W0320: Direct Methods Applied to Large MAD Problems

W0320

Direct Methods Applied to Large MAD Problems. Janet L. Smith & Joseph M. Krahn, Dept. of Biological Sciences, Purdue University, West Lafayette, IN 47907-1392 USA

Multiwavelength anomalous diffraction (MAD) is a highly successful new phasing method in macromolecular crystallography. MAD exploits physical changes to a sample crystal during the data collection experiment to derive phase information. The anomalous scattering factors of selected atoms in the crystal change with the incident X-ray energy in the vicinity of an atomic resonant frequency. The attractiveness of MAD is due to the possibility of deriving accurate phase estimates from data measured from one sample crystal in one experiment. MAD phasing depends critically on determination of the partial structure of anomalous scatterers. In cases where their number is small or the crystal symmetry is simple, Patterson methods are sufficient for establishing the sub-structure of anomalous scatterers. However, this is often not the case. Much of the success of MAD is due to the Se label in selenomethionine (SeMet) because of its biological incorporation in place of the amino acid methionine. Thus, for SeMet problems, the complexity of the anomalous scatterer partial structure is proportional to the size of the protein, on average one SeMet for every 50-60 amino acids. Direct methods have been applied to MAD data to solve a number of complex Se partial structures having asymnmetric units with up to 30 Se sites and up to 130 kDa protein. MAD problems differ from the small-molecule crystal structures to which direct methods have traditionally been applied. The structure amplitudes (F_A) of the partial structure are not the F_obs, but rather are obtained from the often-weak Bijvoet and dispersive differences. Despite the additional challenge of MAD problems, there is no indication that the upper limit has been reached in size of protein or anomalous-scatterer partial structure to which direct methods can be applied successfully.

This work has been supported by NIH grant DK 42303 to JLS.