Local elasticity of strained DNA studied by all-atom simulations

Genomic DNA is constantly subjected to various mechanical stresses arising from its biological functions and cell packaging. If the local mechanical properties of DNA change under torsional and tensional stress, the activity of DNA-modifying proteins and transcription factors can be affected and regulated allosterically. To check this possibility, appropriate steady forces and torques were applied in the course of all-atom molecular dynamics simulations of DNA with AT- and GC-alternating sequences. It is found that the stretching rigidity grows with tension as well as twisting. The torsional rigidity is not affected by stretching, but it varies with twisting very strongly, and differently for the two sequences. Surprisingly, for AT-alternating DNA it passes through a minimum with the average twist close to the experimental value in solution. For this fragment, but not for the GC-alternating sequence, the bending rigidity noticeably changes with both twisting and stretching. The results have important biological implications and shed light upon earlier experimental observations.

💡 Research Summary

This paper presents a systematic all‑atom molecular dynamics (MD) investigation of how external mechanical stresses—steady tensile forces and torques—affect the local elastic properties of short DNA fragments with alternating AT and GC sequences. The authors modeled 14‑base‑pair (bp) duplexes, which correspond to a full helical turn, and applied “steady‑load” forces and torques that are distributed over selected atom groups so that the net external force and torque on the system are zero. This approach mimics the physiological range of supercoiling and tension experienced by DNA in vivo.

Simulations employed the AMBER98 force field, TIP3P water, and PME electrostatics, with an internal‑coordinate MD (ICMD) scheme allowing a 0.01 ps timestep. Each stress condition was sampled for a total of 164 ns (≈2.15 million frames), providing robust statistics. Elastic parameters were extracted using the worm‑like chain (WLC) formalism: stretching persistence length (l_s), torsional persistence length (l_t), and bending persistence length (l_b). Three distinct definitions of DNA length were examined—end‑to‑end distance (L′), the 3‑D zig‑zag path through reference‑frame origins (L′′), and the sum of local rise values from 3DNA (L′′′). The authors argue that L′ best corresponds to experimental observables because, for very short DNA, length fluctuations are dominated by angular flattening rather than true contour extension.

Key findings: (1) The apparent stretching rigidity increases both with applied tension and with increasing twist, evidencing a twist‑stretch coupling. Notably, the sign of this coupling depends on the length definition; using L′′ or L′′′ yields a coupling opposite to that measured experimentally, whereas L′ reproduces the correct sign. (2) Torsional rigidity is essentially insensitive to stretching but varies strongly with applied torque, and the response is sequence‑dependent. For the AT‑alternating fragment, l_t reaches a pronounced minimum when the average twist is close to the canonical ~34° per base pair observed in solution, suggesting a mechanical “sweet spot” that could be exploited by cellular processes. (3) Bending rigidity shows a marked dependence on both twist and tension for the AT fragment but remains nearly constant for the GC fragment, reflecting differences in base‑stacking and hydrogen‑bonding patterns.

Biologically, these results imply that modest supercoiling or transcription‑generated forces can modulate local DNA elasticity in a sequence‑specific manner, potentially providing an allosteric mechanism for transcription factor binding and gene regulation. The study also clarifies why earlier experimental measurements of DNA torsional stiffness have been inconsistent: they likely probed different effective length measures and did not account for twist‑stretch coupling. By demonstrating that local elastic parameters are non‑harmonic and sequence‑specific under physiologically relevant stresses, the work bridges the gap between coarse‑grained polymer models and the molecular details needed to understand DNA‑protein interactions in vivo.