Systemic delivery of full-length dystrophin in Duchenne muscular dystrophy mice

Duchenne muscular dystrophy (DMD) is a lethal X-linked disorder caused by mutations in the DMD gene leading to loss of dystrophin and progressive degeneration of skeletal, cardiac, and respiratory muscles. While AAV-mediated micro-dystrophin replacement can be packaged within AAV and provides partial benefit, it lacks approximately two-thirds of the full-length (FL) dystrophin coding sequence, omitting critical rod and hinge domains needed for full dystrophin-glycoprotein complex (DGC) function, particularly in the heart. The limited cargo capacity of AAV (~4.5 kb) precludes delivery of the >11 kb dystrophin cDNA in a single vector. The research question is whether a triple AAV vector strategy using orthogonal split inteins can assemble and express FL-dystrophin systemically at therapeutically meaningful levels in vivo, thereby overcoming limitations of micro-dystrophin and restoring muscle integrity and signaling in both skeletal and cardiac tissues.

Prior strategies to exceed AAV cargo limits include dual-vector approaches: overlapping fragments enabling homologous recombination, trans-splicing using concatemerization across ITRs, and hybrid systems combining these mechanisms. Although these methods can expand payload size, expression efficiency remains limited by low recombination rates, concatemer formation, and trans-splicing efficiency, which constrain clinical utility. A triple trans-splicing approach previously demonstrated only very low efficiency for FL-dystrophin expression. Split inteins, which mediate protein trans-splicing to reconstitute mature proteins, have been used to deliver oversized genome editors in vivo via dual AAV vectors. These precedents motivate testing orthogonal split inteins to assemble FL-dystrophin from three AAV-delivered fragments.

Design and optimization of split-intein dystrophin constructs: The >11 kb FL-dystrophin cDNA was split into three fragments (N, M, C) guided by fragment size, domain structure, and intein junction compatibility. Initial split sites: between spectrin repeats (SR) 8 and 9 for Cfa intein (requiring Cys-W at the junction), and within Hinge 3 (H3) for Gp41-1 intein (serine-rich extein). Constructs Dys-N1, Dys-M1/2, Dys-C1/2/3 were expressed under a mini-CMV promoter with MCK enhancer (meCMV). HEK293 transfections and western blots with N-, M-, and C-recognizing anti-dystrophin antibodies assessed assembly efficiency. Optimization steps included: (1) refining the H3 split site and extein residues (e.g., IGA-SPT with -1 A→Y to favor Gp41-1); (2) removing a small intron in C (Dys-C3); (3) adjusting Kozak sequences and junction residues (SPT→SSS) and adding a ubiquitin-dependent proteolysis signal (PB29) to modulate C-fragment abundance (Dys-C4/C5); (4) switching intein pairs for M–C splicing to IMPDH-1 (with extein changes to better match native GGG-SIC) generating Dys-M3/C6; (5) tuning C expression with upstream polyadenylation signals (Dys-C7/C8). Atypical inteins (Cat, Vidal) were also tested but were less efficient. The optimized in vitro combination was Dys-N1/M3/C6. AAV vectorization and in vivo delivery: The N and M fragments were driven by a synthetic muscle-specific promoter Spc5-12; the C fragment used Spc2-26 (v1) or Spc5-12 (v2). All were packaged in the myotropic AAV capsid MyoAAV4A. Male mdx or mdx4cv mice (C57BL/6J background) aged 3–4 weeks received retro-orbital injections of a total dose 2×10^14 vg/kg (N:M:C at 2:1:1) or a lower dose 8×10^13 vg/kg for dose-response. Micro-dystrophin comparators (µ-v1, µ-v2) under Spc5-12 were delivered at 8×10^13 vg/kg using the same capsid. Outcome assessments: Immunofluorescence using N-, M-, and C-specific anti-dystrophin antibodies quantified dystrophin-positive fibers in gastrocnemius (GA), diaphragm, and cardiomyocytes. Western blotting in GA, diaphragm, and heart compared FL-dystrophin levels to human skeletal muscle controls (50%, 25%, 10% loads). Restoration of DGC components (α-/β-sarcoglycan, α-/β-dystroglycan, nNOS, α-dystrobrevin) was assessed by immunostaining. Functional measures included serum creatine kinase (CK), in vivo maximum plantarflexion tetanic torque (150 Hz tibial nerve stimulation), and wire hanging time. Histopathology included H&E for centrally nucleated fibers (CNF), fiber size distribution, and Masson’s Trichrome for fibrosis. Cardiac cavin-4 localization and ERK phosphorylation were analyzed by immunofluorescence and western blot. Methods detail AAV production/titration (iodixanol purification, qPCR titers), cell culture/transfection (AD293, PEI), and animal ethics.

• In vitro assembly: Systematic optimization (split site/extein mutations, intron removal, degron addition, intein swap to IMPDH-1) yielded Dys-N1/M3/C6 with substantially improved FL-dystrophin assembly versus earlier versions. Switching to IMPDH-1 for M–C splicing and extein tuning increased the FL-dystrophin signal by approximately 86% and reduced unassembled C by ~76%, with low accumulation of unassembled/partial intermediates.
• In vivo expression (MyoAAV4A, 2×10^14 vg/kg, N:M:C=2:1:1): Robust sarcolemmal FL-dystrophin detected in GA of mdx mice; 83.4 ± 2.7% dystrophin-positive GA fibers; heart: 78.3 ± 2.7% positive cardiomyocytes; diaphragm lower at 8.6 ± 2.6%. Western blots confirmed predominantly FL-dystrophin with minimal unassembled N/M; some residual unassembled C.
• DGC restoration: α-/β-SG, α-/β-DG, nNOS, and α-dystrobrevin were severely reduced in mdx GA but substantially restored after FL-dystrophin delivery.
• Functional and histological improvement: Serum CK (IU/L) decreased from mdx 3946.0 ± 341.1 (n=12) to 1060.0 ± 229.4 (n=13; p<0.0001); WT 220.6 ± 109.1 (n=14). Tetanic torque increased from 298.2 ± 10.6 mN-m/kg (mdx, n=10) to 452.5 ± 12.0 (treated, n=11; p<0.0001); WT 544.7 ± 9.8 (n=8). GA CNF reduced from 58.2 ± 1.7% to 25.1 ± 1.5%. Fibrosis markedly attenuated in GA and diaphragm.
• Promoter and dose effects: Using Spc5-12 for the C fragment (AAV-FL-v2) achieved comparable GA expression to v1 but significantly improved diaphragm and heart FL-dystrophin levels. Lower total dose (8×10^13 vg/kg) produced lower GA expression but smaller differences in diaphragm/heart.
• Quantitative restoration versus normal: In GA, FL-dystrophin restored to ~32.6–49.5% (v1) and ~36.9–43.4% (v2) of normal human skeletal muscle. In comparison studies, FL-dystrophin reached ~46.3% in GA (high dose), ~10.0% in diaphragm, and ~69.3% in heart, whereas micro-dystrophins (µ-v1/µ-v2) were overexpressed at ~2.6–3.7× normal in GA and ~2.6–7.8× in diaphragm/heart.
• Efficacy vs micro-dystrophin: All treatments (FL and µDys) reduced CK and improved tetanic torque and wire hang time to near WT levels; µDys groups showed CK closer to WT. Histological improvements (CNF, fibrosis) were similar across groups.
• Cardiac signaling superiority of FL: Only FL-dystrophin restored cavin-4 membrane localization and improved ERK phosphorylation in mdx hearts; micro-dystrophins did not correct these defects.

This study demonstrates that a triple AAV system using orthogonal split inteins can assemble and express full-length dystrophin systemically at therapeutically meaningful levels in dystrophic mice. The approach overcomes AAV cargo limits and addresses known shortcomings of micro-dystrophin constructs, particularly for cardiac signaling. Optimization of intein selection, split-site junctions, fragment expression balance, and degradative signals was critical to enhance assembly efficiency and reduce unassembled intermediates. In vivo, MyoAAV4A capsid enabled robust skeletal and cardiac transduction at clinically relevant total vector doses, yielding substantial restoration of FL-dystrophin, DGC components, muscle function, and histopathology. While micro-dystrophins provided comparable functional and histological benefits at the doses tested, only FL-dystrophin rescued cavin-4 localization and ERK signaling in the heart, underscoring the importance of the C-terminal domain and complete DGC interactions. Promoter choice affected tissue specificity, with stronger performance of Spc5-12 in diaphragm and heart, highlighting the need to tailor regulatory cassettes for fragment stoichiometry and tissue targeting. These findings support the feasibility and potential superiority of FL-dystrophin gene replacement as a mutation-independent therapy for DMD.

A split intein-mediated triple AAV system packaged in a myotropic capsid enables systemic delivery and assembly of full-length dystrophin in mdx mice, restoring DGC components, improving muscle pathology and function, and correcting cardiac cavin-4/ERK signaling defects not addressed by micro-dystrophins. This work provides a path toward clinically translatable FL-dystrophin gene therapy. Future studies should optimize promoters and enhancers for fragment stoichiometry (including diaphragm), refine intein pairs and split sites to further increase assembly efficiency and purity, enhance fragment stability, evaluate long-term efficacy and safety, and develop immune modulation strategies to prevent anti-dystrophin and anti-capsid responses. The platform may extend to other large genes exceeding AAV capacity (e.g., LAMA2, RYR1, DSP, FLNC, DYSF, OTOF, MYO7A, F8).

• Diaphragm expression was relatively low, particularly with the Spc2-26 promoter; further regulatory cassette optimization is needed for this tissue.
• Presence of unassembled C fragments persisted in vivo; long-term effects of unassembled or partially assembled products require investigation.
• The triple-vector approach depends on co-transduction of three vectors; stochastic delivery and expression stoichiometry may limit efficiency across tissues.
• Total vector doses used are relatively high; safety, biodistribution, and immunogenicity at scale need evaluation.
• Immune responses to restored FL-dystrophin (especially in patients with large deletions) and to intein fusion junctions were not assessed; mitigation strategies will be necessary.
• Studies were conducted in male mouse models and short to mid-term time frames; long-term cardiac outcomes and durability remain to be established.
• Micro-dystrophin groups exhibited lower CK than FL despite similar functional gains, suggesting tissue-specific expression differences (e.g., diaphragm) that merit further optimization.