Protein crystallization by design: chymotrypsinogen without precipitants.

Protein crystals are usually obtained by an empirical approach based on extensive screening to identify suitable crystallization conditions. In contrast, we have used a systematic predictive procedure to produce data-quality crystals of bovine chymotrypsinogen A and used them to obtain a refined X-ray structure to 3 A resolution. Measurements of the osmotic second virial coefficient of chymotrypsinogen solutions were used to identify suitable solvent conditions, following which crystals were grown for approximately 30 hours by ultracentrifugal crystallization, without the use of any precipitants. Existing structures of chymotrypsinogen were obtained in solutions including 10-30 % ethanol, whereas simple buffered NaCl solutions were used here. The protein crystallized in the tetragonal space group P4(1)2(1)2, with one molecule per asymmetric unit. The quality of the refined map was very high throughout, with the main-chain atoms of all but four residues clearly defined and with nearly all side-chains also defined. Although only minor differences are seen compared to the structures previously reported, they indicate the possibility of structural changes due to the crystallization conditions used in those studies. Our results show that more systematic crystallization of proteins is possible, and that the procedure can expand the range of conditions under which crystals can be grown successfully and can make new crystal forms available.

Structure of a thermostable disulfide-bridge mutant of phage T4 lysozyme shows that an engineered cross-link in a flexible region does not increase the rigidity of the folded protein.

A disulfide bond introduced between amino acid positions 9 and 164 in phage T4 lysozyme has been shown to significantly increase the stability of the enzyme toward thermal denaturation [Matsumura, M., Becktel, W.J., Levitt, M., & Matthews, B. W. (1989) Proc. Natl. Acad. Sci. U.S.A. 86, 6562-6566]. To elucidate the structural features of the engineered disulfide, the crystal structure of the disulfide mutant has been determined at 1.8-A resolution. Residue 9 lies in the N-terminal alpha-helix, while residue 164 is located at the extreme C terminus of T4 lysozyme, which is the most mobile part of the molecule. The refined structure shows that the formation of the disulfide bond is accompanied by relatively large (approximately 2.5 A) localized shifts in C-terminal main-chain atoms. Comparison of the geometry of the engineered disulfide with those of naturally observed disulfides in proteins shows that the engineered bridge adopts a left-handed spiral conformation with a typical set of dihedral angles and C alpha-C alpha distance. The geometry of the engineered disulfide suggests that it is slightly more strained than the disulfide of oxidized dithiothreitol but that the strain is within the range observed in naturally occurring disulfides. The wild-type and cross-linked lysozymes have very similar overall crystallographic temperature factors, indicating that the introduction of the disulfide bond does not impose rigidity on the folded protein structure. In particular, residues 162-164 retain high mobility in the mutant structure, consistent with the idea that stabilization of the protein is due to the effect of the disulfide cross-link on the unfolded rather than the folded state.(ABSTRACT TRUNCATED AT 250 WORDS)

The effect of crystal packing on oligonucleotide double helix structure.

One of the questions that constantly is asked regarding x-ray crystal structure analyses of macromolecules is: To what extent is the observed crystal structure representative of the molecular conformation when free in solution, and to what degree is the structure perturbed by intermolecular crystal forces? This can be assessed with DNA oligomers because of an unusual aspect of crystallization self-complementary oligomers should possess a twofold symmetry axis normal to their helix axis, yet more often than not crystal of such oligomers do not use this internal symmetry. The two ends of the helix are crystallographically distinct though chemically identical. Complexes of DNA oligomers with intercalating drugs such as triostin A tend to use their twofold symmetry when they crystallize, whereas complexes with non-intercalating, groove-binding drugs ignore this symmetry unless the drug molecule is very small. A detailed examination of crystal packing in the dodecamer C-G-C-G-A-A-T-T-C-G-C-G provides an explanation of all of the foregoing behavior in terms of the mechanism of nucleation of DNA or DNA-drug complexes on the surface of a growing crystal. Asymmetry of the ends of the DNA helix is the price that is paid for efficient lateral packing of helices within the crystal. The actual end-for-end variation in standard helix parameters is compared with the experimental noise level as gauged by independent re-refinement of the same oligonucleotide structure where available, and with the observed extent of variation of these same parameters along the helix. Oligomers analyzed are the B-DNA dodecamer C-G-C-G-A-A-T-T-C-G-C-G, the A-DNA octamer G-G-T-A-T-A-C-C, and the phosphorothioate analogue of the B-DNA hexamer G-C-G-C-G-C. End-for-end variation, presumably the result of crystal packing is typically double the experimental noise level, and half the variation in the same parameter along the helix. Analysis of crystal packing in the phosphorothioate hexamer, which uses the same P212121 space group as the dodecamer, shows that the highly unsymmetrical B1 vs. BII backbone conformation probably is to be ascribed to crystal packing forces, and not to the sequence of the hexamer.

Binding of Hoechst 33258 to the minor groove of B-DNA.

An X-ray crystallographic structure analysis has been carried out on the complex between the antibiotic and DNA fluorochrome Hoechst 33258 and a synthetic B-DNA dodecamer of sequence C-G-C-G-A-A-T-T-C-G-C-G. The drug molecule, which can be schematized as: phenol-benzimidazole-benzimidazole-piperazine, sits within the minor groove in the A-T-T-C region of the DNA double helix, displacing the spine of hydration that is found in drug-free DNA. The NH groups of the benzimidazoles make bridging three-center hydrogen bonds between adenine N-3 and thymine O-2 atoms on the edges of base-pairs, in a manner both mimicking the spine of hydration and calling to mind the binding of the auti-tumor drug netropsin. Two conformers of Hoechst are seen in roughly equal populations, related by 180 degrees rotation about the central benzimidazole-benzimidazole bond: one form in which the piperazine ring extends out from the surface of the double helix, and another in which it is buried deep within the minor groove. Steric clash between the drug and DNA dictates that the phenol-benzimidazole-benzimidazole portion of Hoechst 33258 binds only to A.T regions of DNA, whereas the piperazine ring demands the wider groove characteristic of G.C regions. Hence, the piperazine ring suggests a possible G.C-reading element for synthetic DNA sequence-reading drug analogs.