Animal models play a critical role in screening AAVs designed to target specific tissues. Recent studies have highlighted a key finding: AAVs can perform differently across species and animal strains, underscoring the importance of cross-species testing [Tabebordba et al., 2021; Gonzalez et al., 2022]. For instance, AAV-PHP.B displayed the ability to cross the blood-brain barrier (BBB) in specific mouse strains but not in non-human primates (NHPs) [Hordeaux et al., 2018; Matsuzaki et al., 2019]. To identify variants that are more likely to be effective in humans, it is essential to conduct variant screening in a range of animal models and human-based systems, ensuring the selection of AAVs with translatable performance.

The directed evolution of a random peptide insertion library of AAV9 (VR-VIII) led to the development of the AAV-PHP family, with AAV-PHP.B showing significantly enhanced BBB-crossing capabilities in mice. Subsequent engineering of AAV-PHP.B resulted in the creation of AAV-PHP.eB, which retains the same peptide insertion but incorporates additional flanking substitutions. AAV-PHP.eB demonstrated efficient central nervous system (CNS) transduction in mice, achieving 55–76% neuron transduction depending on the region—more than 2.5 times that of AAV-PHP.B. Further modifications led to liver de-targeted variants of AAV-PHP.eB, including AAV.CAP-B10 and AAV.CAP-B22. AAV.CAP-B10 maintained CNS targeting in mice while exhibiting reduced transduction of peripheral organs and a 50-fold decrease in liver tropism compared to AAV-PHP.eB, and over 100-fold compared to AAV9. This variant showed specific neuronal targeting within the CNS, with fewer transduced astrocytes and oligodendrocytes compared to AAV-PHP.eB. Notably, AAV.CAP-B10 achieved broad and robust transgene expression in the CNS of adult marmosets, with a 4-fold increase in CNS transduction over AAV9 and a 17-fold reduction in liver expression compared to AAV9. However, when administered as a pool in infant rhesus macaques, AAV.CAP-B10 produced only slightly higher CNS enrichment compared to AAV9. AAV.CAP-B22, another variant, demonstrated even higher CNS transduction in marmosets, achieving a 12-fold increase compared to AAV9. However, it also showed greater astrocyte transduction than both AAV9 and AAV.CAP-B10. While AAV.CAP-B22 showed similar liver tropism to AAV9, it did not effectively translate in newborn rhesus macaques.