Computational Modeling of PROTAC Ternary Complexes and Linker Design

Proteolysis targeting chimeras (PROTACs) are heterobifunctional ligands that promote targeted protein degradation (TPD) by reprogramming E3 ubiquitin ligase activity. By affixing an E3-recruiting moiety onto an existing target-binding warhead, many known inhibitors can now be repurposed as degraders of their protein targets. Historically, developing a PROTAC has been an empirical process, requiring extensive medicinal chemistry optimization to achieve efficient target degradation. A key hurdle has been identifying the specific chemical linker to use in tethering the two functional components of the PROTAC to one another (referred to as “linkerology”). Given that multiple E3 ligases can be used for building a PROTAC, coupled with a vast diversity of linker lengths and compositions, the challenge to explore the huge potential chemical space available in PROTAC design quickly becomes apparent. To address this, multiple computational approaches have recently been developed: these can be used to rapidly screen the vast chemical space of potential PROTACs for degrading a given target protein. These methods typically aim to model the structure of the PROTAC-induced ternary complex; formation of this complex is thought to be the key step in effective degradation. In this chapter, we summarize computational approaches that have proven effective for retrospective ternary complex modeling in benchmark experiments, and we describe emerging deep learning/artificial intelligence methods for de novo linker construction. Our perspective emphasizes the biophysical underpinnings of ternary complex formation and how these inform PROTAC design. In light of multiple PROTACs' rapid advance approaching and into the clinic, improved methods for designing effective degraders are expected to accelerate the development of chemical tools for research and new classes of therapeutics in the near future.