ABSTRACT
To overcome limitations in the generalizability and efficiency of current AAV vectors, in this current study, we constructed an AAV variant library by the insertion of random heptapeptide sequences in the receptor-binding domain of the AAV9 capsid gene. We then applied a recently developed organ-on-a-chip in vitro model of the human blood-brain barrier (BBB) to iteratively enrich for variants that efficiently cross the BBB and transduce astrocyte cells. Through multiple rounds of screening, we obtained two candidate AAV variants, AAV-M6 and AAV-M8, which showed significantly higher BBB penetration efficiency than AAV9 or AAV-PHP.eB. Quantitative PCR (qPCR) assay showed that AAV-M6 could accumulate to a 5 times higher titer, while AAV-M8 reached a 3 times higher titer, than AAV-PHP.eB in the neural chamber of the model. The transduction assay further verified that the AAV-M6 candidate vector was able to infect HA-1800 cells after crossing the BBB, suggesting it could potentially transduce brain parenchymal cells after crossing the hCMEC/D3 layer at higher efficiency than AAV-PHP.eB. Molecular simulations suggested that the human receptor proteins, LY6D and M6PR, could bind the AAV-M6 heptapeptide insertion with high affinity. This study provides two promising candidate AAV vectors and demonstrates the use of this in vitro BBB model for scalable, high-throughput screening of gene therapies. These tools can drive investigations of the mechanisms underlying BBB permeability and the cell-type specificity of virus vectors.
