#apaperaday: Adenine base editing-mediated exon skipping restores dystrophin in humanized Duchenne mouse model

In today’s #apaperaday, Prof. Aartsma-Rus reads and comments on the paper titled: Adenine base editing-mediated exon skipping restores dystrophin in humanized Duchenne mouse model

Today’s pick is from Nature Communications by Lin et al on base editing to restore dystrophin in a humanized animal model. DOI: 10.1038/s41467-024-50340-x

Duchenne is caused by lack of dystrophin. It is possible to let Duchenne patients make partially functional dystrophins as are found in Becker patients by manipulating transcripts (exon skipping) or the dystrophin gene (genome editing).

Note that authors in their introduction do not mention Becker or exon skipping – only gene editing. They argue that much work has been done with Cas9, which causes a double stranded break, while the more recently developed base editors cause a single stranded break (=safer).

I agree, but am a bit upset authors mention that work has been done with humanized animal models in the gene editing space. Eric Olsen’s work is cited, but not Melissa Spencer or Charles Gersbach (using our humanized mouse model). See Creation of a Novel Humanized Dystrophic Mouse Model of Duchenne Muscular Dystrophy and Application of a CRISPR/Cas9 Gene Editing Therapy.

While this may be a petty ego thing from my end, I think our mouse model (with the full-length dystrophin integrated into the mouse genome) predated the humanized models the Olsen lab and this group have made where only one exon is humanized.

We freely share the humanized model AND the deletion models (deletion of exon 52 is published, but we presented deletions of exon 44, 45, 51, and 53 at conferences) with academics under an MTA. So no reason to make new mice! Anyway, back to the paper:

Authors made a humanized mouse by replacing mouse dystrophin exons 50 and 51 with the human exon 51. The resulting model has no dystrophin as expected. Histology shows muscle damage, CK is elevated, and muscle strength decreased. Good that authors check this!

Then, authors introduce the adenine base editor, which is too large to fit in an AAV vector. If you read the last 2 #apaperaday s you will know the answer to this problem: inteins! Indeed, authors use inteins to split the base editor over 2 AAV vectors.

In one of the 2 vectors, they also add 3 guide RNAs (3 times the same ones), which they optimized in HEK cells. Local treatment confirmed that the base editor can do the targeted editing, leading to exon 51 skipping and dystrophin restoration.

Authors noticed no off-target effects at predicted sites where there was overlap with the guides. However, as expected, they notice a bystander effect (edits close to the target site). As the goal is to disrupt the splice site of exon 51, the extra edits are not a problem.

Authors confirmed no missplicing, as expected (but good to check). Then authors did intraperitoneal and intravenous injections with the AAV base editors. They do not mention which AAV serotype they used, but given that heart transduction is very high, I’m guessing AAV9.

How the reviewers did not ask for this information is beyond me. Authors notice about 16-20% editing in heart and less in diaphragm and skeletal muscle (4-6%). Interestingly, this leads to very high levels of exon 51 skipping, and almost all fibers became dystrophin positive.

Also, on western blot, dystrophin levels are very high. This improved muscle strength and rotarod running and decreased CK levels. On heart function at 10 months, authors saw no difference between treated and untreated and wild-type mice (hooray wild-type controls) as expected.

Authors discuss again the need for humanized models, again not mentioning work from colleagues who used them. They also discuss the surprisingly high dystrophin restoration with low gene editing levels. They assume this is because they measure also editing in non-muscle cells.

I can understand this may explain some of the discrepancy, but not all. From the methods, the exon skipping levels were assessed by deep sequencing, and the western blotting seems to have been done properly as well. Maybe nonsense-mediated decay helped.

Furthermore, the dystrophin will accumulate over time and has a long half-life, so this can explain the unexpectedly high dystrophin restoration levels. I agree with the authors that more work is needed.

Also, I hope authors will be more cautious in future publications as they suggest the gene editing can alleviate existing symptoms. Based on the mechanism of action, this is not expected: a slower disease progression yes, but not restoration or reduction of existing function loss.