In this paper, a structure-function analysis of B-DNA self-fitting is reviewed in the light of recent oligonucleotide crystal structures. Their crystal packings provided a high-resolution view of B-DNA helices closely and specifically fitted by groove-backbone interaction, a natural and biologically relevant manner to assemble B-DNA helices. In revealing that new properties of the DNA molecule emerge during condensation, these crystallographic studies have pointed to the biological importance of DNA—DNA interactions.

In the years that have passed since the publication of Wolfram Saenger's classic book on nucleic acid structure (Saenger, 1984), a considerable amount of new data has been accumulated on the range of conformations which can be adopted by DNA. Many unusual species have joined the DNA zoo, including new varieties of two, three and four stranded helices. Much has been learnt about intrinsic DNA curvature, dynamics and conformational transitions and many types of damaged or deformed DNA have been investigated. In this article, we will try to summarise this progress, pointing out the scope of the various experimental techniques used to study DNA structure, and, where possible, trying to discern the rules which govern the behaviour of this subtle macromolecule. The article is divided into six major sections which begin with a general discussion of DNA structure and then present successively, B-DNA, DNA deformations, A-DNA, Z-DNA and DNARNA hybrids. An extensive set of references is included and should serve the reader who wishes to delve into greater detai.

Growing interest in gene targeting drugs has inspired the development of a multitude of nucleic acid analogues, many of which feature substitutions in the phosphodiester moiety of the backbone (reviewed by Mesmaeker et al. 1995 and Nielsen, 1995). Peptide nucleic acid (PNA) is an example of a more radical redesign of DNA. The entire sugar-phosphate backbone is substituted by a chain of peptide-like N-(2-aminoethyl)glycine units so that an achiral and uncharged DNA-mimic is obtained (Fig. 1; Nielsen et al. 1991). The synthesis is based on standard peptide chemistry (Christensen et al. 1995) and has been automated. PNA can relatively easily be modified to include modifications of the backbone as well as of the bases (Hyrup & Nielsen, 1996). PNA is chemically stable and, in contrast to natural nucleic acids and peptides, PNA is expected to remain intact in living cells since it is not a substrate for natural hydrolytic enzymes and is not degraded by cell extracts (Demidov et al. 1994).

The passage of molecules and information across cell membranes is mediated largely by membrane-spanning proteins acting as channels, pumps, receptors and enzymes. These proteins perform many tasks: they control electrochemical gradients across the membrane, receive signals from the environment or from other cells, convert light energy into chemical signals, transport small molecules into and out of cells, and harness proton gradients to generate the energy consumed in metabolism. Indeed, of the estimated 50000–100000 genes in the human genome, fully 20–40 % are thought to encode integral membrane proteins. If one also includes membrane-associated proteins, which are attached to the membrane surface through fatty acyl chains or electrostatic interactions, this percentage is likely to be much higher.