Evaluation of a Human Neurovascular Model to Complement a Parallel Non-human Primate Selection for Blood–Brain Barrier Penetrant AAV Capsids


  • Delivery of genomic medicine to the central nervous system (CNS) is a major hurdle for clinical applications of gene therapy; the blood–brain barrier (BBB) limits the brain distribution of virtually all intravenously administered macromolecules.
  • Several adeno-associated virus (AAV) serotypes, most notably AAV9, distribute to the brain after intravenous (IV) administration but require high doses to achieve limited expression.
  • AAV capsid engineering has produced novel variants that are superior to their parental serotypes and have progressed into the clinic for several indications. However, translation of clinical programs from preclinical models to humans remains a challenge for the entire gene therapy field, including capsid engineering efforts.
    • Two factors for a stringent selection campaign have emerged: library designs that incorporate functional cellular transduction pressure, and selection of appropriate in vitro and/or in vivo models.
  • In this study, we employed SIFTER™ (Selecting In vivo For Transduction and Expression of RNA) to engineer capsids with improved CNS transduction following IV administration in Cynomolgus macaque (non-human primates [NHPs]). This was followed by implementation of an all human cell model of the BBB that recapitulates many key BBB properties to address discordant capsid performance observed in vitro vs in vivo and between species.