We describe a very accurate addition (called structure X here) to the B-DNA dodecamer family of X-ray structures. Our results confirm the observation of Drew and Dickerson [(1981) J. Mol. Biol. 151, 535-556] that the spine of hydration in AT tract DNA is two layers deep. However, our results suggest that the primary spine is partially occupied by sodium ions. We suggest that many sequence-dependent features of DNA conformation are mediated by site specific binding of cations. For example, preferential localization of cations, as described here within the minor groove of structure X, is probably the structural origin of AT tract bending and groove narrowing. The secondary spine, which does not interact directly with the DNA, is as geometrically regular as the primary spine, providing a model for transmission of sequence information into solvent regions. A fully hydrated magnesium ion located in the major groove of structure X appears to pull cytosine bases partially out from the helical stack, exposing pi-systems to partial positive charges of the magnesium ion and its outer sphere. A partially ordered spermine molecule is located within the major groove of structure X. Dodecamer structures are derived from crystals of [d(CGCGAATTCGCG)]2 in space group P212121 (a = 25 A, b = 40 A, and c = 66 A). On average, those crystals diffracted to around 2.5 A resolution with 2500 unique reflections. Structure X, with the same space group, DNA sequence, and crystal form as the "Dickerson dodecamer", is refined against a complete, low-temperature, 1.4 A resolution data set, with over 11000 reflections. Structure X appears to be conformationally more ordered than previous structures, suggesting that at least a portion of the conformational heterogeneity previously attributed to DNA sequence in fact arises from experimental error.

The potassium form of d(CGCGAATTCGCG) solved by X-ray diffraction to 1.75 A resolution indicates that monovalent cations penetrate the primary and secondary layers of the "spine of hydration". Both the sodium [Shui, X., McFail-Isom, L., Hu, G. G., and Williams, L. D. (1998) Biochemistry 37, 8341-8355] and the potassium forms of the dodecamer at high resolution indicate that the original description of the spine, only two layers deep and with full occupancy by water molecules, requires substantive revision. The spine is merely the bottom two layers of a four layer solvent structure. The four layers combine to form a repeating motif of fused hexagons. The top two solvent layers were not apparent from previous medium-resolution diffraction data. We propose that the narrow minor groove and axial curvature of A-tract DNA arise from localization of cations within the minor groove. In general, the results described here support a model in which most or all forces that drive DNA away from canonical B-conformation are extrinsic and arise from interaction of DNA with its environment. Intrinsic forces, originating from direct base-base interactions such as stacking, hydrogen bonding, and steric repulsion among exocyclic groups appear to be insignificant. The time-averaged positions of the ubiquitous inorganic cations that surround DNA are influenced by DNA bases. The distribution of cations depends on sequence. Regions of high and low cation density are generated spontaneously in the solvent region by heterogeneous sequence or even within the grooves of homopolymers. The regions of high and low cation density deform DNA by electrostatic collapse. Thus, the effects of small inorganic cations on DNA structure are similar to the effects of proteins.

Direct methods of crystal structure solution are greatly facilitated in centrosymmetric space groups where the complexity of the phase-problem is reduced. For most peptides and proteins, crystallization in a centrosymmetric arrangement is precluded by an intrinsic dissymmetry due to the constituent chiral amino acid residues. The synthetic accessibility of peptide sequences containing amino acids of either chirality offers the possibility for co-crystallization of racemic crystals. We report here the first use of such an approach for the de novo structure determination of a medium-sized molecule, trichogin A IV, which is a constituent of a fungal lipopeptaibol mixture possessing membrane-modifying properties of biological interest.

Nucleic acid structure, stability, and reactivity are governed substantially by cations. We propose that magnesium and other biological inorganic ions unstack bases of DNA and RNA. This unstacking function of cations opposes their previously accepted role in stabilizing DNA and RNA duplexes and higher assemblies. We show that cations interact favorably with pi-systems of nucleic acid bases. These cation-pi interactions require access of cations or their first hydration shells to faces of nucleic acid bases. We observe that hydrated magnesium ions located in the major groove of B-DNA pull cytosine bases partially out from the helical stack, exposing pi-systems to positive charge. A series of critical cation-pi interactions contribute to the stability of the anticodon arm of yeast-tRNAphe, and to the magnesium core of the Tetrahymena group I intron P4-P6 domain. The structural consequences of divalent cation-pi interactions are clearly distinct from, and some cases in opposition to, cation-electron lone pair interactions. These observations of cation-pi interactions suggest a number of new mechanistic roles for cations in DNA bending, DNA-protein recognition, base-flipping, RNA folding, and catalysis.

The application of detailed structural data bases has now culminated in the successful design of a new generation of bisanthracyclines that form ultratight DNA complexes [Chaires, J. B., Leng, F., Przewloka, T., Fokt, I., Ling, Y. H., Perez-Soler, R., & Priebe, W. (1997) J. Med. Chem. 40, 261-266]. Daunomycin dimers were designed to bind to DNA in complexes resembling those of monomers intercalated at adjacent sites. The goal of the work described here was to determine, with X-ray crystallography, if a potent member of this newly designed and synthesized class of bisanthracyclines (WP631) binds as intended. WP631 is composed of two daunomycin molecules, linked N3' to N3' by a xylyl group. We have solved the 2.2 A X-ray crystal structure of a complex of WP631 bound to [d(CGATCG)]2. We demonstrate, on a detailed molecular level, that the WP631 design strategy is a success. The structures of WP631 and two daunomycin molecules bound to [d(CGATCG)]2 provide the unprecedented opportunity for detailed comparison of mono- and bis-intercalated complexes of the same chromophore, allowing us to distinguish effects of mono-intercalation from those of bis-intercalation. Differences are focused primarily in the centers of the complexes. DNA unwinding and other helical distortions propagate more efficiently to the center of the WP631 complex than to the center of the daunomycin complex.