Programmable Protein Stabilization with Language Model-Derived Peptide Guides

Area of Science:

• Biochemistry
• Molecular Biology
• Drug Discovery

Background:

• Dysregulated protein degradation is implicated in diseases like cancer and neurodegeneration.
• Targeted protein stabilization (TPS) is a therapeutic strategy, but current methods have limitations.
• Existing TPS modalities use limited chemical linkers and warheads, hindering application to new or
• undruggable
• targets.

Purpose of the Study:

• To engineer a novel, modular system for targeted protein stabilization (TPS).
• To expand the range of druggable targets for protein stabilization therapies.
• To develop a programmable, CRISPR-analogous strategy for therapeutic intervention.

Main Methods:

• Engineered deubiquibodies (duAbs) by fusing computationally designed peptides to the OTUB1 deubiquitinase catalytic domain.
• Validated duAb efficacy in stabilizing exogenous and endogenous proteins in human cells.
• Demonstrated modularity by designing new target-binding peptides using generative language models.

Main Results:

• Deubiquibodies (duAbs) effectively stabilized target proteins in a deubiquitinase-dependent manner.
• The duAb system demonstrated modularity, enabling stabilization of diverse targets including p53, WEE1, and PAX3::FOXO1.
• Successfully stabilized key tumor suppressors and disordered oncoproteins, expanding TPS applicability.

Conclusions:

• The engineered deubiquibody (duAb) system provides a versatile and programmable platform for targeted protein stabilization (TPS).
• This approach overcomes limitations of small-molecule-based TPS, offering a new strategy for diseases driven by protein dysregulation.
• Deubiquibodies represent a significant advancement in developing novel therapeutics for challenging diseases.