Optimizing cell-type specificity of AAV gene therapy vectors for the human brain

Introduction: Molecular-based therapies promise to improve outcomes for our most frustrating pathologies. To date, however, we have no method that can deliver such sequence-based therapies selectively to the cell types involved. Virtually all prior work on adeno-associated viruses (AAVs), the most common vector for neurological gene therapies, has studied capsid tropism in model organisms, primarily mice, despite abundant data that demonstrates a lack of translation across species. For this reason, we have developed an ex vivo organotypic system that uses surgically resected brain tissue to screen libraries of AAV variants.

Methods: We consent patients undergoing resective neurosurgical procedures for any tissue not needed for clinical use. After sectioning the tissue, transducing with virus libraries, and allowing time in culture, we use both imaging and single-nuclei sequencing to characterize each variant’s tropism.

Results: Resected human brain tissue is viable for at least two weeks. We find that transduction with 14 known AAV variants, including AAV2, the most common vector used in clinical trials, show quite low transduction of human neurons, all less than 5%. Engineered capsids DJ8, AAV2-retro, and PHP.s show the most reliable neuronal transduction at around 4%, with less than 1% of all cells transduced.
Using resected glioblastoma tissue, we find that an engineered AAV capsid, AAV2-retro, transduces ~40% of GFAP+ glioblastoma cells in astrocytic GBMs. Future work will iteratively optimize variants of this capsid to increase transduction efficiency as well as specificity, to permit delivery of any sequence-based therapy into tumor cells.

Conclusion : There is no prior work characterizing the cellular AAV tropism in human brain tissues, even those variants currently FDA-approved. This work shows the limited ability of animal models to predict the efficiency and specificity of AAV transduction, which relies on cell surface molecules that are known to be widely divergent between species and even strains of model organisms. These results provide the foundation for large-scale AAV gene therapy vector screening using human brain tissue, especially from diseases for which no durable therapy is currently available.