2.1 Gene therapy approaches

Over the last few decades, numerous new treatment options have been researched and developed, transitioning from bench to bedside. Among these, gene therapies have emerged that can persistently adjust RNA transcription levels or directly target the DNA sequences of pathogenic genes to modify gene product expression.²¹ These approaches—gene replacement, gene suppression, gene supplementation, and gene editing (Figure 1)—aim to correct gene mutations of various profiles with limited or even single treatments.^22-26 This significantly expands the scope of gene therapy for different types of genetic disorders, restoring the expression of normal genes to not only modify but potentially cure the diseases.

2.2 Delivery systems

Another crucial step in gene therapy is ensuring the accurate and safe delivery of the aforementioned cargos to targeted organs in vivo. Several delivery systems, which rely on either viral or nonviral vectors, are currently under investigation for the gene therapy of genetic diseases.⁴⁰

3.1 Leber congenital amaurosis

LCA is a severe inherited retinal dystrophy that is present from birth, characterized by significant visual impairment or blindness in early infancy. More than 20 gene mutations are currently known to be associated with LCA. Research on population genetic screening for these mutations in a cohort of Danish LCA probands revealed that RPE65 was the most frequently mutated gene, with biallelic variants found in 16% of the population.⁵⁹ RPE65, expressed in the retinal pigment epithelium (RPE), encodes an all-trans-retinyl ester isomerase that maintains the function of both rod and cone photoreceptors by participating in the biochemical recycling of chromophores in the visual cycle.⁶⁰ Mutations in RPE65 lead to the accumulation of detrimental precursor molecules and the degeneration of RPE cells, resulting in a progressive loss of photoreceptor cells.⁶¹

LUXTURNA (voretigene neparvovec-rzyl) was the first gene therapy medicine approved by the US FDA for rare genetic diseases, providing meaningful improvements in visual function for patients who previously had limited options.¹⁶ This therapy consists of human RPE65 cDNA with a modified Kozak sequence, driven by a cytomegalovirus (CMV) immediate early enhancer/chicken β actin (CAG) promoter, which enhances the efficiency of expression profiling and transcription. It is delivered via an AAV2 vector through subretinal surgery.^{62, 63} The mean bilateral multiluminance mobility testing (MLMT) score after 1 year for patients who received the therapy was significantly higher than that of the control group, with a difference of 1.6 (95% CI: 0.72–2.41, p = 0.0013).⁶⁴ Results from a 3-year follow-up of delayed intervention (DI) patients and a 4-year follow-up of original intervention (OI) patients in the Phase III trial (NCT00999609) indicated mean bilateral MLMT score changes of 2.4 and 1.7, respectively, with 71% of patients passing the MLMT at the lowest light level (1 lux) by year 3. Improvements in light sensitivity and visual field were also observed in both DI and OI groups, suggesting that the benefits of the therapy persist for up to 4 years.⁶⁵ The serious adverse events that occurred were related to the administration procedure rather than the gene therapy itself, and no harmful immune responses associated with either the viral capsid or the delivered gene were observed.⁶⁵ Nonetheless, as the first US FDA-approved gene therapy for rare diseases, LUXTURNA has been subject to ongoing controversy, much of which is common to other approved therapies. This includes concerns regarding the lack of full restoration of normal vision, insufficient long-term efficacy data, and significant barriers to accessibility and broader implications due to the high personal and societal costs, with a price tag of up to 425,000 US dollars per eye.⁶⁶

3.2 Hematological diseases

Hematological disorders, such as hemophilia, sickle cell disease (SCD), and thalassemia, impose a significant disease burden, surpassing that of rare diseases affecting other solid organs. This situation underscores the urgent need for new treatment options. In comparison with other cell types, hematopoietic stem cells (HSCs) are relatively straightforward to collect, culture, expand ex vivo, manipulate, and reinfuse, making them particularly suitable for gene therapy.⁶⁷ Furthermore, systemic treatment is inherently necessary for hematological diseases. Consequently, more than half of the medications on the approved list are directed toward genetic diseases of the hematological system.

3.3 Dystrophic epidermolysis bullosa

VYJUVEK (beremagene geperpavec) is currently the first and only approved topical gene replacement therapy for dystrophic epidermolysis bullosa (DEB).¹⁶ DEB is a rare disorder caused by genetic variations affecting cell adhesion, resulting in extremely fragile skin that is prone to blisters, wounds, and scars.⁹⁸ The major form of DEB is attributed to mutations in the COL7A1 gene, which encodes collagen type VII (COL7), a protein expressed in the basement membrane of the dermis that contributes to the maintenance of skin structural integrity and the development of keratinocytes and fibroblasts.⁹⁹

To deliver and replace COL7 in the cells surrounding the wounds of DEB patients and promote healing, VYJUVEK utilizes a genetically modified, replication-defective HSV-1 as the vector to express human COL7 as the therapeutic agent. The medication is combined with a sterile gel to enhance adherence to the wound, thereby stabilizing drug administration.¹⁰⁰ A double-blind Phase III trial (NCT04491604) evaluated the efficacy of VYJUVEK in 31 DEB patients, revealing that 46% (95% CI: 24–68, p = 0.002) more patients in the treatment group achieved complete wound healing, with only mild erythema considered a treatment-related adverse event after 6 months. No significant immunologic reactions were detected, even though the study design permitted readministration of the medication up to four times.¹⁰¹ An additional Phase I/II trial (NCT03536143) further demonstrated the expression of COL7 in the wounds, providing hope for an improved quality of life for DEB patients.¹⁰⁰

In parallel with the development of VYJUVEK, similar redosable topical approaches are under investigation, including HSV-1 delivered TGM1 and SPINK5 for autosomal recessive congenital ichthyosis and Netherton syndrome, respectively. Additionally, a strategy involving the treatment of patient-derived dermal cells with gene therapy, followed by ex vivo expansion and autotransplantation, is also undergoing clinical research.¹⁰²

3.4 Neurological disorders

3.5 Duchenne muscular dystrophy

Duchenne muscular dystrophy (DMD) is the most prevalent genetic muscular disorder, arising from genetic frame-shift mutations in the dystrophin gene, which is the largest gene in the human genome.¹²¹ This gene contains hotspot regions that are particularly susceptible to deletion; specifically, the distal hotspot region spans exons 44–55, while the proximal hotspot region encompasses exons 2–19, accounting for 85.7 and 14.3% of deletions, respectively.¹²² These mutations lead to the production of an abnormal and truncated protein, which hinders cell regeneration and induces myocyte necrosis.¹²³ Similar to SMA, several AONs targeting the mutated exons to induce exon skipping have been approved for clinical application in the treatment of DMD. However, these approaches face consistent challenges, including the need for repeated administration, poor delivery efficiency, ambiguous evidence, and high costs.^{22, 124-126}

ELEVIDYS (SPR9001, delandistrogene moxeparvovec) was a gene replacement method developed for ambulatory patients (4–5 years) with DMD, making it the only US FDA approved gene therapy for genetic muscular disorders at the time of manuscript completion.¹²⁷ This approach employs rAAVrh74 as the vector and the MHCK7 gene regulatory component as the regulatory element. The MHCK7 component is optimized from murine muscle creatine kinase (MCK), comprising a creatine kinase (CK) 7 promoter and an α-myosin heavy chain enhancer, which contributes to its specificity and high activation efficiency in muscle tissue.¹²⁸ The therapeutic cargo delivered consists of a shortened, functional, codon-optimized microdystrophin, which includes spectrin repeats 1–3 and 24, as well as hinges 1, 2, and 4.¹²⁹ Clinical evidence indicates the safety and potential for long-term improvement in muscle function following a single dose of SPR9001 treatment.^130-134  The expression of microdystrophin, detected by western blot, exhibited robust sarcolemmal localization at week 12 following the single intravenous infusion, with a mean (SD) change from baseline (CFBL) of 54.2% (42.6) (p < 0.0001).¹³² Several clinical studies concluded that, compared with a propensity-score-weighted external control (EC) cohort, the mean North Star Ambulatory Assessment (NSAA) scores significantly increased for up to 4 years, suggesting a positive effect of this treatment on disease progression.^{130, 134}

However, controversial opinions suggest that these studies did not achieve a statistically significant difference in NSAA compared with the placebo (95% CI: −0.45, 1.74; p = 0.24). Additionally, the correlation between the expression of microdystrophin and functional improvement remains uncertain, highlighting the need for more statistically significant primary endpoints or more robust subgroup analysis results to evaluate therapeutic efficacy.¹³⁵

The most common treatment-related adverse events observed were mild to moderate vomiting and hepatotoxicity, which occurred within 70 days posttreatment and subsequently resolved.¹³⁴ Drawing from the experiences of these trials, practical considerations have been outlined in a report recommending that anti-AAV antibody levels, liver function, platelet count, and troponin-I should be monitored before and after AAV-based gene therapy. Furthermore, the maintenance of corticosteroid therapy postinfusion is deemed important for managing adverse events.¹³¹

Moreover, it is evident that numerous therapeutic approaches are currently under investigation in both preclinical and clinical studies.¹⁴ Among these, the advancements in gene therapy for genetic muscular disorders serve as an exemplary case illustrating the developmental trends in this field. These disorders share significant gene therapy requirements with diseases affecting other systems, owing to the complexity of the molecular pathogenesis underlying muscle diseases, the extensive number of muscle cells that necessitate therapeutic targeting, and the specific targeting requirements of various muscle types.^17-20

4.1 Mutational spectrum of genetic muscular diseases

Current molecular biology research has highlighted that numerous genetic muscular disorders are attributed to monogenic diseases, suggesting that gene therapy could serve as an effective treatment strategy (Figure 3).

For instance, dystrophin, as mentioned above, is one of the major components of the dystrophin-associated protein complex (DAPC).¹⁴⁰ Monogenic mutations in other subunits of the DAPC can also directly affect dystrophin function, leading to various types of muscular dystrophies, including certain forms of LGMDs.¹⁴¹ Alternatively, mutations in other related genes can indirectly contribute to muscular dystrophies; for example, OPMD is caused by a loss-of-function mutation in an RNA binding protein,^{142, 143} and excessive activation of DUX4 triggers FSHD.²³

Congenital myopathy, however, encompasses a group of more complex disorders that are less well understood. These diseases can arise from either monogenic or multigenic mutations, but their subtypes are classified according to the morphological features observed in muscle biopsies. Notably, the same mutation may result in different muscle pathologies. Recent knowledge indicates that, irrespective of the distinct morphologies, these mutations commonly disrupt the function of either actin thin filaments, the excitation–contraction coupling apparatus, sarcoplasmic reticulum Ca²⁺ stores, or myosin.¹⁴⁴

Pompe disease, also known as acid maltase deficiency or glycogen storage disease type II, affects muscle function through a distinctly different mechanism. This metabolic disorder is caused by mutations in the glucosylceramidase (GAA) gene, which encodes the lysosomal enzyme acid alpha-glucosidase, responsible for catalyzing the conversion of glycogen into glucose. A deficiency of this enzyme results in lysosomal glycogen accumulation, leading to significant pathology in both skeletal and cardiac muscle.¹⁴⁵

With deepening understanding of these molecular pathogenic mechanisms and advancements in gene therapy methods, numerous genetic muscular disorders are currently undergoing clinical trials involving gene therapy (Table 2), while new treatment options continue to emerge through preclinical research (Table 3). Among these approaches, gene replacement remains the most prevalent strategy in clinical trials. Concurrently, as gene editing technology continues to mature, both in vivo and stem cell-based ex vivo gene editing techniques are being rapidly implemented in clinical trials for specific types of muscular dystrophies.

4.2 Clinical research progress of genetic muscular disorders

4.3 The development of preclinical research in genetic muscular disorders

The success of a gene therapy product in clinical trials depends on various preliminary data obtained from appropriate pre-clinical settings. In addition to the preclinical programs associated with the aforementioned clinical solutions, numerous promising gene therapy research initiatives are underway to broaden the range of treatable muscle-related disorders and enhance treatment techniques (Table 3).

5.1 Novel strategy of AAV package

To address the limitations of packaging capacity during AAV-mediated therapy, mini-proteins have been developed for the treatment of DMD. However, this strategy may not be universally applicable to other genetic disorders characterized by large genes. In many gene therapy approaches that utilize long fragment nucleic acids as delivery vehicles—such as base editing, prime editing, or gene replacement that requires full-length cDNA—the packaging capacity of AAV vectors is often exceeded.^{188, 224, 225}

To overcome this limitation, a dual AAV delivery system has been established. For instance, a study divided CBE or ABE components into N-terminal and CT halves. Each half was fused with a fast-splicing split-intein and packaged into separate AAVs. When coexpressed through the coinfection of AAV particles, the components spliced in trans to reconstitute the full-length base editor.^{257, 258} The therapeutic editing efficiency was demonstrated in the mouse brain, liver, retina, heart, and skeletal muscle, presenting an innovative approach for AAV-mediated in vivo gene therapy utilizing relatively large editing tools.²⁵⁸ However, it is important to note that the nature of dual vector systems necessitates higher titers of viral vectors, which may be less effective than single virus systems and could pose increased immunological and toxicological concerns.

Spliceosome-mediated RNA trans-splicing (SMaRT) offers an alternative therapeutic strategy for monogenic diseases characterized by mutations near the 5' or 3' exons.^{259, 260} This approach targets RNA at the pre-mRNA level by utilizing pre-trans-spliced molecules (PTMs) that incorporate the wild-type coding sequence, a robust splicing site, and a binding domain to ensure the specific binding of PTMs to the splicing site of the targeted pre-mRNA. The PTMs facilitate the incorporation of the WT coding sequence into the corresponding mRNA by obstructing the RNA splicing of the mutant exon on the pre-mRNA. An in vitro study utilizing primary myotubes derived from a mouse model of LMNA-related CMD demonstrated that a PTM targeting intron 5 of Lmna pre-mRNA effectively rescued the mutant phenotype. However, further development is necessary to improve the efficiency of such AAV-delivered in vivo therapies.²⁶¹

5.2 Immunotoxicity of the AAV capsids

Achieving durable and efficient gene therapy in patients with muscular disorders necessitates high doses of AAV vectors as the high proportion of muscle mass. However, this requirement for elevated vector dosages presents significant challenges regarding the safety of AAV-based gene therapies, as evidenced by serious adverse events reported in several clinical trials.^{149, 151} Such adverse events, which range from transient vomiting and nausea to more severe occurrences such as hepatotoxicities, thrombocytopenia, renal impairment, and anemia, were frequently observed in the early stages of multiple AAV trials. These events are highly likely to be associated with innate immune reactions related to systemic doses of AAV capsids.^{113, 262} Nevertheless, further research is needed to investigate the detailed mechanisms involved, given the varying immunosuppression strategies employed in different trials, gene therapy modalities, and diseases. Additionally, wide variations in the quantification of vectors may still exist in current clinical trials.²⁶³ It is essential to develop standardized protocols for collecting and analyzing higher quality data sets across different sponsors to facilitate the development of safe and effective clinical trials.

On the other hand, many humans possess pre-existing NAbs to AAVs that are induced by natural infections during childhood. The seroprevalence rates of these antibodies vary significantly according to different serotypes, geographic locations, and across various reports.²⁶⁴ In a study that defined seropositivity as a neutralizing titer of greater than or equal to 3.1, antibodies against AAV serotypes 2, 5, 6, and 8 were detected in 100, 36, 91, and 90% of subjects, respectively.²⁶⁵ Another study, which analyzed a broader range of serotypes, revealed that the prevalences of total immunoglobulin G (IgG) against anti-AAV1 and -AAV2 were higher (67 and 72%) compared with those against anti-AAV5 (40%), anti-AAV6 (46%), anti-AAV8 (38%), and anti-AAV9 (47%).²⁶⁶ These pre-existing NAbs can potentially inhibit the efficacy of AAV-based gene therapy. Furthermore, the administration of rAAV vectors in immunologically naive patients can lead to the neutralization of a subsequent injection of the AAV vector due to the generation of NAbs from acquired immunity.^{55, 56} Additionally, some NAbs have been reported to cross-react among different AAV serotypes, further limiting options for sequential gene transfer.²⁶⁷ Several strategies are currently under investigation to address these challenges, including immune suppression to block antibody formation, depletion of NAbs, and the engineering of novel vectors.

5.3 Nonviral vectors

Alternatively, the development of novel nonviral vectors, which possess a high capacity for carrying genetic material, offers improved safety, convenient accessibility, and the potential for large-scale production. These delivery approaches may provide an effective means to address the immunotoxicity and packaging limitations associated with viral vectors. Recent discussions have highlighted various aspects of nonviral vectors applied in gene therapy for genetic muscular disorders.⁵⁸ Generally, polymer nanoparticles, LNPs, inorganic nanocarriers, and nucleic acid conjugates are promising candidates for delivering gene therapy cargos to muscle tissues.^292-297 Some recent advancements are outlined below as examples.

The cell-internalization SELEX (Systematic Evolution of Ligands by Exponential Enrichment) approach can be employed to select RNA aptamers that bind to and internalize within cells, thereby providing a valuable tool for recognizing and delivering linked cargos into muscle cells.²⁹⁸ After 15 cycles of skeletal muscle cell-internalization SELEX, a screened RNA aptamer labeled A01B was found to preferentially internalize in skeletal muscle cells both in vitro and in vivo while exhibiting decreased affinity for off-target cells.²⁹⁹  This finding suggests a novel strategy in which the A01B RNA aptamer, when fused to short therapeutic AONs,³⁰⁰ enhances the targeting capability of AONs against muscular dystrophies. Similarly, the improved targeting ability of AONs fused with cell-penetrating peptides and antibodies, or linked with stereochemical backbones, is also under investigation.^301-303 These naked oligonucleotides encounter significant challenges in their delivery to the heart, diaphragm, and CNS. A recent study indicated that the use of lipid-ligand-conjugated complementary strands, when hybridized with these cargos, markedly enhanced delivery efficiency in a DMD mouse model, effectively normalizing cardiac and CNS abnormalities without any adverse effects.²⁴⁹

Furthermore, nonviral vectors have demonstrated the capability to deliver relatively long DNA cargos in gene therapy targeting muscular disorders. A pH-dependent ionizable lipid with three hydrophobic tails has been reported to formulate a LNPs delivery system that successfully delivered a CRISPR/Cas9-mediated editing tool into the skeletal muscle of a humanized DMD mouse model, effectively rescuing the loss of dystrophin protein. Importantly, the LNP system can be repeatedly administered intramuscularly, leading to cumulative recovery of dystrophin protein.²⁵² Extracellular vesicles (EVs), defined as heterogeneous populations of membrane-bound vesicles released by various cell types, are also considered candidate gene therapy vectors due to their low toxicity, low immunogenicity, and high loading capacity.^{304, 305} A recent study demonstrated the specific and rapid disruption of miR-29b expression using an artificially engineered EV-based delivery system, which employed CRISPR/Cas9 to target miR-29b (EVs–Cas9–29b), and observed a protective effect in an induced muscle atrophy mouse model.³⁰⁶ The delivery capacity of other nanoparticles has also been investigated. A calcium phosphate (CaP) biomineralization system has been reported to efficiently load ribonucleoprotein complexes of Cas9 protein and sgRNA (RNPs) to form nanoparticles with good serum stability.³⁰⁷ Cellular uptake, nuclear transport, and subsequent DMD gene editing were observed in this study, suggesting a stable and safe nonviral strategy for gene therapy in muscular disorders.³⁰⁸ Moreover, nonviral vectors can be combined with AAV capsids to mitigate the clearance of AAV caused by NAbs in vivo. EVs containing packaged AAV have demonstrated the ability to reach the CNS and exhibit greater resistance to NAbs compared with unprotected AAV.³⁰⁹ In another study, AAV capsids were modified with unnatural amino acids to which oligonucleotides were ligated. These oligo-AAVs could be identified by specific sequences and, importantly, were shielded by lipid-based cloaks that effectively protected them from neutralizing serum while retaining full functionality in cell transduction.³¹⁰

Despite these advancements, challenges in nonviral delivery approaches remain significant and cannot be overlooked. Enhancing the stability and increasing the circulation time of these medicines within the body are critical to ensuring treatment efficacy. Neutral auxiliary lipids, such as dioleoylphosphatidylethanolamine and cholesterol, can improve the colloidal stability of cationic LNPs.³¹¹ Additionally, PEGylation modifications can reduce nanoparticle aggregation and enhance their stability, thereby improving immune evasion from the reticuloendothelial system.³¹² Furthermore, these vectors must navigate multiple biological barriers, including the endothelium, the extracellular matrix of skeletal muscles, and the cell membrane.³¹³ By precisely regulating the physicochemical properties of nonviral vectors—such as surface potential, shape, size, and hardness—the internalization efficiency of cells toward these vectors can be influenced.³¹⁴ Concurrently, the incorporation of cell-specific ligands on the surface of these vectors can significantly enhance their binding to cell surfaces, a topic that will be explored in a subsequent section.³¹⁵ Moreover, there are intracellular barriers that nonviral vector-delivered cargos must overcome, including lysosomal escape, intracellular dissociation, and appropriate localization. Selecting an appropriate vector and modifying it with reducible sensitive groups, such as disulfide bonds, may facilitate lysosomal escape and the intracellular release of cargo.³¹⁶ Additionally, cargo linked with nuclear localization sequences (NLS), or covalently modified with NLS peptides and other molecules, can enhance nuclear localization and therapeutic efficacy.²⁵³

5.4 Immunotoxicity of the cargos

The Cas protein-specific immune response is a significant concern in gene therapy, primarily due to the bacterial and archaeal origins of Cas proteins.³¹⁷ Research has shown that the introduction of SaCas9 into the liver elicits a robust memory T cell response in mice, leading to the near-complete elimination of gene-edited hepatocytes.³¹⁸ Similar immune responses have been observed in a dog model of DMD, where gene editing was performed through Cas9-mediated exon skipping. Although initial assessments indicated substantial dystrophin restoration, the treatment also resulted in muscle inflammation accompanied by Cas9-specific humoral and cytotoxic T-lymphocyte responses, despite the muscle-specific expression of the Cas9 protein.³¹⁹ Consequently, the combination of immunosuppressive therapy during CRISPR/Cas-based gene editing, such as the introduction of corticosteroids or Cas9-reactive regulatory T cells, may be considered in clinical practice.³²⁰ Enhancements to the CRISPR/Cas system represent another potential strategy. Optimized forms of Cas9 proteins with reduced immunogenic epitopes have been developed.³²¹ However, further evidence regarding antibody responses to these highly variable epitopes is necessary. Additionally, self-deleting CRISPR/Cas systems, which utilize sgRNA that targets themselves, along with more specific and efficient editing tools, may further mitigate the immune response associated with CRISPR/Cas-based gene therapy.³²²

In addition to the foreign protein, it has long been established that microdystrophin can also elicit immune responses during gene replacement therapy, as evidenced by the detection of dystrophin-specific T cells.^{323, 324} Ongoing analyses across several studies suggest that patients with N-terminal deletions of the dystrophin gene, particularly in regions corresponding to sequences represented in the microdystrophin, may exhibit immune naivete or cross-reactive immunologic material negativity for these N-terminal epitopes. This implies that the potentially harmful T-cell response targeting specific peptides is likely encoded by exons 8–11.³²⁴ Redesigning the corresponding constructs of microdystrophin to be less likely to provoke an adaptive immune response may offer a viable solution. Alternatively, the codon-optimized miniaturized utrophin, which serves as the autologous homologue of dystrophin, has been reported to prevent the muscular dystrophy phenotype in mouse and dog models.³²⁵ Further comparison between microutrophin and microdystrophin, when directly injected into the muscle of a dog model, revealed that, in contrast to the strong T-cell response observed against microdystrophin, no evidence of cell-mediated immunity was detected in the microutrophin injected model.³²⁶

5.5 Tissue specifical delivery

Given the toxicity concerns associated with high-dose systemic delivery of viral vectors, various strategies have been evaluated for their potential to enhance delivery specificity, thereby reducing the necessary viral dosage and improving therapeutic efficacy. The fusion of membranes, directly mediated by fusogens, is a critical step that enveloped viruses utilize to facilitate entry into new hosts.³²⁷ Additionally, the cell-cell fusion process occurs during skeletal muscle development and regeneration, with Myomaker and Myomerger acting as fusogens to regulate the fusion of progenitor cells into multinucleated myofibers.³²⁸ It is particularly promising that muscle cell fusogens could potentially guide gene therapy vectors, significantly enhancing muscle cell delivery specificity. This concept was supported by a recent study demonstrating that the local and systemic injection of an engineered lentivirus, incorporating Myomaker and Myomerger on its membrane, resulted in specific transduction of skeletal muscle. This system exhibited both the capability and specificity for delivering mDystrophin to skeletal muscle in a DMD mouse model, thereby alleviating pathology.²⁵⁰ This reported strategy serves as an impressive example, encouraging further research into the development of novel receptor-binding ligands or protein-based adaptor molecules, such as Designed Ankyrin Repeat Proteins (DARPin), nanobodies, or single-chain antibodies, to restrict cell transduction toward target cells in various delivery vehicles, including nonenveloped viral and nonviral vectors.³¹⁵ For instance, a class of AAV capsids containing an arginine-glycine-aspartic acid (RGD) motif, as well as a myotropic peptide-fused capsid, has demonstrated superior efficiency and selectivity for muscle transduction following intravenous injection in both mice and nonhuman primates.^{251, 329}

Another concern regarding muscle fiber cell-targeting gene therapy is that the therapeutic effects may be diminished due to transgene loss, which can occur alongside ongoing muscle fiber degeneration and regeneration, as well as during muscle growth in treated children.³³⁰ Since satellite cells, also known as skeletal muscle stem cells, play a crucial role in the formation of primary and secondary myofibers during muscle development and injury repair,³³¹ targeting these cells with gene therapy holds the potential to provide a renewable pool of “corrected” cells for existing muscle fibers. A study utilizing a novel dual reporter mouse model has demonstrated that AAV9 and AAV8 are suitable for transducing satellite cells through both local and systemic injections. Furthermore, the AAV-delivered Cas9-based gene editing system possesses the capability to edit satellite cells using muscle-specific promoters, including CK8e, SPc5-12, and MHCK7.³³² These findings underscore the promise of gene therapy, particularly gene editing directed at satellite cells; however, further investigation into delivery methods and expression specificity in large animal models before human trials is necessary.

Conversely, enhancing the cellular or subcellular targeting abilities of the delivered cargos may also improve the tissue specificity and therapeutic efficiency of gene therapy. The myonuclear propagation capability of these cargos can augment overall editing efficiency, as the pathogenic products from nontransduced nuclei in multinucleated muscle cells may dilute the transgenic products.³³³ By screening various combinations of NLS, nuclear export signals, and other elements, a set of short peptide sequences termed “Myospreader” was identified. This sequence promotes Cas9 myonuclear propagation across myofibers, enhances protein stability, and increases muscle gene editing efficiency in vivo.²⁵³ This underscores the significance of considering the spatial dimension in gene regulation, editing, and therapy.

The key to addressing the muscle-related issues associated with Pompe disease lies in the effective delivery of glucosylceramidase to muscle tissue. To enhance the targeting of glucosylceramidase to muscles, a HSC-mediated lentiviral gene therapy was developed, which expresses an insulin-like growth factor 2 (IGF2) peptide fused to glucosylceramidase (designated LV-IGF2.GAAco). This approach is based on the principle that ex vivo gene-modified autologous HSPCs can stably overexpress the therapeutic transgene, while the IGF2 peptide increases binding affinity to the cation-independent mannose-6-phosphate/IGF2 receptor (M6PR/IGF2R).^{255, 334} Preclinical data demonstrated nearly complete phenotypic and proteomic correction in the primary affected organs of a mouse model when compared with LV-GAA alone, indicating its potential for further clinical applications.^{255, 256}

5.6 Tissue specifical expression

Successful gene therapy depends not only on the selection of an appropriate vector but also on the specific expression of the appropriate exogenous genes in targeted tissues. A properly selected promoter for transgene expression can facilitate long-term, sustained high expression in muscle tissues while demonstrating limited activity in other areas. This is crucial for the gene therapy of muscular disorders. Specifically, a short regulatory sequence that effectively drives high and robust expression of a therapeutic transgene in various muscle types, while exhibiting minimal activity in nontarget tissues, is vital for multiple gene therapy strategies addressing genetic muscular disorders.

An earlier review summarized the commonly used muscle-specific promoters based on skeletal muscle α-actin, MCK, and desmin genes, and provided information on current gene therapy candidates that utilize these promoters.³³⁵ A more recent study compared some frequently used muscle-specific promoters and demonstrated that the Desmin and MHCK7 promoters exhibited stronger reporter gene expression levels in myogenic cell lines and also promoted gene expression in cardiac cells, whereas miR206 and CAPN3 promoter expression was restricted to skeletal muscle.³³⁶ Despite advancements in the understanding of these natural promoters, there remains a demand for higher therapeutic gene expression levels under specific treatment conditions. The construction of a novel muscle hybrid (MH) promoter may address these additional requirements. Research data indicated that when combined with AAV2/9, the MH promoter—composed of the Des gene enhancer, Ckm gene enhancer, modified Ckm gene core promoter, and a small intronic enhancer derived from the Ckm gene—demonstrated higher in vivo expression levels in skeletal muscle and the heart compared with CMV, desmin, or CKM-based promoters, regardless of whether delivery was intramuscular or systemic.³³⁷ Furthermore, a multistep computational approach employing a genome-wide data-mining strategy has been reported to design robust muscle-specific transcriptional cis-regulatory modules, resulting in a substantial increase in muscle-specific gene transcription (up to 400-fold) when delivered with AAV in mice, without any discernible immune complications.³³⁸

5.7 CNS delivery

Several types of genetic muscular disorders, including DMD and Pompe disease, are associated with CNS involvement.^{19, 20} However, the ability of gene therapies to traverse the blood–brain barrier is limited when delivered systemically. To address this limitation, numerous studies have proposed IT delivery methods, which target the CNS and peripheral organs via cerebrospinal fluid, significantly enhancing cargo delivery to the brain and spinal cord. In two studies involving DMD mouse models, AONs delivered via ICV and intra-cisterna magna routes demonstrated detectable exon 51 skipping in various brain regions. This was linked to partial improvements in anxiety traits, unconditioned fear responses, and Pavlovian fear learning and memory.^{247, 248} Regarding viral-based delivery, a single IT administration of AAV–hGAA in 1-month-old GAA–KO 6neo/6neo mice was performed, resulted in reduced glycogen levels in the brainstem, spinal cord, and left cardiac ventricular wall. Neurological correction, neuromuscular improvement, and the resolution of hypertrophic cardiomyopathy were monitored at 4, 9, and 12 months posttreatment.³³⁹

5.8 Off-target issues

Beyond the immunogenicity of the vector and Cas protein discussed, the potential for off-target activity in CRISPR/Cas systems also presents a barrier to their clinical application. While most off-target cleavage events have been reported in highly proliferating cells in culture, they were believed to be minimal in postmitotic skeletal and cardiac cells.³⁴⁰ However, the unexpectedly high rate and diverse distribution of AAV gene insertion into the host genome via CRISPR-induced DSBs during gene therapy in muscle cells raises concerns about off-target effects in therapies for muscular disorders.²²² Consequently, the risk of deleterious off-target mutations in susceptible cells in vivo cannot be disregarded, particularly with systemic administration. Various strategies have been developed to mitigate off-target mutagenesis, including the careful selection and design of sgRNAs, the development and use of high-fidelity Cas enzymes, and the implementation of less mutagenic tools, alongside the application of muscle cell-specific vectors and promoters as discussed above. Furthermore, the delivery of RNPs containing recombinant Cas9 protein and guide RNA, rather than transgenes, has demonstrated immediate cleavage while reducing off-target sites across multiple cell types due to the rapid degradation of these particles within cells.³⁴¹ This approach offers another potential delivery method in conjunction with nonviral vectors for gene editing.

In conjunction with traditional biochemical methods, the enriched databases related to gene editing and the advancements in artificial intelligence technology over the past few years have led to the emergence of data-driven machine learning methods as a powerful new approach for predicting both on-target and off-target effects of the CRISPR/Cas system.³⁴² Currently, both traditional machine learning models and deep learning models are being utilized in this field. Numerous tools have been developed using traditional machine learning models to predict off-target sites, design optimized sgRNA libraries that maximize on-target activity while minimizing off-target effects,³⁴³ and extract gRNA knockdown efficiency based on specific sequence features.³⁴⁴ As gene editing datasets continue to expand, deep learning models have demonstrated significant advantages in processing large volumes of data characterized by complex nonlinear patterns. Reports indicate that deep learning models yield superior prediction results, as measured by AUPRC and AUCROC, when employing novel sgRNA–DNA sequence encoding strategies compared with traditional models.³⁴⁵ Furthermore, biological features beyond sequence patterns can also be incorporated into deep learning models for predictive purposes. For instance, the binding energy estimations of the gRNA-DNA hybrid as well as the Cas9–gRNA–DNA hybrid have been successfully utilized to predict on-target activity and off-target sites.³⁴⁶ Additionally, deep learning models have been applied to address data imbalance issues that hinder off-target prediction,³⁴⁷ as well as to enhance feature explainability for accurate on- and off-target predictions,³⁴⁸ underscoring their emerging importance in predicting gene editing effects.