Refinement of NMR-determined protein structures with database derived distance constraints

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Protein structures determined by Nuclear Magnetic Resonance (NMR) spectroscopy often lack the detail and accuracy of X-ray crystallography.
• This underdetermination stems from insufficient distance data in NMR experiments, limiting applications like homology modeling and drug design.

Purpose of the Study:

• To develop and validate a computational approach for enhancing the accuracy of NMR-determined protein structures.
• To leverage existing databases of high-quality protein structures to derive additional distance constraints.

Main Methods:

• A survey of 462 NMR structures was conducted to analyze inter-atomic distance distributions.
• A computational method was applied to derive additional distance constraints from databases of known protein structures.
• Refinement was performed on 10 selected NMR structures using these derived constraints.

Main Results:

• Many inter-atomic distances in the surveyed NMR structures significantly deviated from their expected database distributions.
• Refinement of selected NMR structures, by constraining distances within high-probability ranges, led to significant improvements in structural accuracy.
• The computational approach demonstrated the potential to enhance the quality of NMR-derived protein models.

Conclusions:

• Computational enhancement using database-derived distance constraints is a viable strategy to improve NMR protein structures.
• This method can overcome limitations of traditional NMR data, expanding the utility of NMR structures in structural biology and drug discovery.
• The findings suggest a pathway to more accurate and reliable protein structure models from NMR data.