Investigations on the dynamic structures of adenine- and thymine-containing DNA

The structures of poly(dA-dT), poly(dA-dBr5U) and of poly(dA).poly(dT) have been investigated in solution and in fibers, by Raman spectroscopy. Both the alternating poly(dA-dT), poly(dA-dBr5U) and non-alternating poly(dA).poly(dT) exhibit, in the region of sugar phosphate backbone vibrations, two bands of almost equal intensity at about 841 cm-1 and 817 cm-1. The analysis of the characteristic bands of thymine residues that are sensitive to sugar puckers gives indication of a significant displacement from the C(2')-endo conformer suggesting the adoption of alternative conformers such as O(4')-endo. In contrast, the diagnostic Raman bands for the sugar pucker of adenine residues suggest, instead, predominant adoption of C(2')-endo conformations. These Raman results are compatible with rapid dynamic changes of sugar puckers between C(2')-endo and O(4')-endo for the thymidine (and uridine) residues, whereas in adenine residues the sugar puckers fluctuate around the C(2')-endo pucker in all synthetic DNA molecules studied. Molecular dynamics simulations, performed on six different starting models using two distance-dependent dielectric functions epsilon(r) = 4 r and a sigmoidal dependence), all gave similar dynamic behavior in agreement with these Raman data and their interpretation. The mean calculated pseudorotation phases of the adenine residues are systematically higher (around C(2')-endo) than those of the thymine residues (close to O(4')-endo-C(1')-exo). Besides, the mean lifetimes of the thymine residues are 1.5 to 2.0-fold higher in the O(4')-endo than in the C(2')-endo domain, while those of the adenine residues are two to threefold higher in the C(2')-endo than in the O(4')-endo domain. In the Raman spectra of the alternating poly(dA-dBr5U), the splitting of a band into two components arising from the two contributions of ApBr5U and Br5UpA provides strong evidence for a repeating dinucleotide structure in solution. The calculated twist values averaged over the simulation runs are also systematically higher in the 5'T-A3' step (39 degrees) than in the 5'A-T3' step (33 degrees). Simultaneously, the calculated roll values are positive in the 5'T-A3' step (6 degrees) and negative in the 5'A-T3' step (-9 degrees), while the propeller twist values are about the same (-11 degrees to -16 degrees). On the other hand, in the homopolymer, the average twist value is close to 36 degrees with the roll angle close to 0 degrees and large propeller twist values (-20 degrees).