Physics-guided design of intrinsically disordered proteins

Area of Science:

• Biochemistry
• Computational Biology
• Molecular Biology

Background:

• Intrinsically disordered protein regions (IDPs) lack stable 3D structures, exhibiting conformational plasticity.
• This plasticity enables diverse functions but challenges rational protein design.
• Physics-based models aid in predicting IDP properties from sequence and interactions.

Purpose of the Study:

• To develop a machine learning framework for de novo design of IDPs.
• To enable IDPs to perform complex signal processing and form tailored biomolecular condensates.
• To engineer IDPs for specific localization within biological condensates in living cells.

Main Methods:

• Utilized a machine learning framework to invert physics-based models for IDP design.
• Programmed IDPs to sense and respond to biophysical cues, enabling signal processing.
• Engineered multicomponent IDP mixtures to control emergent condensate properties.
• Demonstrated IDP design for selective partitioning in cellular condensates.

Main Results:

• Designed IDPs capable of complex signal processing, including threshold detection and Boolean logic.
• Engineered IDP mixtures exhibiting tunable condensate properties like layering and RNA-dependent remodeling.
• Successfully demonstrated designed IDPs that selectively enter or leave cellular condensates.

Conclusions:

• The developed framework offers a scalable strategy for designing IDPs with specific ensemble-derived and collective properties.
• This approach advances the engineering of dynamic biomolecules for complex cellular functions.
• Enables precise control over protein behavior and interactions within cellular environments.