Integrative Identification of Anti-Photoaging Peptides From Stress-Tolerant Microorganisms via Machine Learning and KEAP1-NRF2 Docking.

Excess reactive oxygen species generated by ultraviolet exposure cause photoaging by degrading collagen and inhibiting its synthesis. This study presents a comprehensive strategy connecting the biological stress responses of γ-irradiated microorganisms to the discovery of novel anti-photoaging peptides. We profiled the radiation-regulated transcriptomes of Deinococcus radiodurans and Cryptococcus neoformans, focusing on DNA repair and oxidative stress responses. From these datasets, peptide libraries were generated in silico, filtered for biochemical properties, and prioritized using a seven-classifier machine-learning algorithm. Structural validity was established using Rosetta FlexPepDocking against the KEAP1-NRF2 pocket, which identified 48 docking-positive sequences. We then synthesized the top 21 peptides and subjected them to in vitro validation. Seven of these candidate peptides inhibited collagenase activity at 200 μM. Among them, four peptides dose-dependently increased the procollagen type I C-peptide level in ultraviolet B-induced fibroblasts. Furthermore, these peptides significantly elevated COL1A1 mRNA levels while simultaneously reducing MMP1 and MMP9 transcript and protein levels. In summary, this study provides an integrated strategy that combines omics, machine learning, and docking to discover promising peptide candidates, which were validated through in vitro assessments. This approach offers promising anti-photoaging candidates that can be applied to other oxidative stress pathways and biological resources.