Structural characterization of an intrinsically disordered protein complex using integrated small-angle neutron...

Area of Science:

• Biophysics
• Structural Biology
• Computational Biology

Background:

• Characterizing intrinsically disordered proteins (IDPs) and their regions (IDRs) is crucial for understanding protein function.
• Small-angle neutron scattering (SANS) with selective labeling is a powerful technique for studying dynamic protein structures.
• Experimental SANS data often requires advanced methods for disentangling complex conformational information.

Purpose of the Study:

• To develop and demonstrate an integrated computational and experimental method for elucidating the structural ensemble of a complex formed by two IDRs.
• To leverage selective deuterium labeling in SANS experiments to gain additional insights into protein structural ensembles.
• To validate molecular dynamics (MD) force fields and assess the utility of deep learning (DL) algorithms in structural analysis.

Main Methods:

• Utilized SANS experiments with full contrast and contrast matching, incorporating residue-specific deuterium labeling.
• Performed microsecond all-atom molecular dynamics (MD) simulations using four different molecular mechanics force fields.
• Employed an autoencoder-based deep learning (DL) algorithm to analyze and integrate data from SANS and MD simulations.

Main Results:

• Selective deuteration in SANS experiments provided valuable information for characterizing structural ensembles.
• The a99SB-disp and CHARMM36m force fields demonstrated the best agreement with experimental SANS and NMR data.
• The DL algorithm effectively differentiated between NMR and MD structures and presented an ensemble superior to single-force-field MD ensembles.

Conclusions:

• The integrated approach offers a novel and effective strategy for characterizing the structural ensembles of IDPs.
• The study identified three distinct conformational clusters within the studied IDR complex.
• The findings highlight the synergy between advanced experimental techniques, computational simulations, and machine learning for structural biology.