Novel pharmacological maps of protein lysine methyltransferases: key for target deorphanization

Abstract

Epigenetic therapies are being investigated for the treatment of cancer, cognitive disorders, metabolic alterations and autoinmune diseases. Among the diferent epigenetic target families, protein lysine methyltransferases (PKMTs), are especially interesting because it is believed that their inhibition may be highly specifc at the functional level. Despite its relevance, there are currently known inhibitors against only 10 out of the 50 SET-domain containing members of the PKMT family. Accordingly, the identifcation of chemical probes for the validation of the therapeutic impact of epigenetic modulation is key. Moreover, little is known about the mechanisms that dictate their substrate specifcity and ligand selectivity. Consequently, it is desirable to explore novel methods to characterize the pharmacological similarity of PKMTs, going beyond classical phylogenetic relationships. Such characterization would enable the prediction of ligand of-target efects caused by lack of ligand selectivity and the repurposing of known compounds against alternative targets. This is particularly relevant in the case of orphan targets with unreported inhibitors. Here, we frst perform a systematic study of binding modes of cofactor and substrate bound ligands with all available SET domain-containing PKMTs. Protein ligand interaction fngerprints were applied to identify conserved hot spots and contact-specifc residues across subfamilies at each binding site; a relevant analysis for guiding the design of novel, selective compounds. Then, a recently described methodology (GPCR-CoINPocket) that incorporates ligand contact information into classical alignment-based comparisons was applied to the entire family of 50 SET-containing proteins to devise pharmacological similarities between them. The main advantage of this approach is that it is not restricted to proteins for which crystallographic data with bound ligands is available. The resulting family organization from the separate analysis of both sites (cofactor and substrate) was retrospectively and prospectively validated. Of note, three hits (inhibition>50% at 10 µM) were identifed for the orphan NSD1.