Active sites without restraints: high-resolution analysis of metal cofactors.

For most three-dimensional structures of biological macromolecules, the factual accuracy of atom positions by far exceeds the resolution of the experimental data, although the refinement problem presented by a protein structure is substantially underdetermined. This is achieved through using restraints that precisely define protein geometries and thus reduce the degrees of freedom of the refinement problem. If such information is not available or when unusual geometries or particular ligand states complicate structural analysis, possible pitfalls arise that not only concern the precise definition of spatial arrangements, but also the identification of atom types and bond distances. Prominent examples include CO dehydrogenase, hydrogenase, acetylene hydratase and nitrogenase, all of which employ unique active sites that turned out not to be what they seemed upon first inspection.