Using sitemap to aid in the identification of cryptic binding pockets.

Abstract

Identifying binding sites on protein surfaces is pivotal to structure-based drug design. A wide variety of conformational changes on proteins may happen upon ligand binding due to receptor flexibility. Consequently, certain proteins have binding sites that are not formed in the unbound state and such binding sites are considered as cryptic binding pockets. The identification and characterization of cryptic binding pockets is an attractive approach in drug discovery because it allows for the targeting of proteins which were previously considered as undruggable. Due to the cryptic nature of such binding pockets, they are often challenging to identify using rational drug discovery approaches based on structural experiments. Computational modeling approaches such as molecular dynamics (MD) simulations can provide atomic level details of binding pocket opening and possible binding pathways but the timescale that MD can achieve at a reasonable cost often limits its application for this purpose. Enhanced sampling techniques can improve the efficiency of MD simulations for sampling of interested degrees of freedom, but a prior knowledge of the studied system is required. In this work, we explored the ability of SiteMap, a widely used commercial package for binding site identification, to identify cryptic binding pockets and elucidate features to help distinguish true sites. The information gained from this analysis enables the use of enhanced MD sampling techniques to confirm cryptic binding pockets predicted by SiteMap. We used a combination of the PocketMiner dataset, a novel set of cryptic pockets, and the Cryptosite Set, a classic set of cryptic pockets. In total, we examined SiteMap results on 134 known cryptic pockets. Our findings demonstrate the promise of using SiteMap for efficiently screening the protein and guiding further MD simulations to efficiently identify cryptic binding pockets.