Novel computational pipeline to identify target sites for broad spectrum antiviral drugs

Emerging viruses pose an ongoing threat to human health. While certain viral families are common sources of outbreaks, predicting the specific virus within a family that will cause the next outbreak or pandemic is not possible, creating an urgent need for broad spectrum antiviral drugs that are effective against a wide array of related viral pathogens. However, broad spectrum drug development is hindered by the lack of detailed knowledge of compound binding sites that are structurally and functionally conserved between viral family members and are essential for virus replication. To overcome this limitation, we developed an in silico approach that combines AI-driven protein structure prediction, computational fragment soaking, multiple sequence alignment, and protein stability calculations to identify highly conserved target sites that are both solvent-accessible and conserved. We applied this approach to the Togaviridae family, which includes emerging pandemic disease threats such as chikungunya and Venezuelan equine encephalitis virus for which there are currently no approved antiviral therapies. Our analysis identified multiple solvent accessible and structurally conserved pockets in the alphavirus non-structural protein 2 (nsP2) protease domain, which is essential for processing the viral replicase proteins. Mutagenesis of key solvent accessible and conserved residues identified novel pockets that are essential for the replication of multiple alphaviruses, validating these pockets as potential antiviral target sites for nsP2 inhibitors. These findings highlight the potential of artificial intelligence-informed modeling for revealing functionally conserved, accessible pockets as a means of identifying potential target binding sites for broadly active direct acting antivirals.