Researchers at Baylor College of Medicine have combined high-throughput proteomics with AI-driven structural modeling to discover molecular glues that degrade VAV1, an immune-cell signaling protein implicated in blood cancers and autoimmune disease. The study, published in Nature Communications, was led by Jin Wang, PhD, director of Baylor's Center for NextGen Therapeutics, with first and co-corresponding author Hanfeng Lin, PhD, a postdoctoral researcher in Wang's laboratory.

The work matters for two distinct reasons. It delivers a series of VAV1-targeting molecular glue degraders, including an optimized analog with measured activity in primary human T cells. And it demonstrates a computational-plus-proteomics workflow — GluePlex — that any discovery group can apply to degrade targets that lack experimental ternary structures.

"Many scientists are increasingly exploring a new way to treat disease: instead of blocking harmful proteins, they aim at eliminating them entirely," Wang said. Molecular glues recruit disease-linked proteins to the cell's disposal machinery, and because degradation removes the whole protein rather than inhibiting one function, the approach could yield a more complete therapeutic effect.

The team first screened a compound library using high-throughput proteomics, a technology that assesses thousands of proteins simultaneously. "This unbiased analysis revealed a series of compounds, including NGT-201-12, that caused VAV1 levels to drop while affecting relatively few other proteins," Lin said. Follow-up studies showed the degradation depended on the proteasome and on cereblon (CRBN), a component of the cell's protein-degradation pathway.

From screening hits to structure-guided optimization

The researchers then built GluePlex, a computational workflow that integrates AI-based protein-structure prediction with physics-based modeling. Without an experimental structure of the ternary complex, GluePlex modeled how VAV1, CRBN, and the glue assemble.

"The model identified a specific region of VAV1, known as the SH3-2 domain, as being essential for degradation. Experimental tests confirmed the prediction and pinpointed the exact spot the glue uses: a small surface loop on VAV1 that acts as a degradation signal, or 'degron,'" Lin said. Notably, this loop differs from the degradation signals commonly associated with cereblon-targeting molecular glues — a finding that widens the known engagement space for CRBN-recruiting chemistry.

The workflow's ranking layer is Free Energy Perturbation (FEP) applied to predicted ternary structures. "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," the authors wrote. For teams weighing where to spend computational budget, that claim is worth scrutiny: cooperativity metrics that rank analogs prospectively, starting from weak binders, would shorten design-make-test cycles that docking alone leaves ill-resolved.

Structure-guided optimization produced results. Adding halogen substitutions restricted molecular flexibility and improved degradation efficiency, yielding NGT-201-18, a more potent degrader that formed a stronger degradation complex. In primary human T cells, NGT-201-18 reduced VAV1 levels and suppressed T-cell activation.

Off-target degradation argues for proteome-wide profiling

Dose-response proteomics identified VAV1 as the principal target but also revealed degradation of LIMD1, an off-target carrying a canonical G-loop degron. The result carries a practical lesson for portfolio decisions: a single molecular glue can engage structurally distinct degrons, so unbiased proteome-wide profiling remains necessary even when a compound appears selective in narrower assays.

The caveats are clear. The compounds remain preclinical, and the study does not yet address pharmacology, safety, selectivity, or activity in disease models; those studies are still ahead. The dataset's cellular evidence comes from primary human T cells, and no animal or clinical data exist yet.

Even so, the work provides a starting point for therapies aimed at VAV1-driven autoimmune disorders and hematologic malignancies, plus a transferable discovery strategy. "This work introduces a series of VAV1-targeting molecular glues and, just as importantly, shows how artificial intelligence, structural modeling and proteomics can work together at the earliest stage of a project," Wang said. Whether GluePlex generalizes to other ternary systems without experimental structures will be the next test for groups adopting the approach.