Targeting Aggregating Proteins with Language Model-Designed Degraders

Rio Watson, Kishan Patel, Tong Chen, Pranam Chatterjee

Track: Biology: datasets and/or experimental results

Nature Biotechnology: Yes

Keywords: Peptide design, language models, cancer, neurodegenerative disease

TL;DR: Language model-designed peptides enable degradation of aggregating proteins, offering a new TPD strategy for cancer and neurodegenerative diseases.

Abstract: Protein aggregation drives several neurological diseases and pediatric cancers, yet current inhibitors fail to directly target aggregating proteins or provide long-term disease modification. Advances in generative artificial intelligence (AI), particularly protein language models (pLMs), have enabled the design of peptide binders for disordered and oncogenic targets. Using these models, we designed peptide binders for mutant GFAP (Alexander Disease) and PAX3::FOXO1 (Alveolar Rhabdomyosarcoma). When fused to E3 ubiquitin ligase domains, these binders selectively degrade their targets, and in the case of GFAP, reducing aggregation. Our results demonstrate that pLM-designed peptide-guided degraders provide a powerful strategy for treating aggregation-driven diseases.

Anonymization: This submission has been anonymized for double-blind review via the removal of identifying information such as names, affiliations, and identifying URLs.

Presenter: ~Rio_Watson1

Format: Yes, the presenting author will attend in person if this work is accepted to the workshop.

Funding: No, the presenting author of this submission does *not* fall under ICLR’s funding aims, or has sufficient alternate funding.

Submission Number: 30