De Novo Design of Multivalent α/β-Peptides Mimicking Transcription Factors Targeting the CBP KIX Domain.

Protein-protein interactions that regulate gene expression in the nucleus are increasingly recognized as potential therapeutic targets but present unique challenges. Effective targeting requires molecules capable of interacting with large protein surfaces while remaining small enough to pass through the nuclear pore, for which peptidomimetics are promising candidates. De novo design strategies often focus on projecting hot-spot residues using well-defined, rigid scaffolds; however, other critical properties-such as solubility, charge, or local flexibility-are often overlooked. Our approach employs α/β-peptidomimetics to enable fine-tuning of these properties by generating manageable-sized soluble libraries. Using the KIX domain of the coactivator proteins p300/CBP as a model system, we demonstrate that our strategy allows modular optimization of hot-spot, solvent-exposed, and structure-inducing residues, thereby tuning affinity and binding-site selectivity. Coactivator proteins often use multivalency to enhance affinity and selectivity, which we exploit by creating dimers from our initial hits. We show that both static libraries and template-directed dynamic covalent chemistry facilitate the screening of multivalent ligands that closely mimic the native interaction partners. Our design strategy, built on α/β-peptide building blocks, represents a promising approach to develop de novo ligands against proteins characterized by high plasticity and multivalent interactions.