The continuing epidemic of overweight and obesity faces a lack of appropriate pharmaceutical treatment options to support health care systems. Although studied for decades, currently available drugs only show low efficacy but serious or at least unpleasant side effects. Multi-target drugs hold promise to overcome the limitations of traditional pharmaceutics by modulating several nodes of the disease-relevant biological network. In this thesis, the multi-target concept was applied to macromolecular targets involved in obesity. In order to identify target pairs useful as starting points for multi-target drug design, we applied a systematic data mining approach employing publicly available bioactivity data of small molecules binding to targets involved in obesity. The target pair with the highest molecular similarity among known active ligands was found to comprise of histamine H3 receptor (H3R) and melanin-concentrating hormone receptor 1 (MCHR1). Both proteins are part of the G-protein coupled receptor (GPCR) family and were extensively studied as potential obesity targets. Although antagonizing either receptor was efficient in rodent models of obesity, drug candidates failed to proof efficacy in clinical studies. To test the potential of H3R and MCHR1 in multi-target drug development, a shape-based virtual screening campaign was conducted resulting in the selection of three small molecules. A subsequent in vitro evaluation revealed nanomolar affinity for all three molecules at both receptors. Lead optimization against multiple targets can dramatically benefit from integration of structural data. Since H3R and MCHR1 lack experimental structural data for structure-based drug design, two novel methods were developed that support drug design campaigns based on homology models. H3R is part of the aminergic family of GPCRs, which share a conserved charged interaction between ligand and protein. This crucial interaction was incorporated into a ligand-guided homology modeling campaign revealing valuable insights into side chain conformations critical for appropriate ligand placement in H3R. A subsequent virtual screening campaign followed by in vitro validation revealed two novel ligands with nanomolar affinity at H3R. MCHR1 is less well characterized and was found to contain several highly flexible residues in the ligand binding pocket, which hindered the translation of the ligand-guided homology modeling strategy to MCHR1. To include the high flexibility of binding site residues, the protein environment of water molecules in molecular dynamics simulations was analyzed to derive 3D pharmacophores for virtual screening. Generated 3D pharmacophores were highly successful in a retrospective virtual screening campaign in discriminating active MCHR1 ligands from decoys. This method was translated into a Python package (PyRod), where the source code was released publicly. The results and methods developed in this thesis provide valuable tools to support the development of more efficient and safe anti-obesity medications. We show that the simultaneous inhibition of H3R and MCHR1 with a single high affinity binder is possible. We developed novel computational methods to support structure-based virtual screening campaigns against both receptors.
