Establishing a combined rational design protocol for the discovery of novel peptide binders of FGF7.

Designing peptide binders is a widely used strategy for developing potential therapeutic agents. Fibroblast Growth Factor 7 (FGF7) plays a critical role in cell proliferation and tissue repair, and its dysregulation is associated with various diseases. Here, we established an integrated computational-experimental workflow to identify peptide inhibitors targeting FGF7. We first generated a library of 100,000 random 8-mer peptides and progressively narrowed it using peptide toxicity analysis and binder prediction via PepBind-SVM. These methods eliminated 75.8% of non-viable candidates, enabling rapid library refinement. Next, we applied a sequence-based machine learning approach incorporating principal component analysis to classify the remaining peptides. The random candidates from three identical cluster were selected and subjected to molecular docking using Rosetta FlexPepDock. Peptides with the highest predicted binding affinity were synthesized and experimentally validated using isothermal titration calorimetry (ITC). Eight peptides demonstrated measurable binding to recombinant human FGF7 (rhFGF7), with three peptides exhibiting notably higher affinities of 43-67 µM. While these affinities are relatively weak and may limit immediate biological relevance, they nevertheless confirm binding and highlight both the potential and current limitations of the pipeline. Further molecular dynamics simulations revealed that key FGF7 residues, including R65, R67, and N149 play significant roles in stabilizing peptide interactions. This study presents an integrated in silico-to-in vitro pipeline for identifying preliminary peptide binders of FGF7 and provides mechanistic insights that may inform subsequent optimization and rational peptide design.