Mapping Protein-Protein Interactions with Combinatorial Peptides

Since its invention in 1985 by George Smith, phage display has been an extremely beneficial tool in proteomics and drug discovery. Screening of phage-displayed libraries of random peptides can be employed to identify short binding motifs recognized by a specific protein domain, allowing one to predict new interacting partners for specific protein domains. Isolated ligands also have potential as the starting point in the drug discovery process, (i.e., peptidomimetic inhibitor). A complete set of protocols necessary to identify peptides in the library that bind selectively to a given target protein and to determine the common binding motif shared by the peptides has been first presented. To illustrate the utility of this approach, I affinity selected one human (Lyn SH3) and one yeast (Cbk1, kinase domain) protein target. Sequence alignment of Lyn SH3 binding peptides revealed the well-established PxxP binding motif. Sequence alignment of Cbk1 binding peptides revealed a novel motif, Y/FxFP, separately shown to be docking motif for Cbk1. In addition, one of the peptide ligands contained the motif, FKFP, present in Ssd1, a known interacting partner of Cbk1. To expand on the applications of the phage display approach to map protein-protein interactions, I affinity selected peptide ligands to the SH3 domain of mixed-lineage kinase 3 (MLK3). One unique ligand, named MLK3 SH3-Interacting Peptide (MIP), containing single proline, was isolated. Alanine scanning of the phage-displayed MIP(1-19) revealed 10 residues to be potentially important for its binding; interestingly, the proline residue turned out to be nonessential for this interaction. Also, we solved the crystal structure of both the apo MLK3 SH3 domain and the MLK3 SH3/MIP complex at 1.5 Å and 1.2 Å resolution respectively, which revealed a novel binding pocket on the surface of MLK3 SH3, defined by an extended n-Src loop that opens up to accommodate MIP binding. The discovery of a functional binding surface on the SH3 domain of MLK3 suggests the existence of a cellular ligand that binds in the same manner as our peptide ligand. We predict that the MIP ligand shares a binding motif with a cellular ligand(s).