Increasing RNA Editing Target Space Through Chemical Modification 5’ of Target A

Not all target adenines (A) are created equally.  It is well documented (Schneider et al. 2014; Picardi et al 2015) that sequence context plays an important role in rendering A ripe for editing. So to realize the full potential of the platform, it is important to optimize the editing oligos taking sequence context into account.  Chemical optimization will play a critical role here.

ProQR and Wave Life Sciences have demonstrated most advances in this regard and I have previously discussed the promising dZ modification at the orphan base that the Beal lab in collaboration with ProQR have developed.  Today, I will shine light on the recent chemical optimization work by the prolific Beal lab on improving editing when the base 5’ of the target A is an unfavored G (Doherty et al. 2022).

Funding for this work came from ProQR and the Rett Foundation.  Rett is a neurological genetic disorder where RNA Editing looks like a great option for many patients. As another aside, Dr. Beal holds equity in both ProQR (as do I) and CRISPR-based editing company Beam Therapeutics.

 

Structural analysis of guide-target RNA

Using a target RNA derived from the alpha-L-iduronidase (IDUA) gene, Doherty and colleagues first confirmed that when the base 5’ of the target A is a G, editing efficiency is very low when this G is Watson-Crick base-paired to C (known non-preferred context).  As observed before for ADAR deaminase domains (Schneider et al. 2014), the efficiency could be improved when C was replaced by G for full-length ADAR2,  or by either G or A for full-length ADAR 1 p110.

The Beal lab is the leading laboratory when it comes to structurally analyzing ADAR enzymes in complex with guide and target RNAs and then translating these insights to chemically optimized guide RNAs.  They therefore were interested in what was going on in this G-G context and found that the G 5’ of target A was in a syn conformation as opposed to the anti conformation seen in Watson-Crick interactions.

This conformation was stabilized by the 5’ G hydrogen-bonding with its Hoogsteen face to Ganti in the editing oligo.  As a result, the clash of the 2-amino group of the 5’ G base with an amino acid residue in the deaminase flipping loop was averted.  To wit, the flipping loop plays a critical role in editing as it facilitates the rate-limiting flipping of the target A out of the substrate duplex into the deaminase reaction pocket.

Still, there were apparent structural adjustments that had to be made by ADAR2 to accommodate G relative to what is observed in an ideal U:A context.  They therefore replaced G with various A- and G-derived modified nucleosides.  Interestingly, Gà3-deaza dA was found to result in the best editing efficiency, ~2-fold higher than a simple G.

Like G:G, the 5’G was in syn when paired to 3-deaza dA.  However, this conformation was additionally stabilized by not one, but two interactions of Gsyn with its phosphodiester backbone thus rationalizing the improved editing efficiency.

It is great to see the RNA Editing field make progress in elucidating the rules so that most theoretical target sites also become addressable ones with sufficient potency.  Certainly the bases around the to be edited A which are expected to interact with the deaminase domain of ADARs are a hotspot of such research activity.  For example, in this year’s Nature Biotech paper by Wave Life Sciences, an inosine (I) was taking the place of 3-deaza dA thereby doing away with the loop-clashing 2-amino group altogether.   It would have been nice to quickly add on a GalNAc and confirm the findings in animals.  I am guessing though that ProQR has done that experiment already.