Modulating fast skeletal muscle contraction protects skeletal muscle in animal models of Duchenne muscular dystrophy

Duchenne muscular dystrophy (DMD) is a lethal muscle disease caused by absence of the protein dystrophin, which acts as a structural link between the basal lamina and contractile machinery to stabilize muscle membranes in response to mechanical stress. In DMD, mechanical stress leads to exaggerated membrane injury and fiber breakdown, with fast fibers being the most susceptible to damage. A major contributor to this injury is muscle contraction, controlled by the motor protein myosin. However, how muscle contraction and fast muscle fiber damage contribute to the pathophysiology of DMD has not been well characterized. We explored the role of fast skeletal muscle contraction in DMD with a potentially novel, selective, orally active inhibitor of fast skeletal muscle myosin, EDG-5506. Surprisingly, even modest decreases of contraction (<15%) were sufficient to protect skeletal muscles in dystrophic mdx mice from stress injury. Longer-term treatment also decreased muscle fibrosis in key disease-implicated tissues. Importantly, therapeutic levels of myosin inhibition with EDG-5506 did not detrimentally affect strength or coordination. Finally, in dystrophic dogs, EDG-5506 reversibly reduced circulating muscle injury biomarkers and increased habitual activity. This unexpected biology may represent an important alternative treatment strategy for Duchenne and related myopathies.


Introduction
DMD is a lethal, inherited muscle myopathy caused by the absence of dystrophin and destabilization of the dystrophinglycoprotein complex in the cell membrane (1). Dystrophin provides a structural link between the contractile elements of the sarcomere and the basement membrane of muscle (2). When dystrophin is dysfunctional, mechanical stress (force applied over the muscle area) can lead to the opening of membrane stress channels (such as the TRPC family) and the influx of calcium (3).
Ectopic calcium initiates muscle fiber breakdown through several mechanisms, including hypercontraction, activation of proteases, and initiation of mitochondrial apoptotic and necrotic pathways (4). Once muscle fiber breakdown has occurred, skeletal muscle regeneration is possible through the activation of inflammatory pathways and mobilization of muscle stem cells. However, chronic inflammation, cell stress, and stem cell exhaustion reduce the fidelity of this process as patients with DMD get older, leading to fatty and fibrotic tissue accumulation within muscle and compromised physical function (5). Continued muscle degeneration eventually leads to mortality by either cardiac or respiratory muscle failure by the third to fourth decade of life (6).
The connection between mechanical stress and muscle breakdown in DMD has been the subject of scientific research for several decades. Dystrophin is not required for normal sarcomeric organization and contractile activity, as experiments in isolated fibers using dystrophic mdx mouse muscle in which membranes have been removed demonstrate normal force and injury resistance. In contrast, contraction of intact mdx muscle causes sarcolemmal Duchenne muscular dystrophy (DMD) is a lethal muscle disease caused by absence of the protein dystrophin, which acts as a structural link between the basal lamina and contractile machinery to stabilize muscle membranes in response to mechanical stress. In DMD, mechanical stress leads to exaggerated membrane injury and fiber breakdown, with fast fibers being the most susceptible to damage. A major contributor to this injury is muscle contraction, controlled by the motor protein myosin. However, how muscle contraction and fast muscle fiber damage contribute to the pathophysiology of DMD has not been well characterized. We explored the role of fast skeletal muscle contraction in DMD with a potentially novel, selective, orally active inhibitor of fast skeletal muscle myosin, EDG-5506. Surprisingly, even modest decreases of contraction (<15%) were sufficient to protect skeletal muscles in dystrophic mdx mice from stress injury. Longer-term treatment also decreased muscle fibrosis in key disease-implicated tissues. Importantly, therapeutic levels of myosin inhibition with EDG-5506 did not detrimentally affect strength or coordination. Finally, in dystrophic dogs, EDG-5506 reversibly reduced circulating muscle injury biomarkers and increased habitual activity. This unexpected biology may represent an important alternative treatment strategy for Duchenne and related myopathies.
Modulating fast skeletal muscle contraction protects skeletal muscle in animal models of Duchenne muscular dystrophy The inhibitory activity and specificity of EDG-5506 was measured using myofibril preparations of different myosin composition. EDG-5506 completely inhibited fast myofibril ATPase from rabbit psoas muscle (96% IIx/d; ref. 19) with an IC 50 of 0.2 μM but was inactive against cardiac and slow bovine masseter skeletal myofibrils (100% type I) with an IC 50 greater than 100 μM (Figure 1B). EDG-5506 was also a potent inhibitor of other mixed fast myosin myofibrils (mouse gastrocnemius, IC 50 0.4 μM, 21% IIa, 15% IIx, 56% IIb; mouse TA IC 50 0.4 μM, mixed IIa, IIx, IIb; ref. 20 and Supplemental Figure 1B), suggesting equal potency against all fast skeletal myosin proteins. In contrast, EDG-5506 was a partial inhibitor of mixed-composition fast/slow human muscle myofibrils (Supplemental Figure 1C, additional selectivity data in Supplemental Table 1). In addition to EDG-5506, we also developed a second myosin inhibitor, EDG-4131 for confirmation of results (for details of the structure and biochemical activity of EDG-4131, see Supplemental Figure 1D and Supplemental Table 2).
The complex nature of these muscle preparations does not allow determination of compound mechanism. Myosin inhibition was confirmed by measuring the actin-activated ATPase of a purified S1 myosin motor subfragment (21). EDG-5506 inhibited the enzymatic activity of fast skeletal S1 from rabbit psoas but not porcine ventricular cardiac S1 or the more unrelated smooth muscle, S1 with an IC 50 of 0.11 μM and more than 100 μM, respectively ( Figure  1C). Other myosin inhibitors, including fast-selective BTS (22) and cardiac-selective mavacamten (23), inhibit inorganic phosphate (Pi) release, the rate-limiting step in the actin-myosin ATPase cycle (Supplemental Figure 1E). We compared the effect of EDG-5506 and BTS on Pi release with actin. Preincubation of EDG-5506 or BTS and myosin (minus nucleotide) for 30 minutes prior to addition of actin decreased the rate of Pi release (Supplemental Figure 1F).
The specificity of EDG-5506 for fast skeletal myosin over other, unrelated proteins was confirmed using two off-target binding and activity screens. EDG-5506 (at 20 μM) did not exhibit any binding affinity with 90 kinase proteins in vitro (KINOMEscan, Eurofins DiscoverX). A second off-target panel (SAFETYscan E/IC 50 ELECT, Eurofins DiscoverX) tested in vitro enzyme activity against 78 proteins, including GPCR, ion channel, kinases, nuclear hormone receptors, and transporters. In all assays, the IC 50 of EDG-5506 was more than 10 μM (Supplemental Data File 1).
We next tested the ability of EDG-5506 to reduce force in intact muscle. In detergent-treated single-fiber preparations (24), EDG-5506 selectively reduced force in fast rabbit psoas muscle fibers ( Figure 1D) with an IC 50 of 0.7 μM (Supplemental Figure 1G). In contrast, EDG-5506 had no effect on slow skeletal rat soleus (100% type I) or rat cardiac muscle fibers (Supplemental Figure 1, H and I). In ex vivo assays using mouse extensor digitorum longus (EDL; 100% fast, majority IIb myosin, refs. 20, 25), EDG-5506 reduced force in a concentration-and time-dependent manner, completely inhibiting force at 10 μM ( Figure 1E). Submaximal force was also similarly inhibited (Supplemental Figure 1J). Similar experiments with mixed fast/slow mouse soleus (31% type I, 60% type IIa/IIx; ref. 20) muscle yielded partial force inhibition (Supplemental Figure 1K). EDL muscles from mdx mice have an exaggerated injury response to lengthening (eccentric) contraction, exhibiting strength loss with repeated contractions (8). EDL muscles were preincubated with EDG-5506, and peak strain and strength loss was measured rupture and force loss (7), particularly under lengthening (eccentric) conditions. Passive lengthening is not sufficient to cause exaggerated injury (8), and removal of contraction by denervation (9) or treatment with high concentrations of the myosin inhibitor, N-benzyl-p-toluene sulfonamide (BTS), also prevents injury (10). The specific nature of structural rearrangements caused by contraction that lead to membrane stress and degeneration in dystrophic muscle is not currently understood.
Adult skeletal muscle consists of two main fiber types, "slow" (type I) and "fast" (types IIa and IIx/d; type IIb is also present in other mammals but not human muscle), defined by the myosin isoform that they express (11). Muscles enriched in fast fibers are more susceptible to mechanical stress in dystrophic mice whereas muscles enriched with slow fibers are more resistant (12). This also appears to be the case in individuals with DMD. Histological studies of young patients with DMD display fiber-type imbalances in the colocalization of fast and embryonic myosin, a marker of regenerating muscle fibers (13). Circulating biomarkers of muscle injury specific to fast but not slow fibers, such as fast troponin I, are also enriched in the plasma of individuals with DMD (14). Histological observations of fast fiber susceptibility have also been made in dog and pig models of DMD (15,16). With age, slow skeletal muscle fibers also show evidence of injury, including coexpression of embryonic myosin (17). The cause of fast fiber susceptibility to injury and whether it has any influence on slow fiber injury or overall progression of disease is not well understood.
In this study, we used a pharmacological approach to explore the role of muscle fiber contraction in dystrophic muscle stress and degeneration. To achieve this goal, we developed a compound, EDG-5506, which inhibits the ability of myosin to hydrolyze ATP and develop force by decreasing strong binding between myosin and actin within the sarcomere. The high selectivity of this compound for fast but not slow skeletal, cardiac, or smooth muscle allowed us to dissect the role of fast muscle contraction in muscle breakdown and disease progression in models of DMD. Using EDG-5506 ex vivo, in situ and in vivo, we came to the unexpected conclusion that small amounts of fast myosin inhibition result in almost complete protection against skeletal muscle membrane injury, force loss, and longer-term fibrosis. These results point to the possibility that fast myosin inhibition may represent an alternative treatment modality for DMD and other myopathies that are exacerbated by mechanical stress.

Results
Contraction via fast skeletal myosin is coupled to force drop and membrane injury in dystrophic muscle. To explore the contribution of skeletal muscle contraction to injury in dystrophic muscle, we sought to identify an inhibitor of skeletal contraction that was appropriate for both in vitro and in vivo experiments and was not inhibitory to cardiac or smooth muscle. Contraction of muscle is directly coupled to ATP hydrolysis (18), so we used a high-throughput small-molecule screen measuring fast skeletal muscle myofibril ATPase to identify promising leads (Supplemental Figure 1A; supplemental material available online with this article; https:// doi.org/10.1172/JCI153837DS1). Chemical leads were optimized for potency, selectivity, physiochemical and pharmacokinetic properties, leading to the identification of EDG-5506 ( Figure 1A). traction lengthening injury model (29). Oral administration of EDG-5506 decreased isometric force in a dose-dependent manner ( Figure 2F and Supplemental Figure 2D). Lengthening injury yielded an exaggerated strength drop in mdx but not WT TA muscle, and it was dramatically improved in mdx muscle with EDG-5506 ( Figure 2G). Like ex vivo results, protection was achieved at doses associated with small decreases in isometric force (<10%), with force drop similar to that in WT mice. Protection was specific to dystrophic muscle, as there was no decrease in force drop in WT mice ( Figure 2G). Unlike ex vivo results, lengthening peak stress was not altered by treatment (Supplemental Figure 2E). EDG-4131 was also protective against in situ injury (Supplemental Figure  2F). Muscle injury is commonly associated with increased circulating creatine kinase (CK). Treatment of mdx mice with EDG-5506 prior to in situ injury lowered plasma CK activity compared with that in control mice ( Figure 2H). There was no short-term effect of EDG-5506 on CK in the absence of lengthening injury (Supplemental Figure 2G).
Membrane stress is an important component of eccentric injury in dystrophic muscle (30). We next performed ex vivo lengthening contraction of mdx EDL muscle using a membrane-impermeable dye to assess sarcolemmal disruption. EDG-5506 effectively reduced dye uptake (Figure 3, A and B). There was no effect of the dye itself on force (Supplemental Figure 3A). One consequence over 10 rounds of lengthening contractions (Supplemental Figure  2A). EDG-5506 lowered preinjury isometric force and lengthening peak strain in a concentration-dependent manner in both mdx and WT muscle ( Figure 2A). However, peak strain was inhibited to a lower extent compared with isometric force, leading to an increase in the eccentric/isometric force ratio with EDG-5506 (Supplemental Figure 2B). Strength drop with repeated contractions in mdx muscle was dramatically lowered with EDG-5506 ( Figure 2, B and C), protecting against isometric and peak force drop (Figure 2, D and E). Both metrics approached but did not achieve WT strength drop, perhaps suggesting other contributing factors. The dose response for protection had a surprisingly nonlinear relationship with near maximal protection at ≤1 μM, a concentration associated with approximately 15% inhibition of isometric force (Figure 2A). Similar protection was also observed with EDG-4131 (Supplemental Figure 2C). Several signaling pathways have been implicated in protection from eccentric injury, including Akt (26), IGF1 (27), and LTBP4/TGF-β (28). To assess whether any of these pathways were directly or indirectly modulated by EDG-5506, EDL muscle from mdx mice was incubated with vehicle or 5 μM EDG-5506 for 1 hour prior to analysis with a phosphoprotein array. There were no consistent changes in implicated pathways (Supplemental Data File 2).
We next examined muscle responses in situ, measuring mouse TA muscle force after stimulation of the sciatic nerve with a 2-con- Myofibril ATPase activity curves for EDG-5506, with myofibrils isolated from rabbit fast skeletal muscle, bovine cardiac ventricle, and slow bovine masseter muscle. (C) Purified myosin S1 ATPase activity curves for EDG-5506, with rabbit fast skeletal muscle (psoas muscle), pig cardiac muscle, and smooth muscle myosin S1 isolated from chicken gizzard. ATPase activity in the myofibrils is measured at the pCa50 (calcium concentration where ATPase activity is 50% of maximum) value for free calcium for each muscle type (n = 2). (D) Representative force-calcium curve in single permeabilized fast skeletal muscle fibers (rabbit psoas) with EDG-5506. (E) Percentage of initial force with time after addition of EDG-5506 in WT mouse EDL muscle ex vivo. Force was recorded at 250 Hz. Each point represents mean peak force ± 1 SEM (n = 4). *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Significance was calculated by 1-way ANOVA with Dunnett's multiple-comparison test. Significance was calculated by 1-way ANOVA with Dunnett's multiple-comparison test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. All indicated comparisons were made against results obtained after treatment with vehicle (0 μM EDG-5506).
Supplemental Figure 3B). Contraction caused a strength drop that was completely prevented by EDG-5506 ( Figure 3D). EDG-5506 also reduced accumulation of resting calcium and tension (Figure 3, E and F, and Supplemental Figure 3C). Calcium also drives destructive hypercontraction and clotting of fiber myoplasm (31). EDG-5506 completely prevented clot formation ( Figure 3, G and of membrane disruption is extracellular calcium uptake, a major contributing factor to muscle fiber necrosis (4). We used an ex vivo mouse lumbrical model (31) with fura-2 calcium indicator dye to measure the effect of EDG-5506 on calcium dynamics and uptake with contraction in mdx lumbrical muscle. EDG-5506 reduced preinjury force without altering calcium transients ( Figure 3C and . Data are shown as the mean ± SEM. Significance was calculated by 1-way ANOVA with Dunnett's multiple-comparison test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. H, and Supplemental Videos 1-3). Highlighting the importance of calcium influx to muscle injury (32), we measured a linear relationship between intercontraction calcium and force drop that was corrected with EDG-5506 (Supplemental Figure 3D). Consistent with previous results, maximal efficacy was achieved at 0.3 μM, where specific tension before injury was reduced by 15%.
EDG-5506 protects skeletal muscle in mdx and DBA/2 mdx mice in vivo. Mice have more than 90% fast muscle, and its contraction is essential to physical activity (33). We next tested whether protective levels of myosin inhibition impaired strength, coordination, or activity in vivo. Single doses of EDG-5506 up to 10 mg/kg had no effect on grip strength or rotarod endurance in WT or mdx mice ( Figure 4A). EDG-5506 also had no effect on wheel running (Supplemental Figure 4A). Physical exercise causes enhanced muscle injury in mdx mice, particularly with respect to plasma CK (34). We assessed plasma CK after strength and coordination tests. EDG-5506-treated mdx mice had significantly lower postexercise CK compared with controls ( Figure 4B). There was no effect of single doses of EDG-5506 on CK without exercise (Supplemental Figure 4B). Membrane permeability can also be visualized in vivo via Evans blue dye (EBD) muscle uptake (35). Three-week treatment of mdx mice with EDG-5506 reduced blue fibers close to the level seen in WT mice, demonstrating muscle protection with longer compound treatment without formal exercise ( Figure 4, C and D, and Supplemental Figure 4C).
Muscle degeneration in mdx mice leads to fibrosis, particularly in the diaphragm (36). We next tested whether chronic treatment with EDG-5506 would benefit fibrosis in mdx mice. Seven-week-old mdx mice were dosed with EDG-5506 for 8 weeks. EDG-5506 increased average grip strength ( Figure 5A) and decreased diaphragm fibrosis compared with placebo ( Figure 5B) and returned the skeletal muscle transcription profile closer to that of WT mice ( Figure 5C and Supplemental Data File 3). EDG-5506 had no effect on body or muscle weight, and there was a trend toward fewer small, fragmented fibers in the TA muscle but not the diaphragm (Supplemental Figure  5, A-D). A rotarod test administered midstudy confirmed sustained membrane protection with decreased exercise-associated plasma CK (Supplemental Figure 5E). To better understand the effect of EDG-5506 on muscle degeneration, we performed a second study in young mdx mice before muscle necrosis that peaked at weaning (3-4 weeks), with the soleus muscle particularly affected (37). EDG-5506 decreased central nucleation, with a trend toward decreased embryonic myosin-positive fibers in the soleus ( Figure 5D). Specific tension was also higher in the soleus of treated WT and mdx mice ( Figure 5E). Muscle size, fiber size, and proportion of fast (IIA) fibers was unaffected (Supplemental Figure 5, F-H).  has been documented in DMD dogs (16), we confirmed sustained fast myosin from biopsies of age-matched WT/GRMD gastrocnemius muscle (Supplemental Figure 6A). EDG-5506 concentration was measured from a muscle biopsy at the end of the dosing period (4,951 ± 217 ng/g muscle). EDG-5506 administration was associated with a more than 50% decrease in CK, returning to pretreatment levels after compound washout ( Figure 6A). Physical activity normally increases CK in GRMD (41), raising the question of whether EDG-5506 lowered CK via decreasing physical activity. We sought to test the effect of EDG-5506 on habitual activity in the same dogs using an activity monitor (42). EDG-5506 improved average daily activity by more than 30% ( Figure 6B and Supplemental Figure 6B) with reversible increases in time active and decreases in time resting ( Figure 6C). One GRMD used in the biomarker study was too severely progressed to participate in this study. Activity in this dog did not change over the same dosing period (Supplemental Figure 6C).
Somascan plasma proteomics has been used in patients with DMD to generate a distinguishing protein signature compared with Unlike patients with DMD, mdx mice exhibit limited muscle fibrosis (36). Therefore, we performed studies in the more fibrotic DBA/2J mdx mouse (38). Beyond fibrosis, DBA/2J mdx mice exhibit more fragmented, smaller fibers (39). EDG-5506 had no effect on body or muscle weights, but the diaphragm muscle had fewer small fibers after 12 weeks of dosing (Supplemental Figure 5, I-K). Treatment also reduced fibrosis in TA and diaphragm muscles, with a trend toward lower fibrosis in the left ventricle (Figure 5, F and G). Studies in DBA/2J mdx mice with EDG-4131 confirmed this pattern, including significant decreases in left ventricular fibrosis (Supplemental Figure 5, I-M).
EDG-5506 lowers disease biomarkers and improves physical function in DMD dogs. We viewed data in mice as encouraging but were interested in testing EDG-5506 in dystrophic golden retriever (GRMD) dogs; their muscle composition and function more closely resemble that of humans (40). Older (7 months of age), disease-stable GRMD were dosed in 2-week periods with vehicle, then EDG-5506, followed by a vehicle washout. As fast fiber atrophy  4) before, during, and after 14 days oral gavage with EDG-5506. Each data point represents CK activity from an individual blood draw (2-3 draws during the vehicle baseline, 5-8 draws during the dosing period, and 2 draws during the vehicle washout). In a separate study, the same 4 dogs were dosed daily with vehicle, and blood CK was at regular intervals for 14 days (indicated as vehicle repeat, 6-9 blood draws per dog). Data are shown as the mean ± SEM. (B) Average daily activity measures from an electronic activity monitor (FitBark) in the same DMD dogs (n = 3, 15 months old) before, during, and after an 11-day oral gavage with EDG-5506 (2 mg/kg daily for 4 days, then every other day). Each data point represents the daily average activity for the 3 dogs. (C) Timed function data from the same activity monitors. Average daily active time, average daily play time, and average daily rest time. Each point represents average activity for 1 day (2 mg/kg represents approximately 68.2 μmol EDG-5506/dog). Significance was calculated by 1-way ANOVA with Tukey's multiple-comparison correction. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. egy for muscle relaxation in myotonia or spasticity (44). Our findings uncover a connection between muscle contraction and dystrophic muscle injury and expand the possible uses for this class of compounds into DMD, a progressive, grave disease. This relationship may also apply to other myopathies where muscle contraction causes fiber breakdown, including types of limb girdle muscular dystrophy, Becker muscular dystrophy, and metabolic myopathies such as McArdle's disease (45).
Protective effects from myosin inhibition build on earlier reports of dystrophic muscle protection by ablation of muscle contraction by unloading, denervation, or immobilization with the myosin inhibitor BTS (10). In contrast to these approaches, we demonstrate that full immobilization is not required for protection and that modest force inhibition (10%-15%) is sufficient to prevent dystrophic muscle injury. Our initial assumption was that muscle protection via this mechanism would occur through decreased contraction peak strain, a key contributor to injury in dystrophic muscle (8,30). The results of the present study are inconsistent with this hypothesis, as EDG-5506 had only modest effects on peak strain ex vivo and no effect in situ. Examination of the ratio between isometric force healthy controls (43). As CK was decreased with EDG-5506, we asked whether other proteins associated with DMD were also changed. Somascan comparison of GRMD plasma with healthy dog plasma revealed many protein differences (Supplemental Data File 4). We examined the effect of EDG-5506 on the proteomic signature of GRMD. Treatment reversibly lowered GRMD-elevated proteins and increased GRMD-downregulated proteins ( Figure 7A and Supplemental Data File 5). Gene ontology term analysis revealed decreases in pathways associated with dystrophic muscle, including apoptotic, cellular signaling, metabolic, and immune responses (Supplemental Table 3). Comparison of the GRMD proteomic signature with that from DMD (43) revealed a common set of 40 elevated and 9 depleted proteins ( Figure 7B). EDG-5506 also significantly reversed this common DMD signature ( Figure 7C and Supplemental Data File 6).

Discussion
These studies reveal a surprising sensitivity of dystrophic muscle to sarcomeric and membrane stabilization by inhibition of force by fast skeletal muscle myosin. Following the experiments reported here, reports have also proposed fast myosin inhibition as a strat- Possible factors include lower levels of the dystrophin-related protein utrophin (53), lower oxidative and mitochondrial capacity, and differences in organization of the z-disc, making it more susceptible to disruption (54). Muscle injury studies in healthy volunteers have shown elevation of circulating troponin I from fast but not slow muscle (55), suggesting that fast fiber sensitivity is perhaps normal physiology amplified in the absence of dystrophin.
Older patients with DMD (≥7 years) exhibit muscle fiber dysfunction in both slow and fast fibers, including expression of embryonic myosin (17). It is unclear whether fast fiber damage indirectly affects slow fibers or whether injury occurs directly over a longer time. One nuance of the biochemical profile of EDG-5506 is that it inhibits both fast and embryonic myosin (Supplemental Table 1). Embryonic myosin, present with fiber regeneration, is present in approximately 30% of fibers from patients with DMD (56) and should be targeted by EDG-5506, whether they coexpress fast or slow myosin. As embryonic myosin is expressed during muscle regeneration, we confirmed that EDG-5506 did not interfere with muscle repair using a cardiotoxin model. A surprising finding of this study was that muscle size was larger with EDG-5506 treatment (Supplemental Figure 7). Young mdx mice undergo large-scale muscle degeneration/regeneration at 3-4 weeks of age (37). EDG-5506 was protective against this degeneration, reducing central nuclei. While neither muscle nor fiber size was increased in this study, specific tension occurred in both mdx and WT mice and was higher in these animals than in controls (that exhibit no degeneration). Why myosin inhibition would increase either muscle size or function in these dynamic situations is not currently understood.
Our results in older GRMD help understanding of compound response in older patients with DMD. GRMD exhibit early disease onset that stabilizes at 6 months. Results from studies of 7-to 18-month-old GRMD can be compared with those of patients older than 10 years with DMD, with muscle weakness and increased occurrence of mixed composition fast/slow fibers (40). In this light, it is encouraging that we observed both decreased CK and increased activity with EDG-5506. There are few precedents for acute functional improvements in GRMD. The closest exemplar in DMD dogs would be glucocorticoids, which increased some aspects of physical function and lowered CK (in combination with cyclosporine) but had longer-term adverse effects, including muscle atrophy and calcification (57).
Our use of Somascan proteomics on samples from GRMD facilitated a detailed analysis of disease-associated biomarkers in this model. One caveat to these data is that Somascan uses aptamers for human proteins with unknown cross-reactivity against dog. Given this, we saw consistent patterns between GRMD and DMD signatures with overlap between published DMD signatures (43), further validating GRMD as a close comparator to DMD. Treatment with EDG-5506 reversed this overlapping signature, speaking to a universal proteomic correction of disease in GRMD and, by extension, DMD. Limitations of this study include in vivo differences between mdx mice, GRMD, and patients with DMD, and the short duration and lack of force measures in our GRMD study.
Our studies demonstrating common beneficial effects of EDG-5506 across preclinical models are unusual and speak to its fundamental nature. EDG-5506 protects muscle ex vivo and in situ from contraction injury with a response that compares favorably to AAV and peak strain, another factor influencing dystrophic muscle injury (8), revealed an increase with treatment (Supplemental Figure 2C). This is consistent with reports showing decreased isometric force but increased force enhancement with lengthening in skeletal muscle fibers treated with myosin inhibitors such as BDM and vanadate (46,47). Mechanistically, it has been proposed that inhibitors that increase the weakly bound myosin ADP.Pi state convert to a strongly bound state with lengthening. Functional consequences of this lengthening enhancement are unclear.
The magnitude of eccentric stress in our injury models is not commonly seen in sedentary mice, yet we measured significant decreases in EBD incorporation with EDG-5506. A hypothesis for protection in sedentary mice is based on an established model (31) where muscle fiber stress causes localized increases in calcium and nonuniform contraction of serial sarcomeres. In dystrophic muscle, the lack of costameres amplifies stress along serial sarcomeres, resulting in further membrane stress and calcium entry. Under these conditions, EDG-5506 would provide protection against local stress by limiting uneven contraction.
A caveat with any small-molecule approach is that off-target activities might contribute to positive pharmacology. Documented pathways that also confer protective effects in dystrophic muscle include sarcolemmal nNOS (48) and membrane excitability (49). Consistency between ex vivo (where blood supply is severed) and in situ results with EDG-5506 suggest that blood perfusion and NOS activity are not essential for protection. Similarly, decreases in force but not calcium in treated lumbrical muscles suggest EDG-5506 does not decrease membrane excitability. Additional studies also explored the promiscuity of EDG-5506 via in vitro binding, enzymatic, and phospho-antibody arrays, confirming low off-target activity for EDG-5506. Regardless, there is still a possibility that other unknown activities of EDG-5506 might contribute to its protective effects.
Protective doses of EDG-5506 had no negative effects on strength, coordination, or habitual activity in mice, suggesting functional accommodation or excess capacity is present in skeletal muscle. Mouse muscles are composed of primarily fast fibers while human muscles are composed of an approximately even split of slow and fast fibers (50). Muscle contraction is controlled by both motor neuron stimulation rate and coordinated recruitment of populations (units) of fast or slow fibers (51). As physical demands increase, stimulation rates and recruitment increases with slow units usually recruiting before faster units (51). This system is designed to maintain physical performance even under extreme stress. Inhibition of fast myosin via EDG-5506 would be expected to reduce fast fiber force but overall muscle performance might still be maintained via force adjustment provided by stimulation and recruitment of motor units. Genetic evidence for both accommodation and adverse effects from total fast fiber inhibition exists in individuals with null mutations in MYH2, the gene encoding myosin IIa, which makes up 85% of human fast fibers (52). MHY2-null skeletal muscles have few fast fibers, but individuals remain ambulatory although they manifest mild-to-moderate proximal limb and oculofacial weakness and dysfunctional extraocular muscles.
Although selective injury of fast skeletal muscle fibers in DMD was first documented over 30 years ago (13), its cause and relevance to disease progression is poorly understood. Fast fibers differ in many ways from slow fibers that might heighten susceptibility. sion, Perkin Elmer) for 10 minutes per plate at room temperature. Hits were reconfirmed, and secondary assays were used to evaluate selectivity, including bovine cardiac myofibrils (pCa 5.86) and enzyme coupled hits with a hexokinase screen (Sigma-Aldrich).
Stopped-flow analysis of Pi release from myosin. The rate of Pi release from myosin with actin utilized a fluorescent phosphate binding protein, as described previously (63). Experiments were performed in 10 mM MOPS, pH 7.0; 5 mM KCL; 1 mM EGTA; 1 mM DTT; 0.5 mM MgCl 2 . ATP and ADP were removed by incubation with apyrase (64). Prior to mixing with 1 mM ATP in a stopped flow device, compound was incubated with 6 μM psoas S1 at 25°C for 30 minutes.
Skinned fiber force measurements. Measurement of isometric force in skinned fibers from rabbit psoas, rat soleus, and rat cardiac trabeculae were performed as described previously (65) with an 802D fiber system (Aurora Scientific). Muscle bundles were thawed from -80°C and placed in relaxing solution. Single fibers were transferred to the testing apparatus with steel pins connected to a force transducer and motor (model 403A and 322C, Aurora Scientific), and collodion glue was applied. Temperature was set to 15°C for skeletal fibers and 30°C for rat cardiac trabeculae. The wells of the 802D were filled with different ratios of relaxing and activating pCa solutions as detailed previously (65). All control solutions contained 1% final DMSO. Traces were compared between control and compound runs to evaluate isometric force and calcium sensitivity.
Ex vivo force measurement. Isometric force was measured ex vivo in mouse EDL and soleus muscle. EDL muscle eccentric injury was based on published protocols (30), with degree of lengthening and stimulation modified to yield a uniform drop in mdx isometric and peak force over 10 contractions. After isolation, muscle was mounted via 5-0 suture to a combination servomotor/force transducer (300C-LR, Aurora) between 2 electrodes connected to a stimulator (701C, Aurora) in Ringers solution at 27°C with continuously infused CO 2 /O 2 gas mixture. Experiments were recorded with custom MATLAB software via a National Instruments PCIe-6321 board and 2090A BNC connector. Muscles were preincubated for 60 minutes with DMSO (final 0.1%)/compound before 10 eccentric contractions, 1 minute apart, with 100 milliseconds of isometric stimulation at 100 Hz and active lengthening of 0.1 L 0 at a rate of 2 L 0 /s. After lengthening, muscles were held at 1.1 L 0 with stimulation for 100 milliseconds (Supplemental Figure 2A).
In summary, these findings represent an alternative strategy for the treatment of DMD. The mechanism of myosin inhibition is structural in nature and, as such, is independent of inflammation or the specific genetic lesions conferring disease. As a result, we predict that it should be complimentary with therapeutic approaches that alter inflammation or modify/replace the dystrophin gene, such as antisense oligonucleotides or AAV microdystrophin. Further studies will explore whether EDG-5506 confers additional therapeutic benefit when combined with glucocorticoids or genetic strategies. Importantly, short-term clinical studies have now been completed in healthy volunteers and individuals with Becker muscular dystrophy using EDG-5506 (NCT04585464), and longer-term studies are underway in Becker muscular dystrophy and DMD (NCT05291091 and NCT05540860).

Methods
Actin and myosin S1 preparation. Smooth muscle chicken gizzard S1 was purchased from Cytoskeleton Inc. Other myosin S1 was prepared via published protocols (21). Briefly, fresh rabbit psoas and frozen porcine ventricular muscle (Pel-freez) were extracted in 0.3 M KCl, 0.15 M KH 2 PO 4 , 1 mM EGTA, 1 mM ATP, 1 mM DTT, pH 6.5, at 4°C. Fulllength myosin was purified from each extract via low-ionic precipitations with final precipitation into PM12 buffer (12 mM PIPES, pH 7.0, 2 mM MgCl 2 ). Purified myosin was resuspended in PIPES/EDTA buffer (20 mM PIPES, pH 6.8, 10 mM EDTA, 1 mM DTT) and reacted with 0.2 mg/ mL α-chymotrypsin for 30 minutes at room temperature. Proteolysis was quenched with PMSF (final concentration 100 μM), and the reaction was centrifuged for 45 minutes at 230,000g. S1 fragment in the supernatant was dialyzed into PM12 with 0.02% sodium azide and clarified by centrifugation. Sucrose was added to 10% w/v and flash-frozen in liquid N 2 and stored at -80°C. Actin was purified from minced porcine cardiac muscle after 10-minute incubation in extraction buffer (0.5 M KCl, 0.1 M K 2 HPO 4 , pH 7.0). Centrifuged pellets were washed 3-4 times with carbonate buffer (Na 2 CO 3 , pH 8.2-8.5) and then 3 times with cold acetone. Residues were collected and dried to yield acetone powder. Actin was purified using published methods (21). Actin was polymerized with 0.1 volumes of 10× polymerization buffer (50 mM PIPES, pH 7.0, 550 mM KCl, 22 mM EGTA, 22 mM MgCl 2 , 10 mM ATP) and incubated at room temperature for 60 minutes.
Myofibril preparation. Myofibrils were prepared from various animals and tissues: rabbit psoas muscle and porcine cardiac muscle were purchased from Pel-Freez Biologicals. Human bicep and soleus muscle were purchased from BioIVT, and neonatal rat muscle was obtained in house. All myofibrils were prepared using published methods (21).
Myofibril and myosin S1 enzymatic activity. A lactate dehydrogenase/ pyruvate kinase enzyme-coupled system (62) was used to monitor ATP hydrolysis rates in various muscle preparations. Rabbit psoas myofibrils were used for high-throughput screening at pCa 6.25 at 0.25 mg/mL. A multidrop cassette (Thermo Fisher Scientific) dispensed 10 μL of the coupled enzyme system, calcium, and myofibrils to 10 μL ATP, NADH and PEP to initiate the reaction prior to reading on a plate reader (Envi-accelerated 1 rpm every 3 seconds to 12 rpm and a maximal time of 500 seconds. This was repeated 3 times with a 3-minute rest and total time reported. Voluntary wheel running performance was assessed with a customized system. Timestamped RPMs were measured 24 hours per day by infrared sensors, and a custom software calculated distance. Whole-body EBD was visualized after intravenous administration (35). EBD in PBS at 10 mg/mL was sterile filtered, and 50 μL/10 g BW was injected via tail vein. After 24 hours, animals were sacrificed, skin was removed and fixed in 4% formalin for 24 hours, and photographed using a Leica/Lumix camera.
Plasma CK activity. Plasma from mice and dogs were analyzed for CK activity using standard protocols (Pointe Scientific, Thermo Fisher Scientific). Note that experiments performed in Supplemental Figure  2D and Supplemental Figure 4B were performed with separated batches of mdx and CK kits and exhibited higher background CK activity compared with previous experiments ( Figure 2H and Figure 4B).
Histological and immunohistochemical analysis. For sterile muscle injury, mice were anesthetized with isoflurane and 50 μL cardiotoxin (12 × 10 -6 M, 217503, Millipore) injected in the TA muscle. Muscles 8 days after injury were snap frozen in nitrogen-chilled isopentane. 8 μm cryosections were cut and stained with H&E. For each analysis, more than 10 slides (per condition/group) containing 6 muscle sections/sample were used, and myofibers in the injured area were counted and measured with a Mirax digital scanner and HALO software (68). For adult mouse studies, muscles were fixed in 10% formalin and embedded in paraffin for sectioning. Laminin staining (RB-082, Thermo Fisher Scientific) was used to outline fibers, and picrosirius red was used for collagen accumulation. Image analysis was performed using ImarisX64 software (Bitplane AG), and fiber sizing was calculated using minimal Feret's diameter (69). For after weaning juvenile mouse studies, cryosections from soleus muscles were mounted in ProLong antifade mountant with DAPI (Thermo Fisher Scientific). Fiber type, size, and central nucleation were quantified with analysis via SMASH (70) and ImageJ (NIH). Antibodies used were mouse anti-MHC2A (1:50, SC-71, Developmental Studies Hybridoma Bank) and mouse anti-eMHC (1:10, BF-45, Developmental Studies Hybridoma Bank).
RNA-Seq analysis of gastrocnemius muscle. RNA-Seq was performed with isolated single nuclei using the 10X Genomics Chromium Single Cell 3' Protocol, Accessories, and Kits (CG000183 Rev A). Data were then analyzed by Rosalind (https://rosalind.onramp.bio/), with a HyperScale architecture developed by OnRamp BioInformatics Inc. Data have been submitted to the Gene Expression Omnibus (GEO) database at the NIH (accession GSE227510).
Myosin isoform analysis. Biopsies of dog gastrocnemius muscle (GRMD, n = 5; age, 6-9 months; WT, n = 5; age 6-7 months), rabbit psoas, and bovine masseter were snap frozen on liquid nitrogen. Muscle was thawed in lysis buffer (30 mM Tris, pH 7.5, 1% SDS) and homogenized in a bead beater for 3 minutes at 58 cycles per second. Myosin-separating gels were prepared and run as described previously (71) using 1 μg protein lysate for gastrocnemius and 3.5 μg psoas/masseter (at 80 V for 8 hours and 85 V for 18 hours) at 4°C. Protein was quantified using Sypro Total Protein stain (rapid protocol) and visualized on a LiCor Odyssey M. Quantification of fast and slow MHC was performed using Empiria Studio 2.2 software (LiCor Inc.).
Habitual activity and circulating biomarker measurement in GRMD. Studies were performed in two phases. In the first phase, GRMD (n = 4, female, 7 months old) were dosed PO daily with vehicle for 2 weeks, surface component algorithm was used to detect procion-labeled fiber boundaries. Nonspecific procion staining of the peripheral tendon and outer layer of cells (where handling could cause injury) was excluded by applying a size and location filter (>2× average fiber size and located adjacent to laminin-positive fibers at the edge of the muscle). Each fiber was identified with an ID and color coded before calculation.
Ex vivo analysis of mdx mouse lumbrical muscle contraction. Intracellular calcium quantification and isometric injury were described previously (31). Lumbrical muscles were transferred to a custom-built chamber and mounted horizontally to a force transducer (model 400A, Aurora Scientific) at 25°C. Sarcomere length was adjusted to 2.5 μm by microscope. Contraction was elicited with 0.2-millisecond stimuli and injury via 1-second stimuli at 125 Hz.
For fluorescence experiments, a 75W xenon lamp was used with wavelengths selected using a diffraction grating monochromator (model DeltaRAM, Photon Technology International). Muscles were loaded with mag-fura-2 AM (10 μM in Tyrode) for 30 minutes at 25°C, excited at 344 and 375 nm (bandwidth 10 nm), and passed through a 510 nm emission filter (bandwidth 40 nm). Fluorescence responses were recorded alternately during 32 twitch contractions at 20-second intervals; background fluorescence was subtracted, and the response was averaged. High-affinity fura-2 was used to detect membrane breaches between tetanic contractions. Muscles were incubated with fura-2 AM (15 μM in Tyrode) for 30 minutes before 12 injurious contractions, separated by 1 minute with (10/s) alternating excitation wavelengths of 340 and 375 nm (bandwidth 1.25 nm) with a 510 nm emission filter (bandwidth 40 nm) and preloading background subtraction. Still and video images were captured using a digital camera (Nikon D750).
In situ force measurement. Measurement of isometric force and eccentric injury response in situ was performed in mouse TA muscle based on published protocols (30). Mice were anesthetized with 2% isoflurane (0.7%-1.5% maintenance) on a 37°C platform. The distal tendon to the TA and EDL muscle was attached with 4-0 suture a servomotor (Aurora Scientific). The sciatic nerve was exposed at the knee for hook electrodes. For compound assessment, baseline isometric contraction was assessed at 25-175 Hz (9 mA, 300 ms) followed by 100 Hz stimuli every 5 minutes. After 15 minutes, compound was dosed PO, and recording continued for 4 hours followed by a second force frequency. For in situ eccentric injury, compound was administered PO, muscle was stimulated every 5 minutes at 100 Hz for 3.5 hours, and then 2 lengthening contractions were performed (10 min apart, 100 ms at 150 Hz, then lengthening 0.2 L 0 at 2 L 0 /s and hold for 100 ms). Isometric force before and 10 minutes after the injury contractions was recorded.
Phosphoprotein array analysis. mdx mouse EDL muscle was prepared for ex vivo analysis as described (n = 5). Muscles were incubated for 1 hour with DMSO or EDG-5506 (5 μM) and then flash frozen in liquid nitrogen. Tissue preparation and analysis were performed as recommended by the manufacturer (Cell Signaling Phospho Antibody Array, Full Moon Biosystems).
In vivo assessments in mice. Prior to in vivo assessments, mice were dosed PO after a 2-hour fast with 10 mL/Kg suspensions. Grip strength and rotarod assessments were performed masked, 4 hours after dosing using published protocols (66,67). Grip strength was performed with a Columbus Instruments Rodent Force Meter. Forepaw strength represented the average of 5 trials with 30-second rest periods, normalized by BW. Rotarod endurance was performed with a Columbus Instruments RotaMax. Mice were placed on the device drum at 1 rpm, which