Melting of a self-complementary DNA minicircle

Abstract

Melting curves are calculated for the 16-base-pair duplex DNA sequence 5′ GTATCCGTACGGATAC 3′ linked on the ends by TTTT single-strand loops. The equilibrium statistical thermodynamic theory of DNA melting is modified to include effects of end-loops on the melting transition. An excellent fit of the experimental melting curve in 0.2 m-NaCl is obtained using two adjustable parameters, one for end-loop formation and the other for formation of the complete 40-base single-strand loop. The best-fit calculated melting curve permits evaluation of these parameters. The free energy to close a TTTT end-loop is 2.12 kcal/mol (1 cal = 4.184 J). A TTTT end-loop or hairpin loop is significantly more stable than an internal loop of comparable size sandwiched between two helical regions, even after allowing for the different stacking contributions. Reasons for this increased stability are presented. The loop free energy of the 40-base single-strand open minicircle is evaluated to be +1.27 kcal/mol, thus favoring the melting of two end-loops into the large open minicircle. The present results are compared with those of others for d(T-A) oligomers. The sequence TTTT forms a more stable end-loop, or hairpin, than TATA by about 2.0 kcal/mol.

Theoretical rate constants for the proton-transfer step in the standard hydrogen-exchange model are calculated by extending the theory of diffusion-controlled reactions to take account of the electrostatic potential of the DNA. The predicted ratios of rate constants for different pairs of catalysts exchanging an A · T proton agree satisfactorily with the available experimental data for a 14-base-pair linear duplex, which confirms the diffusion-control of the proton-transfer step. Data presented here for the 16 base-pair duplex of the minicircle are consistent with catalysis-limited exchange in which the proton-transfer step is likewise diffusion-controlled. Under catalysis-limited conditions, the imino proton exchange rates are predicted from the catalytic rate constants, prevailing buffer catalyst concentrations, and the equilibrium constants to form the unstacked open state of optical melting theory. The observed exchange rates of the A · T base-pairs show no sign of the strong predicted end-melting trend, and exceed the predicted values by factors of 10 to 400. Moreover, the succession of “melting” in the nuclear magnetic resonance line-broadening deviates from that predicted by optical melting theory. In particular, the central base-pairs (T · A(8) and A · T(9) are out of order by 25 deg.C. These and other considerations indicate that the solvent-accessible open state of the standard model for imino proton exchange is most probably not identical with the unstacked open state of optical melting theory. Considerations concerning the recently reported exchange in the absence of added catalyst (AAC) indicate that the structures and kinetics of the solvent-accessible open states may differ for A · T base-pairs in certain locales, as well as A · T pairs in different duplex DNAs. AAC exchange is inferred to be much less important for many DNAs, including ours, than for others, and to be much more important in poly[d(A-T)] · poly[d(A-T)].