{
  "id": 10836186,
  "title": "AI and Proteomics Accelerate VAV1 Molecular Glue Discovery",
  "url": "https://urgent.news/2026/09/30/ai-and-proteomics-accelerate-vav1-molecular-glue-discovery",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-30T02:06:22.000Z",
  "source": {
    "name": "GEN Biotechnology",
    "slug": "gen-biotechnology",
    "url": "https://www.genengnews.com/topics/drug-discovery/ai-and-proteomics-accelerate-vav1-molecular-glue-discovery/"
  },
  "original_language": "en",
  "account": "Researchers have harnessed the power of high-throughput proteomics and artificial intelligence (AI) to identify and optimize molecular glues that target VAV1, an immune-cell signaling protein associated with blood cancers and autoimmune diseases. Led by Jin Wang, PhD, of Baylor College of Medicine, the study published in Nature Communications details how the collaborative approach led to the discovery of potent VAV1 degraders.\n\nMolecular glues serve as matchmakers, bringing disease-linked proteins into contact with the cell's disposal machinery, marking them for elimination. Unlike traditional methods that merely inhibit protein functions, degradation through molecular glue removes the entire protein, potentially yielding a more comprehensive therapeutic effect. To discover compounds capable of degrading VAV1, the team screened a library of molecules using proteomics, which assesses thousands of proteins simultaneously. This unbiased analysis revealed several candidates, with NGT-201-12 showing promise by lowering VAV1 levels with minimal off-target effects.\n\nFurther investigation revealed that degradation relied on the proteasome and cereblon (CRBN), a crucial component of the cell's protein-degradation pathway. To streamline the discovery process, the researchers developed GluePlex, a computational workflow that merges AI-based protein-structure prediction with physics-based modeling. This approach modeled how VAV1, CRBN, and the molecular glue assemble, identifying a key region on VAV1—the SH3-2 domain—as essential for degradation. Experimental validation confirmed this prediction, pinpointing a specific surface loop on VAV1 acting as a degradation signal, or 'degron.'\n\nBy applying Free Energy Perturbation (FEP) to predicted ternary structures, the team established a method to correlate degradation potency with computational metrics, surpassing the limitations of standard docking techniques. This innovation facilitated the ranking of molecular analogs, even from weak initial binders. Modifications with halogen substitutions were found to restrict molecular flexibility and enhance degradation efficiency, resulting in NGT-201-18, a more potent degrader that formed a stronger complex with VAV1.\n\nIn primary human T cells, NGT-201-18 effectively reduced VAV1 levels and suppressed T-cell activation. While VAV1 emerged as the primary target, proteome-wide profiling also uncovered LIMD1, an off-target protein involved in canonical G-loop degradation, highlighting the versatility of the molecular glue. However, these promising compounds are currently in preclinical stages, and further research is necessary to evaluate their pharmacology, safety, selectivity, and efficacy in disease models. Despite this, the study presents a valuable starting point for developing therapies targeting VAV1-related autoimmune disorders and hematologic malignancies, as well as a novel discovery strategy leveraging AI, structural modeling, and proteomics.",
  "summary": "High-throughput proteomics combined with AI-based structural modeling helped researchers discover molecular glues that degrade VAV1. The approach could support new therapies for blood cancers and autoimmune diseases. The post AI and Proteomics Accelerate VAV1 Molecular Glue Discovery appeared first on GEN - Genetic Engineering and Biotechnology News .",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}