Prevention of early-onset cardiomyopathy in Dmd exon 52–54 deletion mice by CRISPR-Cas9-mediated exon skipping

Duchenne muscular dystrophy (DMD) is a disease with a life-threatening trajectory resulting from mutations in the dystrophin gene, leading to degeneration of skeletal muscle and fibrosis of cardiac muscle. The overwhelming majority of mutations are multiexonic deletions. We previously established a dystrophic mouse model with deletion of exons 52–54 in Dmd that develops an early-onset cardiac phenotype similar to DMD patients. Here we employed CRISPR-Cas9 delivered intravenously by adeno-associated virus (AAV) vectors to restore functional dystrophin expression via excision or skipping of exon 55. Exon skipping with a solitary guide significantly improved editing outcomes and dystrophin recovery over dual guide excision. Some improvements to genomic and transcript editing levels were observed when the guide dose was enhanced, but dystrophin restoration did not improve considerably. Editing and dystrophin recovery were restricted primarily to cardiac tissue. Remarkably, our exon skipping approach completely prevented onset of the cardiac phenotype in treated mice up to 12 weeks. Thus, our results demonstrate that intravenous delivery of a single-cut CRISPR-Cas9-mediated exon skipping therapy can prevent heart dysfunction in DMD in vivo.


INTRODUCTION
Duchenne muscular dystrophy (DMD) is the most prevalent X-linked pediatric neuromuscular disease with an incidence of around 1 in 5,000 males. 1-3 DMD exerts a high disease burden, with those afflicted experiencing progressive, systemic muscle wasting early in life. Throughout childhood, independent ambulation is gradually lost, with upper extremity weakness closely following. 2,[4][5][6][7] Skeletal and cardiac muscle are affected, impairing the diaphragm and heart from carrying out their essential functions. Cardiorespiratory failure typically emerges in a patient's mid-teens to late twenties, although advancements in non-invasive ventilation have enabled DMD pa-tients to live, on average, into their thirties and forties. 2,[4][5][6][7] In turn, heart dysfunction resulting from cardiomyopathy is now the leading cause of death among DMD patients because current treatment methods are of limited effectiveness over time. [7][8][9][10] Dystrophin is a subsarcolemmal protein encoded by the DMD gene, which is essential to the association and integrity of linkages of intracellular cytoskeletal elements to the sarcolemma and extracellular matrix of muscle via the dystrophin-associated protein complex (DAPC). 11,12 The DAPC is critical for resilience of muscle against contraction-induced damage/stress. 13,14 Mutations in DMD that prevent expression of functional dystrophin cause DMD. In the absence of dystrophin the DAPC cannot form, and myofibers accumulate damage during regular muscle contractions. Moreover, essential molecular pathways, such as nNOS signaling, are interrupted. [11][12][13][14] These events conclude in widespread muscle necrosis. Fibrotic and adipose tissue then infiltrate and replace the degenerated muscle tissue, leading to gradual weakening of the muscle group. 11 While thousands of unique mutations have been identified in DMD patients, approximately 70% are exonic deletions, which represent an enormous segment of the population. 15,16 The vast majority of DMD mutations occur in two hotspot regions at exons 2-20 and exons 45-55 in the DMD gene. 15,16 To investigate therapeutic avenues applicable to the majority of DMD patients, our lab and others previously generated several mouse models recapitulating DMD deletion mutations. We recently published the generation and characterization of a multiexonic 52-54 Dmd deletion (D52-54) mouse, which recapitulates the genomic architecture of a DMD patient. 17 D52-54 mice lack dystrophin because of disruption of the open reading frame (ORF) in exon 55. Hallmarks of DMD, such as elevated fibrosis in muscle tissue, progressive muscle degeneration, impaired motor function, and elevated serum creatine kinase (CK) are readily observed in the D52-54 mouse. 17 Most notably, the D52-54 mouse model exhibits prominent early-onset cardiac hypertrophy and tachycardia. 17 It is important to note that, while these are symptoms commonly presented by DMD patients, the cardiac phenotype of D52-54 mice does not manifest as dilated cardiomyopathy but instead hypertrophic cardiomyopathy. 17 Nonetheless, the early onset of cardiac dysfunction is a novel feature with immense utility for investigating the efficacy of therapeutics.
One of the most explored therapeutic approaches for treating DMDcausing deletions is exon skipping, which removes the frameshifted exon to restore expression of a shorter, partially functional dystrophin protein. [18][19][20] This essentially converts the DMD phenotype into a Becker muscular dystrophy (BMD)-like one. BMD is more variable, but patients typically exhibit milder symptoms than DMD patients, mostly because of in-frame deletions that permit expression of a partially functional dystrophin protein.
Antisense oligonucleotides (AONs) are a clinically validated strategy for exon skipping in DMD patients. 11,21 AONs mask the chosen exon from the splicing machinery, resulting in its exclusion from the final mature transcript. The efficacy of AONs has been demonstrated to marginally restore production of a truncated dystrophin protein for various DMD mutations, but the effect was enough to improve clinically relevant outcomes. 21,22 Presently, eteplirsen, golodirsen, viltolarsen, and, most recently, casimersen have been approved by the US Food and Drug Administration (FDA) for clinical use. However, AONs are limited by their transient nature, requiring regular readministration, and current inability to be delivered to the heart via systemic delivery. 21,23,24 Thus, while AONs may be able to lessen the disease burden of DMD to some degree, they are unlikely to stave off cardiomyopathy by current delivery methods.
Another therapeutic strategy with several clinical trials underway and more currently enrolling is micro-or mini-dystrophin gene therapy. 11,25,26,27 Here the dystrophin coding sequencing is truncated to its most essential elements to enable its most essential functions while fitting within the limited packaging size of adeno-associated viruses (AAVs). As with exon skipping, the objective is to convert the DMD phenotype to a BMD-like one, with preliminary clinical trial results demonstrating widespread levels of dystrophin expression and possible improvements to muscle function. 26,27 While promising, the transgene will remain as an episome, which will likely be lost due to muscle turnover, thus reducing the therapy's efficacy over time.
With the discovery of CRISPR-Cas9 as a genome editing tool, advancements have progressed rapidly, and in less than a decade, the first therapies utilizing this technology have been approved for treating human genetic disease. 28 The main advantage of genome editing is the potential for permanent correction of genetic mutations, thus addressing the primary cause of the disease rather than its symptoms. Exon skipping can be achieved with CRISPR-Cas9 via two approaches: (1) paired, flanking single-guide RNAs (sgRNAs) that excise the out-of-frame exon and (2) a solitary sgRNA approach harnessing the non-homologous end joining (NHEJ) DNA repair pathway, primarily used by post-mitotic cells such as myofibers, when repairing the double-strand breaks induced by Cas9. 20 NHEJ repair typically results in random insertions and deletions (indels) that can disrupt sequences at and immediately adjacent to the Cas9 cut site. By targeting the splice donor or acceptor site of the out-offrame exon with an sgRNA, the resulting indels from NHEJ can sufficiently disrupt these critical sequences to permanently exclude the exon from the DMD transcript. 20 These exon skipping strategies have significantly restored the dystrophin protein in several in vivo pre-clinical DMD studies using viral and non-viral delivery methods. 20 The most common approach evaluated is the use of paired guides for excision of the mutated exon in the mdx mouse model. 20 To treat our D52-54 mice, we employed AAV9s packaged with Staphylococcus aureus Cas9 (SaCas9) and accompanying sgRNAs, systemically delivered via the temporal vein into D52-54 neonates, to exclude exon 55 from the final Dmd transcript and restore the ORF. We compared the efficacy of three different approaches. First, a pair of sgRNAs flanking exon 55 were used to excise the exon in its entirety (dual guide approach). The second utilized a solitary sgRNA targeting the exon 55 splice donor site to enable exon skipping (single guide approach). Finally, we increased the sgRNA dosage from the single guide approach relative to the SaCas9 (high guide approach). We demonstrated that our dual guide approach resulted in partial recovery of dystrophin expression, notably in the heart. Moreover, a solitary guide at a higher stoichiometry restored dystrophin in the heart at levels sufficient to prevent early-onset cardiac dysfunction in D52-54 mice. However, our single guide and high guide strategies did not significantly increase dystrophin levels in the heart. Our results reveal that single-cut exon skipping is a potential therapeutic avenue for treating the cardiac phenotype, which is the current leading cause of death in DMD patients.

RESULTS
The dual guide approach can excise exon 55 and restore dystrophin expression in D52-54 mice In D52-54 mice, splicing of exon 51 to exon 55 disrupts the ORF and introduces a premature stop codon whereas splicing between exons 51 and 56 does not. 17 Thus, excluding exon 55 from the mature Dmd transcript can restore the ORF and produce a truncated but functional dystrophin protein, converting DMD to BMD. A strategy was devised to employ SaCas9 and a pair of intronic targeting sgRNAs flanking exon 55 to excise this entire coding sequence from the genome ( Figure 1A). In vivo delivery of SaCas9 and sgRNAs was accomplished with a pair of AAV9 vectors. Each sgRNA, driven by a human U6 RNA polymerase III promoter, was packaged into a separate AAV9 alongside a copy of SaCas9 expressed from the constitutively active cytomegalovirus (CMV) promoter ( Figure 1B). Both AAV9s were administered systemically via temporal vein injection into post-natal day 2 (P2) D52-54 neonates, with a GFP-packaged AAV9 serving as a negative control. Six weeks post injection, muscle tissues were collected for analysis.
To evaluate the efficiency of excision of the 5.6-kb fragment encompassing exon 55, we employed a digital droplet PCR (ddPCR) assay quantifying the presence of the anticipated deletion junction. Detection of the deletion junction resulting from SaCas9 cleavage demonstrated an average editing efficiency of 0.75% at the genomic level among the four treated mice ( Figure S1).
To evaluate editing at the transcript level, RT-PCR was performed on cDNA derived from the gastrocnemius, triceps, diaphragm, and cardiac tissues of GFP and dual-guide-treated D52-54 mice. Amplifica-tion of the region between exon 51 and exon 56 enabled edited transcripts to be discerned from the wild type by a reduction in amplicon size from 420 bp to 230 bp. Exon 55-null transcripts were primarily restricted to the heart, and Sanger sequencing validated the presence of the desired deletion ( Figures 1C and 1D). Quantification by qRT-PCR concluded that, on average, among all treated mice, 11% of transcripts in the heart were edited correctly ( Figure S2). Immunofluorescence staining and western blot analysis confirmed the expression of dystrophin protein in cardiac tissue, albeit at relatively sparse levels ( Figures 1E and 1F). While this dual sgRNA strategy successfully excised exon 55, overall editing efficiency and dystrophin recovery were notably low and restricted to cardiac tissue.
Skipping of exon 55 in D52-54 mice by a single guide approach targeting the splice donor enhances dystrophin recovery While successful in restoring dystrophin expression, the low editing rates observed with our dual guide strategy prompted us to devise an alternate strategy. Here a solitary sgRNA was designed to target the splice donor of exon 55. Leveraging the random indels NHEJ yields at Cas9 cut sites, we aimed to disrupt the splice donor site at the genomic level to prevent inclusion of exon 55 in the mature Dmd transcript. This single guide approach has been validated previously in several studies investigating CRISPR-Cas9 as a therapeutic avenue for treating DMD deletions. 29,30-36 A U6-driven sgRNA, which cuts exactly in the splice donor site consensus sequence of exon 55, was packaged alongside SaCas9 expressed under a CMV promoter into a single AAV9 vector (Figures 2A and 2B).
This AAV9 was administered systemically via the temporal vein into P2 D52-54 neonates. An AAV9 encoding GFP was used as a negative control. Six weeks post injection, muscle tissues from these mice were collected. Genomic editing rates at the splice donor site were validated with the online Inference of CRISPR Edits (ICE) tool, using indel formation as an estimate of on-target Cas9 activity. The indel rates were 0.6%, 3.2%, 3.2%, and 8.6% for the gastrocnemius, triceps, diaphragm, and heart, respectively ( Figure S3A). In the heart, all identified indels were generated in proximity to the Cas9 cut site and impacted the splice donor consensus sequence, likely prompting exon 55 exclusion ( Figure S3B).
RT-PCR of cDNA derived from skeletal and cardiac muscle revealed that the largest population of edited transcripts was still in the heart; faint but detectable levels were present in the tibialis anterior, triceps, gastrocnemius, and diaphragm ( Figure 2C). Compared with the dual sgRNA strategy, there was a 1.5-fold increase in edited transcripts (16% of transcripts vs. 11% with the dual guide approach) in the heart with our single sgRNA treatment, as assessed by qRT-PCR; however, this increase did not reach statistical significance ( Figure S2). Dystrophin protein was again detectable by immunofluorescence in the heart and, sparsely, in the gastrocnemius ( Figure 2D). Western blot analysis demonstrated an average of 4.7% protein recovery in the heart ( Figure 2E). These results demonstrate that disrupting the splice donor site of exon 55 with a single guide enhanced editing efficiency The plasmid expressing the in55g1_SD sgRNA (purple arrow) and SaCas9 was packaged into AAV9 and administered systemically through the temporal vein into D52-54 neonates. Treated mice were analyzed 6 weeks post injection. (C) RT-PCR of RNA derived from tibialis anterior (TA), triceps (Tri), heart (H), gastrocnemius (G), and diaphragm (D) tissues from GFP and single-guide-treated mice. (D) Immunofluorescence staining of heart and gastrocnemius tissues detected dystrophin-positive fibers in single-guide-treated mice. Scale bar, 100 mm. (E) Western blot probing for the presence of dystrophin (anti-dystrophin), SaCas9 (anti-HA), and vinculin (anti-vinculin) in GFP-and single-guidetreated mice. The percentage of restored dystrophin relative to wild-type (WT) levels for each treated sample is included in the lane's label. and dystrophin recovery relative to a dual guide exon excision approach.
The high guide approach to increase sgRNA dosage improves some editing outcomes for exon skipping While the single sgRNA strategy exhibited improved outcomes in the hearts of D52-54 mice, editing levels and recovered dystrophin remained relatively low; thus, we sought to further enhance our strategy. Work by Min et al. 29 on CRISPR/Cas9-mediated exon skipping in Dmd exon 44 deletion mice demonstrated that increasing the ratio of sgRNA to Cas9 significantly augmented editing and dystrophin levels. 29 To incorporate this strategy into our own, we employed a second AAV9 vector encoding three additional copies of the splice donor targeting sgRNA driven by three U6 promoters and an accompanying GFP stuffer sequence ( Figure 3A).
P2 D52-54 neonates were injected with both AAV9s via the temporal vein. Two separate cohorts were further treated with a GFP-negative control and, for direct comparison, the previous single sgRNA strategy. To optimize the window for editing to occur, we collected muscle tissue 12 weeks post injection. ICE analysis concluded that the indel formation rate was 2-fold higher in the heart with the high guide approach (7.5% indels) compared with the single guide approach (3.6% indels) and that all indels impacted the splice donor sequence ( Figures S4A  and S4B). High-sgRNA-dosage DNA editing rates trended upward in the gastrocnemius (increase of 0.5%), triceps (increase of 2.3%), and diaphragm (increase of 1.3%) relative to the single guide strategy but did not reach statistical significance ( Figure S4A).
Analysis of cDNA derived from mature mRNA unveiled exon skipping within several skeletal muscles groups as well as cardiac tissue ( Figure 3B). As observed previously, the heart still harbored the largest proportion of edited transcripts, with 19.7% on average omitting exon 55 compared with 16.5% with the single sgRNA method ( Figure 3C). The triceps also demonstrated an increase in edited transcripts from 1.6% to 2.4% ( Figure 3C). However, the increases observed in the heart and triceps lacked statistical significance. Dystrophin-positive myofibers were observed by immunofluorescence in mice treated with the high guide strategy, although these remained sparse in skeletal muscle, with the heart notably more populated (Figure 3D). Quantification showed a 2-fold increase in dystrophin-positive myofibers in the heart with the high guide approach (24%) compared with the single guide approach (12%) ( Figure 3E). Maximum levels of dystrophin protein in the hearts of treated mice were similar between the high guide approach (5.6%) and the single guide approach (5.6%), as determined by western blot ( Figure 3F). Overall, augmenting the sgRNA dosage led to an increase in editing and dystrophin-positive fibers but similar dystrophin protein levels as the single guide approach, which were restricted to the heart.
High-guide-dose-treated D52-54 mice show no heart dysfunction at 12 weeks of age We proceeded to analyze disease phenotypes in D52-54 mice treated with the high guide approach 12 weeks after treatment administra-tion. First, we screened cardiac phenotypes via echocardiography. Increased thickness of the left ventricular wall, suggesting cardiac hypertrophy, tachycardia, and elevated fractional shortening, characteristic of D52-54 mice, returned to wild-type ranges in the high-guidetreated group (Figure 4). Then, we evaluated whether the editing that occurred in skeletal muscle was sufficient to lead to any functional improvement. Forelimb/hindlimb grip strength and contractile assays revealed no difference between high-guide-dose treated D52-54 mice compared with GFP-treated D52-54 controls ( Figures S5A  and S5B). Serum CK levels were not significantly reduced following treatment, although they trended downward, with GFP-and highguide-treated mice having levels 1.7-and 1.3-fold higher than wildtype mice, respectively ( Figure S5C). Taken together, our results show how the high guide approach, while not improving the motor phenotype, was able to restore sufficient dystrophin expression to prevent onset of the cardiac complications associated with our D52-54 Dmd mice.

DISCUSSION
In this study, three CRISPR-Cas9 strategies were applied to restore truncated dystrophin expression in the D52 D52-5454 deletion mouse by skipping Dmd exon 55, thus restoring the ORF. We employed a dual guide, single guide, and high guide approach that showed efficacy in recovering dystrophin expression in the hearts of D52 D52-5454 mice. Moreover, we showed that the optimized high guide could prevent the early-onset cardiac phenotype characteristic of this model.
From the presented data, simplifying the delivery system and editing mechanism resulted in a boost to efficiency and dystrophin recovery. Requiring delivery of two individual but essential AAVs and two coordinated cutting events to excise exon 55 was far too inefficient at restoring dystrophin to be therapeutically impactful. Reduction to a single AAV9 and genomic target greatly improved efficacy with the single guide approach.
The high guide strategy saw an increase in edited nuclei and dystrophin transcripts in the heart because of the elevated sgRNA dosage. However, the maximal dystrophin protein recovered was similar to the single guide approach, suggesting that the increase in skipped transcripts was insufficient to notably improve the levels of expressed dystrophin protein. The on-target efficiency of our sgRNA is likely a limiting factor here, and improved dystrophin recovery may be possible with a more active sgRNA. While this strategy did reintroduce a second AAV9, only additional sgRNA copies were encoded by it. Editing and, thus, dystrophin expression would still occur with transduction of only the first AAV9 encoding the SaCas9 and sgRNA; the second AAV9 should only enhance the degree of editing by increasing sgRNA abundance. The mechanism underpinning the beneficial effect of elevated sgRNA dosage on CRISPR-Cas9 editing is not concretely understood. Min et al. 29 utilized a dual AAV9 CRISPR-Cas9 exon skipping strategy to treat an exon 44 deletion mouse. The second AAV9 encoded three copies of their sgRNA, but they also utilized a 10-to-1 sgRNA AAV-to-Cas9 AAV ratio.
Substantial increases in systemic dystrophin recovery were noted in all analyzed muscle groups. One postulation put forth suggests that Cas9 activity is enhanced in the presence of higher sgRNA levels. 29 Another suggests that Cas9 protein translated from cytoplasmic Cas9 transcripts is capable of translocating to any nuclei in the myofiber, whereas transcribed sgRNAs are restricted to the nuclei their AAV transduced. 29 Therefore, increasing sgRNA abundance with a separate AAV may enable sgRNA delivery to a greater proportion of myonuclei. Unbound sgRNAs are also noted to be prone to degradation, but increasing sgRNA levels may maximize their availability to complex with Cas9. 37 A more likely reason is the preferential loss of single-stranded AAVs encoding multiple sgRNAs because this phenomenon has been published previously. 38,39 This AAV depletion can be mitigated by employing a double-stranded self-complementary AAV (scAAV) instead. 40 If this phenomenon is caused by specific AAV vector loss, then we likely did not see significant improvements in dystrophin restoration with our high guide treatment because we did not employ an scAAV; using scAAVs for this purpose was published after our in vivo experiments concluded. Additionally, we likely established a high editing floor because of delivery of the sgRNA and SaCas9 together in an AAV alongside the second AAV, unlike Min et al. 29 where no sgRNA was packaged in the same AAV as their SpCas9. Dedicated studies are necessary to scrutinize these hypotheses and elucidate conclusions; however, the results of improved editing using increasing sgRNA dosages was observed in this study, albeit to a lesser extent than observed previously.
To the best of our knowledge, this is the first report to demonstrate single-cut CRISPR-Cas9-mediated exon skipping via intravenous delivery to neonatal DMD mice. All other therapeutic, single-cut exon skipping studies for DMD have so far relied on intraperitoneal injections for systemic delivery. 29,32,34,35,[40][41][42][43] This delivery method, while convenient, is not feasible for translation to DMD patients, and such high dystrophin restoration in various skeletal muscles is likely due to intraperitoneal delivery introducing a high concentration of AAVs to the center of mass, particularly near the diaphragm and other proximal muscle groups, which is not currently possible with intravenous routes. Injection into the murine temporal vein, however, is analogous to intravenous injection in humans, better reflecting the treatment regimens applicable to DMD patients.
None of our strategies significantly improved the motor function or skeletal muscle integrity of D52-54 mice because of poor editing rates and dystrophin recovery in skeletal muscle. These results align with several other studies that utilized intravenous injections to systemically deliver CRISPR-Cas9-packaged AAVs to treat DMD murine models. 34,38,[44][45][46][47][48][49][50][51][52][53] All of them evaluated a dual sgRNA approach to excise the out-offrame exon in mature adult mice or neonates, and results were consistent across these studies. The heart typically experienced the greatest degree of editing and dystrophin restoration, whereas these outcomes ranged from minute to undetectable in skeletal muscle. Based on previous studies together with our results, it seems that, when AAV-packaged CRISPR-Cas9 genome editing therapies are delivered intravenously, the heart is the primary muscle group that is edited in mice. Only systemic administration by intraperitoneal injection has consistently demonstrated broad, high levels of dystrophin restoration across cardiac and skeletal muscle tissue. This is likely a limitation of the AAV vector, particularly AAV9, used across all of these studies, with intraperitoneal injection being more efficient than intravenous injection for systemic skeletal muscle delivery. For early efficacy studies, intraperitoneal delivery is useful because it is a less technically demanding approach and, as shown by the literature, is good at achieving high levels of skeletal muscle transduction. 29,32,35,[40][41][42][43]54 However, employing the intended clinical route of administration is strongly recommended by the FDA for preclinical development of gene and genome editing therapies. 55 Additionally, no clinical trials employing AAVs have utilized intraperitoneal delivery. [56][57][58][59] For studies with clinical translatability in mind, intraperitoneal delivery should be avoided and intravenous administration used instead to reflect the approach that would be considered for potential use in DMD patients.
Because our strategy could maintain wild-type heart function up to 12 weeks of age, we demonstrated that, in principle, it is feasible to prevent the onset of cardiac dysfunction present in DMD patients. However, to confirm whether the cardiac phenotype has been fully prevented or only delayed, analysis of our treatment in late stages of the disease (24-52 weeks of age) would be necessary. Improvements are also still necessary to restore sufficient dystrophin in cardiac and skeletal muscle to significantly reduce the overall disease burden of DMD.
Future work should seek to improve AAV delivery to peripheral muscles via intravenous delivery because this seems to be a major limitation of CRISPR-Cas9 therapy effectiveness for treating DMD. scAAVs have also been demonstrated to prevent depletion of sgRNA-containing AAVs in muscle and should be further evaluated for their potential to enhance dystrophin recovery. Increasing muscle tropism via chemical modifications to AAVs is another avenue to pursue. A recent study demonstrated that a dendrimer nanoparticle coating greatly enhanced systemic delivery of AAV9s to skeletal and cardiac muscle of a DMD pig model. 60 Utilization of promoters with specificity for cardiac and skeletal muscle, such as CK8e and MHCK7, may further enhance efficacy while improving the safety profile of a DMD CRISPR-Cas9 therapy. One of the biggest leaps forward in improving AAV delivery has been from Tabebordbar et al., 61 who employed random mutagenesis of a motif within the AAV9 capsid protein, yielding several "MyoAAV" variants following selection. These MyoAAVs improved skeletal and cardiac muscle targeting while detargeting the liver in mice and non-human primates. When used to intravenously deliver a dual guide CRISPR-Cas9 exon skipping therapy into adult mdx mice, dystrophin restoration in numerous skeletal muscles was notably improved with MyoAAV over AAV9. 61 Further increasing the sgRNA-to-Cas9 ratio by delivering a large excess of the second sgRNA-encoding AAV may additionally boost editing outcomes, as Min et al. 29 demonstrated; however, they employed intraperitoneal delivery, and thus the efficacy of a large excess of sgRNAs delivered intravenously remains unevaluated. Increasing the number of sgRNA copies on the SaCas9-encoding AAV could further boost editing but would necessitate miniaturizing every possible feature, such as the promoters, poly(A) signal, and possibly the Cas9 because AAVs have an extremely limited packaging size of $4.7 kb. We additionally did not conduct off-target analysis following treatment because of the objective of this study being efficacy and not safety, with the murine sgRNAs employed not suitable for direct translation to human DMD sequences. This limitation of our study will need to be thoroughly addressed in any future CRISPR-Cas9 therapeutic work seeking to provide pre-clinical safety data.
Here we evaluated several approaches to exclude exon 55 from the final Dmd transcript in a D52-54 mouse model of DMD. Targeting the splice site for exon skipping with a solitary sgRNA substantially improved outcomes relative to a dual sgRNA method of excising the entire exon. Editing was notably enhanced by increasing the sgRNA dosage with a second AAV encoding multiple copies of the exon skipping sgRNA; dystrophin restoration remained similar between both single-cut strategies. While we were unable to improve the motor phenotype because of poor dystrophin recovery in skeletal muscle, sufficient dystrophin was restored in the heart to prevent early-onset heart dysfunction of D52-54 mice after 12 weeks. This work adds to the growing compendium of knowledge on CRISPR-Cas9 strategies for correcting DMD deletions. In this study, we demonstrate application of single-cut exon skipping delivered via a clinically relevant intravenous route into Dmd deletion neonates possessing an early-onset cardiac phenotype. From our findings, we anticipate that CRISPR-Cas9 could be beneficial for preventing and/or mitigating heart dysfunction in DMD patients.

MATERIALS AND METHODS
sgRNA design and cloning S. aureus sgRNAs were designed using the Benchling tool (Table 1). For the dual guide experiment, sgRNAs targeting intron 54 (in54g2) and intron 55 (in55g4) were chosen based on the highest specificity score (in54g2, 83.9; in55g4, 79.5) The sgRNA used for single guide and combo guide experiments was chosen because it targets the Dmd exon 55 splice donor site.
AAV9 production and in vivo delivery AAV9 packaging and titering was conducted by Vigene Biosciences, and viruses were stored at À80 C. AAVs packaged with SaCas9/ sgRNAs or GFP were delivered systemically at 7.5 Â 10 11 genome copies (GCs) per vector for all dual, single, and combo guide experiments via the temporal vein. In brief, P1-P2 male Dmd D52-54 mice were anesthetized on ice and injected with the AAV up to 50 mL. Functional tests were performed 12 weeks post injection, and mice were euthanized by CO 2 inhalation. Tissues were dissected, coated with OCT, and frozen in nitrogen-cooled isopentane, followed by storage at À80 C.

Animal husbandry
All mice were housed at The Centre for Phenogenomics (TCP; Toronto, ON, Canada) under environmental regulation of a 12-h light/ dark cycle with food and water provision in individual units (Techniplast). All animal procedures were conducted in compliance with the Animals for Research Act of Ontario and the Guidelines of the Canadian Council on Animal Care. Animal protocols performed at TCP were reviewed and approved by the local animal care committee.

Grip strength tests
Forelimb and hindlimb grip strength tests were performed by TCP based on TREAT-NMD: DMD_M.2.2.001. Age-matched 12-weekold C57BL/6J and Dmd D52-54 mice were lowered over the grid of the grip strength meter (Bioseb) with the torso parallel to the grid. Forepaws and hindpaws were allowed to attach to the grid before pulling the mouse back by the tail, and the maximal grip strength value of the mouse was recorded. The test was done in triplicates, where the average grip strength value was corrected by the mouse's body weight.

Echocardiography
For echocardiography, male mice were scanned using the Vevo2100 ultrasound machine (VisualSonics, Toronto, ON, Canada) with a 30-MHz transducer as described previously. 62 All mice were scanned under 1.5% isoflurane anesthesia for $20-30 min with careful monitoring of the body temperature to maintain it at 37 C-38 C (TREAT-NMD: DMD_M.2.2.003).

In vivo contraction assay
In vivo contraction tests were performed as described previously. 63 Briefly, contractile activity was measured using the 1300A 3-in-1 Whole Animal System and analyzed using the Dynamic Muscle Control/Analysis 5.5 and 5.3 high-throughput software (Aurora Scientific). The mice were anesthetized with ketamine-xylazine solution at 100 mg/kg and 10 mg/kg to body weight, respectively, through intraperitoneal injection. Percutaneous electrodes were placed in the tibialis anterior and contractile output was measured. Specific tetanic force at 200 Hz was recorded and normalized to body weight.

Genomic DNA isolation
Genomic DNA was isolated using the DNeasy Blood and Tissue Kit (QIAGEN) according to the manufacturer's protocol.

ddPCR
Removal of the 5.6-kb fragment using the dual guide approach was quantified using ddPCR with the QX200 system (Bio-Rad), which was performed at TCAG at the Hospital for Sick Children according to the manufacturer's protocol. The junction resulting from successful deletion and Dmd exon 51 was amplified to quantify either the presence of editing or the wild-type allele, respectively. TaqMan hydrolysis probes labeled with fluorescein amidites (FAM) and hexachloro-fluorescein (HEX) were designed to target the deletion junction and exon 51, respectively. All oligonucleotides and probe sequences can be found in Table 2. To determine the level of editing, droplets with FAM fluorescence were quantified and normalized to the total amount of droplets. A g-block sequence (sequence included below) spanning the intron 54-tointron 55 junction and wild-type C57BL/6 female mouse DNA were used as controls for the dual junction and exon 51 probes, respectively. ICE CRISPR-Cas9 editing using the single and combo guide strategies was evaluated using the online ICE CRISPR analysis tool (Synthego). The 337-bp region encompassing the in55g1_SD-mediated cleavage site was amplified using the primers indicated in Table 2 using tissue-specific DNA from treated and untreated Dmd D52-54 mice. The amplicons were purified using the QIAquick PCR Purification Kit (QIAGEN) and sequenced with Sanger sequencing using the sa_ex55g1_fw primer. Sanger sequencing data from treated mice and untreated mice (control) were uploaded into ICE, which determined the frequency and nature of indel formation.

RT-PCR
For RNA isolation, mouse tissues were sectioned in 30-mm slices and collected in 1.4-mm zirconium bead pre-filled tubes (OPS Diagnostics) and homogenized using a MagNA Lyser (Roche Diagnostic) for two 20-s cycles at 7,000 rpm with 3-min incubation on ice between cycles. TRIzol chloroform (Thermo Fisher Scientific) RNA extraction was conducted on the homogenized tissue. One microgram of RNA with random hexamers was used for cDNA synthesis using the SuperScript III First-Strand Synthesis System (Thermo Fisher Scientific). The cDNA was used for subsequent RT-PCR experiments using the primers in Table 2 to detect unedited and edited transcripts after exon skipping.
qRT-PCR qRT-PCR was performed using PowerUp SYBR Green Master Mix (Thermo Fisher Scientific) on a QuantStudio 3 system (Bio-Rad) ( Table 2). All samples were run in triplicate, and data were analyzed using QuantStudio analysis software (Bio-Rad). Edited transcripts were identified using mus_Dmd_ex51-F2 and mus_Dmd_ex51_56-R primers, which only amplified when Dmd exon 55 was absent and exon 51 was joined to exon 56. Edited transcripts were normalized to the amount of Dmd transcripts expressed using the primers ex 19 mus cdna_F and ex 19 mus cdna_R.

Clinical chemistry
Mice were euthanized using cervical dislocation, and whole blood was collected into tubes from the chest cavity immediately after heart dissection. Blood was centrifuged at 10,000 Â g at 4 C for 5 min. Clear serum was extracted and stored at À80 C. Serum was measured using the Liquid Creatine Kinase Reagent Kit (Pointe Scientific) according to the manufacturer's protocol. In brief, serum was diluted in 1Â PBS at a 1:4 ratio and incubated with the reagent for 2 min. Absorbance was measured at 340 nm, and readings were recorded every 2 min two more times. Final serum CK was calculated based on the manufacturer's protocol, and serum CK was plotted relative to an average of serum CK levels of seven wild-type mice.

Statistical analysis
GraphPad Prism v.7 was used to conduct Student's t test for all statistical analyses.

DATA AND CODE AVAILABILITY
All original data are available from the authors without any restrictions.

DECLARATION OF INTERESTS
The authors declare no competing interests.