Generative Peptide Drug Design for AD/ADRD

About This Grant

PROJECT SUMMARY Even with small-molecule drugs and antibodies on the market, there remains a pressing need to develop more effective, affordable, and safer treatments for Alzheimer's Disease (AD) and related dementias (ADRD). To fill this need, this project enables generative artificial intelligence (AI) technologies to advance peptide drugs for AD and ADRD treatments. This approach is significant because peptide drugs can combine the key advantages of small-molecule drugs and antibodies, potentially leading to improvements in effectiveness, safety, biocompatibility, manufacturability, and delivery. However, peptide drugs face significant design challenges, particularly with limited permeability and stability. Our central hypothesis is that generative AI can address these fundamental challenges by designing more pharmaceutically relevant peptide candidates. With appropriate design and training, generative AI provides unmatched power to explore the vast chemical space of peptides and identify novel peptides with drug-like properties. Our goal is to develop an integrated platform that combines generative AI with cutting-edge high-throughput data generation and in vitro/in vivo assays for designing AD- targeting peptides. We have obtained strong preliminary data, including the development of generative AI models for peptide design (AMP-GAN and Pep-Diff), initial success in creating an innovative microdroplet-based peptide selection platform, and peptide candidates targeting AD/ADRD-related protein aggregates. We will create effective, transferable AI models for sequence- and structure-based peptide designs targeting AD and ADRD. These models build on our previous successes and are carefully designed to overcome the challenges of exploring the chemical complexity and to address the limitation of scarce AD-targeting peptide data. We will also develop multimodal AI models to predict peptide drug-like properties. Moreover, high-throughput selection methods and advanced mass spectrometry will produce extensive data to enhance generative AI models. Finally, in vitro and in vivo assays will be performed with the AI-generated peptides, providing insights for further enhancements of the AI models. The successful completion of this project is expected to provide new tools for academic and industrial researchers to discover peptides that intervene with various AD targets. Overall, our study will pave new paths for discovering peptide therapeutics for AD/ADRD and accelerate the transition from design to clinical application by shortening the design-development cycle. The resulting insights will be widely disseminated within the scientific community to advance AD/ADRD research and drug development, transforming the landscape of peptide-based therapeutics through generative AI-driven design for enhanced accuracy, diversity, and target selectivity. The methodologies and tools developed in this study will enable the discovery of novel peptide therapeutics, thus helping to meet the increasing demand for effective AD/ADRD treatments. The general platform technology developed for peptides inhibiting tau and α-synuclein aggregation will also apply to other less-studied AD/ADRD targets in the future.