Commentary on: ribozyme-activated RNA trans-ligation: an emerging strategy for large gene delivery in muscular dystrophies

INTRODUCTION

Large genes implicated in monogenic disorders such as Duchenne muscular dystrophy (DMD) and dysferlinopathies have long challenged gene therapy development due to their size exceeding the packaging capacity of adeno-associated virus (AAV) vectors. Although split-intein approaches have enabled the reconstitution of full-length proteins from dual AAV vectors, the intein strategy relies on post-translational repair at the protein level. In a recent Science publication, Lindley et al. introduced Stitch RNA, or “StitchR”, a ribozyme-mediated RNA trans-ligation platform that restores large gene expression by repairing RNA precursors before translation^[1]. The name evokes the concept of “stitching” RNA fragments together without leaving residual sequences. This commentary explores the innovation and translational potential of StitchR, and contrasts it with clinically advancing split-intein systems, and discusses remaining challenges for clinical applications.

NOVELTY AND IMPACT OF THE STITCHR PLATFORM

To assess the feasibility of ribozyme-mediated, scarless RNA trans-ligation in mammalian cells, the investigators engineered two plasmids encoding nonoverlapping N-terminal (Nt-GFP) and C-terminal (Ct-GFP) fragments of the green fluorescent protein (GFP) reporter gene. Each transcript was designed with flanking self-cleaving ribozymes - a 3’ hepatitis delta virus (HDV) ribozyme for Nt-GFP and a 5’ hammerhead (HH) ribozyme for Ct-GFP - to precisely excise the ribozyme sequences and generate ligation-competent RNA ends. Co-transfection of Human embryonic kidney 293T cells (HEK293T) with both plasmids resulted in robust green fluorescence in the absence of sequence homology or splint oligonucleotides, indicating efficient trans-ligation. Reverse transcription polymerase chain reaction (PCR) and Sanger sequencing across the predicted junction confirmed seamless ligation at the ribozyme cleavage sites. Full-length GFP protein was detected exclusively in cells co-transfected with both constructs, demonstrating restoration of functional gene expression via ribozyme-facilitated RNA trans-ligation.

The StitchR platform employs a two-vector design in which the target gene is divided into N- and C-terminal halves. The Nt vector contains a splice donor (SD) sequence, and the Ct vector carries a splice acceptor (SA) sequence, forming an engineered intron architecture that defines the ligation junctions. Each construct includes small engineered catalytic ribozymes positioned at the 3’ end of the Nt vector and the 5’ end of the Ct vector. Upon transcription, these engineered ribozymes self-cleave messenger RNA (mRNA) precursors at defined sites, generating precise RNA termini compatible with endogenous RNA ligases such as RNA terminal phosphate cyclase-like protein B (RNA ligase RtcB). Following ribozyme self-cleavage, the resulting RNA termini are recognized by RtcB, which mediates scarless and accurate trans-splicing between the two transcripts. During this process, the engineered intron is completely excised, and the joined mRNA sequence is identical to the target transcript. Consequently, translation produces a full-length protein identical in amino acid sequence to the target, without any additional residues or exogenous sequences. Using the StitchR system, two separate RNA fragments, encoded by independent AAV vectors, are scarlessly ligated inside cells to regenerate full-length transcripts. Through in vitro systematic optimization - including ribozyme selection and the incorporation of spliced introns - Lindley et al. improved expression levels to 75-95% of full-length single vector controls^[1]. To optimize the reconstitution efficiency, Nt and Ct constructs containing luciferase (Luc) were transiently co-transfected at equimolar concentrations into COS7 [CV-1 in Origin Simian-7 cells (African green monkey kidney fibroblast-derived cell line)] cells under the cytomegalovirus immediate-early 94 promoter (CMV IE94). Cells were harvested 18-48 h post-transfection for Luc assays. The relative Luc activity of different StitchR versions, each incorporating either active ribozymes or catalytically inactive mutant ribozymes, was compared with Luc expressed from single-vector constructs containing the full coding sequence, with or without an intron. Optimization of StitchR from version 1.0 to 4.0 resulted in an approximately 900-fold increase in activity, with the optimized version achieving 75% of the Luc activity of the single-vector construct^[1].

Importantly, the StitchR platform was used to restore full-length dysferlin protein and a functional, but truncated dystrophin (∆H2-R15) protein in murine models, leading to the membrane localization of therapeutic proteins, normalization of muscle architecture, and biochemical correction of disease markers such as serum creatine kinase (CK). These results highlight the potential of StitchR to overcome a major limitation in large gene therapy applications.

COMPARISON TO CLINICALLY ADVANCED SPLIT-INTEIN-MEDIATED GENE RECONSTITUTION

In contrast to the StitchR platform, the split-intein approach for muscular dystrophies, such as those developed by Tasfaout et al., utilizes post-translational protein splicing to reconstitute large proteins from dual- or triple-AAV delivered fragments^[2,3]. The split-intein strategies have shown promising protein reconstitution results in various preclinical rodent models^[2] and have been proven effective in the non-human primate (NHP) models^[4,5]. Notably, the split-intein approach has now reached the clinical stage in the United States (U.S.). In 2024, the United States Food and Drug Administration (U.S. FDA) granted Investigational New Drug (IND) clearance for SpliceBio’s split-intein-based gene therapy, SB-007, to initiate a Phase 1/2 clinical study for Stargardt disease^[6,7]. Table 1 highlights and contrasts features of these two gene splicing methodologies. StitchR and split-intein platforms use different mechanisms for large-gene reconstitution. StitchR performs ribozyme-mediated RNA ligation to produce a single scarless mRNA, whereas split-intein-mediated protein trans-splicing reconstitutes the full-length protein from separate fragments following translation. In preclinical mouse models, StitchR achieved near wild-type (WT) dystrophin expression after intraperitoneal injection, reaching ~95% in vitro and up to 95% in quadriceps, 42% in tibialis anterior (TA), and 75% in heart in vivo^[1]. Split-intein dual-AAV approaches also restored high dystrophin levels, about 75% in TA and up to 178% in cardiac muscle via systemic administration, along with improved muscle morphology and recovery of contractile force^[2]. Corresponding in vivo force data for StitchR have not yet been reported. For both platforms, the serum CK levels were reduced near WT levels, indicating effective stabilization of the muscle membrane, while the proportion of centrally nucleated myofibers was markedly reduced, reflecting enhanced regeneration and structural integrity. From a safety standpoint, StitchR may introduce novel junctional epitopes; however, these seams are minimal, whereas neoepitopes from split-intein constructs may pose immunogenicity. Thus, while StitchR offers a conceptually elegant alternative, split-intein approaches have demonstrably entered human clinical testing, underscoring their current translational advantage.

POTENTIAL UNKNOWNS AND RISKS FOR CLINICAL TRANSLATION

Despite impressive preclinical results, several key unknowns must be addressed before StitchR can be advanced to clinical development:

• Immunogenicity of stitched mRNAs: Although scarless in design, new junctional sequences could generate cryptic epitopes presented on major histocompatibility complex (MHC) complexes. The immunogenicity of stitched mRNA can be assessed by combining in silico MHC-binding predictions, in vitro T-cell activation or cytokine release assays, and in vivo immune profiling, including Enzyme-linked immunospot assay (ELISpot), flow cytometry for cellular immune responses, and serum cytokine panel analysis following AAV administration^[8-10].

• Variable Efficiency Across Cell Types: The activity of endogenous RNA ligases such as RtcB may differ between tissues, disease states, and species. Tissue-dependent kinetics can be quantified across key organs, including skeletal muscle, heart, liver, kidney, spleen, and brain, using enzyme activity assays to assess RtcB kinetics and targeted proteomic analysis to assess its abundance^[11].

• Persistence and Fidelity: The long-term stability, translation fidelity, and potential for errors in stitched transcripts are not yet established. Fidelity and stability can be monitored by long-read RNA sequencing (RNA-seq; PacBio or Nanopore) to detect mis-ligated transcripts and by mass spectrometry-based proteomics to confirm accurate junctional peptide formation and the absence of truncated translation products^[12-14].

• Off-Target or Aberrant RNA Ligation: Although not detected in initial studies, there remains a theoretical risk of aberrant RNA recombination events, and unligated RNAs may persist to elicit immune responses or off-target effects. Off-target trans-ligation events can be assessed using transcriptome-wide and long-read RNA-seq. RNA-seq provides a quantitative context for expression changes, whereas long-read sequencing enables direct identification of aberrant or mis-ligated transcripts. To further explore the molecular basis of these events, cross-linking immunoprecipitation sequencing (CLIP-seq) can be employed to map RNA-protein interaction sites potentially involved in aberrant ligation^[15,16].

• Dose and Stoichiometry Requirements: Achieving consistent co-transduction and optimal stoichiometry of N- and C-terminal AAV vectors poses challenges, especially in large animals and humans. Fluorescently barcoded AAVs or Droplet digital polymerase chain reaction (ddPCR) quantification of vector genomes across targeted tissue can be used to quantify vector stoichiometry and guide optimization of dose ratios in vivo.

• Regulatory authorities may require detailed safety characterization of the novel RNA splicing mechanism, adding complexity to future translational pathways. From the regulatory perspective, early implementation of (i) good laboratory practice (GLP)-compliant biodistribution studies, to define tissue persistence/clearance and exposure-toxicity relationships; (ii) off-target transcriptomic profiling, to detect unintended ligation events and aberrant junctions; and (iii) immunogenicity assessments in rodents and NHPs, to support translational risk evaluation, will be critical to facilitate future IND submissions^[17,18].

• Effectiveness in ligating three or more fragments: Extra-large proteins, such as full-length dystrophin^[3] and Ryanodin receptor, are three or even four times larger than the AAV packaging capacity. Although StitchR showed high efficacy in ligating two fragments delivered via dual AAV infusion, the feasibility of efficiently ligating multiple RNA fragments using two or more ribozymes remains unknown.

EMERGING LITERATURE AND FUTURE DIRECTIONS

Among the earliest dual AAV strategies developed to overcome the limitation of cargo size was the overlapping vector strategy, where homologous sequences facilitated concatemerization or homologous recombination between the vector genomes^[19-22]. Despite its conceptual simplicity, this method has typically yielded low to moderate reconstitution efficiencies. In recent years, however, advances in dual AAV technologies have propelled research beyond cargo capacity constraints. Increasingly, emphasis has focused on optimizing reconstitution efficiency and systematically evaluating the risks and benefits of each method. Below are the recent publications that extend the landscape of dual AAV technologies.

Protein trans-splicing:

Beyond therapeutic protein reconstitution, split-intein system has been employed to reconstitute large gene-editing enzymes such as prime editors (PEs), base editors, and CRISPR-associated protein 9 (Cas9) variants, enabling dual-AAV delivery of oversized genome-engineering systems.

• Davis et al. developed and optimized a PE delivery system using a split-intein AAV approach. This strategy enabled efficient in vivo prime editing, achieving editing efficiency ranging from 11% to 46% across various mouse organs^[23]. Similarly, Muller et al. demonstrated a base editing method via split-intein dual AAV vectors, which resulted in high editing efficiency of up to 87% in retinal pigment epithelial cells of a NHP model^[24].

mRNA trans-splicing:

Similar to split-intein-mediated reconstitution, the StitchR approach is not confined to large-gene therapeutic applications; it has also been adapted as a prime editing platform. In addition, other mRNA trans-splicing-based methods, including RNA trans-splicing enhancer systems, have been investigated to enhance trans-splicing efficiency.

• Riedmayr et al. described RNA trans-splicing enhancer strategies to improve dual-AAV expression, though these approaches yielded more modest gains relative to StitchR’s reported efficiencies^[25].

• The StitchR platform introduced by Lindley et al. for delivering the large PE retained approximately 82% of the in vitro editing activity of the optimized PE (PEmax). Additionally, StitchR showed approximately twice the editing activity compared to the split-intein system when both methods were evaluated under the same experimental conditions in a cell-based assay^[1].

DNA recombination:

DNA recombination between overlapping AAV genomes enables reconstitution of the full-length gene in preclinical models. This strategy has been extensively evaluated in animal studies and represents a foundational approach for overcoming the AAV packaging limit in large-gene delivery.

• Datta et al. described a novel Cre recombinase-loxP recombination system (CRE-lox) DNA recombination method that allows for the delivery of a large gene using up to four AAV vectors. This method facilitates the reconstitution at the DNA level, providing greater flexibility and minimizing the production of truncated proteins compared to the protein trans-splicing method. However, directly comparing the reconstitution efficiency of the CRE-lox DNA recombination system with other dual AAV strategies is limited due to the lack of quantitative data from the study that examined the same target gene^[26].

At present, independent replication of StitchR’s results across larger animal models, different tissues, and additional disease contexts remains to be seen. Longitudinal studies of immunogenicity, biodistribution, and durability will be essential to inform regulatory pathways and future clinical trials.

CONCLUSION

The StitchR platform represents a compelling technological innovation, demonstrating for the first time that endogenous RNA repair mechanisms can be robustly harnessed for large gene reconstitution in vivo. The StitchR platform enabled full-length protein expression accompanied by substantial muscle regeneration and improved structural integrity in the mouse model. Taken together, these findings support the strong potential of StitchR as a gene therapy for muscular dystrophy. However, its current stage of development remains preclinical, while split-intein strategies validated by Tasfaout et al.^[2]. and advanced clinically by SpliceBio^[7] have already crossed into clinical trials. Thus, while StitchR holds the promise of achieving more natural full-length protein expression with fewer truncated byproducts, considerable validation, safety profiling, and manufacturing optimization are required before clinical deployment can be contemplated.

Continued head-to-head comparisons between StitchR and split-intein systems across rigorous translational models will ultimately determine whether this novel RNA ligation approach fulfills its therapeutic potential.