Experimental Inferential Structure Determination of Ensembles for Intrinsically Disordered Proteins

Area of Science:

• Biophysics
• Structural Biology
• Computational Chemistry

Background:

• Intrinsically disordered proteins (IDPs) lack stable structures, posing challenges for traditional structural determination.
• Nuclear Magnetic Resonance (NMR) spectroscopy provides valuable data (chemical shifts, J-coupling) for characterizing protein dynamics.
• Accurate back-calculation from structural models to experimental observables is crucial for ensemble refinement.

Purpose of the Study:

• To develop a robust Bayesian framework for determining the most probable structural ensemble of IDPs.
• To address limitations in existing methods by optimizing nuisance parameters and improving data integration.
• To enable rapid ranking of large conformational ensembles for IDPs.

Main Methods:

• A Bayesian approach integrating NMR chemical shifts and J-coupling data with their associated errors.
• Optimization of experimental and back-calculation nuisance parameters as random variables.
• Development of the Experimental Inferential Structure Determination (EISD) method with O(N) scaling for rapid ensemble ranking.

Main Results:

• The EISD method efficiently ranks large structural ensembles (tens of thousands of conformations).
• Application to folded and misfolded proteins highlights issues with Boltzmann-weighted priors.
• Identified poor back-calculation accuracy as a primary limitation in current IDP ensemble determination.
• Demonstrated that reducing back-calculation error uncertainty improves discrimination of IDP ensembles, using amyloid-beta peptide as an example.

Conclusions:

• The EISD method provides a powerful tool for analyzing NMR data and ranking structural ensembles of IDPs.
• Improving the accuracy of structure-to-observable back-calculation is critical for advancing IDP structural studies.
• This work facilitates more accurate characterization of dynamic protein structures.