Identification of novel umami peptides from oyster hydrolysate and the mechanisms underlying their taste characteristics using machine learning

Excessive sodium consumption poses considerable health risks, prompting the exploration of umami peptides as potential alternative for reducing sodium intake. This research investigated umami peptides (PQFAPEED, EEHPVLLTEA and DQAIPNKPEE) using machine learning, determining sensory thresholds to be 0.24, 0.30 and 0.29 mg/mL. Molecular docking studies revealed hydrogen bonds and hydrophobic interactions are vital for their binding to umami receptors T1R1/T1R3, with key residues identified as Val714, Leu852, Gln853 and Glu855. Combination of DQAIPNKPEE with 3 mg/mL sodium chloride (NaCl) closely mimicked the salinity perception of 5 mg/mL NaCl. Additionally, DQAIPNKPEE and PQFAPEED were recognised as salt-enhancing peptides, with Ala283, Glu284, Glu286, Arg294, Arg330 and Arg583 identified as critical amino acid residues of human transmembrane channel-like 4 (TMC4). These peptides substitute chloride ions to activate TMC4, resulting in sensation of saltiness. This study highlights efficacy of machine learning in rapid identification of umami peptides from oysters and taste receptors interactions.