Accurate protein structure modeling using sparse NMR data and homologous structure information.

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Homologous structure information is crucial for X-ray crystallography but underutilized in NMR structure determination due to potential inaccuracies.
• Inaccurate evolutionary inferences or significant structural divergence can limit the utility of homologous data in NMR.

Purpose of the Study:

• To develop a robust method for modeling protein structures up to 225 residues using NMR data and homologous structures.
• To overcome limitations of using homologous information in NMR structure determination.

Main Methods:

• Combines backbone chemical shift data ((1)H(N), (13)C, (15)N) with distance restraints from homologous structures.
• Employs a physically realistic all-atom energy function for modeling.
• Validates models by comparing their energies and structural convergence against unrestrained calculations and experimental data.

Main Results:

• Accurate protein models are achieved by ensuring lower energies and structural convergence compared to unrestrained calculations.
• The method successfully models protein structures with remote homology, yielding backbone RMSDs of 1.2-1.9 Å compared to NMR ensembles.
• Achieves backbone RMSDs of 0.9-1.6 Å relative to X-ray structures for well-defined regions.

Conclusions:

• This approach enables accurate protein structure modeling using backbone chemical shift data without requiring side-chain assignments or extensive NOESY analysis.
• Facilitates robust protein structure determination even with limited or distant homology information.
• Offers a valuable tool for structural biologists and computational chemists.