The assumption that crystal contacts reflect natural macromolecular interactions makes a basis for many studies in structural biology. However, the crystal state may correspond to a global minimum of free energy where biologically relevant interactions are sacrificed in favour to unspecific contacts. A large-scale docking experiment was performed in order to assess the extent of misrepresentation of natural (in-solvent) protein dimers by crystal packing. As found, the failure rate of docking may be quantitatively interpreted if both calculation errors and misrepresentation effects are taken into account. The failure rate analysis is based on the assumption that crystal structures reflect thermodynamic equilibrium between different dimeric configurations. The analysis gives an estimate of misrepresentation probability, which suggests that weakly bound complexes with KD ≥ 100 µM (some 20% of all dimers in the PDB) have higher than 50% chances to be misrepresented by crystals. ...
Evgeny B. Krissinel