| Title : Machine learning prediction, molecular dynamics simulation, and in vitro studies of putative leads from citrus essential oil metabolites as potential dipeptidyl peptidase-4 inhibitors - Lukman_2025_Comput.Biol.Med_200_111375 |
| Author(s) : Lukman HY , Sabiu S |
| Ref : Computers in Biology & Medicine , 200 :111375 , 2025 |
|
Abstract :
Citrus spp., including their essential oils, have been studied for their antidiabetic potential; however, their modulatory mechanism on dipeptidyl peptidase-4 (DPP4) remains elusive. This study profiled citrus essential oils' (CEOs) constituents using gas chromatography-mass spectrophotometry and evaluated the DPP4 inhibitory potential of the oils and their metabolites using in vitro, machine learning (ML), and molecular dynamics simulation methods. Of the six CEOs investigated, only Citrusreticulata (18.21 microg/mL) had lower half maximal inhibitory concentration (IC(50)) than the reference standard [sitagliptin (22.88 microg/mL)]. A total of 64 metabolites were curated into a library from the chemical profiles of the oils. Machine learning analysis highlighted the principal component and hierarchical clustering of the top 35 abundant metabolites and correlated metabolites abundances with in vitro DPP4 inhibition of the oils. The ML characterization of DPP4 inhibition using docking scores and molecular descriptor of the metabolites yielded a regression model (R(2)) of 0.682 (root mean square error of 0.409), with strong correlations (Pearson r = 0.82, Spearman ro = 0.93) between docking scores and binding free energy values. Structural activity analysis of the metabolites revealed that while beta-terpineol (-35.78 kcal/mol) presented superior binding free energy than sitagliptin (-34.80 kcal/mol) and the other CEO metabolites, it also had better binding stability and interaction with the catalytic residues of DPP4, which is critical for modulating the enzyme activity. Kinetic analysis indicated beta-terpineol's competitive inhibition pattern against DPP4. Findings from this study show that relative abundance of metabolites does not directly translate into their biological activities and CEOs and their metabolites, especially beta-terpineol, possess potential anti-DPP4 effects. Nevertheless, further studies such as in vivo validation and lead optimization are necessary to fully establish the antidiabetic potential of the investigated CEOs and their metabolites. |
| PubMedSearch : Lukman_2025_Comput.Biol.Med_200_111375 |
| PubMedID: 41370954 |
Lukman HY, Sabiu S (2025)
Machine learning prediction, molecular dynamics simulation, and in vitro studies of putative leads from citrus essential oil metabolites as potential dipeptidyl peptidase-4 inhibitors
Computers in Biology & Medicine
200 :111375
Lukman HY, Sabiu S (2025)
Computers in Biology & Medicine
200 :111375