Gene therapy is a promising therapeutic avenue for thousands of human diseases. Adeno-associated virus (AAV) is a benign virus that has shown tremendous promise as a DNA delivery vehicle for gene therapy. Although this virus has already been co-opted for human use in approved therapeutics, our ability to tune the vector’s cell targeting abilities and enhance gene delivery is in its infancy. AAV’s multifunctional protein capsid protects the therapeutic DNA cargo on its journey to targeted cells, facilitates cell entry through receptor interactions, and ferries DNA into the cell nucleus before coming apart to permit gene expression. Computational approaches may allow us to identify which portions of the capsid are crucial for viral formation and gene therapy delivery, informing targeted engineering to design the next generation of viral vectors.
The cover image of the February 2 issue of Biophysical Journal is a visualization of the AAV capsid exterior structure (PDB: 1LP3). The structure is colored according to a computational prediction of the change in local frustration of each protein residue as protein subunits assemble into the viral capsid shell. This prediction was made using the AWSEM-MD Frustratometer, an algorithm based on energy landscape theory that quantifies energy distribution in proteins and protein assemblies and predicts how this distribution shifts upon conformational change. Blue regions of the capsid are predicted to favor the assembled state, while yellow regions of the capsid are predicted to favor the non-assembled state. We hypothesized that these regions may play crucial roles in viral assembly, stability, and genetic cargo release.
Our work applying the Frustratometer and an evolution-inspired computational technique called Direct Coupling Analysis to AAV suggests that these algorithms may have complementary strengths in capturing features of the protein energy landscape to make predictions about viral function. With this study and image, we hope to encourage the development of reliable computational tools that integrate predictive models and the growing body of data from high-throughput AAV screening experiments to enable rapid development of improved gene therapy vectors.
- Nicole Thadani, Qin Zhou, Kiara Reyes Gamas, Susan Butler, Carlos Bueno, Nicholas Schafer, Faruck Morcos, Peter G. Wolynes, Junghae Suh