At the Core of Highly Active RNA Editing Oligos

 Increasing the inherent potency of ADAR guide RNAs (AgRNAs) through chemical and structural means is critical for the success of RNA Editing. For activity, the AgRNAs need to pair with the elements around the target Adenosine (A) and activate resident cellular ADAR enzymes.

There are 3 ADAR enzymes that are relevant for therapeutic RNA Editing: 2 isoforms of ADAR1 (p110, and the longer p150) and ADAR 2.

ADARs fundamentally comprise of N-terminal double-stranded RNA binding domains (dsRBDs) and a C-terminal deaminase domain which itself can also recognize structural features of the double-strand RNA elements around the target 'A'.  

ADAR1 p110 is fairly uniformly expressed in the nucleus across tissues (possibly least in muscle; Picardi et al), but appears somewhat less effective in oligo-guided RNA editing compared to its interferon-induced cytoplasmic bigger brother p150.  Of note, the lung, one of the initial attractive target organs for this new therapeutic modality is an exception as it naturally expresses more p150 than p110.  Finally, ADAR 2 (mostly nuclear) is most highly expressed in the CNS and also quite effective in oligo-mediated RNA editing.

ADAR expression levels are important as more ADAR means more effective RNA Editing.  It is therefore important to use clever chemistry in the design of AgRNAs to make the most of what ADAR is present in a cell. 

ADAR structure

A breakthrough step in that direction was the structural elucidation by the Beal laboratory at UC Davis of hADAR2 in complex with its dsRNA substrate (Matthews et al 2016).  Although the structures of the hADAR1 isoforms have not been solved yet, the high sequence similarity between the ADARs as well as mutation experiments with hADAR1 (Park et al. 2020) suggest that the deamination reaction is highly similar between the enzymes.  Chemical strategies successful for one enzyme should therefore be directly applicable to the other one.

Having said this, it remains an open question whether one ultimately ought to tailor AgRNAs depending on which ADAR is most relevant in a given target tissue and disease setting.

After capturing the structure of human ADAR2 with substrate dsRNA in which the target 'A' (actually a more easily trapped nucleoside analogue replacement 8-azanebularine) is flipped out of the double-strand into the catalytic deaminase pocket of ADAR2 ready for deamination, Matthews and colleagues noted that this rate-limiting step was apparently facilitated by hydrogen bonding between a glutamate residue of ADAR2 with the base originally opposite target A.  This base is also referred to as the orphan base at this stage of the reaction and cytosine is thought to be preferred at this position.

The structure provides an explanation for the base preference as the nitrogen 3 (N3) in the cytosine base ring can hydrogen bond with the acidic side chain of ADAR2 glutamate 488.  This functions as a firm handshake thus displacing and keeping target A out of the double-strand and pushed into the deaminase pocket.

This model also neatly explains the 60x increased activity of a hADAR2 mutant in which glutamate 488 has been replaced with a glutamine amino acid residue as the amide group of glutamine very happily provides a hydrogen for hydrogen bond formation.

Cytidine analogs

While co-delivery of (engineered) ADAR was strongly considered in the early ADAR RNA Editing days, it suffered from widespread off-targeting and delivery challenges.  So to still take advantage of this structural insight for improving therapeutic RNA Editing efficiency, the Beal group, this time in collaboration with Dutch RNA Editing pure-play ProQR investigated whether nucleoside analogues, in particular cytidine analogues with improved hydrogen donating ability could similarly enhance AàI editing (Doherty and colleagues 2021).

I can only imagine the excitement in the lab when it was found that, indeed, cytidine analogues Benner’s base Z and pseudoisoC could increase the rate of deamination.  Importantly, it was shown for Benner’s base Z that this benefit translated to improved AàI editing in living cells.

It should be noted that the deoxy forms of the nucleoside analogues were used as this is known to be tolerated in the orphan position and should make the AgRNAs more stable. 

Of interest, a bulkier adenosine analogue that should be a ready hydrogen donor impaired deamination in the test tube.  Curiously, Wave Life Sciences successfully used this 8-oxodA in their efficacious alpha-1-antitrypsine AgRNA in the recent Nature Biotech paper discussed in this blog (Monian et al 2022).  It remains to be seen whether the discrepancy is explained by sequence and modification context or whether the in vitro deamination assay is not always predictive of performance in cells.

Regardless, 8-oxodA is one of the analogues covered in the patent applications by UC Davis and ProQR (WO 2020/25237641).

Disclosure: I own shares in ProQR.