Overcoming the inadequacies or limitations of experimental structures as drug targets by using computational modeling...

X-ray crystallography, NMR spectroscopy, and cryoelectron microscopy stand out as powerful tools that enable us to obtain atomic detail about biomolecules that can be potentially targeted by drugs. This knowledge is essential if virtual screening or structure-based ligand-design methods are going to be used in drug discovery. However, the macromolecule of interest is not always amenable to these types of experiment or, as is often the case, the conformation found experimentally cannot be used directly for docking studies because of significant changes between apo and bound forms. Furthermore, sometimes the desired insight into the binding mechanism cannot be gained because the structure of the ligand-receptor complex, not having been time-resolved, represents the endpoint of the binding process and therefore retains little or no information about the intermediate stages that led to its creation. Molecular dynamics (MD) simulations are routinely applied these days to the study of biomolecular systems with the aims of sampling configuration space more efficiently and getting a better understanding of the factors that determine structural stability and relevant biophysical and biochemical processes such as protein folding, ligand binding, and enzymatic reactions. This field has matured significantly in recent years, and strategies have been devised (for example activated, steered, or targeted MD) that allow the calculated trajectories to be biased in attempts to properly shape a ligand binding pocket or simulate large-scale motions involving one or more protein domains. On the other hand, low-frequency motions can be simulated quite inexpensively by calculation of normal modes which allow the investigation of alternative receptor conformations. Selected examples in which these methods have been applied to several medicinal chemistry and in silico pharmacology endeavors are presented.