Computational Design of High-Affinity Binders for Intrinsically Disordered Protein Regions Using Deep Learning and Structural Induction

A groundbreaking computational method has been developed to design proteins that specifically bind to intrinsically disordered regions (IDRs)—flexible protein segments lacking a fixed structure. Unlike traditional approaches that struggle with such dynamic targets, this innovation turns their flexibility into an advantage. The designed proteins guide IDRs into specific, binding-competent shapes, enabling highly specific and strong interactions.

The pipeline combines physics-based modeling with deep learning (notably RFdiffusion) to build a library of protein scaffolds tailored to recognize diverse disordered peptide conformations. Using this approach, researchers successfully designed high-affinity binders for 39 out of 43 diverse IDRs, including therapeutically important ones related to cancer and GPCR signaling.

These binders demonstrated remarkable specificity, even among similar sequences, and functioned effectively in living cells—for example, by enriching rare proteins for analysis or altering protein localization. Structural studies, including co-crystal and NMR data, validated that the binders induce and stabilize specific conformations of their disordered targets, confirming the method’s precision and effectiveness.

More reading

Design of intrinsically disordered region binding proteins

DOI: 10.1126/science.adr8063

SHARE

Related Post

Fascinating advances in time-resolved structural biology

Researchers from the University of Oxford, SLAC National Accelerator Laboratory, and Lawrence...

From A Single Protein to a Revolution in Neuroscience

This year’s Nobel Prize in Physiology or Medicine recognizes Karl Deisseroth, Peter...

GOOSE: Understanding and Designing Intrinsically Disordered Protein

We often read papers about the three-dimensional (3D) structures of proteins, particularly...