps, and ADAR evolutionary homologs, and reveal that it is encoded by the differential RNA binding domain architecture. We demonstrate that this offset-enhanced editing can allow an improved design of ADAR2-recruiting therapeutics, with proof-of-concept experiments suggestive of increased on-target and potentially decreased off-target editing. Our findings provide novel insight into the determinants guiding ADAR2 substrate selectivity and into the roles of the RNA binding domains of ADAR1 and ADAR2 in mediating differential targeting, and should facilitate the design of improved ADAR-recruiting therapeutics.

Adenosine deaminases that act on RNA, then and now

“…Since the reaction stopped before all adenosines were deaminated, it was proposed that the reaction reached completion when the dsRNA was too single-stranded to be recognized as dsRNA; this process, whereby only a subset of the preferred adenosines are deaminated, is called "selectivity" (Bass, 1997). Recent high-throughput analyses, discussed in more detail below, corroborate the idea that progressive disruption of a completely base-paired A-form double helical structure leads to lower editing levels (Zambrano-Mila et al, 2023).…”

“…How do these structural disruptions affect ADAR editing? While answering this question seems daunting, recent studies using a massively parallel synthetic approach provide a clear advance in our understanding (Uzonyi et al, 2021;Zambrano-Mila et al, 2023).…”

“…Editing would be enhanced at a specific adenosine within the binding site, but many sequence independent sites would be occluded, limiting additional editing sites. Indeed, the recent studies show that structural disruptions increase editing of an adenosine target at a distance, while blocking editing in the intervening region (Uzonyi et al, 2021;Zambrano-Mila et al, 2023).…”

“…Thus, it is possible that structural disruptions promote sequencespecific interactions of the dsRBDs of ADARs in either the major or minor groove. Fascinatingly, it is observed that editing at an AU base-pair, which creates an IU mismatch and thus another structural disruption, propagates the effect by also enhancing editing at a distance, leading to a reiterative model for ADAR editing (Uzonyi et al, 2021;Zambrano-Mila et al, 2023). Not surprisingly, editing of A:C mismatches, which restore basepairing, do not propagate additional editing sites.…”