Identifying representative kinases for inhibitor evaluation via systematic analysis of compound-based target relationships

Abstract

We have investigated the question if kinases could be identified whose compound binding characteristics represent those of many other kinases. Therefore, compound-based relationships between kinases were systematically explored on the basis of a large curated collection of inhibitors with multi-kinase activity. With the aid of network analysis, individual kinases were identified that shared at least 50 inhibitors with more than 100 other kinases. Combinations of three to four kinases with many compound-based relationships were found to represent ∼150 kinases distributed across the human kinome. These findings have implications for the practice of medicinal chemistry because small numbers of kinases can be prioritized for compound testing that are expected to account for binding characteristics of many others. Hence, choosing three or four representative kinases may be sufficient for a first-path control evaluation of new candidate compounds to assess their potential for promiscuity in kinase inhibition.