Technical Highlight - December 2010
Short description: Low-resolution diffraction data hampers structural determination of macromolecular complexes. There is a way round it.
Large complexes are a challenge to structural biologists. It is extremely difficult to grow crystals of these assemblies and, even when successful, they often diffract weakly, producing a resolution of 4 Å or lower. An electron density map at this resolution is hard to interpret, and it is almost impossible to build an accurate model of the atomic structure.
High-resolution crystal structures typically have a coordinate accuracy that is roughly ten times better than the resolution of the data (we call this 'super-resolution'). It is achieved by using the Pauli exclusion principle, and has been used with small-molecule crystals to solve the phase problem. Using an entirely different approach, Schröder et al. now show that super-resolution can be achieved for macromolecular complexes.
Theoretically, a 5-Å diffraction data set contains all the information that is necessary to produce a higher-resolution structure, but the computational effort required to extract it is, at present, so huge that it is not feasible. Instead, the team used a homologous structure to add in known information. To allow for differences between the homologous model and the actual structure, they used a deformable elastic network (DEN). This method tolerates large deformations, such as a hinge, to occur between the model and the real structure.
Testing this approach with the protein penicillopepsin, the authors generated synthetic data sets of between 3.5 Å and 5 Å to show that it was successful. DEN significantly improved refinement, with and without phase information. By cross-checking against R free the best parameters can be established to yield the most accurate models.
Further developments of the DEN approach could allow families of homologous structures to be used as the template or perhaps aid modeling of the loop conformations.
G.F. Schröder et al. Super-resolution biomolecular crystallography with low-resolution data.
Nature 464, 1218-1222 (2010). doi:10.1038/nature08892