Sequence specific conformation of a DNA decamer containing an adenine tract studied in solution by H-NMR spectroscopy.

The decanucleotide duplex d(AAAACGTTTT)2 and a variety of phase-sensitive two-dimensional (2D) NMR experiments have been used to investigate the solution conformation of an adenine-tract and its junction with another DNA sequence. 2D nuclear Overhauser effect data confirm that the oligonucleotide has a general B-type DNA morphology but an array of unusual correlations implies that the adenine tract and the 5'-ApC junction have conformations more compatible with the modified X-ray structures recently reported for DNAs of similar sequence (Nelson, H.C.M., Finch, J.T., Luisi, B.F. and Klug, A. (1987) Nature 330, 221-226). The pattern and magnitude of interstrand NOEs from the adenine H2s to the sugar H1's of the complementary base to the 5'-neighbouring residue indicate that the A-T basepairs are highly propeller twisted and that the minor groove is narrowed, showing its greatest compression at the 3'-end of the tract at the 5'-ApC step. Quantifying spin-coupling interactions within the deoxyribose rings by analysing both 1D and high-resolution 2D DQF-COSY data reveals that the conformation of the purines is predominantly C2'-endo, with the pseudorotation phase angle P lying in the range 140-180 degrees. For the pyrimidines, however, there are distortions away from this standard B-type geometry with the data being best described by P values lying in the range 90-130 degrees (i.e., O4'-endo, C1'-exo). The sugar puckers of A1, T9 and T10 are dynamically distorted no doubt as a consequence of their positions at, or close to, the ends of the duplex. Thus the conformation of the adenine and thymine sugars within the oligo(dA) and oligo(dT) strands are different with an abrupt change in sugar puckering occurring at the 5'-ApC (5'-GpT) step. Peculiar chemical shifts values for A4H2, T7CH3 and sugar C5 H1', H2' and H2", together with a number of interresidue NOEs with unusual intensities, imply that there are also substantial modifications to basepair stacking interactions at this step. Taken as a whole, our data are consistent with the view that the conformational dislocation at the 5'-ApC dinucleotide results from a combination of slide and roll manoeuvres and that the junction between the AAAA and CG sequences is a potential nucleation site for DNA bending.