Algorithm and software for predicting the structure and degrading efficiency of PROTACs

The use of PROteolysis TArgeting Chimeras (PROTACs) has become an important technology for modulating a target protein by degradation. PROTACs are heterobifunctional molecules that connect a ligand bound to a protein of interest (POI) to an E3 ubiquitin ligase (E3) recruiting ligand with an optimal linker. Degradation is initiated when PROTAC promotes the POI and the E3 ligase to form a ternary complex. Since the two proteins are expected to favorably interact with each other and also with a flexible linker, the design of the optimal linker that yields the desired POI-PROTAC-E3 complex is a challenge, and current computational tools provide limited help. We address three important problems leading to improved PROTAC design. First, we have recently developed and tested an effective protocol for predicting high accuracy models of POI-PROTAC-E3 ligase ternary complexes. Our Aim 1 is to implement the protocol as an easy-to-use computer program for licensing. The method is based generating two large ensembles (“clouds”) of half-linker conformations, one linked to the warhead bound to the POI and the other to the “anchor” ligand bound to the E3 ligase, and use a modified version of our efficient protein-protein docking program PIPER to directly sample conformations that have a favorable relative orientation of the two proteins and in which the end points of some half-linkers are close to each other. After local refinement, each resulting conformation yields a feasible structure of the ternary complex, including the linker. After removing poses that do not assure accessibility to ubiquitination sites, the new method consistently generates high accuracy models among the top few predictions. It was shown that in most cases the best model can be selected by clustering a set of models generated for different PROTACs with the same E3 ligase and target protein. In Aim 2 we explore the use of both Monte Carlo and molecular dynamics methods to generate ensembles of structures around the models in order to determine the stability of the ternary complex and to predict its dissociation constant Kd and the level of binding cooperativity. Increasing the latter tends to reduce the hook effect. A highly efficient Monte Carlo method, specific to applications to PROTAC ternary complexes, is also developed. In Aim 3 we compare maximal degradation percentage (Dmax) and half maximal degradation concentration (DC50) values, measured in cellular essays available in the literature, to calculated Kd and cooperativity values in order to separate the impact of bioavailability related linker properties such as membrane permeability and solubility. Methods from the three aims will enable predicting POI-PROTAC-E3 structures, Kd and cooperativity values, and the level of cellular degradation on the basis of the two proteins, their ligands, and the chemical structure of the linker. Thus, the software package will be useful for rational PROTAC design as demonstrated by enquiries from companies working on development of related technologies.