HENRY M. SOBELL
MOLECULAR STRUCTURES OF THE B-A AND B-B PREMELTONS
Using the technique of linked-atom nonlinear least squares (Smith and Arnott, 1978), it has been possible to calculate structural intermediates that lie on the minimal energy pathway connecting B- with A- DNA. This was accomplished by calculating a series of uniform transitions along the polymer, in which the sugar puckering of alternate deoxyribose residues was altered and the structure then energy minimized, subject to a series of constraints and restraints.
In this way, we have discovered the existence of a minimal energy pathway connecting B- with A- DNA that passes through the lowest energy beta-DNA form. A detailed description of these calculations, together with the final coordinates of all twenty-five structures, can be found in "Kink-Antikink Bound States in DNA Structure", Henry M. Sobell, in Biological Macromolecules and Assemblies, Volume 2, Nucleic Acids and Interactive Proteins, edited by F.A. Jurnack and A. McPherson, John Wiley & Sons, New York, 192-216 (l985).
To simulate the bound states structures, base-paired dinucleotide elements from each structure in this sequence were then pieced together using a least squares procedure. The decision to compute twenty-five intermediate structures that connect B-DNA with A-DNA without a detailed knowledge of the total energies involved is somewhat arbitrary, but does not alter the basic conclusions presented here.
It is seen that, whereas B-A premeltons are topological, B-B premeltons are nontopological. The formation of these two different classes of premeltons demonstrates the concept of a bifurcation, which has also led to a mechanism to understand the B- to A- structural phase transition (this is described in the main section).
Although B-A and A-B premeltons are moveable boundaries, B-B and A-A premeltons are not, experiencing "trapping potentials" that determine the likelihood they form at specific DNA regions (see below). B-B and A-A premeltons also nucleate other phase transitions, as discussed in the main section and in APPENDIX A.
It is evident that sequentially homogeneous DNA polymers are not adequate models to understand the properties of naturally occurring DNA because they lack sufficient information in their nucleotide base sequence to give rise to this site specificity. In the context of soliton models, this poses the problem of describing soliton behavior in the presence of locally altered potentials. A general theory for this has been developed (Fogel, M.B., Trullinger, S.E., Bishop, A.R. and J.A. Krumhansl (1976). Phys. Rev. Lett. 36, 1411-1414).
The theory shows that solitons either move nonuniformly or are trapped by locally favorable potentials. It remains to extend this theory to DNA structure to predict the localization of kink-antikink bound states at specific nucleotide base sequences. Because the kink-antikink bound state is multiple base pairs in extent (our calculations suggest this number to be about fifty), it may be that the effective trapping potential involves the recognition of an extended sequence rather than being determined by any single base pair energetics or its immediate neighbors.