Enhancements to the Rosetta Energy Function Enable Improved Identification of Small Molecules that Inhibit Protein-Protein Interactions

Abstract

Protein-protein interactions are among today’s most exciting and promising targets for therapeutic intervention. To date, identifying small-molecules that selectively disrupt these interactions has proven particularly challenging for virtual screening tools, since these have typically been optimized to perform well on more “traditional” drug discovery targets. Here, we test the performance of the Rosetta energy function for identifying compounds that inhibit protein interactions, when these active compounds have been hidden amongst pools of “decoys.” Through this virtual screening benchmark, we gauge the effect of two recent enhancements to the functional form of the Rosetta energy function: the new “Talaris” update and the “pwSHO” solvation model. Finally, we conclude by developing and validating a new weight set that maximizes Rosetta’s ability to pick out the active compounds in this test set. Looking collectively over the course of these enhancements, we find a marked improvement in Rosetta’s ability to identify small-molecule inhibitors of protein-protein interactions.

“Solvent hydrogen‐bond occlusion”: A new model of polar desolvation for biomolecular energetics

“…As we noted in our previous studies involving this benchmark, some of the docked decoy complexes include steric clashes that can be easily identified by Rosetta. We therefore surmised that the ability to discard many decoys on the basis of sterics alone might have partly obscured some underlying differences in performance between SHO and EEF1.…”

“…Given that the inhibitors in the former case remain quite exposed to solvent when bound to their protein partners, we surmised that here too a more accurate modeling of polar solvation might help discriminate native‐like models. As a pragmatic application, we therefore sought to explore the effect of replacing EEF1 with SHO for distinguishing active versus inactive compounds when docked to a protein of interest; this is precisely the task that one carries out in the final step of virtual screening, and is a problem that is still very challenging for PPI inhibitors …”

“…We note that the active compound is used in the native pose, rather than itself being docked. This simplification focuses the benchmark fully on the discrimination step: using instead the redocked native poses would have added noise, since misdocked inhibitors should not be considered “correct” for a given target at the scoring stage we consider here . Also, we note that the use of multiple protein targets entails a distinct active compound for each complex (Supporting Information Fig.…”

Ultra-High-Throughput Structure-Based Virtual Screening for Small-Molecule Inhibitors of Protein–Protein Interactions

“…Previously we compiled a set of 18 complexes, each corresponding to a crystal structure of a small-molecule inhibitor bound at a protein interaction site 77 ; in 11 of these cases the binding site is only composed of the biological unit (i.e. the stoichiometry of binding is clearly 1:1), and a structure of the unbound protein is also available.…”

Rotamer Libraries for the High-Resolution Design of β-Amino Acid Foldamers

“…To this end, superior β-amino acid rotamer libraries will require more sophisticated interpolation methods. Further efforts to create a framework suited for general heteropolymer modeling and design will include scoring function optimization to include alternative solvation and electrostatics models (Haberthur and Caflisch, 2008; Bazzoli et al, 2015), reference energies for design, and better methods for modeling hydrogen and halogen bonds. Re-refinement of structures containing β-amino acids will provide an experimental basis against which that scoring function might be optimized.…”