While the plant kingdom harbors numerous natural products, many exhibit insufficient antibacterial efficacy in their native state. Herein, strategic ester-bond grafting of amino acids substantially amplifies the antimicrobial activity of these underutilized natural compounds, with the resulting NP-g-Amino conjugates advantageously degrading into environmentally benign amino acids and natural products under ambient conditions. In this investigation, three Graft-QSCR models and three Graft-QSAR models were established utilizing distinct deep learning methodologies to predict both antibacterial potential and minimum inhibitory concentration (MIC) of NP-g-Amino conjugates, respectively. The Graft-QSCR1 and Graft-QSAR1 models developed using the Multi-Layer Perceptron (MLP) algorithm demonstrated excellent accuracy in predicting antibacterial potential (AUC = 0.98, ACC = 0.96) and optimal MIC prediction performance (R2 = 0.77, RMSE = 0.37). The combined application of Graft-QSAR1 and Graft-QSCR1 models guided amino acid grafting on three plant-derived natural products (Diosgenin, Stigmasterol, Protopanaxadiol), enhancing antibacterial activity by 60-235-fold. Taking diosgenin from Chinese yam (Dioscorea) as an example, successful amino acid grafting yielded the Do-g-Amino conjugates, whose structures were confirmed via multiple advanced spectroscopic techniques (FTIR, HR-MS, 1H NMR, and 13C NMR). Antibacterial assays revealed potent inhibitory effects of Do-g-Arg (MIC = 64 μg/mL) and Do-g-His (MIC = 128 μg/mL) against Staphylococcus aureus, while lactate dehydrogenase cytotoxicity assessments validated their exceptional biosafety profiles. The underlying antibacterial mechanisms were further elucidated through scanning electron microscopy observations complemented by molecular simulations. This study not only establishes an innovative strategy for discovering novel antimicrobial agents but also illuminates promising opportunities for enhancing the economic valorization of plant-derived compounds.
扫码关注我们
求助内容:
应助结果提醒方式:
