Engineered AAV Capsids: Muscle-Specific Transduction and Liver Detargeting

Adeno-associated virus (AAV) vectors are widely used due to their broad cell tropism and low immunogenicity. However, capsid limitations can hinder gene therapy efficacy in specific cell types or tissues. Capsid engineering optimizes gene delivery efficiency and specificity by enhancing transduction, modulating tropism, and improving immune evasion, which are critical factors in overcoming limitations in gene therapy research. Ongoing research in this area holds significant promise for treating diverse genetic disorders.

PackGene has developed the π-Icosa system, an AAV capsid engineering platform, to create and screen capsid variants with enhanced organ targeting, reduced off-target effects, or other customized features. This platform employs a three-phase process of AAV capsid library construction and in vivo screening to identify top-performing variants for gene and cell therapy applications. Further details on the π-Icosa system are available here. Leveraging this platform, we developed a series of chimeric variants that effectively minimize liver targeting while significantly enhancing muscle tissue specificity.

 

Identification of muscle-targeting capsids through AAV9 library screening

To enhance the specificity and transduction efficiency of AAV vectors for muscular tissues, we constructed an initial AAV9 capsid library by inserting a random 4-mer peptide adjacent to an RGD motif between residues N583 and A591 within hypervariable region VIII. Following a first-round in vivo screening in C57BL/6 mice and subsequent validation (Fig. 1A), we identified four enriched capsid variants—designated AAV.1R1, AAV.1R2, AAV.1R3, and AAV.1R4—that exhibited significantly enhanced luciferase expression in muscle tissues compared to wild-type AAV9 (Fig. 1B). However, these variants also displayed residual transduction in the liver, indicating incomplete de-targeting from hepatic tissue (Fig. 1B).

Figure 1. A. Diagram illustrating the AAV9 capsid-modified virus library and the screening cascade for optimized variants. B. Fluorescence images showing quadriceps and liver cross-sections from C57BL/6 mice after systemic injection with 2E+11 vg AAV.

Optimization of liver detargeting characteristics by capsid modification

To reduce hepatic tropism in AAV gene therapy while maintaining transduction in target tissues, we engineered capsids using structural insights from AAV2 (PDB: 6IH9) and AAV9 (PDB: 3UX1). We identified loop IV as a key protrusion on the three-fold symmetry axis, influencing receptor binding (Fig. 2A), and swapped AAV2 loops IV (R447-Q461) and V (K490-D494) with AAV9 loops IV (K449-K462) and V (T491-Q495), creating AAV.Zero1. Additional mutants, AAV.Zero2 (R585A) and AAV.Zero3 (R585-G586 deletion), were designed to disrupt HSPG-mediated liver uptake. We evaluated these capsids by injecting C57BL/6J mice with AAV9-, AAV2-, AAV.Zero1-, AAV.Zero2-, or AAV.Zero3-CAG-luciferase-P2A-EGFP (Fig. 2B). Three weeks post-injection, luciferase imaging and tissue analysis (qPCR, Western blotting) assessed transgene expression across liver, heart, muscle, and brain.

Figure 2. A. Ribbon diagram of AAV.Zero3 VP1 monomer; red indicates Loops IV and V；B. Schematic of virus administration and biodistribution assessment.

Whole-body bioluminescence imaging revealed that AAV.Zero1 and AAV.Zero3 exhibited significantly reduced luciferase activity in the liver compared to AAV9 (Fig. 3A), with luminescence intensities approaching levels observed in the control group. Quantification of liver luminescence confirmed a remarkable reduction for both AAV.Zero1 and AAV.Zero3 relative to AAV9. Consistent with these findings, quantitative PCR (qPCR) analysis of luciferase mRNA in liver tissue showed markedly lower expression in AAV.Zero1 and AAV.Zero3 groups compared to AAV9 (Fig. 3B). Western blotting further corroborated these results, demonstrating that luciferase protein levels in the livers of AAV.Zero1- and AAV.Zero3-injected mice were significantly diminished (Fig. 3C). Collectively, these data indicate that while AAV.Zero1 and AAV.Zero3 effectively minimize hepatic transduction, their overall transgene expression is reduced across all tested tissues, suggesting they may serve as promising backbones for further optimization to enhance specificity and efficiency in target tissues.

Figure 3. Chimeric capsid variants show marked reductions in liver transduction. A. In vivo bioluminescence data showing whole-body images and liver luciferase quantification. B. Quantitative analysis of fold change in Fluc mRNA levels in liver, quadriceps, abdomen, and heart of C57BL/6J mice. C. Western blot analysis of Luciferase and GAPDH protein expression in multiple tissues.

 

Chimeric capsid muscle specificity enhanced by RGD incorporation

To evaluate the potential of AAV.Zero3 capsid backbones for tissue-specific targeting, we inserted three previously characterized myotropic RGD-containing peptides—2A, 4E, and 4A (ref. 1)—at residue 585, creating new serotypes AAV.Zero3-2A, AAV.Zero3-4E, and AAV.eM (PG007), respectively. Adult C57BL/6J mice were intravenously injected with 2×10^11 vg/mouse of AAV-CAG-luciferase-P2A-EGFP vectors. Three weeks post-injection, we assessed in vivo luciferase activity and transgene expression across multiple tissues. Whole-organ bioluminescence imaging revealed that PG007 exhibited luciferase expression in skeletal muscle comparable to MyoAAV 4A (Fig. 4A, 4B), a finding supported by elevated luciferase mRNA levels (Fig. 4C), increased transgene protein expression via Western blotting (Fig. 4D), and enhanced EGFP fluorescence in quadriceps (Fig. 4E). These data indicate that RGD-peptide insertion significantly enhances AAV.Zero3 muscle transduction efficiency. Moreover, PG007 retained the non-hepatotropic profile of AAV.Zero3, as demonstrated by minimal transgene mRNA and protein expression in the liver (Fig. 4E).

Figure 4. RGD-containing peptides confer muscle-specific targeting and high transduction efficiency to chimeric capsids in C57BL/6J mice. A. Whole-body in vivo bioluminescence images. B. Firefly luciferase luminescence quantification in mouse hindlimbs. C. Fluc mRNA quantification in targeted muscles (quadriceps, abdomen, heart). D. Western blot images of Luciferase and GAPDH in quadriceps and liver. E. Fluorescence cross-sectional images of quadriceps and liver.

 

Conclusion

The π-Icosa capsid engineering platform is designed to generate and screen AAV capsid variants with enhanced properties, including improved tissue specificity, increased infectivity, and reduced off-target effects, for gene therapy applications. This platform addresses key limitations of natural AAV serotypes, which often lack the precision and efficiency required for clinical success across diverse diseases.

In this study, we utilized the π-Icosa platform to screen and identify a novel capsid that exhibits reduced liver targeting and increased muscle tropism. The π-Icosa platform integrates rational design, directed evolution, and high-throughput screening, incorporating computational approaches such as bioinformatic prediction and machine learning (ML) to refine AAV capsid engineering. This enables the creation of AAV capsid variants for potent and safe gene delivery to any target tissue or cell type.

The platform is built upon three core features for a robust and versatile AAV capsid engineering pipeline:

• • Comprehensive Evaluation: Assessing capsid variants for both target transduction efficiency and reduced off-target effects.
  • In Vivo Mammalian Validation: Identifying vector systems directly translatable from preclinical animal models to human applications.
  • Versatile Library Compatibility: Supporting diverse capsid library screening and validation, and fully compatible with mechanism-guided directed evolution.

 

We will provide further data characterizing this specific π-Icosa platform-generated capsid in our next publication, including cross-species expression profiles, enhanced gene delivery efficacy, and functional restoration within a DMD mouse model.

 

 

References

1. Tabebordbar, M., Lagerborg, K.A., Stanton, A., King, E.M., Ye, S., Tellez, L., Krunnfusz, A., Tavakoli, S., Widrick, J.J., Messemer, K.A., Troiano, E.C., et al. (2021). Directed evolution of a family of AAV capsid variants enabling potent muscle-directed gene delivery across species. Cell 184, 4919-4938 e4922. 10.1016/j.cell.2021.08.028