Design of Peptide-Guided Protein Degraders with Structure-Agnostic Language Models

• DOI: 10.5281/zenodo.7542327
• Publication Date: 2022-10-06
• Authors: Brixi, Garyk; Ye, Tianzheng; Palepu, Kalyan; Li, Xinning; Hong, Lauren; Yudistyra, Vivian; Bhat, Suhaas; Christopher, Jayani; Monticello, Connor; Lopez-Barbosa, Na...; Vincoff, Sophia; Liu, Tong; Peterson, Lillian; Zhao, Lin; Zhang, Sue; Ponnapati, Manvitha; Tysinger, Emma; Stan, Teodora; Koseki, Sabrina; DeLisa, Matthew P.; Chatterjee, Pranam

Targeted protein degradation of pathogenic proteins represents a powerful new treatment strategy for multiple disease indications. Unfortunately, a sizable portion of these proteins are considered “undruggable” by standard small molecule-based approaches, including PROTACs and molecular glues, largely due to their disordered nature, instability, and lack of binding site accessibility. As a more modular, genetically-encoded strategy, designing functional protein-based degraders to undruggable targets presents a unique opportunity for therapeutic intervention. In this work, we integrate pre-trained language models with protein interaction databases to devise a unified, sequence-based framework to develop peptide-guided degraders without structural information. We create a Structure-agnostic Language Transformer & Peptide Prioritization (SaLT&PepPr) module that efficiently selects peptides from natural protein interaction interfaces for downstream screening. We experimentally fuse SaLT&PepPr-derived peptides to E3 ubiquitin ligase domains and reliably identify candidates exhibiting robust intracellular degradation of diverse pathogenic targets in human cells, including those with minimal structural information. We further show that our peptide-guided degraders have negligible off-target effects via whole-cell proteomics and demonstrate degradation of endogenous β-catenin and subsequent downregulation of Wnt signaling in cellular models of colorectal cancer. In total, by integrating language model-based design with our unique protein-targeting architectures, this study lays the foundation for programmable and broad-ranging proteome editing applications.<br/>Other funding sources: This work was supported by the National Science Foundation (grant CBET-1605242 to M.P.D.), the Defense Threat Reduction Agency (grant HDTRA1-20-10004 to M.P.D.), and institutional start-up funds from Duke University (to P.C.).<br/>