DNA: sequence, structure and supercoiling. The twentieth Colworth medal lecture.

Abstract

DNA structures It is now 30 years since Watson & Crick (1953) proposed the base-paired double helical structure for DNA. Although the details of the structure in terms of the precise torsion angles adopted along the deoxyribose-phosphate backbone have undergone considerable revision and refinement, the essential nature of the structure was of course correct, and has subsequently stimulated the growth of the new science of molecular biology to an almost incalculable degree. Clearly, a molecule with the central importance of D N A must equally be of recurrent interest to structural biologists, and indeed the area has undergone something of a renaissance in recent years. This has been a result chiefly of the advent of X-ray crystallographic studies of oligonucleotides of defined sequence. Previous structural studies were limited to fibre diffraction measurements of high molecular weight DNA, and was therefore limited both by the nature of the sample and the manner of the collection of the data. Recent single-crystal studies have allowed for the first time atomic, or nearly atomic, resolution structures to be elucidated, giving information on each nucleotide in a molecule rather than an averaged structure. Both A (Conner et al., 1982) and B (Wing et al., 1980) forms of DNA structure have been solved at this resolution without presenting any great surprises, but the solution of a d(CpG)3 high salt structure by Rich and coworkers (Wang et al. 1979) turned out to be very much less conventional. The most obvious difference was that the new structure, termed Zform DNA, was left-handed, achieved by a CpG dinucleotide symmetrical unit in which the deoxyguanosine has unusual glycosidic bond angle (syn) and deoxyribose pucker (C3’-endo) compared with R-DNA. This was really the first indication that the three-dimensional structure of DNA might reflect its base sequence. Another example of this behaviour emerged from the refined structure of a d(CGCGAATTCGCG) dodecamer by Dickerson and colleagues (Dickerson & Drew, 1981). Whilst the overall structure is clearly recognizable as the B-form, the variation in individual torsion angles is considerable. Each nucleotide differs in its detailed ‘geometry from the others of the dodecamer. The underlying message is very clear: DNA structure exhibits a large element of sequence dependence. Proteins are capable of DNA sequence recognition with exquisite precision, obvious to anyone who has ever performed a restriction enzyme reaction. The problem of understanding the specificity of DNA-protein interaction is fundamental to molecular biology, and the basis may well reside in the fine details of base-specific backbone structure. These interactions may well give clues as to such structural variation, and enzyme probing has been used with considerable success in revealing structural polymorphism in DNA. Experiments of this kind have been performed at varying levels of precision, and with various aims. DNase I has been used as a probe of DNA bound to solid supports such as calcium phosphate, in order to measure helical periodicity, and the results obtained have indicated that this parameter may vary with the sequence of the DNA studied (Rhodes & Klug, 1980, 1981). An alternative endonuclease,