SAD manganese in two crystal forms of glucose isomerase.

Glucose isomerase from Streptomyces rubiginosus was crystallized in two forms: I222, with one molecule of 44 kDa in the asymmetric unit, and P2(1)2(1)2, with two unique molecules. The I222 structure is known, but the P2(1)2(1)2 form has not been solved before. X-ray diffraction data for the P2(1)2(1)2 form were collected at a wavelength of 1.54 A and data for the I222 form were collected at three different wavelengths: 1.34, 1.07 and 0.98 A. The amount of anomalous signal from one Mn and eight S atoms in these data sets varies from 1.24% to as low as 0.56%. The dual-space direct-methods program SHELXD, run against the Bijvoet differences, gave a clear solution of all anomalous scatterers for all data sets. The Mn positions only were used for SAD phasing of all four data sets. The electron-density map after density modification, resulting from the phasing of a single-wavelength data set and based purely on the anomalous deltaf" contribution, was clearly interpretable; an almost complete model of the protein was built by wARP without human intervention in all four cases. As far as is known, this is the first time that an anomalous signal as low as 0.6% has successfully been used to determine the structure of a macromolecule.