Identification of Distant Drug Off-Targets by Direct Superposition of Binding Pocket Surfaces

Abstract

Correctly predicting off-targets for a given molecular structure, which would have the ability to bind a large range of ligands, is both particularly difficult and important if they share no significant sequence or fold similarity with the respective molecular target (“distant off-targets”). A novel approach for identification of off-targets by direct superposition of protein binding pocket surfaces is presented and applied to a set of well-studied and highly relevant drug targets, including representative kinases and nuclear hormone receptors. The entire Protein Data Bank is searched for similar binding pockets and convincing distant off-target candidates were identified that share no significant sequence or fold similarity with the respective target structure. These putative target off-target pairs are further supported by the existence of compounds that bind strongly to both with high topological similarity, and in some cases, literature examples of individual compounds that bind to both. Also, our results clearly show that it is possible for binding pockets to exhibit a striking surface similarity, while the respective off-target shares neither significant sequence nor significant fold similarity with the respective molecular target (“distant off-target”).

A Novel Antiviral Target Structure Involved in the RNA Binding, Dimerization, and Nuclear Export Functions of the Influenza A Virus Nucleoprotein

“…Predicting possible side effects is an important issue when designing new drugs. The potential utility of a molecule as a targetable drug can be examined by comparing the structure of its target binding pocket with the structures of host proteins [ 49 ]. Molecular targets that show high similarity to host molecules may cause severe medical problems; such molecules should not be pursued.…”

CAVIAR: a method for automatic cavity detection, description and decomposition into subcavities

“…The cavity detection field has been prolific in the last three decades (Simões et al, 2017a;Volkamer et al, 2018;Macari et al, 2019). Successful applications include the prediction of target ligandability (Volkamer et al, 2012;Le Guilloux et al, 2009;Halgren, 2009;Nayal and Honig, 2006;Desaphy et al, 2012), identification of offtargets (Ehrt et al, 2016;Xie et al, 2011;Möller-Acuña et al, 2015;Schumann and Armen, 2013;Schirris et al, 2015), functional annotation (Kuhn et al, 2006;Kinoshita et al, 2002;Konc et al, 2013;Anand et al, 2011) and ligand design and drug repurposing (Al-Gharabli et al, 2006;Willmann et al, 2012;Kooistra et al, 2015;Weber et al, 2004). Structure-based cavity detection methods can be grouped into two general families: energy-based algorithms and geometry-based (Simões et al, 2017a;Weisel et al, 2007;Volkamer, Griewel, et al, 2010).…”

Automatic Cavity Identification and Decomposition into Subpockets with CAVIAR

“…Binding pockets can be characterized empirically by analyzing holo structures of the target in complex with a ligand, but the analysis of the entire PDB, including structures without ligand, requires automatic algorithms to perform that task. The cavity detection field has been prolific in the last three decades, [2][3][4] with some successful applications for the prediction of target ligandability, [5][6][7][8][9] identification of off-targets, [10][11][12][13][14] functional annotation, [15][16][17][18] ligand design and drug repurposing. [19][20][21][22] Structure-based cavity detection methods can be grouped into two general families: energy-based algorithms and geometry-based.…”