Bending and curvature calculations in B-DNA

Area of Science:

• Molecular Biology
• Biophysics
• Computational Biology

Background:

• DNA sequence influences its three-dimensional structure.
• Understanding DNA bending is crucial for processes like DNA packaging and protein binding.
• Existing DNA bending models vary in their assumptions about sequence-dependent flexibility.

Purpose of the Study:

• To develop a computational tool (BEND) for analyzing DNA local bending and curvature.
• To evaluate and compare the performance of six different DNA bending models.
• To identify a model that accurately predicts DNA curvature across diverse sequences, including A-tracts and GGGCCC regions.

Main Methods:

• Developed the BEND program to calculate DNA bending based on user-defined models specifying twist, roll, and tilt.
• Applied BEND to six distinct DNA bending models: two non-A-tract, three A-tract, and one junction model.
• Validated model predictions against experimental data from gel retardation and cyclization kinetics, and structural data from X-ray crystallography.

Main Results:

• All six evaluated models successfully predicted experimental A-tract curvature.
• Only the new nucleosome positioning model accurately predicted curvature in GGGCCC-containing sequences.
• The new model demonstrated local bending in mixed sequences, strong bends at GGC sites, and rigidity in A-tracts.

Conclusions:

• The new nucleosome positioning model offers superior accuracy in predicting DNA curvature compared to other models.
• This model integrates solution and structural data, providing a more comprehensive understanding of DNA sequence-structure relationships.
• The BEND program serves as a valuable tool for analyzing DNA bending and testing various sequence-dependent flexibility models.