Changfan Lin, Xinhong Chen, Jonathan D. Hoang, Fiona Ristic, Yujie Fan, Seongmin Jang, Jin Hyung Alex Chung, Erin E. Sullivan, Tomasz Gawda, Bill Kavvathas, Irene Tran, Yitong Li, Andrew D. Steele, Timothy F. Shay, Viviana Gradinaru
Gene therapies using natural adeno-associated virus (AAV) serotypes have restricted applications, particularly in the brain, due to their poor targeting and resulting safety concerns. Directed evolution has identified brain-enhanced engineered AAV capsids for various model organisms, but inter-species differences challenge their translation. Here, we engineer AAVs to target human carbonic anhydrase IV (CA-IV), a recently identified blood-brain barrier transcytosis receptor with several favorable properties. We perform in vitro on-column selection of an AAV library to exclude capsids that do not target human CA-IV. We subsequently screen the top 0.01% of variants in vivo in mice expressing human CA-IV in brain endothelial cells. Notably, two top-performing capsids in "humanized" mice are relatively lower ranked in vitro and outperform several top-ranked in vitro candidates. One of these, AAV-hCA4-IV77, achieves 100-fold-greater brain transduction than AAV9, with robust neuronal and astrocytic coverage across brain regions. These results advance our understanding of receptor-targeted capsid design and support the therapeutic potential of human CA-IV-engaging AAVs.