Promiscuous scaffolds: Friend or foe in fragment-based drug design?

Promiscuous fragments as part of frequent-hitters can certainly be seen as problematic in fragment-based design. In contrast, previous work indicated that promiscuous scaffolds can be part of ligands binding to unrelated targets, whereas the ligands can still show selectivity. In analogy to this work, we present a scaffold-based analysis of bioactivity data with the focus to identify promiscuous scaffolds or fragments on a broader scale. A scaffold network was created based on the ChEMBL database, followed by the analysis of the scaffold decoration using a python based workflow and target similarity calculations. In the end, 3124 promiscuous scaffolds belonging to ligands binding to different unrelated proteins were identified. Within these scaffolds, a phenylthiophene-like scaffold was identified in ligands binding to five unrelated protein targets that show similar hydrophobic subpockets. Interestingly, a subsequent docking study revealed different binding modes of the corresponding ligands, which indicates also a promiscuous subpocket. Subsequent screening by binding site comparisons of the protein ligand complexes available in the PDB revealed Cruzipain of T. cruzi as potential new target for this specific scaffold. The binding site similarity to Cruzipain was confirmed by biochemical validation of the phenylthiophene-like scaffold as well as grown derivatives. A subsequent analysis of the REAL Space showed that these promiscuous scaffolds are also part of the ultra-large chemical space of Enamine. Together with the available commercial derivatives, these promiscuous scaffolds can be seen as a promising sociable fragment library for fragment-based design by catalogue. In contrast, this library can also be excluded from fragment libraries depending on the viewpoint if such promiscuous scaffolds are foe or friends.