Adeno-associated virus (AAV) has undergone a remarkable transformation—from an obscure contaminant discovered in simian adenovirus cultures in the 1960s to a linchpin of clinical gene therapy. The recent review by Suarez-Amaran et al. (2025) offers a sweeping retrospective and future-forward vision of AAV vector evolution, integrating deep insights from virology, molecular engineering, and systems biology. As the therapeutic frontier expands from rare monogenic conditions to broader indications like neurodegeneration, muscle disease, and oncology, AAV engineering is poised for its most pivotal redesign yet.

 

From Contaminant to Platform

The cloning of the AAV2 genome in the early 1980s laid the molecular foundation for modern recombinant AAV (rAAV) systems. Replacing the REP and CAP open reading frames with a therapeutic cassette flanked by inverted terminal repeats (ITRs), rAAV vectors gained traction as a safe and persistent transduction platform, capable of delivering functional transgenes to a wide array of tissues.

Notably, FDA-approved therapies such as Luxturna (voretigene neparvovec) for inherited retinal dystrophy and Zolgensma (onasemnogene abeparvovec) for spinal muscular atrophy validated rAAV’s clinical potential while highlighting the need for better targeting, immune evasion, and scalable production.

 

The Capsid as a Regulatory Hub

Beyond its role in packaging and delivery, the AAV capsid is increasingly recognized as a regulator of transgene expression. Structural subunits VP1, VP2, and VP3 do more than build the icosahedral shell—they contribute critical functions like:

• Endosomal escape via VP1's phospholipase A2 domain.

• Nuclear entry via basic region motifs (BR1–BR5).

• Promoter-transgene epigenetic modulation, as seen in the uVP1/VP2 regions.

Recent findings, such as those by Powell et al. (2021), demonstrate that the capsid itself can influence chromatin accessibility and transcriptional activation of the vector genome—reshaping our understanding of how to engineer capsid-transgene synergy for sustained therapeutic expression.

 

Engineering the Next Generation: Libraries, Logic, and AI

Directed Evolution: Peptides Meet Selection

Platforms such as CREATE, DELIVER, and TRACER use saturation mutagenesis (e.g., of VR-VIII loops), in vivo barcoding, and transcript-guided selection to evolve novel capsids. These strategies have yielded capsids like:

• AAV-PHP.B – enhanced CNS tropism in rodents.

• AAVMYO – efficient systemic muscle delivery across species.

• AAV-HN1 – superior brain transduction in non-human primates.

These libraries leverage viral biodiversity, synthetic peptide diversity, and selection pressure to find high-performing variants—sometimes with >100× improvements in tropism or transduction efficiency.

Rational Design: Structure and Specificity

By targeting known surface loops, receptor-binding domains, and PTM hotspots (e.g., tyrosine residues), rational design enables capsid reprogramming for tissue selectivity and immune escape. Successful examples include:

• AAV2i8 – a rational chimera with redirected tropism to muscle and reduced neutralization.

• AAV2.5 – a vector combining AAV1 tropism with AAV2 structure for Duchenne muscular dystrophy.

• Anc80L65 – an ancestral reconstruction demonstrating potent multi-tissue gene transfer.

Importantly, these approaches often build on structural data from cryo-EM and X-ray crystallography to predict mutational tolerance and functional gains.

 

PTMs and the Protein “Ecosystem” of AAV

Capsid function is increasingly understood as a product not just of amino acid sequence but of contextual post-translational modifications (PTMs)—including phosphorylation, ubiquitination, and glycosylation. These modifications:

• Occur during vector production, purification, and even post-administration.

• Alter capsid stability, receptor binding, intracellular trafficking, and immunogenicity.

• Are influenced by production system (e.g., HEK293 vs. insect cells), storage buffers, and administration route.

These findings challenge the industry’s traditional focus on “purity” alone and support a move toward functional vector characterization, including protein corona profiling and electrostatic modeling.

 

Manufacturing Meets Synthetic Biology

Efforts to reduce cost and improve consistency are driving development of VP3-only virus-like particles via E. coli and cell-free protein synthesis.​ While these systems are scalable and fast, they often lack functional domains from VP1/VP2, reducing infectivity.

The future lies in hybrid production strategies that retain function while leveraging synthetic scalability—necessitating careful CMC harmonization and new release criteria from regulators.

 

Regulatory and Translational Considerations​​

As vectors move into larger patient populations, pre-existing immunity, biodistribution variability, and potency consistency take center stage. Regulatory authorities like the FDA and EMA increasingly emphasize:

• Capsid characterization beyond sequence, including PTM mapping and host protein interactions.

• Assays for potency and vector performance, not just genome copies.

• CMC scalability, especially for commercial readiness.

Platforms such as SEBIR, which integrates transcriptional readout with capsid-genome interactions, and AI-driven modeling (e.g., APPRAISE, SCHEMA) will be vital to meeting these expectations.

 

Conclusion: Toward AAV 3.0

Suarez-Amaran et al. (2025) argue that AAV biology has entered a new phase—one that fuses synthetic design, directed selection, computational modeling, and systems-level understanding. Future vectors will no longer be defined by serotype but by function: tropism, stability, expression, and manufacturability, engineered for each patient and disease.

This new paradigm—AAV 3.0—will depend on deep integration across disciplines: structural biology, epigenetics, machine learning, and translational medicine. The question is no longer whether AAV will remain central to gene therapy, but rather: how far can it go?

 

 

 

Key References & Further Reading

• Suarez-Amaran L, Song L, Tretiakova A, Mikhail SA, Samulski RJ. (2025). AAV vector development, back to the future. Molecular Therapy, 33(5), 1903–1914. https://doi.org/10.1016/j.ymthe.2025.03.064

• Powell SK, et al. (2021). Capsid-promoter interactions regulate AAV gene expression. Nature Biotechnology, 39(5), 670–681. https://www.nature.com/articles/s41587-021-00871-6

• Gradinaru V, et al. (2016). CREATE platform for CNS AAV evolution. Cell, 167(2), 382–396. https://doi.org/10.1016/j.cell.2016.08.024

• Tabebordbar M, et al. (2021). DELIVER platform for muscle-specific AAVs. Nature Biotechnology, 39(8), 1005–1015. https://doi.org/10.1038/s41587-021-00959-z

• Kelsic ED, et al. (2021). Deep learning enables design of AAV capsid libraries. Cell Systems, 12(6), 704–717. https://doi.org/10.1016/j.cels.2021.03.003

• Church GM, et al. (2019). Fitness landscape analysis of AAV. Nature Biotechnology, 37(6), 667–678. https://doi.org/10.1038/s41587-019-0096-4

• Marques JP, et al. (2020). Machine learning predicts AAV assembly. Molecular Therapy - Methods & Clinical Development, 17, 349–359. https://doi.org/10.1016/j.omtm.2019.12.012

• Xie Q, et al. (2002). The atomic structure of adeno-associated virus. Journal of Virology, 76(16), 7910–7919. https://doi.org/10.1128/jvi.76.16.7910-7919.2002

• Samulski RJ, et al. (1982). Cloning of AAV genome into plasmid vector. PNAS, 79(6), 2077–2081. https://doi.org/10.1073/pnas.79.6.2077

• Westhaus A, et al. (2022). AAV-p40 and HTE platforms for capsid screening. Gene Therapy, 29, 56–67. https://doi.org/10.1038/s41434-021-00239-4

• Vandenberghe LH, et al. (2015). Ancestral AAVs for gene delivery. Nature Biotechnology, 33(9), 941–946. https://doi.org/10.1038/nbt.3567

• Zolotukhin S, et al. (2002). Production and purification of AAV vectors. Gene Therapy, 9(14), 958–963. https://doi.org/10.1038/sj.gt.3301753

• O'Donnell J, et al. (2009). Electrostatic prediction of AAV2 heparan binding. J Virol, 83(18), 8966–8978. https://doi.org/10.1128/JVI.00485-09

• Srivastava A, et al. (2008). PTMs enhance AAV transduction. Gene Therapy, 15(12), 902–909. https://doi.org/10.1038/gt.2008.42