Computational Rational Design of Larger AAV Icosahedral Capsids

Adeno-associated virus (AAV) is the preferred viral vector platform in gene therapy. Yet its packaging capacity, about 4.7 kb (kilobases), limits its therapeutic potential and represents a major bottleneck in the field. The packaging capacity of AAV is constrained by its small capsid, which forms a 26-nm-diameter shell assembled from 60 capsid proteins in a T=1 icosahedral architecture. Here, we propose increasing the cargo capacity of AAV vectors by engineering the next possible icosahedral architecture, T=3 (180 capsid proteins), which is predicted to provide a fivefold increase in volume capacity. Oligomers of VP3, the main capsid protein of AAV, were folded using AI-based methods. This identified triangular trimers as the optimal multimer compatible with the tiles of icosahedral lattices in the geometrical theory of capsids. The VP3 trimers were assembled into a T=3 architecture and coarse-grained at 5[A] resolution. It was necessary to introduce 15 deletions (VP3{Delta}15) to accommodate the T=3 curvature. Molecular simulations under physiological conditions demonstrated the stability of the 45 nm-diameter T=3 capsid. Structural analysis measured a five- to sixfold increase in internal volume and estimated a potential upper cargo limit of 35 kb. The engineered VP3{Delta}15 could enable delivery of multicistronic constructs, larger regulatory elements, and CRISPR systems beyond the reach of current AAV vectors. Additionally, the introduced generalized protein design framework could be used to engineer capsids with larger T-numbers and to modify the capacity of other icosahedral delivery systems.