Gene Editing for Duchenne Muscular Dystrophy

Duchenne muscular dystrophy (DMD) is a muscle wasting disease that results from a lack of dystrophin protein, which is an essential musculoskeletal protein. Patients are typically non-ambulatory by their teenage years and suffer prematurely fatal respiratory and/or cardiac complications by the third decade of life. DMD is caused by deleterious mutations in the dystrophin gene, which creates an out-of-frame shift leading to a lack of dystrophin protein and manifestation of DMD. Although scientists have had an understanding of the genetic basis of DMD for decades there has been only modest advancement in improving quality of life for these patients.

Becker muscular dystrophy (BMD) is an allelic disease; BMD is also caused by mutations in the dystrophin gene, although these mutations maintain the translational reading frame and thus a truncated, partially functional dystrophin protein is created. BMD patients have a wide range of symptoms, but BMD typically has much less severe symptoms than DMD. Thus, a common approach to creating a therapy for DMD is to shift the DMD genotype to a BMD genotype. One therapy targeting the genetic cause of the DMD by shifting the messenger RNA (mRNA) and thus protein product to one of BMD has been conditionally approved by the US Food and Drug Administration (FDA), but the treatment is transient and thus far has not demonstrated reliable clinical benefit. DMD presents some unique challenges for developing gene therapies. First, the full-length gene is so large that exogenous delivery in size restricted viral vectors is not an option. Second, popular strategies being explored are transient and would require lifelong administration. The work presented in this dissertation utilized gene editing technology. Building on prior proof-of-concept studies, we show a CRISPR-Cas9 system utilizing Staphylococcus aureus Cas9 (SaCas9) can be used to create permanent changes to the dystrophin gene. This technique overcomes the main challenges presented, as editing the native locus does not require delivering the gene exogenously, and CRISPR-Cas9 mediated DNA double stranded breaks result in permanent changes of the genome. Here we further the proof-of-concept body of work for utilizing CRISPR-Cas9 to treat DMD by targeting exon 51 for excision in a humanized mouse model.

Initially, we recognized the need for a relevant small animal model. A majority of DMD in vivo work is done in the mdx mouse or variants of the mdx mouse, which contains a mutated mouse dystrophin gene such that it does not produce dystrophin protein and displays a mild dystrophic phenotype. While this is a useful research tool, in order to move genome editing closer to the clinic we need to be able to test guide RNAs (gRNAs) that target the human dystrophin gene in a small animal model. As the gRNAs target exact sequences of the genome they must be designed to the human DMD gene. These human DMD targeting gRNAs would not match the mouse Dmd gene, and thus there was a clear need for a preclinical humanized small animal model of DMD. We obtained an hDMD/mdx mouse that contains the full-length, healthy, wild type human DMD gene on mouse chromosome 5. Although this mouse has the human DMD gene, it is ultimately a healthy mouse. Thus, we utilized Streptococcus pyogenes CRISPR-Cas9 (SpCas9) to excise exon 52 of the human DMD gene in the mouse zygotes. We identified a founder mouse that lacked exon 52 in the genomic DNA (gDNA) and bred that mouse with the mdx mouse line. Thus, using genome editing, we created the hDMDΔ52/mdx mouse, which lacks both human and mouse dystrophin protein expression. We confirmed this biochemically by sequencing the gDNA to ensure lack of exon 52 between the gRNA targeted sites, lack of exon 52 in the cDNA, and lack of dystrophin protein by both immunohistochemistry (IHC) staining and Western blot. The hDMDΔ52/mdx mouse also displayed a mild dystrophic phenotype compared to its healthy counterpart, the hDMD/mdx mouse. We have characterized this hDMDΔ52/mdx mouse and shown it lacks dystrophin and has a mild dystrophic phenotype, and this mouse will be a meaningful tool for testing potential DMD therapies.

Next, we created a CRISPR-SaCas9 system that would target the human DMD gene for exon 51 excision. While our lab has previously shown efficacy of this method utilizing SpCas9, we switched to the smaller SaCas9 in order to better accommodate the small packaging limit of adeno-associated virus (AAV). gRNAs were designed to target conserved regions in the intronic area flanking exon 51 of dystrophin in both humans and rhesus macaques. gRNAs were tested individually for on-target activity in HEK293T cells and those with on-target activity were assessed for off-target activity in silico. One gRNA upstream of human dystrophin exon 51 and one gRNA downstream of exon 51 were selected based on distance from the exon, percent modification measured by the Surveyor nuclease assay, and potential for off-target activity in humans and rhesus macaques. Those chosen two gRNAs were tested as a deletion pair in both HEK293T cells and immortalized myoblasts from a DMD patient, lacking exons 48 through exon 50 that is correctable by removing exon 51, and shown to create the desired deletion. Currently there is a lack of rules about what makes an effective gRNA, and in particular even the length of the gRNA protospacer sequence for SaCas9 can have effects on on-target activity. Thus, the two chosen gRNAs were tested with protospacer lengths varying from 19 to 23 base pairs (bp) both individually and as deletion pairs in HEK293T cells. The most effective on-target pair was with both gRNA protospacer sequences at 23 bp long. These 23 bp length gRNAs were re-tested in HEK293T cells and DMD patient immortalized myoblasts and shown to be effective at creating deletions in the genome, having that edit carry over in the mRNA of differentiated myoblasts resulting in the loss of exon 51 and the junction of exon 47 to exon 52 when Sanger sequenced, and restored dystrophin protein expression in the differentiated myoblasts by Western blot. Off-target sequences of these 23 bp length protospacers were assessed in silico and ten of the predicted off-target sites for each gRNA were tested in vitro in HEK293T cells by deep sequencing. Although the upstream gRNA did have two off-target sites that had notable small insertion or deletion (indel) rates measured by treated gDNA/untreated gDNA, ultimately all measurable off-target activity was at least two orders of magnitude lower than the on-target rate of indel formation.

Finally, we created a CRISPR-SaCas9 system with gRNAs that target human DMD for exon 51 removal, and these exact gRNAs tested in vitro were tested in vivo in our previously characterized hDMDΔ52/mdx mouse. Initially we did a small proof-of-concept study by packaging our system in AAV8 and performed local injections into the tibialis anterior (TA) muscle of adult hDMDΔ52/mdx mice. 8 weeks after treatment the TA was analyzed. We noted deletion of exon 51 between the gRNA targeted sites in the gDNA, as well as dystrophin protein restoration by IHC and Western blot. While promising, DMD is a systemic disease that affects all skeletal and cardiac muscles. Thus, we next delivered our CRISPR-SaCas9 system using AAV9 systemically by tail vein injections in adult hDMDΔ52/mdx mice or temporal vein injections in neonatal hDMDΔ52/mdx mice. At 16 weeks of age mice were sacrificed for biochemical analysis. Deep sequencing of gDNA at each gRNA target site showed measurable indel formation above the limit of detection in all tissues assayed in mice treated as both adults and neonates. There were a few trends that emerged in this data and hold true throughout analysis of on-target editing: the upstream gRNA is generally more effective at on-target activity than the downstream gRNA, the mice treated as neonates show more on-target activity than mice treated as adults, and there is much more on-target activity in the heart than in the skeletal muscles. Indels are a measure of on-target activity, but we delivered a system to create a deletion and not just individual cuts. Thus, gDNA from the heart and TA of mice treated as adults was assayed by linear amplification sequencing, which revealed approximately 4% deletions of exon 51 in gDNA from the heart and about 1% deletions of exon 51 in gDNA from the TA. Through this method we are also investigated inversions of the targeted sequence and AAV integrations into the targeted cut site, both of which were much more prominently present in the heart gDNA than the TA gDNA. Confident we were able to edit the genome at low, although measurable, levels, we examined changes in mRNA. In the mRNA from hearts of both mice treated as adults and neonates we see clear deletions of exon 51 by endpoint polymerase chain reaction (PCR). Sanger sequencing the deletion band revealed the exact junction of exon 50 to exon 53 as expected. We performed quantitative droplet digital PCR (ddPCR) on cDNA from the heart, TA, diaphragm, and gastrocnemius, and similar to the indel formation we saw the highest amount of exon 51 deletions in the heart cDNA at about 20% in both mice treated as adults and neonates. The deletions in skeletal muscles varied from about 0.15% to about 1.5% and were all measurable above the limit of detection as defined by the average of samples from untreated mice. Lastly, we examined dystrophin protein expression. By Western blot we saw mouse to mouse variability in intensity, but largely some degree of dystrophin protein expression restoration in protein extracted from hearts and gastrocnemius muscles from mice treated as both adults and neonates, although qualitatively the mice treated as neonates have more dystrophin protein expression than those treated as adults. IHC on hearts and TA muscle sections similarly showed variable but nonetheless present dystrophin protein expression restoration in both mice treated as adults and neonates. Consistent with prior data, we saw more dystrophin expression in the heart than in the TA, and this difference is exacerbated in the mice treated as adults.

In sum, the objective of this dissertation was to create a clinically relevant CRISPR-SaCas9 system and test it in vitro and in vivo in a diseased humanized mouse model. This work is an incremental step to propel forward methods to permanently correct the dystrophin gene by gene editing technology to treat DMD. We created a useful mouse model for the field to test preclinical therapies in vivo and make the most of the rapidly advancing gene editing tools. Collectively this work is significant in extending early proof-of-principle studies to a translational strategy for gene editing as a potential treatment for DMD.