Characterization Of Ligand-binding Properties Of PDZ Domain

Research over the past decade have revealed that a fairly large set of protein are accomplished by peptide-binding domains that recognize short peptide motifs in their binding partners, such as SH2, WW, SH3, etc. Further research on the binding property characterization of peptide-binding domains will help us well understand the mechanisms of protein interaction, as well as find all potential native binding proteins of targeted domain in proteomic scale. Therefore ligand- binding property characterization of domains is a valuable aspect in the studies of protein interaction.PDZ domain is one of the most abundant protein interaction domains which can commonly be found in various organisms. It plays important roles in critical cellular functions, such as subcellular localization, signal transduction, and multiprotein complexes assembling. Typically the characteristics of a PDZ domain are determined by the extreme four C-terminal residues of its target proteins. Based on the C-terminal sequence, PDZ binding motifs are classified conventionally into four classes: ClassⅠ, X[S/T]XΦ-COOH: ClassⅡ, XΦXΦ-COOH; ClassⅢ, X[D/E/K/R]XΦ-COOH; ClassⅣ, XXC-COOH or XΦ[D/E]-COOH, where x is any amino acid,Φis a hydrophobic residue. However, many PDZ domains target on a rather broad ligand sequence, which cannot be explained by simple rules and conventional classifications.This study focused on three PDZ domains of three polarity proteins: ZO-1 PDZ1, Veli3 PDZ, and PAR3L PDZ1. The ligand-binding characteristics of the targeted PDZ domains have been investigated by screening random peptide library using yeast two-hybrid technology. With these ligand-binding characteristics the potential native binding proteins of these domains in proteomic scale base have been identified.The three PDZ domains showed different binding properties. Their C-terminal consensus sequences of ligands are [S/T/V/E] [Ψ/K/R] [V/I/L/C]-COOH, [E/X] [S/T] X [V/I/L]-COOH, and [S/X] [Φ/K/H/E] [Φ/Q/E]Φ-COOH (whereΨis an aromatic amino acid) respectively. ZO-1 PDZ1 can bind classical typeⅠ,Ⅱ, andⅢligands simultaneously with class typeⅠas its dominant selection. Different from the traditional classification its typeⅠligands have a strong preference for aromatic residues at position -1. Yeli3 PDZ domains also bind ClassⅠligands. Unlike ZO-1 PDZ1 domain Glu was overwhelmingly preferred at position -3 of Veli3 PDZ ligands, and, hydrophobic amino acids showed higher affinity than others at position -5. In contrast to ZO-1 PDZ1 and Veli3 PDZ domain, PAR3L PDZ domains bind both classⅡand classⅢligands with dominant preference of hydrophobic amino acids at position -1 of its ligands. These results show more complicated binding properties of PDZ domain-binding motifs. The diversity of the binding properties suggests that various domains contribute to difference cellular functions through binding distinct domain-binding partners.Moreover, with the application of bioinformatics all the potential ligands in human proteome have been predicted by searching several human databases with the consensus-binding sequences, which include, not only most ligands reported previously but also many new potential ligands which are biological relevant with targeted domain and the optimal ligand sequences. For example, the optimal ligand of PAR3L PDZ1 domain, SAFF-COOH, does not exist in the C-terminus of any native protein. It might be as a candidate sequence for PAR3L protein blocker in future. For ZO-1 PDZ1 domain, eleven native human proteins predicted as ligands were chosen by their cellular locations and biological functions for validating protein interaction through yeast two-hybrid method. Four of them were confirmed positive. Till now few PAR3L proteins interaction have been reported. These results provide significant clues for identifying more ligands of the targeted domains and discovering biological functions of these PDZ proteins. We also established a method for screening ligand library of PZD domain to be able to obtain ligand-binding characteristics of the targeted domain and native binging ligands directly. By which more accurate preference of amino acid residues at given position also can be acquired. Our Ligand library was consisted of several PDZ binding ligands obtained, the synthesized potential PDZ binding peptides, and PDZ domains themselves. For ZO-1 PDZ1 domain, 80 negative sequences were identified by screening ligand library. The comparative analysis of both positive and negative sequences showed that any amino acid except Glu can be the candidate at position -3. The new ligands obtained in present or future studies will be added to ligand library to expand our ]igand library. This ligand library will sever as valuabie resource for the further study of PDZ domain-binding properties.To summarize, the results of our present study enriched the traditional classification of PDZ domains. The strategy of screening random peptide library in combination with bioinformatics established in this work is an effective approach to theoretically identify all the potential ligands in the databases, especially those low-abundant ligands rarely obtained by screening a cDNA library. Hence it is a powerful tool for the studying peptide-binding domain interactions.