Abstact

In an earlier study [Smith, Davies, Dodson & Moore (1995). Biochemistry, 34, 415-425] the crystal structure of the d(TGATCA)-nogalamycin complex was determined to 1.8 A and refined with PROLSQ to R = 19.5% against 4767 reflections with F> 1sigma(F). A low-temperature crystallographic study on this complex has now been performed. Native data collection at liquid-nitrogen temperature (120 K) improved the resolution to 1.4 A. The structure has now been refined against these new diffraction data in the resolution range 8-1.4 A using NUCLSQ, PROLSQ, SHELXL93 and X-PLOR, in order to determine to what extent the resulting DNA conformation and associated solvent structure would differ and to examine the suitability of these programs for the refinement of oligonucleotide structures. With the advent of more DNA-protein structure determinations, it is of interest to see how well the protein-refinement packages, PROLSQ and X-PLOR, and the small-molecule program, SHELXL93, are able to accommodate DNA. Comparisons are made between the dictionaries, weights and restraints used and the final models obtained from each program. Although the final R values, using all data in the resolution range 8.0-1.4 A, from PROLSQ (22.8%), SHELXL93 (R1 =21.7% after isotropic refinement) and X-PLOR (24.4%) are higher than the R value from the NUCLSQ refinement (21.2%), the root-mean-square deviations between the four final models are very small. Using this high-quality 8.0-1.4 A data set neither the dictionary nor the refinement program leave an imprint on the final fully refined complex. Likewise, the helical parameters and backbone conformation including sugar-puckering modes are not influenced by the refinement procedure used. Although a different number of water molecules is found in each refinement, varying from 62 (X-PLOR) to 86 (NUCLSQ), the first hydration sphere is well conserved in all four models.