Drug Development · global
Clearing an immune signaling protein: Insilico Medicine selects AI-assisted VAV1 degrader candidate
The oral molecular glue ISM9546 aims to combine immune modulation with low brain exposure and has shown signs of activity in cells and animal models of colitis. However, short-term toxicology findings and tissue improvements remain far from the long-term safety evidence needed for chronic autoimmune diseases.
Can a drug calm an uncontrolled immune response by having cells directly clear a signaling protein? Insilico Medicine announced on October 13 that it had selected ISM9546, an AI-assisted oral VAV1 molecular glue degrader, as a preclinical candidate, with development targeting inflammatory bowel disease and rheumatoid arthritis. This development gives an immune modulation strategy a concrete candidate molecule, although the current evidence of efficacy still comes from cell and animal experiments.
VAV1 is involved in immune cell activation, proliferation, and cytokine secretion, making it a potential target for intervention in immune responses. According to the company, the protein lacks a typical small-molecule binding pocket, posing difficulties for conventional drug design. The molecular glue approach seeks to have the target protein removed by the cell’s own clearance machinery, thereby altering downstream signaling.
AI’s role in this development effort was to help refine molecular structures and properties. Using the Chemistry42 platform, the research team designed molecular fragments based on protein and ligand structures and interactions, while optimizing degradation activity and reducing brain exposure. ISM9546 was then selected through multiple rounds of synthesis, testing, and analysis. This was an iterative process of computational design and experimentation; the announcement did not provide comparative data sufficient to quantify AI’s contribution relative to other methods.
According to results released by the company, ISM9546 demonstrated subnanomolar VAV1 degradation activity in multiple cell assays. Following oral administration, the disease activity index in an animal model of acute colitis declined within days, and histopathology scores improved by approximately 60%, accompanied by reduced inflammatory cell infiltration and intestinal damage. These results support further research, but changes in pathology scores cannot be directly translated into remission rates in human patients.
Another design priority was to have the drug act primarily in tissues outside the central nervous system. The company said tissue distribution studies showed low brain exposure; 14-day toxicology assessments in rodents and nonhuman primates showed good tolerability, and proteomic analysis did not detect depletion of non-target proteins. However, low brain exposure is only one indicator in assessing nervous system risks, and short-term studies cannot capture all the risks of long-term treatment.
This evidence comes primarily from a company announcement. The material carried by PR Newswire was also issued by Insilico Medicine and does not constitute a separate, independent validation. The announcement did not fully disclose animal study sample sizes, doses, statistical methods, or the scope of detection in the proteomic analysis. Although rheumatoid arthritis was listed as a development target, the specific descriptions of efficacy focused on intestinal inflammation models.
Selecting a preclinical candidate means the company has decided to commit resources to further development; it does not mean approval has been granted to begin human trials. For autoimmune diseases requiring long-term control, the questions ahead include whether effective doses can sustain disease suppression, whether immune modulation increases infection risk, and whether low brain exposure can be maintained in humans. The value of ISM9546 will ultimately have to be established through more comprehensive toxicology and human studies.