结构蛋白质组学与人工智能助力PROTAC理性设计

Proximity-inducing drugs are an emerging class of protein interaction modulators, as many cellular processes rely on the coordinated activity of proteins to support cell growth, homeostasis, regulation and communication. This is most famously exemplified by proteolysis-targeting chimera (PROTAC). PROTACs are heterobifunctional compounds consisting of two ligands joined by a linker to bring together an E3 ligase and a protein of interest and induce ubiquitination and subsequent degradation. Dozens of PROTAC drugs are being investigated in clinical trials, most notably targeting androgen receptor (AR) and BTK. The efficacy of a PROTAC molecule is driven by ternary complex formation and stability, rather than relying solely on the binary affinity of a single molecule to a single protein. Therefore, the major challenge in rational development of PROTACs is deep understanding of the dynamics and functions of complex assembly. However, protein complexes and interactions have traditionally been hard to model and are not well represented in existing protein structure resources, both experimental and computational.In this work, we present the AIMS™ platform, which integrates structural mass spectrometry (MS) data and AI modeling tools. We apply our AIMS™ platform to multiple PROTAC systems, allowing us to improve the confidence of predicted structures compared to AI alone and support rational PROTAC design.