(371c) Computational Design of Peptides to Detect Human Health Biomarkers

The goal of this research is to develop an efficient computational algorithm that searches for the best peptide binder to a particular target. The biomarker chosen for initial study is Troponin I, a 210-amino acid protein that is elevated in the blood of heart attack victims and is a general indicator of muscle fatigue. The approach is based on a search algorithm already developed in the PI’s group. An iterative procedure, which involves sequence mutation moves during which the peptide’s backbone conformation is held fixed, and peptide backbone conformation moves during which the peptide sequence is held fixed, is used to arrive at the binding sequence and conformation that has the lowest energy. The search algorithm is supplemented by explicit-solvent atomistic simulations to determine the free energies of the top scoring peptides. The search algorithm was used to identify sixteen 12-residue peptides that bind to Troponin I. The starting sequence was a peptide, P0, discovered using phage display. Explicit-solvent simulations of the peptide-Troponin I complex were used to further screen the candidate peptides. A variety of experimental techniques (Surface Plasmon Resonance, Quartz Crystal Microbalance, Dot Blots) were used by our AFRL collaborators to assess the binding capabilities of the top three candidate peptides, P2, P3 and P5 as well as the original peptide, P0. The binding affinity of P2 was extremely strong at 1.53x 10^-12(M), compared to that of the original peptide, P0, at 6.15 x 10^-9(M).