Abstract
Purpose :
Adeno-associated virus (AAV) mediated gene augmentation therapy is the gold standard for treating inherited retinal dystrophies (IRDs). While, wild-type AAV integrates into a specific location in the human genome, other studies have shown extensive integration of recombinant AAV in other organs, but little is known about integration into the retinal genome. Due to the potential consequences of integration and increasing interest in AAV-mediated gene therapy, we used a high-throughput base-resolution approach to investigate this phenomenon.
Methods :
Eyes of four one-month old mice were injected with 1 x 109 viral particles of pAAV-AcGFP. At 30 days, fluorescent fundus imaging was performed, mice sacrificed, eyes enucleated, and retinas removed and flash frozen. Retinal DNA was isolated and subject to the Agilent SureSelect XT HS2 DNA System using custom baits for the pAAV-AcGFP. Libraries were prepared and 2 x 150 bp sequenced on an Illumina HiSeq4000. Reads were aligned to the mm39 mouse genome with BWA-MEM, and integration sites were identified from discordant and split reads with at least 20 soft-clipped bases. Integration sites were RT-PCR validated.
Results :
On average, 60 million reads were generated per sample, with 2.5 million discordant or split. We found 904 to 8,195 unique integration sites per animal, with 327 to 5,290 sites unique to an individual eye in one animal. In all, 105 exons and 1,072 introns contained AAV plasmid sequence, with the remaining sites intergenic; integrated plasmid sequence was typically a few hundred bases. Aside from suspected homologous recombination of the plasmid’s β-globin intron in mouse hemoglobin genes, no other region was over-represented in integration sites, though we identified sequence fragments from the entire plasmid. Pathway analysis revealed various significant pathways among genes with plasmid integration, but none related to cell-cycle or other cancer-related pathways.
Conclusions :
There is little overlap between integration sites in different animals, or between eyes of the same animal, suggesting AAV integration is largely random. Random integration opens the possibility that sites may occur in deleterious regions. Further, this raises concerns regarding multiple injections in IRD patients. Finally, we do not know the consequence of intragenic integration on expression or translation, though exonic integration likely causes frameshifts.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.