Recently determined structures of class C G protein-coupled receptors (GPCRs) revealed the location of allosteric binding sites and opened new opportunities for the discovery of novel modulators. In this work, molecular docking screens for allosteric modulators targeting the metabotropic glutamate receptor 5 (mGlu₅) were performed. The mGlu₅ receptor is activated by the main excitatory neurotransmitter of the nervous central system, L-glutamate, and mGlu₅ receptor activity can be allosterically modulated by negative or positive allosteric modulators. The mGlu₅ receptor is a promising target for the treatment of psychiatric and neurodegenerative diseases, and several allosteric modulators of this GPCR have been evaluated in clinical trials. Chemical libraries containing fragment-(1.6 million molecules) and lead-like (4.6 million molecules) compounds were docked to an allosteric binding site of mGlu₅ identified in X-ray crystal structures. Among the top-ranked compounds, 59 fragments and 59 lead-like compounds were selected for experimental evaluation. Of these, four fragment-and seven lead-like compounds were confirmed to bind to the allosteric site with affinities ranging from 0.43 to 8.6 μM, corresponding to a hit rate of 9%. The four compounds with the highest affinities were demonstrated to be negative allosteric modulators of mGlu₅ signaling in functional assays. The results demonstrate that virtual screens of fragment and lead-like chemical libraries have complementary advantages and illustrate how access to high-resolution structures of GPCRs in complex with allosteric modulators can accelerate lead discovery.

Molecular modeling is increasingly being integrated into the drug discovery process. Although computational methods are already an essential part of the process today, these techniques have vast potential to evolve and refine drug design in the future. The integration of modeling is further catalyzed by the rapidly growing computational power available to both academia and pharmaceutical companies. These resources extend the reach of the computational methods to new time scales in physics-based simulations of biomolecular systems and increase the number of molecules that is possible to dock to a receptor by several orders of magnitude. However, there is also a need for further development of methods that utilize the increased computational power to increase the level of detail in modeling of molecular interactions. The work presented in this thesis has contributed to method development in two different areas and demonstrated how virtual screening can be applied to identify ligands of proteins. The first main project investigated modeling of ordered water molecules in protein binding sites to understand the role of water in molecular recognition and the impact of ligands on hydration networks. In a second project, an approach for discovery and optimization of fragment-sized ligands based on virtual screening was developed and used to identify inhibitors of a cancer drug target. Finally, molecular docking was applied to assess proposed substrates of cytochrome b₅₆₁, a recently crystallized membrane protein.

Water plays a major role in ligand binding and is attracting increasing attention in structure-based drug design. Water molecules can make large contributions to binding affinity by bridging protein-ligand interactions or by being displaced upon complex formation, but these phenomena are challenging to model at the molecular level. Herein, networks of ordered water molecules in protein binding sites were analyzed by clustering of molecular dynamics (MD) simulation trajectories. Locations of ordered waters (hydration sites) were first identified from simulations of high resolution crystal structures of 13 protein-ligand complexes. The MD-derived hydration sites reproduced 73% of the binding site water molecules observed in the crystal structures. If the simulations were repeated without the cocrystallized ligands, a majority (58%) of the crystal waters in the binding sites were still predicted. In addition, comparison of the hydration sites obtained from simulations carried out in the absence of ligands to those identified for the complexes revealed that the networks of ordered water molecules were preserved to a large extent, suggesting that the locations of waters in a protein-ligand interface are mainly dictated by the protein. Analysis of >1000 crystal structures showed that hydration sites bridged protein-ligand interactions in complexes with different ligands, and those with high MD-derived occupancies were more likely to correspond to experimentally observed ordered water molecules. The results demonstrate that ordered water molecules relevant for modeling of protein-ligand complexes can be identified from MD simulations. Our findings could contribute to development of improved methods for structure-based virtual screening and lead optimization.

Superoxide is a reactive oxygen species produced during aerobic metabolism in mitochondria and prokaryotes. It causes damage to lipids, proteins and DNA and is implicated in cancer, cardiovascular disease, neurodegenerative disorders and aging. As protection, cells express soluble superoxide dismutases, disproportionating superoxide to oxygen and hydrogen peroxide. Here, we describe a membrane-bound enzyme that directly oxidizes superoxide and funnels the sequestered electrons to ubiquinone in a diffusion-limited reaction. Experiments in proteoliposomes and inverted membranes show that the protein is capable of efficiently quenching superoxide generated at the membrane in vitro. The 2.0 Å crystal structure shows an integral membrane di-heme cytochrome b poised for electron transfer from the P-side and proton uptake from the N-side. This suggests that the reaction is electrogenic and contributes to the membrane potential while also conserving energy by reducing the quinone pool. Based on this enzymatic activity, we propose that the enzyme family be denoted superoxide oxidase (SOO).

Fragment-based lead discovery is becoming an increasingly popular strategy for drug discovery. Fragment screening identifies weakly binding compounds that require optimization to become high-affinity leads. As design of leads from fragments is challenging, reliable computational methods to guide optimization would be invaluable. We evaluated using molecular dynamics simulations and the free energy perturbation method (MD/FEP) in fragment optimization for the A(2A) adenosine receptor, a pharmaceutically relevant G protein-coupled receptor. Optimization of fragments exploring two binding site subpockets was probed by calculating relative binding affinities for 23 adenine derivatives, resulting in strong agreement with experimental data (R-2 = 0.78). The predictive power of MD/FEP was significantly better than that of an empirical scoring function. We also demonstrated the potential of the MD/FEP to assess multiple binding modes and to tailor the thermodynamic profile of ligands during optimization. Finally, MD/FEP was applied prospectively to optimize three nonpurine fragments, and predictions for 12 compounds were evaluated experimentally. The direction of the change in binding affinity was correctly predicted in a majority of the cases, and agreement with experiment could be improved with rigorous parameter derivation. The results suggest that MD/FEP will become a powerful tool in structure-driven optimization of fragments to lead candidates.

Fragment-based lead discovery has emerged as a leading drug development strategy for novel therapeutic targets. Although fragment-based drug discovery benefits immensely from access to atomic-resolution information, structure-based virtual screening has rarely been used to drive fragment discovery and optimization. Here, molecular docking of 0.3 million fragments to a crystal structure of cancer target MTH1 was performed. Twenty-two predicted fragment ligands, for which analogs could be acquired commercially, were experimentally evaluated. Five fragments inhibited MTH1 with IC50 values ranging from 6 to 79 mu M. Structure-based optimization guided by predicted binding modes and analogs from commercial chemical libraries yielded nanomolar inhibitors. Subsequently solved crystal structures confirmed binding modes predicted by docking for three scaffolds. Structure-guided exploration of commercial chemical space using molecular docking gives access to fragment libraries that are several orders of magnitude larger than those screened experimentally and can enable efficient optimization of hits to potent leads.

Peptide-recognizing G protein-coupled receptors (GPCRs) are promising therapeutic targets but often resist drug discovery efforts. Determination of crystal structures for peptide binding GPCRs has provided opportunities to explore structure based methods in lead development. Molecular docking screens of two chemical libraries, containing either fragment- or lead-like compounds, against a neurotensin receptor 1 crystal structure allowed for a comparison between different drug development strategies for peptide-binding GPCRs. A total of 2.3 million molecules were screened computationally, and 25 fragments and 27 leads that were top-ranked in each library were selected for experimental evaluation. Of these, eight fragments and five leads were confirmed as ligands by surface plasmon resonance. The hit rate for the fragment screen (32%) was thus higher than for the lead-like library (19%), but the affinities of the fragments were similar to 100-fold lower. Both screens returned unique scaffolds and demonstrated that a crystal structure of a stabilized peptide-binding GPCR can guide the discovery of small-molecule agonists. The complementary advantages of exploring fragment- and lead-like chemical space suggest that these strategies should be applied synergistically in structure-based screens against challenging GPCR targets.