Effects of base sequence on the loop folding in DNA hairpins.

High-resolution NMR and UV-melting experiments have been used to study the hairpin formation of partly self-complementary DNA fragments in an attempt to derive rules that describe the folding in these molecules. Earlier experiments on the hexadecanucleotide d(ATCCTA-TTTT-TAGGAT) had indicated that within the loop of four thymidines a wobble T-T pair is formed (Blommers et al., 1987). In the present paper it is shown that if the first and the last thymines of the intervening sequence are replaced by complementary bases, sometimes base pairs can be formed. Thus for the intervening sequences -CTTG- and -TTTA- with the pyrimidine in the 5'-position and the purine in the 3'-position, a base pair is formed leading to a loop consisting of two residues. For the intervening sequences -GTTC- and -ATTT- with the purine in the 5'-position and the pyrimidine in the 3'-position, this turns out not to be the case. It was found that it made no difference when the four-membered sequence was closed by a G-C base pair or an A-T base pair. Replacement of the two central thymidine residues by the more bulky adenine residues limits the hairpin to a four-membered loop scheme. Very surprisingly, it was found from 2D NOE experiments that the T-A base pair, formed in the loop consisting of the -TTTA- sequence, is a Hoogsteen pair. It is argued that the pairing of the bases in this scheme may facilitate the formation of a loop of two residues, since the distance of the C1' atoms in this base pair is 8.6 A instead of 10.4 A found in the canonical Watson-Crick base pair. Combination of the data obtained for the series of DNA fragments studied shows that the results can be explained by a simple, earlier proposed, loop folding principle which assumes that the folding of the four-membered loop is dictated by the stacking of the double-helical stem of the hairpin.