Multiscale Computational Framework for the Liquid-Liquid Phase Separation of Intrinsically Disordered Proteins.

Area of Science:

• Biophysics
• Computational Biology
• Molecular Biology

Background:

• Intrinsically disordered proteins (IDPs) reversibly assemble into membraneless organelles (MLOs) via liquid-liquid phase separation (LLPS).
• MLOs regulate cellular functions by concentrating molecules but their dynamic assembly is complex and hard to predict.
• Understanding LLPS is crucial for comprehending cellular organization and function.

Purpose of the Study:

• To develop a multiscale computational model for predicting the LLPS behavior of IDPs.
• To simplify the representation of IDPs for efficient simulation.
• To use the Fused In Sarcoma (FUS) protein as a model system.

Main Methods:

• Represented FUS protein as a chain of 'stickers' (Low-Complexity Aromatic-Rich Kinked Segments - LARKS) and 'spacers'.
• Estimated sticker interaction energies using molecular docking and all-atomistic molecular dynamics (AA-MD).
• Employed coarse-grained (CG) modeling and Monte Carlo (MC) simulations with a novel acceptance criteria based on discretized pair potential distributions.

Main Results:

• Successfully modeled FUS protein assembly and disassembly dynamics using the multiscale framework.
• Demonstrated the model's ability to capture the dynamic nature of LLPS through radial distribution functions (RDFs).
• The model efficiently predicts LLPS behavior by considering binding energy states and their probabilities.

Conclusions:

• The developed multiscale computational framework offers an economical and efficient method for predicting IDP LLPS.
• This approach simplifies complex protein interactions, facilitating the study of MLO formation.
• The model provides valuable insights into the spatiotemporal regulation mechanisms within living cells.