Differential base stacking interactions induced by trimethylene interstrand DNA cross-links in the 5'-CpG-3' and 5'-GpC-3' sequence contexts.

Synthetically derived trimethylene interstrand DNA cross-links have been used as surrogates for the native cross-links that arise from the 1,N(2)-deoxyguanosine adducts derived from alpha,beta-unsaturated aldehydes. The native enal-mediated cross-linking occurs in the 5'-CpG-3' sequence context but not in the 5'-GpC-3' sequence context. The ability of the native enal-derived 1,N(2)-dG adducts to induce interstrand DNA cross-links in the 5'-CpG-3' sequence as opposed to the 5'-GpC-3' sequence is attributed to the destabilization of the DNA duplex in the latter sequence context. Here, we report higher accuracy solution structures of the synthetically derived trimethylene cross-links, which are refined from NMR data with the AMBER force field. When the synthetic trimethylene cross-links are placed into either the 5'-CpG-3' or the 5'-GpC-3' sequence contexts, the DNA duplex maintains B-DNA geometry with structural perturbations confined to the cross-linked base pairs. Watson-Crick hydrogen bonding is conserved throughout the duplexes. Although different from canonical B-DNA stacking, the cross-linked and the neighbor base pairs stack in the 5'-CpG-3' sequence. In contrast, the stacking at the cross-linked base pairs in the 5'-GpC-3' sequence is greatly perturbed. The pi-stacking interactions between the cross-linked and the neighbor base pairs are reduced. This is consistent with remarkable chemical shift perturbations of the C(5) H5 and H6 nucleobase protons that shifted downfield by 0.4-0.5 ppm. In contrast, these chemical shift perturbations in the 5'-CpG-3' sequence are not remarkable, consistent with the stacked structure. The differential stacking of the base pairs at the cross-linking region probably explains the difference in stabilities of the trimethylene cross-links in the 5'-CpG-3' and 5'-GpC-3' sequence contexts and might, in turn, account for the sequence selectivity of the interstrand cross-link formation induced by the native enal-derived 1,N(2)-dG adducts.

NMR determination of the conformation of a trimethylene interstrand cross-link in an oligodeoxynucleotide duplex containing a 5'-d(GpC) motif.

Malondialdehyde interstrand cross-links in DNA show strong preference for 5'-d(CpG) sequences. The cross-links are unstable and a trimethylene cross-link has been used as a surrogate for structural studies. A previous structural study of the 5'-d(CpG) cross-link in the sequence 5'-d(AGGCGCCT), where G is the modified nucleotide, by NMR spectroscopy and molecular dynamics using a simulated annealing protocol showed the guanine residues and the tether lay approximately in a plane such that the trimethylene tether and probably the malondialdehyde tether, as well, could be accommodated without major disruptions of duplex structure [Dooley et al. J. Am Chem. Soc. 2001, 123, 1730-1739]. The trimethylene cross-link has now been studied in a GpC motif using the reverse sequence. The structure lacks the planarity seen with the 5'-d(CpG) sequence and is skewed about the trimethylene cross-link. Melting studies indicate that the trimethylene cross-link is thermodynamically less stable in the GpC motif than in the 5-d(CpG). Furthermore, lack of planarity of the GpC cross-link precludes making an isosteric replacement of the trimethylene tether by malondialdehyde. A similar argument can be used to explain the 5'-d(CpG) preference for interchain cross-linking by acrolein.