Technical Highlight - August 2011
Short description: A new method coupled with accurate NMR structures yields highly reliable search models for molecular replacement.
Once diffracting crystals are obtained, the crystallographer must solve the phase problem to determine the structure. The solutions to this problem are either to use additional experimental data from heavy-atom derivatives or to use molecular replacement, a method that relies on an accurate initial model with homology to the unknown structure. As NMR structures are now available for many smaller proteins, a potential solution is to use the available NMR data from either the target protein or a close homolog to generate search models for use in molecular replacement—an approach used with limited success to date.
Montelione and colleagues (PSI NESG) have now leveraged the improved quality of recent NMR structures and advances in molecular replacement software to develop an improved method. Their method is based on an algorithm named FindCore that identifies sets of atoms, mainly in the hydrophobic core of the protein, whose positions are well-defined relative to each other. These core atom sets therefore provide an objective measure of global structural consistency when generating search models from NMR structures.
The authors systematically validated their method using data sets from 25 structures for which both NMR and X-ray structures were previously solved. In most cases the molecular replacement solutions found by FindCore performed best or among the best compared to previous methods, and a total of 22 of the 25 structures were built using a completely automated pipeline, provided the data had a resolution of at least 2.5 Å. An additional two structures required Rosetta refinement of the starting NMR structures to yield reliable search models. The only example that failed to provide a useful search model resulted from genuine pH-dependent differences between the NMR and X-ray structures. The phasing power of search models from FindCore was confirmed by additional calculations on a complex with minimal structural rearrangement and homologous structures with sequence identity above 40% and no more than four copies in the asymmetric unit. The method promises to make the process of generating search models from NMR structures far more reliable.
B. Mao et al. Improved technologies now routinely provide protein NMR structures useful for molecular replacement.
Structure 19, 757-766 (2011). doi:10.1016/j.str.2011.04.005