AUTOSTERE: Systematic Search for Scaffold Replacement Opportunities within Structural Databases

Medicinal chemists often bias toward working with scaffolds with which previously they have had direct experience and successes. In this way, it is often the case that scaffolds which have proven tractable within a research group are "reused" across multiple and sometimes unrelated drug targets. With this concept in mind, we designed a new computer algorithm AUTOSTERE which could systematically assess the opportunities to replace any part of any molecule within an entire database of known ligand structures with a target scaffold and automatically evaluate the potential designs in the context of the original ligand's protein environment. As such, it performs scaffold replacement on an unprecedented scale and suggests new target opportunities for preferred chemistries rather than the conventional reverse situation. The results of this approach for one scaffold, a substituted triazolinone, applied to a set of 10 426 ligand conformations extracted from the PDB are described. This led to the identification of ∼600 novel ligands incorporating the triazolinone scaffolds in complex with their predicted drug targets. From these, design examples are provided for HSP-90, cathepsin K, and TIE-2 kinase. A further study involved the searching for possible drug targets for unusual pyridopyrimidine cores. This process resulted in the identification of potential novel HIV reverse transcriptase inhibitors which were synthesized and shown to exhibit similar in vitro potencies to marketed compounds. Overall, the methodology described provides a powerful new approach to identify new target opportunities for scaffolds of provenance.