Many bioactive proteins interact with collagen, recognizing amino acid sequences displayed on the triple helix. We report here a selection strategy to obtain triple-helical peptides that interact with the proteins from a combinatorial random library constructed in yeast cells. This system enables us to select them using the standard two-hybrid protocol, detecting interactions between triple-helical peptides and target proteins fused to the GAL4-activating and binding domains, respectively. The library was constructed having triple-helical peptides with a “host-guest” design in which host helix-stabilizing regions flanked guest random sequences. Using this system, we selected peptides that bind to pigment epithelium-derived factor (PEDF), a collagen-binding protein that shows anti-angiogenic and neurotrophic activities, from the libraries. Two-step selections from the total random library and subsequently from the second focused library yielded new PEDF-binding sequences that exhibited a comparable affinity to or more potent than that of the native PEDF-binding sequence in collagen. The obtained sequences also contained a variant of the PEDF-binding motif that did not match the known motif identified from the native collagen sequences. This combinatorial library system allows the chemical space of triple-helical peptides to be screened more widely than that found in native collagen, thus increasing the expectation of obtaining more specific and high-affinity peptides.
Read full abstract