A Combined Wormlike-Chain and Bead Model for Dynamic Simulations of Long Linear DNA

✍ Scribed by Hongmei Jian; Alexander V. Vologodskii; Tamar Schlick

• Publisher: Elsevier Science
• Year: 1997
• Tongue: English
• Weight: 414 KB
• Volume: 136
• Category: Article
• ISSN: 0021-9991
• DOI: 10.1006/jcph.1997.5765

✦ Synopsis

sured values for DNA chains (of 367, 762, 1010, 2311 base pairs) shows excellent agreement as well. This lends confidence to the A carefully parameterized and tested simulation procedure for predictive ability of our model and sets the groundwork for further studying the dynamic properties of long linear DNA, based on a work on circular DNA. We conclude with results of such a predictive representation that combines features of both wormlike-chain and measurement, the autocorrelation time, for the end-to-end distance bead models, is presented. Our goals are to verify the model paramand the bending angle as a function of DNA length. Rotational eters and protocols with respect to all relevant experimental data diffusion measurements for different DNA lengths (300 to 2311 base and equilibrium simulations, to choose the most efficient algopairs) are also presented. ᮊ 1997 Academic Press rithms, and to test different approximations that increase the speed of the computations. The energy of the linear model chain includes stretching, bending, and electrostatic components. Beads are asso-

The large-scale dynamic motions of double helical DNA stretching constant) and realistic modeling of the DNA (i.e., small are important for many biological processes, from protein/ deviations of the input contour length); the bead hydrodynamic radius is set to yield agreement with known values of the transla-DNA interactions to higher-order DNA folding and retional diffusion coefficient. By comparing results from both a firstcombination. Several approaches have been developed and a second-order Brownian dynamics algorithm, we find that the during the past decade to model DNA dynamics on the two schemes give reasonable accuracy for integration timesteps in basis of low-resolution models [3, 4, 6, 7, 24, 29, 31, 32, the range 200-500 ps. However, the greater accuracy of the second-

35]. Such approaches allow simulation of slow motions in

order algorithm permits timesteps of 600 ps to be used for better accuracy than the 300 ps used in the first-order method. We develop long DNA molecules that are not possible to capture with a more efficient second-order algorithm for our model by eliminatstandard all-atom simulations, unfortunately limited to ing the auxiliary calculations of the translational diffusion tensor several dozen residues. However, modeling slow motions at each timestep. This treatment does not sacrifice accuracy and in large DNA molecules remains a challenge. In particular, reduces the required CPU time by about 50%. We also show that it is difficult to simulate slow processes in double-stranded an appropriate monitoring of the chain topology ensures essentially no intrachain crossing. The model details are assessed by compar-supercoiled DNA where torsional rotation of the chain ing simulation-generated equilibrium and dynamic properties with segments is important. This is a broad objective of the results of Monte Carlo simulations for short linear DNA (300, 600 simulation protocol developed here. In this paper, we focus base pairs) and with experimental results. Very good agreement is on linear DNA. In a second work [20] we continue to treat obtained with Monte Carlo results for distributions of the end-toclosed circular DNA, where additional terms are required, end distance, bond lengths, bond angles between adjacent links, and translational diffusion measurements. Additionally, compari-and to study biological questions. In particular, our goals son of translational diffusion coefficients with experimentally-meahere are to verify the model parameters and protocols with respect to all relevant experimental data and equilibrium simulations, to choose the most efficient algorithms, and