A team comprising scientists from Baylor College of Medicine and the tech company SandboxAQ have developed a fragment-based method for discovering molecular glues. They say this workflow, published in Nature Communications on Sept. 12, is a less expensive and more accessible way to find the sticky small molecules that bring two proteins together (DOI: 10.1038/s41467-026-77657-z).
The Baylor team started by using high-throughput proteomics to screen a library of phenyl-glutarimide-focused fragments for degradation activity of VAV1, an immune-related signaling protein. The screening process analyzed changes in cellular proteomes after cells were treated with the fragments, and this screening process yielded two molecules.
Those two molecules, dubbed compounds 11 and 12, were initially fairly weak as degraders. But the goal wasn’t to find the best-in-class VAV1 degraders immediately; the strategy was to scan a small library for any kind of hit that could be developed into better glues down the line.
That’s where SandboxAQ came in. It modeled the interactions between the ligands and the VAV1 proteins, giving scientists insight on how to further modify the compounds to make them better glues.
“The reason we like to start with the fragments is because with smaller molecules, it’s relatively easy to cover the chemical space. Then you find weak hits. Then you start to evolve them,” says Jin Wang, the director of the Center for NextGen Therapeutics at Baylor. “I would say this is a lot more cost efficient and more feasible.”
“We have a proof point at least that we can enable fragment-based drug discovery for molecular glues, and that is quite exciting,” says Andrea Bortolato, the vice president of drug discovery at SandboxAQ.
SandboxAQ is an artificial intelligence software company that spun out of Alphabet, Google’s parent company. This isn’t the company’s first foray into AI-guided drug discovery; it has existing partnerships with AstraZeneca, Sanofi, and the University of California, San Francisco.
