Abstract
In magnetic tweezers experiments, we observe that torsional DNA buckling rates and transition state distances are insensitive to base-pairing defects. This is surprising because defects are expected to kink DNA and lower the energy of a localized loop. Nonetheless, base-pairing defects lead to pinning of buckled structures at the defects, which may be important for DNA repair in vivo. We find that the decrease in entropy from pinning roughly balances the decrease in bending energy, explaining why defects have little effect on buckling rates. Our data are generally consistent with elastic rod theory, which predicts that the transition state structure for torsional buckling is a localized wave with a specific shape ("soliton"). The transition state soliton decays to a metastable looped intermediate ("curl") that is separated from the final, fully buckled state by a second, low energy barrier. DNAs with base mismatch defects buckle at lower torque, where elastic rod theory predicts the loop structure is more stable, and manifest an intermediate buckling structure consistent with such a loop. We estimate that, under our high force, high salt experimental conditions, the soliton barrier is approximately 10 kB T and, to reach this transition state from the unbuckled state, the system torque instantaneously decreases by approximately 1 pN·nm for DNA with or without a small defect.