Umami is a fundamental taste sensation, often described as a delicious and pleasant flavor perception. To enhance or complement the original flavor and meet the tastes of diverse regions, umami dipeptides have been extensively utilized in global food manufacturing. Currently, the application and purification techniques of dipeptides are relatively mature, while their umami mechanisms and molecular modification are both scarce. In this work, the 3D structure of the umami dipeptide target T1R1/T1R3 was first obtained through sequence alignment and homology modeling, then followed by the successful construction of a database containing 400 samples of dipeptides. Subsequently, the complex models of T1R1/T1R3, respectively, with DG (Asp-Gly) and EK (Glu-Lys) (i.e., T1R1_DG/T1R3, T1R1/T1R3_DG, T1R1_EK/T1R3, and T1R1/T1R3_EK) were obtained via molecular docking and virtual screening. Finally, based on comparative molecular dynamics (MD) simulation trajectories, the binding free energy was calculated to investigate receptor–ligand recognition and conformational changes, providing some implications for potential modifications of umami dipeptides. T1R1 tends to bind relatively small umami dipeptides, whereas T1R3 does the opposite, both of which favor the recognition of acidic and hydrophilic dipeptides. By comparing strategies such as hydroxyl introduction and chain length alteration, electrostatic effects may be more important than non-polar effects in molecular design. This work not only explores the recognition mechanism of umami dipeptides with the receptor T1R1/T1R3 showing certain theoretical significance, but also provides feasible suggestions for dipeptide screening and modification having certain application value.

Amino acid binding proteins (AABPs) undergo significant conformational closure in the periplasmic space of Gram-negative bacteria, tightly binding specific amino acid substrates and then initiating transmembrane transport of nutrients. Nevertheless, the possible closure mechanisms after substrate binding, especially long-range signaling, remain unknown. Taking three typical AABPs—glutamine binding protein (GlnBP), histidine binding protein (HisJ) and lysine/arginine/ornithine binding protein (LAOBP) in Escherichia coli (E. coli)—as research subjects, a series of theoretical studies including sequence alignment, Gaussian network model (GNM), anisotropic network model (ANM), conventional molecular dynamics (cMD) and neural relational inference molecular dynamics (NRI-MD) simulations were carried out. Sequence alignment showed that GlnBP, HisJ and LAOBP have high structural similarity. According to the results of the GNM and ANM, AABPs’ Index Finger and Thumb domains exhibit closed motion tendencies that contribute to substrate capture and stable binding. Based on cMD trajectories, the Index Finger domain, especially the I-Loop region, exhibits high molecular flexibility, with residues 11 and 117 both being potentially key residues for receptor–ligand recognition and initiation of receptor allostery. Finally, the signaling pathway of AABPs’ conformational closure was revealed by NRI-MD training and trajectory reconstruction. This work not only provides a complete picture of AABPs’ recognition mechanism and possible conformational closure, but also aids subsequent structure-based design of small-molecule oncology drugs.