RNA-guided DNA integration in mouse muscle as a gene replacement platform

Introduction: Gene-editing therapies are entering the clinic for a broad range of genetic diseases. Gene editing may provide life-long correction of genetic disease by correcting causative mutations. However, expanding the applicability of gene editing to heterogenous patient mutations remains a significant challenge. This work aims to develop versatile gene-editing approaches to treat a diverse range of genetic diseases. Specifically, we use RNA-guided techniques to insert gene-sized DNA fragments into precise loci in the genome for long-lasting expression. We “hijack” high-expressing muscle loci to take advantage of the high transcriptional output to turn muscle into a protein factory. The recent description of myotropic variants of adeno associated virus (MyoAAV) has made new gene editing approaches possible in muscle by reducing the effective dose by ~10X. We used MyoAAV to enable efficient targeted integration in vivo.

Materials and

Methods: RNA-guided integration was performed using CRISPR-Cas9 derived from S. Pyogenes. For integration, we used homology-independent targeted integration (HITI). Briefly, HITI functions by creating simultaneous double strand breaks at the target location in the genome and flanking the desired gene of interest (Fig. A). We produced promoterless transgenes encoding green fluorescent protein (GFP) or human factor 9 (hF9) with flanking gRNA target sites by Gibson cloning and used resulting plasmids to create minicircles and adeno-associated virus (AAV) containing the transgene. We deeply characterized mouse C2C12 myoblasts and differentiated myotubes by genomic integration percentage with a custom Tn5-based unbiased sequencing assay, RNA sequencing, and protein expression. Myotropic AAV (MyoAAV1C capsid) was produced at high titer, purified by ultracentrifugation, and injected into the tibialis anterior of mice. Mice were harvested at 3 weeks and 8 weeks post injection. We performed unbiased unidirectional DNA sequencing to quantify DNA integration percentage and precision. For transcriptomic analysis, we performed qRT-PCR and RNA sequencing. Clotting factor and serum cytokines were analyzed from serum collected from cardiac puncture at endpoints. Finally, we used long-read sequencing to further characterize precision of targeted integration at the genomic locus and the resulting chimeric transcript.

Results, Conclusions, and Discussions: Ckm and Mb were selected because they are the highest expressing muscle gene that does not have a disease associated with haploinsufficiency. All constructs were confirmed with whole plasmid sequencing or sanger sequencing. In cultured C2C12 myoblasts, we detected a significant increase in expression of GFP or hF9 that could be sustained for 30 days in differentiated myotubes. We compared expression from integration into the creatine kinase locus (Ckm) and myoglobin (Mb). Increases in hF9 expression ranged depending on the target and transgene design from 3- to 14-fold increase over a promoterless control with a scrambled gRNA. RNA sequencing was performed to characterize alterations in the transcriptome (Fig B). We noted significant differential expression primarily in the myoglobin treated group. In this group, Mb was downregulated ~80% indicating that intronic DSBs could disrupt expression. Ckm, however, generated a similar increase in hF9 expression without significant loss of Ckm or differential expression. We then generated AAV9 and Myotropic AAV to test integration in vivo in mouse skeletal muscle. We first confirmed that MyoAAV showed ~10-fold increase in muscle transduction as previously described. We use MyoAAV capsid 1C containing SpyCas9 driven by a muscle-specific promoter and a vector containing the gRNA and the promoterless transgene flanked by gRNA cut sites. Here we found a profound 89-fold increase in hF9 expression rivaling transgene overexpression. Note that the scrambled group with promoterless vector expresses the gene to some extent owing to self-promotion by the AAV vector genome. We used an “unbiased” unidirectional sequencing method to quantify gene editing integration efficiency and precision. This method showed an overall editing efficiency of 4% in unselected C2C12s with mostly precise integration, however, we detected some vector truncation and reverse integration events. We used nanopore long-read sequencing to profile the target genomic loci and the resulting transcript. This allows detection of larger structural variants and simultaneous methylation profiling. Taken together, this data shows the promise of full-length gene integrations to treat a range of genetic disease. Future work should seek to improve efficiency of integration and precision with next-generation RNA-guided integrases and transposases.

Acknowledgements (Optional): This work was supported by an NIH/NIBIB R00 #EB023979, ASGCT Career Development Award, University of Arkansas Chancellor’s Innovation Grant, and The Arkansas Bioscience Institute. CEN was supported by the 21st Century Chair in Biomedical Engineering. MHP was supported by the Fulbright Indonesia Research Science and Technology (First) Master’s Degree Program and PhRMA Foundation 2023 Predoctoral Fellowship in Drug Discovery. GNB was supported by the University of Arkansas Honors College Undergraduate Research Grant, MSJ was supported by a State Undergraduate Research Fellowship (SURF), the Bodenhamer Fellowship, and the Goldwater Scholarship.