Leveraging high-throughput proteomics and AI-based protein folding to accelerate VAV1 molecular glue discovery

Abstract

This study reports a series of CRBN molecular glues that selectively induce proteasomal degradation of VAV1, a hematopoietic-specific signaling protein and key target in hematological malignancies and autoimmune diseases. Using unbiased global proteomics, we identify phenyl-glutarimide derivatives (e.g., NGT-201-12) as effective VAV1 degraders that engage its C-terminal SH3 domain (SH3-2). We elucidate a non-canonical RT-loop degron (RDxS motif, residues 796–799) in SH3-2, distinct from previously characterized G-loop degrons, supported by the physics- and AI-driven GluePlex workflow and site-directed mutagenesis. Applying Free Energy Perturbation (FEP) to predicted ternary structures yields cooperativity metrics that correlate with degradation potency, overcoming limitations of standard docking and enabling prospective ranking of analogs—even from weak initial binders. We further show that conformational restriction via halogen substitution (e.g., NGT-201-18) potentiates degradation, a principle rationalized by DFT calculations, while comprehensive SAR studies chart a roadmap for next-generation degraders. Dose-response proteomics confirms VAV1 as the primary target and reveals LIMD1, bearing a canonical G-loop, as an off-target, showing that one glue can engage disparate degrons. These findings expand the landscape of CRBN neosubstrate recognition and offer a therapeutic strategy for targeting VAV1. VAV1 is a blood-cell signaling protein linked to leukemia and autoimmune disease. Here, the authors discover CRBN molecular glues that degrade VAV1 via a non-canonical degron, using proteomics and AI-based structural modeling to guide the design of more potent degraders.