Advancing adeno-associated virus for Duchenne muscular dystrophy treatment: Moving beyond rodent models

Main text

In the recent issue of Molecular Therapy, Liu et al. tested the biological properties of three different adeno-associated virus (AAV) serotypes by systemic delivery in 10 Cynomolgus monkeys (Macaca fascicularis).¹ The three AAV serotypes, AAV5, AAV8, and AAV9, were packaged with a proprietary baculoviral system developed by the authors incorporating the heart and skeletal muscle selective promoter (C5-12), to deliver a micro-dystrophin transgene, encompassing a gene replacement therapy for the muscle disorder Duchenne muscular dystrophy (DMD). The goal of this nonhuman primate (NHP) study was to compare the biodistribution into skeletal muscle and heart (myo- and cardio-tropism, respectively) and examine transgene protein levels and specificity of the C5-12 promoter 4 weeks after dosing.

The micro-dystrophin transgene in the study is identical to the commercial gene therapy product Elevidys that is FDA approved for the treatment of DMD. Because the DMD cDNA encoding dystrophin is too large to fit within the 4.7 kb packaging limit of AAV, approximately 70% of the endogenous cDNA has been eliminated. It is known that dystrophin tolerates large in-frame deletions, based on the phenotype of people with Becker muscular dystrophy (BMD). BMD mutations tend to spare the DMD reading frame, resulting in an internally deleted dystrophin that is associated with a milder clinical outcome compared to the more severe DMD (Figure 1). These insights were used by Chamberlain and colleagues to create and systematically evaluate “micro-dystrophins.”^2,3 Micro-dystrophins comprise critical regions of dystrophin needed to maintain the dystrophin glycoprotein complex (DGC) at the muscle membrane and bind to the actin cytoskeleton while eliminating ∼70% of dystrophin.^2,3 However, micro-dystrophins are less functional than full-length dystrophin, devoid of repeats that serve as shock absorbers that protect the muscle membrane. New advancements including CRISPR gene editing⁴ and split intein⁵ technology could facilitate the creation of larger, more functional dystrophins that will likely confer better membrane protection and enhance therapeutic efficacy.

Figure 1.

The dystrophin glycoprotein complex in dystrophic muscles and after delivery of AAV-microdystrophin

(A) In healthy muscle, dystrophin (shown in green) links the dystroglycan and sarcoglycan complexes (often called the dystrophin glycoprotein complex or “DGC”) through its cysteine-rich region at the C-terminus. The N-terminal actin binding domain and spectrin repeats 12–19 bind the actin cytoskeleton. Repeats 1–3 interact with the plasma membrane and repeats 5–9 and 20–21 interact with microtubules. Repeats 16–17 serve as a scaffold for both α-syntrophin and nNOS. (B) Duchenne muscular dystrophy mutations usually lead to loss of dystrophin protein production and leads to failure of the DGC to assemble at the membrane. (C) Becker muscular dystrophy mutations in the DMD gene often to spare the DMD reading frame, resulting in an internally deleted dystrophin that retains some functionality and is often associated with a milder phenotype. (D) Microdystrophin retains the actin-binding and cysteine-rich regions of dystrophin but lacks the majority of spectrin repeats that confer protection to the membrane by serving as a shock absorber that reduces the damage arising from muscle contraction.

NHPs are commonly used for pre-clinical pharmacokinetic (PK) and toxicological assessments of therapeutic AAV vectors under the assumption that AAV behaves similarly in NHPs and humans; however, very few studies have systematically investigated this. Liu et al. adds to the limited body of work evaluating AAV biology in primates, contributing to understanding AAV pharmacokinetics, tropism of different serotypes following systemic administration, and ultimately, how these factors impact the transgene-encoded protein. Dissemination of such data is particularly important as results from clinically treated patients are sparse, sample collection is not extensive in humans, and NHP studies are rarely published. Most published pre-clinical AAV studies use rodent models that are vastly different from humans in their size, metabolism, and immune response. Publicly sharing data from AAV-dosed NHPs is an essential step to advancing our understanding of AAV biology in higher order model systems and provides critical insights toward improving translation of these therapies to humans.

In clinical trials, the PK of AAV is typically assessed by measuring viral genomes in blood, while transgene protein is assayed from muscle biopsies, which by their nature results in only a small sampling of a single muscle, limiting insights into overall AAV distribution. Since AAV uptake can vary within a muscle and across muscle types, broader tissue access provides valuable data. NHPs provide a critical opportunity to establish the relationship between viral genome levels and protein expression in both on- and off-target tissues. In this study, the authors closely examined the relationship between viral genomes, transgene mRNA, and transgene protein in the heart, different skeletal muscles, and in off-target tissues like blood, lung, liver, and kidney. Notably, this study found little correlation between viral genome kinetics in the blood and transgene mRNA and protein expression in primary target tissues, heart and skeletal muscles. In fact, correlations across these measurements were minimal, highlighting the limitations of using blood-based PK data to predict AAV transduction efficiency within target tissues. For example, while the viral genome concentrations of AAV5 and AAV8 in skeletal muscle were similar, the mRNA and protein levels were greater for AAV8. Conversely, across all serotypes, higher viral genomes in the heart did not equate to higher transgene expression. In fact, despite the 7× lower viral genome concentration in skeletal muscle compared to heart, transgene expression was ∼2× higher in skeletal muscle. Surprisingly, this study reported a weak correlation between viral genome concentration and transgene mRNA and protein expression within the same tissues, emphasizing that many factors beyond AAV cellular entry can influence the levels of transgene produced, including AAV endosomal escape, trafficking to the nucleus, mRNA stability, and translation efficiency.

What factors could account for AAV8’s enhanced trafficking following cellular entry? This area of AAV biology has not been well studied for all serotypes; however, it is known that AAVs use receptor-dependent, clathrin-mediated endocytosis to enter cells. For AAV2, it has been shown that the N-terminal domains of VP1 and VP2 contain a phospholipase A domain responsible for endoscomal escape.^6,7 After endosomal escape, three N-terminal sequences (BC1, BC2, and BC3) are exposed, and both BC1 and BC2 serve as nuclear localization signals. BC1 and BC2 are identical between AAV5, 8, and 9, whereas BC3 differs between all three AAV types. It is thus possible that the enhanced nuclear trafficking observed for AA8 compared to AAV5 and 9 may be due to the differences in BC3 across these three serotypes.

Liu et al.’s data also suggest that quantitative immunofluorescence lacks the sensitivity to discriminate small differences in transgene protein levels when compared to ELISA or liquid chromatography-mass spectrometry (LC/MS). For instance, the authors found no detectable dystrophin protein in the diaphragm for any serotype except AAV8 using ELISA. However, quantitative immunostaining revealed comparable microdystrophin in both diaphragm and quadriceps. This discrepancy between these assay results highlights the need for further investigation of the accuracy and sensitivity limitations in preclinical and clinical studies.

The data presented by Liu et al. suggest that AAV9 requires twice the blood concentrations of AAV8 to achieve comparable transgene protein expression in heart and skeletal muscle. It is worth noting that even at 2E14 vg/kg of AAV9, micro-dystrophin expression was undetectable in the diaphragm and was lower in the forearm muscles compared to NHPs dosed with 1E14 vg/kg of AAV8. AAV9 has been one of the most utilized capsids for muscle gene therapies; however, AAV8 is gaining favor based on promising preliminary results from Regenex Bio in their on-going Duchenne trials (public disclosure). As another consideration, AAV9 has been associated with the serious adverse event, thrombotic microangiopathy, in both SMA and DMD clinical gene therapy trials.⁸

A stated goal of the study was to examine the selectivity and activity of the C5-12 artificial promoter in target tissues, skeletal muscle, and heart, compared to off-target tissues like liver, lung, and kidney. While the study demonstrated C5-12 specificity in skeletal muscle and heart, the presented data were not compared to commonly used promoters such as CK8e or MHCK7. Thus, it is not possible to determine the relative activity of C5-12 compared to other commonly used promoters. Based on levels of skeletal muscle dystrophin reported here, the C5-12 promoter seems equivalent to MHCK7⁹ used for Elevidys. Using the more quantitatively accurate LC/MS methods, the authors report micro-dystrophin levels at ∼20% of total NHP dystrophin using the C5-12 promoter. Based on the phase 3 Elevidys trials, the MHCK7 promoter yielded micro-dystrophin at ∼34% of healthy levels in muscle using traditional immunoblotting.⁹ However, the CK8e promoter may be more active; recent disclosures (February 2025) from Solid Biosciences report closer to 100% of healthy dystrophin levels at 90 days post-treatment in three patients, although the capsid used for those studies was screened for its myotropism. It bears mentioning that dystrophic muscles are better transduced with systemically delivered drugs^10,11 compared to healthy muscles due to the inflammatory environment and enhanced capillary permeability. Because the primates used in this study were not dystrophic, their muscle transduction was likely less than would have been achieved in a dystrophic context. Other factors may have also influenced microdystrophin levels, such as the presence of full-length dystrophin, which could compete for binding sites. While tissue-specific promoters enable restricted transgene expression to target tissues, the addition of miRNA target sites can further limit protein expression in certain tissues or cell types.¹²

Further studies on AAV biology in NHPs are needed. Moreover, more research on human patients who have received AAV treatments is essential, as the human immune system differs significantly from that of NHPs. While one study attempted to compare the efficacy of different microdystrophin species in mice,¹³ there are few studies with direct, head-to-head comparisons among AAV serotypes, transgenes, and promoters. The lack of direct comparisons renders it challenging to draw definitive conclusions. Despite this constraint, the data from this study represent a crucial first step in elucidating how variations in manufacturing methods, capsid serotypes, and promoter selection influence AAV performance in NHPs, offering critical insights that may bridge the gap between preclinical models and human clinical trials.

Declaration of interests

The authors declare no competing interests.