STRUCTURE DETERMINATION OF GLUCOSAMINE-6P SYNTHASE: FIGHTING AGAINST PSEUDOSYMMETRY. Alexei Teplyakov¹ and Michael N. Isupov², ¹EMBL, c/o DESY, Notkestr. 85, Hamburg 22603, Germany, ²Department of Chemistry, University of Exeter, Exeter EX4 4QD, UK.

Structure determination of the glutaminase domain of glucosamine-6-phosphate synthase faced difficulties due to pseudosymmetry of the crystals. X-ray diffraction data collected to 1.8 Å resolution indicated strong pseudo C-centering of the primitive orthorhombic lattice which was confirmed by the peak at (0.5, 0.473, 0.0) in the native Patterson map. Molecular replacement with the model of PRPP amidotransferase gave solutions in two possible space groups, P2₁2₁2 and P2₁2₁2₁, indistinguishable due to pseudoabsences. Neither of these solutions could be further refined. Attempts to solve the structure by isomorphous replacement were also unsuccessful. Difference Pattersons for putative heavy atom derivatives were uninterpretable in spite of specific binding in many of them as appeared later. Accidentally, a crystal in the space group P2₁ was found which had the same packing of the molecules but without pseudocentering. While using the PRPP amidotransferase model for molecular replacement in this crystal form failed, the preliminary refined "orthorhombic" model gave a unique solution which was successfully refined at 2.4 Å. The phases were improved by solvent flattening and electron density averaging around a non-crystallographic two-fold axis. However, the back transfer to the orthorhombic form did not work until it was realized that the symmetry of the crystal could be lower than orthorhombic. Finally, the structure was refined using the 1.8 Å data reprocessed in the space group P2₁ (no difference in R-merge). The R-cryst is 18% for all reflections in the range 10-1.8 Å. Analysis of the structure shows that the pseudo-orthorhombic lattice consists of two sublattices shifted to each other by 4 Å.