SAMPLING OF THE CONFORMATIONS OF THE D(CGCTGCGGC) HAIRPIN IN SOLUTION BY 2-DIMENSIONAL NUCLEAR-MAGNETIC-RESONANCE AND THEORETICAL METHODS

Most NMR studies of DNA oligomers have focused on rigid structures that show a strong preference for one or a small set of ground-state conformations. There is an increasing interest in extending NMR methods to investigate DNA systems in which this preference does not exist. A DNA hairpin is one such system where a large number of low-energy structures coexist in solution. In this article we show how 1D/2D NMR data of the d(C1-G2-C3-T4-G5-C6-G7-G8-C9) hairpin are used to map the conformational space of this molecule. First, we characterize the gross morphology of the hairpin by monitoring the exchangeable imino signals in the molecule. Second, we extract a set of inter-proton distances (i.e., the average values and the associated dispersions) for various pairwise interactions by performing full-matrix NOESY simulation with respect to the observed NOESY data for mixing times of 250 and 100 ms. Third, we use these distances as structural constraints to perform a 300-ps molecular dynamics simulation at 500 K. Fourth, we extract 600 snapshots (one after every 0.5 ps) from the MD trajectory and perform constrained energy minimization to map local minima on the sampled energy surface (we call this the rapid temperature quenching step). Fifth, we assign 600 structures to 14 disjoint clusters such that conformationally similar hairpins belong to the same cluster while conformationally distinct hairpins belong to different clusters. Finally, we interpret the NOESY data in terms of conformationally distinct structures by recalculating NOESY contributions taken from representative structures of different clusters. Our analyses clearly demonstrate that the NMR data correspond to an ensemble of distinct structures, i.e., a set of energetically stable but conformationally distinct structures that satisfies the constraints of loop folding in the d(C1-G2-C3-T4-G5-C6-G7-G8-C9) hairpin. Two types of loop folding consistent with NMR data are obtained: (i) a hairpin with two G.C pairs in the stem and four residues in the loop and (ii) a hairpin with two G-C pairs and a reverse wobble G-T pair in the stem plus two residues in the loop.