Critical comments on the one recent DNA bending model and its fit to electrophoretic data

The approach of Calladine, Drew and McCall (J. Mol. Biol 201, 127-137, 1988) for the investigation of DNA curvature has been analyzed. The authors relate their bending parameters to the experimental data on the basis of a new method of predicting electrophoretic anomalies from the trajectory of helical axis. This method of treating the DNA superhelix as a rigid body seems to be hardly applicable to the cases when superhelical turn is longer than or comparable to persistence length of straight DNA (approximately 150 bp). The extrapolation of experimental curves to plateau, done in a number of cases, seems somewhat arbitrary. Besides, alternative angle sets (e.g., with the roll angle on the AA step not equal to zero) having the same or even better fit to experimental data can be found even within the framework of the proposed approach. Another serious drawback of the model is its inability to account for new data (H.-S. Koo and D.M. Crothers. PNAS 85, 1763-1767, 1988), the predicted values of relative anomaly on (N8A6N5A6N4A6N7) multimers amounting to thousands.