Virtual Hydrolysis-Based Screening of Wheat-Derived DPP-IV Inhibitory Peptides: A Mechanistic Analysis Integrating Cell Experiments and Molecular Dynamics Simulations

Dipeptidyl peptidase-IV (DPP-IV) inhibitors play a critical role in the treatment of diabetes and metabolic diseases. This study combines computational simulations with experimental validation to identify peptides with potential DPP-IV inhibitory activity from wheat proteins. A peptide database was constructed through trypsin digestion simulation, and screening was performed using the ConPLex deep learning algorithm, leading to the identification of four promising peptides: TENEWK (Thr-Glu-Asn-Glu-Trp-Lys), NFVSER (Asn-Phe-Val-Ser-Glu-Arg), LDLPSK (Leu-Asp-Leu-Pro-Ser-Lys) and QHEQR (Gln-His-Glu-Gln-Arg). Experimental results showed that the IC₅₀ values of these four peptides were 4.96 mM, 3.07 mM, 2.89 mM, and 4.18 mM, respectively, and they were all competitive DPP-IV inhibitors with significant inhibitory activity. Molecular dynamics (MD) simulations revealed the inhibitory mechanism by which inhibitory peptides led to the disappearance of the α-helix of the DPP-IV active center. Further tau-random accelerated molecular dynamics (tau-RaMD) simulations were used to calculate the binding residence time of each peptide, providing insights into their binding stability and key interacting residues. The combination of computational and experimental approaches significantly improved the accuracy and efficiency of screening peptides with DPP-IV inhibitory potential, providing a promising research foundation for developing peptide-enriched health foods.

Caco-2 cell; DPP-IV inhibitory peptides; deep learning; inhibitory mechanism; molecular dynamics simulation; wheat (Triticum aestivum L.).