Long- and short-read sequencing for AAV-CRISPR in Dmd reveal gene editing outcomes for double-stranded breaks in vivo

Introduction:: Duchenne muscular dystrophy (DMD) is an X-linked muscle wasting disease caused by mutations in the Dmd gene resulting in non-functional dystrophin. Due to the heterogeneous nature of the mutations causing DMD, exon skipping to induce an in-frame mutation can restore functional dystrophin activity. Adeno-associated virus (AAV)-mediated delivery of CRISPR/Cas9 resulting in double-stranded breaks (DSBs) is promising as a curative DMD therapeutic. However, the consequences of AAV-CRISPR require more thorough characterization due to sustained expression post-transduction increasing risks of off- and on-target integration events. Further, PCR-enriched short-read amplicon sequencing at cut sites is the most common method of sequencing CRISPR editing, but PCR bias results in unreliable editing quantification. Third generation long-read sequencing technology such as Oxford Nanopore can sequence large fragments, permitting the detection of large structural variants resulting from DSBs.


Materials and Methods:: We analyze orthogonal long- and short-read sequencing approaches to screen the outcomes of CRISPR technologies at the DNA and RNA-level for DMD. We performed intramuscular injections of AAV-CRISPR or liposomes containing SpCas9 mRNA excising exon 23 using dual gRNAs at intron 22 and 23, in mdx and wild-type mice. Serotypes AAV9 and AAV8 were used for the AAV-CRISPR studies either using an all-in-one SaCas9 or separate SaCas9 and guide RNAs, respectively. Editing in vivo and in vitro with mouse myoblasts was confirmed using short-read indel sequencing. We developed unbiased enrichment methods for short- and long-read sequencing libraries with UMI’s and nanopore Cas9-targeted sequencing (nCATS) to reduce PCR-bias in editing quantification. We additionally examined the impact of AAV-CRISPR on the Dmd transcript in mdx and wild-type mice. We performed 5’RACE enriching for Dmd transcripts and 3’ RACE enriching for AAV to detect off-target AAV integration sites and chimeric transcripts at the 3’ end. These strategies have been replicated in mouse myoblasts using both AAV and electroporation edited samples to assess editing differences in vitro and in vivo. We additionally investigated prime editing-mediated exon skipping or targeted integration by PASTE for DMD in vitro.


Results, Conclusions, and Discussions:: Short-read sequencing validated the editing in three different approaches for AAV-CRISPR and a nonviral alternative excising exon 23 in mdx mice. Nanopore sequencing of a 10kb PCR-enriched amplicon at the target site, in comparison, can detect larger editing events but similarly cannot avoid PCR-bias causing an overrepresentation of large deletions. 5’ RACE nanopore libraries revealed AAV integration at the RNA-level in up to 0.3% of reads, and AAV-Dmd chimeric RNA with the transcriptional start site downstream of the internal CMV promoter of the AAV genome was detected using this enrichment method. However, 3’RACE analyses of transcriptional stop sites seemed to solely enrich for AAV or internal poly A sites within Dmd itself. Further optimizations to the 3’ RACE approach have been used to account for this effect. Prime-editing experiments successfully demonstrated low efficiency exon skipping in vitro, and low editing at the DNA-level, indicating that this approach can achieve exon skipping but requires further efficiency optimization. We additionally were able to use twin-PE to generate a site for Bxb1 integration of a large transgene, which can be used to restore missing or mutated exons. Our results indicate that double-stranded break-free methods are more precise but have a low efficiency at target sites. These results are significant for the translation of AAV-CRISPR for DMD and facilitates the development of unbiased sequencing approaches.


Acknowledgements (Optional): :

References (Optional): :