Miniature Protein Inhibitors of the p53–hDM2 Interaction

We have developed a strategy for the design of miniature proteins that bind DNA or protein surfaces with high affinity and selectivity. This strategy, which is often called protein grafting, involves dissecting a functional recognition epitope from its native a-helical or polyproline type II (PPII) helical context and presenting it on a small but structured protein scaffold (Figure 1A). Here we describe the development and characterization of miniature proteins that bind the human double-minute 2 oncoprotein (hDM2) in the nanomolar concentration range, and inhibit its interaction with a peptide (p53AD15–31) derived from the activation domain of p53 (p53AD). hDM2 is the principal cellular antagonist of the tumor-suppressor protein p53. Elevated hDM2 levels are found in many solid tumors that express wild-type p53 and there is considerable interest in hDM2 ligands that are capable of up-regulating p53 activity in vitro or in vivo. The high-resolution structure of the p53AD–hDM2 complex has revealed a recognition epitope that is composed primarily of three p53AD residues (F19, W23, and L26); these are located on one face of a short a-helix. Although the p53AD peptide possesses little a-helical structure in the absence of hDM2, augmenting the level of intrinsic a-helix structure in p53AD by using constrained, unnatural amino acids dramatically increases affinity for hDM2 in vitro and activity in vivo. In addition, several other scaffolds have been used to display the p53AD epitope, including large proteins, cyclic b-hairpin peptides, retro-inverso peptides, and b-peptides. The first highly active small-molecule inhibitors were reported in 2004 and had IC50 values for inhibiting the p53–hDM2 interaction in the 100–300 nM range. These molecules also resembled p53AD’s primary recognition epitope. Since all these inhibitors appear to preorganize the p53AD epitope to some degree, we reasoned that protein grafting would be a logical route to developing miniature protein hDM2 ligands. In contrast with the previously reported molecules, miniature protein-based inhibitors would be both synthetically tractable and genetically encodable. This would facilitate their use as in vitro and in vivo tools for probing the intricate p53/hDM2 pathway. Avian pancreatic polypeptide (aPP, Figure 1B) is a small, well-folded miniature protein that consists of an eight-residue PPII helix linked through a type I b-turn to an eighteen-residue a-helix. Structure-guided alignment of the a-helical segments of p53AD and aPP (Figure 1B) positions the three critical hDM2 contact residues (F22, W26, and L29 in the aPP-aligned sequence) and five residues important for aPP folding (L17,