Optimization of stealth adeno-associated virus vectors by randomization of immunogenic epitopes

Abstract Therapeutic gene transfer by adeno-associated virus of serotype 2 (AAV-2) vectors is hampered in patients with pre-existing immunity. Molecular engineering was recently used to identify key immunogenic amino acid residues of the viral capsid and generate mutants with decreased antibody recognition. Here we explored the importance of finely tuning amino acid identity at immunogenic sites to optimize vector phenotype. A capsid library was generated by codon randomization at five positions where substitutions were shown to yield antibody evading phenotypes. Screening this library to isolate immune-escaping mutants allowed an exhaustive scan of combinations of the 20 natural amino acids at each position and yielded variants that remained infectious when incubated with serum or IVIG concentrations that completely neutralize AAV-2. Clones obtained replacing different residues at the same positions displayed strikingly different phenotypes, demonstrating that a precise choice of amino acid substitutions is fundamental to optimize immune-escaping, packaging ability, infectivity and tropism.