Abstact

The X-ray crystal structure of squid-type diisopropylfluorophosphatase (DFPase) has been refined to a resolution of 0.85 A and a crystallographic R value of 9.4%. Crystal annealing improved both the mosaicity and resolution of the crystals considerably. The overall structure of this protein represents a six-bladed beta-propeller with two calcium ions bound in a central water-filled tunnel. 496 water, two glycerol and two MES buffer molecules and 18 PEG fragments of different lengths could be refined in the solvent region. 45 of the 314 residues have been refined with alternative orientations. H atoms have been omitted from disordered residues. For the residues of the inner beta-strands, H atoms are visible in a normal F(o) - F(c) difference map of a hydrogen-deficient structure model. The 208 most reliable residues, without disorder or reduced occupancy in their side chains, were finally refined without restraints. A subsequent full-matrix refinement cycle for the positional parameters yielded estimated standard deviations (e.s.d.s) by matrix inversion. The thus calculated bond lengths and bond angles and their e.s.d.s were used to obtain averaged bond lengths and bond angles, which were compared with the restraints applied in the preceding refinement cycles. The lengths and angles of the hydrogen bonds inside the antiparallel beta-sheets of the DFPase structure were compared with data averaged over 11 high-resolution protein structures. Torsion angles were averaged according to angle types used as restraints in X-PLOR and CNS and subsequently compared with values obtained from 46 high-resolution structures. Side-chain torsion angles were also classified into rotamer types according to the Penultimate Rotamer Library. Moreover, precise dimensions for both Ca(2+)-coordination polyhedra could be obtained and the coordination of one Ca(2+) ion by an imidazole N atom was confirmed. This statistical analysis thus provides a first step towards a set of restraints that are founded completely on macromolecular data; however, 10-20 additional protein data sets of comparable accuracy and size will be required to obtain a larger statistical base, especially for side-chain analysis.