Discovering potential anti-skin-aging peptides in collagen: computer-assisted rapid screening and structure–activity relationships

The application of peptides as inhibitors of skin aging is a promising area of research. Previous researches have predominantly focused on extracting anti-aging peptides from the collagen of specific animals, while large-scale rapid screening and analysis of the structure–activity relationships of these peptides have been scarcely reported. In the present investigation, we developed a machine learning model for screening potential anti-skin-aging peptides (PASAPs), achieving a Matthews correlation coefficient (MCC) of 0.927 ± 0.044 and balanced accuracy (BACC) of 0.963 ± 0.022. These metrics surpassed those of the existing PeptideRanker model, which is widely used in bioactive peptide studies. Based on in silico screening, we identified and synthesized six novel PASAPs derived from tilapia collagen: KKHVWFGE, NGTPGAMGPR, PGAAGLKGDR, DGAPGPKGDR, TGPVGMPGAR, and GAPGGAGGVGEPGR. In vitro assays revealed that all six peptides exhibited significant inhibitory activity against aging-related enzymes, with the most pronounced effects on elastase and collagenase. A comprehensive analysis of the C-terminal amino acid residues indicated that the presence of arginine (R) at the C-terminus notably enhanced peptide binding to aging-related enzymes. This enhancement was attributed to an increased number of hydrogen bonds and stronger chemical interactions, which augmented the aging-related enzyme inhibitory activity of the peptides. In summary, this study proposed an effective strategy for discovering PASAPs from collagen and validated the machine learning model through experimental evidence. Structure–activity relationship insights can guide the synthesis of bioactive peptides and the selection of proteases for bioactive peptide production.

Graphical Abstract