Designed Armadillo repeat proteins: library generation, characterization and selection of peptide binders with high specificity

Designed Armadillo repeat proteins (ArmRPs) are a novel class of binding proteins intended for general modular peptide binding and have very favorable expression and stability properties. Using a combination of sequence and structural consensus analyses, we generated a 42-amino-acid designed Armadillo repeat module with six randomized positions, having a theoretical diversity of 9.9×10(6) per repeat. Structural considerations were used to replace cysteine residues, to define less conserved positions and to decide where to introduce randomized amino acid residues for potential interactions with the target peptide. Based on these concepts, combinatorial libraries of designed ArmRPs were assembled. The most stable version of designed ArmRP in library format was the N5C format, with three randomized library repeat modules flanked by full consensus repeat modules on either side and, in turn, flanked by N- and C-terminal capping repeats. Unselected members of this library were well expressed in the Escherichia coli cytoplasm, monomeric and showed the expected CD spectra and cooperative unfolding. N5C libraries were used in ribosome display selections against the peptide neurotensin. Highly specific peptide binders were enriched after four rounds of selections using ribosome display. Four peptide side chains were shown to contribute most of the interaction energy, and single alanine mutants could be discriminated. Thus, designed ArmRP libraries can become valuable sources for peptide binding molecules because of their favorable biophysical properties and with a potential for application in general modular peptide recognition.