27/08/2024 | News release | Distributed by Public on 27/08/2024 14:39
How researchers are using structural information and in silico modelling applications to improve the accuracy of bifunctional protein degraders
Historically, small molecules have been the most common pharmaceutical approach to treating diseases. Despite the growth of biologics, small molecule drugs still represented 52% of the drugsapproved last year by the US Food and Drug Administration.
Still, some biochemical targets, due to their morphology, are intractable to the small molecule approach, laying bare a whole category of proteins that aren't "druggable". Researchers have been trying to get around this limitation by developing larger molecules called targeted protein degraders (TPDs), which activate and exploit the physiological mechanism called degradation that prompts the cell's own protein-clearing machinery to gobble up the proteins.
Protein degradationis an amino acid and peptide recycling process that also influences protein populations within cells. This plays a crucial role in cell regulation and homeostasis. The process involves enzymes called proteases that degrade proteins into small peptides; TPDsare able to induce this process and cause the degradation of the protein of interest (POI).
In addition to enabling access to a wider range of targets, degraders can show advantages over traditional small molecules such as lower dose administration and reduction of off-target effects.
Currently, TPDs are most often used as chemical probes to study the function and cellular importance of proteins, however, there are some TPD drug discovery pipelines in progress. The biotech Arvinaspioneered a TPD pipeline that includes two degraders in clinical trials- ARV110and ARV-471-that target an androgen receptor degrader for prostate cancer and an estrogen receptor for breast cancer respectively. Other biopharmaceutical companies such as C4 Therapeutics, Pfizer, Kymera Therapeutics, AstraZeneca, Bayer, Novartis and Vertex have also begun investing in TPD pipelines.
TPDs can be classified into two groups: bifunctional protein degradersand molecular glues. This article focuses on the ability to accelerate bifunctional degrader development programs using structural information and in silicomodelling applications.
The structure of a bifunctional degrader consists of two ligands called warheads connected via a linker. One warhead targets the POI and the other an E3 ubiquitin ligase. E3 ligases are members of ligase complexes, which ubiquitylate proteins and induce degradation via ubiquitin-proteasome complexes. Often, in an advanced stage of a TPD project, the optimised warheads of bifunctional degraders are not altered. Therefore, the coveted goal of bifunctional degrader development is the optimization of the linker aiming for the right linker-length, contacts and geometry. On the other hand, at an earlier stage of a project, exploration of different POI ligands can offer alternative exit vectors and the selection of different E3 ligase substrates can recruit different ligases. Clearly, the development of TPDs is a very complex process.
Fortunately, the combination of computational and medicinal chemistry disciplines is helping to speed up this process. Since 2017, a growing number of X-ray structures of TPD ternary complexes have been published. Ternary complex structures can accelerate TPD development via in silicostructure-based applications by providing the 3D information key to the efficient design of new linkers. The ternary complex orientation is crucial for the design of a degrader with an appropriate linker as it binds in the protein-protein interface. Properties can be predicted using the appropriate computational methods, as shown in Figure 1.
Figure 1. A. PDB: 5T35- the first TPD complex structure solved by X-ray in 2017. The red protein is the POI (Brd4), in light blue is the E3 ligase (VHL), the bifunctional degrader is highlighted in yellow, and the remaining proteins are transcription factors (green and orange). B. TPD binding mode. C. Bifunctional degrader MZ1 acts as an anticancer agent.
Computational methods focus on the development of modelled ternary complexes as well as linker optimization.
When a structure of the ternary complex of interest is available, modeling strategies mainly focus on optimising the linker via enumerationsusing virtual linker libraries. However, when structures of a specific ternary complex are not available, modelling becomes necessary to generate a modelled ternary complexthat can be used for linker optimization. In the absence of a structure, ligand-based approachescan also be used to optimize a degrader's linker without developing the complex (structure-based modeling).
Overall, it is early days for TPD modeling. Significant limitations must be overcome. The most challenging aspect of structure-based TPD modeling is the accurate prediction of the ternary complex. Many factors influence the quality of the results such as the resolution of the X-ray structure(s) used in modelling, the flexibility of the proteins and the errors of scoring-functions for protein-protein docking contacts etc. In addition, different degrader linkers and/or POI ligands (alternative linker exit vector) are able to induce different ternary complex conformations, as shown in Figure 2.
Figure 2. Overlay of SMARCA2 ternary structures. Grey ribbons: SMARCA2 with phenol containing inhibitor bound. Green ribbons: rest of the ternary structures
Although, TPD modeling is currently challenging for computational chemists, the potential impact on projects and diseases is motivating the research community to continue to support it. Leading software companies like Chemical Computing Group, Schrödinger, Cresset, etc. have been actively involved in optimizing workflows and methods that can improve TPD modeling accuracy and quality.
Just as many computational approaches used routinely today seemed impossible a few decades ago, in time TPD modeling may also become a widely accepted and utilized tool in the modeler's toolbox. Stay tuned!