AI-Guided Design of Cyclic Peptide Binders Targeting TREM2 Using CycleRFdiffusion and Experimental Validation

Abstract

Triggering receptor expressed on myeloid cells 2 (TREM2) plays a central role in regulating microglial function in the central nervous system and has emerged as a promising therapeutic target for Alzheimers disease. Despite advances in antibody-based therapeutics, small molecules and peptides capable of modulating TREM2 remain limited. Here, we present a cyclic peptide design pipeline that integrates CycleRFdiffusion, ProteinMPNN for sequence design, and HighFold for structural prediction and screening. Using the TREM2 structure as input, we generated and screened 1,500 peptide-target complexes, prioritizing four candidates that met structural and energetic criteria. Subsequent biophysical evaluation identified TP4 as a TREM2 binder, demonstrating consistent binding in spectral shift, microscale thermophoresis, and surface plasmon resonance. Pharmacokinetic profiling indicated that TP4 possesses favorable plasma stability and moderate metabolic stability, supporting its tractability for further optimization. This study establishes a generalizable framework for AI-driven cyclic peptide discovery and provides the first proof-of-concept demonstration of TREM2-targeted cyclic peptide binders.