HighPlay2 workflow designs cyclic peptides with non-canonical amino acids, achieving micromolar binding to MDM2 and GABARAP

Developing effective binders for dynamic biological interfaces, particularly in protein-protein interactions (PPIs), remains a significant challenge in drug discovery. Traditional small molecules often lack the specificity or binding affinity for these complex interfaces, while biologics can face issues with cell permeability and oral bioavailability. Cyclic peptides offer a promising alternative due to their conformational preorganization, which can enhance affinity and selectivity. Furthermore, incorporating non-canonical amino acids (ncAAs) expands the chemical diversity beyond the 20 natural residues, potentially improving pharmacokinetic properties and target engagement, addressing limitations of current therapeutic modalities.

Researchers developed HighPlay2, a structure-guided computational workflow for iteratively designing cyclic peptide sequences containing ncAAs. The workflow involved generating candidate sequences, predicting their protein-bound structures, and evaluating these models using a Structure-Constrained Objective Score (SCOS) that integrates structure-prediction confidence with interface geometry. This framework was applied to multiple protein targets, including MDM2 and GABARAP. Metrics derived from structure prediction, Rosetta interface analysis, and molecular dynamics simulations guided candidate selection. Selected peptides were then synthesized and experimentally evaluated for binding affinity using surface plasmon resonance (SPR) assays.

The HighPlay2 workflow successfully identified cyclic peptide candidates incorporating non-canonical amino acids that demonstrated measurable binding to target proteins. For both MDM2 and GABARAP, synthesized peptides exhibited binding in the micromolar range in selected cases. This indicates the feasibility of the HighPlay2 framework for early-stage design and screening. The SCOS proved effective in prioritizing candidates by combining structural prediction confidence with favorable interface geometry, streamlining the design process. The integration of molecular dynamics simulations further refined candidate selection by assessing binding stability and dynamics. While specific numerical affinities beyond "micromolar" were not detailed, the consistent observation of binding across different targets supports the workflow's general applicability. The authors emphasize that this proof-of-concept demonstrates the potential of structure-guided design for novel peptide therapeutics.

Synthesized peptides designed by HighPlay2 showed measurable micromolar binding to MDM2 and GABARAP in selected cases, validating the workflow's early-stage design capabilities.