A systematic molecular dynamics study of nearest-neighbor effects on base pair and base pair step conformations and fluctuations in B-DNA

Abstract

It is well recognized that base sequence exerts a significant influence on the properties of DNA and plays a significant role in protein–DNA interactions vital for cellular processes. Understanding and predicting base sequence effects requires an extensive structural and dynamic dataset which is currently unavailable from experiment. A consortium of laboratories was consequently formed to obtain this information using molecular simulations. This article describes results providing information not only on all 10 unique base pair steps, but also on all possible nearest-neighbor effects on these steps. These results are derived from simulations of 50–100 ns on 39 different DNA oligomers in explicit solvent and using a physiological salt concentration. We demonstrate that the simulations are converged in terms of helical and backbone parameters. The results show that nearest-neighbor effects on base pair steps are very significant, implying that dinucleotide models are insufficient for predicting sequence-dependent behavior. Flanking base sequences can notably lead to base pair step parameters in dynamic equilibrium between two conformational sub-states. Although this study only provides limited data on next-nearest-neighbor effects, we suggest that such effects should be analyzed before attempting to predict the sequence-dependent behavior of DNA.

Moving beyond Watson–Crick models of coarse grained DNA dynamics

“…While the primary structure is typically constant, secondary and tertiary structures fluctuate under thermal energy and are modified by external forces, the action of enzymes, or the presence of a complementary strand. Since the time-scales characterizing these structural dynamics exceed the current capabilities of atomistic simulations, 1 their simulation requires a coarse-grained model. It is reasonable to contend that a minimal model of DNA should at least account for Watson-Crick base pairing and stacking, along with the chemical asymmetry of the backbone (i.e., a 5 -3 directionality).…”

The Role of Unconventional Hydrogen Bonds in Determining BII Propensities in B-DNA

“…The canonical state, referred to as BI, features ε/ζ in a trans/gauche-(t/g-) conformation, while the other state, BII, has ε/ζ in g-/t conformation. To determine BI/BII equilibrium in B-DNA, proton and Phosphate NMR experiments [4][5][6][7] , Molecular Dynamics (MD) simulations [8][9][10] , and data mining of crystal structures from databases [11][12][13] have being historically used as the preferred methods. Following initial observations based on crystal structures showing that BI/BII transitions were associated with base destacking and minor groove widening 14,15 , computer MD simulations have shed light on the influence of water and ion dynamics on the propensity of BI/BII states 8,9,16,17 .…”

Contributions of Sequence to the Higher-Order Structures of DNA

“…If the distributions of parameters in a large sample of known structures are Gaussian, the mean values will provide a good approximation of the equilibrium rest states and the widths of the distributions will provide an estimate of the elastic constants governing the spatial arrangements of neighboring basepairs. The original estimates of DNA sequence-dependent structural deformability, derived nearly 20 years ago from the few available well-resolved, DNA-containing crystal structures (6), have been supplemented over the years by analyses of larger, better-resolved, nonredundant structural data sets (29) and by massive atomic-level simulations with the capability to examine the influence of sequence content (e.g., nonadjacent nearby residues) on double-helical structure and deformability (30)(31)(32).Until very recently, the predicted effects of sequence on DNA structure failed to account for many key features of the double-helical molecule, including the basepair twist-a parameter that can be reliably measured in solution, gels, and the solid state (33-37). In particular, atomic-level simulations based on some of the most widely used nucleic-acid force fields predicted the twist angle of a generic DNA basepair step (a step generated by averaging the twist angles over all possible combinations of adjacent basepairs) to be~1.2 lower on average than the corresponding values compiled from experiments on mixed-sequence DNA (Fig.…”

“…If the distributions of parameters in a large sample of known structures are Gaussian, the mean values will provide a good approximation of the equilibrium rest states and the widths of the distributions will provide an estimate of the elastic constants governing the spatial arrangements of neighboring basepairs. The original estimates of DNA sequence-dependent structural deformability, derived nearly 20 years ago from the few available well-resolved, DNA-containing crystal structures (6), have been supplemented over the years by analyses of larger, better-resolved, nonredundant structural data sets (29) and by massive atomic-level simulations with the capability to examine the influence of sequence content (e.g., nonadjacent nearby residues) on double-helical structure and deformability (30)(31)(32).…”

“…(A) Average sequence-dependent values of the twist angle between successive basepairs grouped by chemical class (pyrimidine-purine, purine-purine, and purine-pyrimidine). Gold, trends deduced from early biophysical studies of DNA in gels and in solution (35); violet, values accumulated over time in crystal structures of protein-bound DNA (6,29) and a mix of ligand-bound and free DNA (31); green, values predicted with increasingly improved atomic force fields (30)(31)(32). Note that the twist of a generic, sequence-averaged DNA step (MN) found in the most recent data corresponds to a helix with a nucleotide repeat close to the~10.5 bp/turn repeat of mixed-sequence DNA (dashed line) (33,34,36) and that large changes in twist at pyrimidine-purine steps underlie this improvement.…”