Gene therapy has been hampered by problems that have limited its clinical progress, including inefficient gene delivery and safety concerns. Designing better delivery vectors is therefore a paramount requirement for this technology to realize its potential. In a paper published in Nature Biotechnology, David Schaffer and colleagues describe a high-throughput method to aid the design of viral vectors with improved functional properties.

Recombinant adeno-associated viral vectors (rAAV), which are non-pathogenic and can establish persistent transgene expression in various cell types, are at the forefront of vector design efforts yet present a number of therapeutic challenges. Problems associated with rAAV delivery stem from the properties of the outer shell of the virus, the capsid, which is a major focus of engineering strategies. In past studies function-enhancing peptide sequences were inserted into specific sites in the capsid gene, providing improved cell-specific targeting. However, modifying more complex vector activities by this approach would require a greater understanding of the relationship between capsid structure and specific properties. This new study bypasses this issue by exploiting an approach that has been successful in protein engineering: directed evolution.

The authors generated more than 106 AAV2 variants with random mutations throughout the capsid gene. The resulting library was screened for variants with desirable functional properties followed by further rounds of mutagenesis and screening to enrich the selection. Proof of principle was demonstrated by applying the technique to two clinically relevant problems. One issue is that binding to heparin sulphate can limit rAAV2 dispersal. Affinity chromatography was used to identify variants with low heparin-binding affinity, additionally confirming the library's high functional diversity. Another major therapeutic limitation is that patients previously exposed to wild-type AAV2 carry antibodies that bind to and inactivate the vector before it can reach its target. Several variants were identified that could avoid antibody neutralization in vitro. These mutants were used to produce a vector with better delivery efficacy and less antibody neutralization than wild-type vector in vivo.

Although this approach cannot address issues such as the possible mutagenic effects of vector integration into a functional gene, it could apply to any properties governed by capsid structure and could help to overcome a variety of problems associated with gene delivery. A directed-evolution approach might lead to improvements in the specificity and efficiency of targeting particular cell types, which are influenced by interactions with cell surface receptors. Additionally, this strategy could help to reduce vector antigenicity. Furthermore, this approach can add to our understanding of the molecular basis of vector properties.