SynGlue: AI-Driven Designer for Clinically Actionable Multi-Target Therapeutics

The rational design of multi-targeting compounds and targeted protein degraders, such as PROTACs, remains a significant challenge in drug discovery. Here, we introduce SynGlue, a modular generative AI platform that integrates fragment-based reasoning with data-driven and structure-guided approaches to design multi-target therapeutics. Its flexible four-module architecture enables the generation of chemically diverse, target-specific, and polypharmacologically optimized compounds. The platform supports user-defined scoring functions and external docking tools, offering an end-to-end pipeline from target structure to synthesizable candidate. Using this platform, we designed and synthesized novel BRD4-VHL PROTACs featuring previously unreported and uniquely structured warheads. In vitro profiling confirmed potent nanomolar growth inhibition across multiple cell lines, driven by selective BRD4 degradation and subsequent c-Myc downregulation. In vivo , the lead compound established favorable pharmacokinetics, pharmacodynamics, and tolerability profiles. Importantly, this lead PROTAC demonstrated significant tumor burden reduction in mouse xenograft models via its targeted degradation mechanism. In summary, this work validates SynGlue as a powerful generative AI platform for the rapid and effective design of multi-targeting therapeutics with promising preclinical anti-cancer activity.