| Food components with cytoprotective function were an important material basis for the body to resist the nutritional and environmental stress,and improve cell resilience.Electrophilic compounds,as active ingredients widely existing in food,can covalently modify nucleophilic sites on biofunctional macromolecules to exert positive or negative biological toxicity.At present,systematic researches on the identification,screening and evaluation of food-derived electrophilic compounds were lacking,and the molecular characteristics of cytoprotective electrophilic compounds were still insufficient.In this study,an independent database of food-derived electrophilic chemicals was created,and a prediction model based on machine learning techniques was developed to distinguish the positive and negative impacts of electrophilic compounds.The molecular docking and Fukui function analysis were used to reveal the binding effect of cytoprotective electrophilic compounds on the Keap1 protein.Then,the cell and animal models were used to verify the effect and molecular mechanism of potential cytoprotective electrophilic compounds.Firstly,a database of food-derived electrophilic compounds containing 332 cases was established using structure search,and prediction models of positive/negative effects were created by the feature selection method(f_classif)and four machine learning methods based on logistic regression,decision tree,support vector machine and semi-supervised K-Nearest Neighbor(KNN),respectively.These results showed that the semi-supervised KNN model had the best prediction effect on the database of food-derived electrophilic compounds with the overall accuracy,precision and recall rates of 91.5%,92.6%and 97.6%,respectively,and the AUC value of learning curve(ROC)was 0.74,suggesting the reliable statistical performance and good predictive ability.Cluster analysis of 29 key eigenvalues showed that the positive/negative effects of food-derived electrophilic compounds were mainly related to the molecular characteristics of groups,electronegativity and lipophilicity in the molecular structure.The analysis of the whole data set found that the negative electrophilic compounds tend to appear aliphatic,ether structures and heteroatoms,which had stronger electrophilic ability,while the positive electrophilic compounds had higher lipophilicity.Secondly,combined with the process of first affinity and then covalent bonding between electrophilic compounds and target proteins,Autodock Vina was used to carried out molecular docking between positive electrophilic compounds and Nrf2 inhibitor(Keap1 Cys151 and Cys434)sites,and then the Fukui function of target compounds was calculated by Multiwfn software,and four electrophilic compounds with cytoprotective potential were finally screened out:4-Hydroxyderricin(4HD),isoliquiritigenin(ISO),Butein and 10-isobutyryloxy-8,9-epoxythymol isobutyrate(XHY-1).The binding energy of Cys151 was from-4.4 to-4.6kal/mol,while the binding energy of Cys434 was from-5.7 to-6.9 kal/mol.The Fukui functions were 1.10,1.12,1.11,and 1.16.The covalent docking analysis showed that the four compounds could form covalent bonds with the cysteine sulfhydryl groups of Keap1 Cys151and Cys434 throughα,β-unsaturated carbonyl or ternary oxygen rings,and simultaneously form secondary bonds such as hydrogen bonds and van der Waals forces,thereby affecting the protein conformation,which had the potential to activate Nrf2 signaling pathway.Thirdly,based on the H2O2-induced Hep G2 cell model,the cytoprotective effects of chalcone 4HD,ISO,and Butein were verified by real-time fluorescent quantitative PCR,Western blot,and si RNA.These results revealed that 4HD,ISO,and Butein,at concentration of 10 or 15μM,could promote the transcription of Nrf2 and downstream related genes(Nqo1,Ho-1,Gsr,Gclc and Gclm),as well as promote the transfer of Nrf2 into the nucleus,resulting in up-regulated expression of downstream Nqo1 and Ho-1 of Nrf2.The activities of endogenous cellular antioxidant enzymes SOD,CAT and GSH-px were enhanced,and the level of ROS was reduced,and the production of MDA was reduced.These changes could moderate the impact of H2O2 on cellular redox homeostasis,and play a protective role in cells.However,after Nrf2silencing,ISO could not upregulate the transcription and translation levels of Nrf2 and phaseⅡmetabolic enzymes,and the antioxidant capacity decreased synchronously with the level of ROS increasing.The protective effect on Hep G2 cells was eliminated.Finally,the protective effect of ISO was evaluated based on the DSS-induced colitis mouse model.The pathological alterations,inflammatory levels and oxidative stress indicators were determined.These results indicated that ISO could increase the activities of antioxidant enzymes SOD and CAT in colon tissue,improve the total antioxidant capacity,and effectively reduce the levels of MPO,MDA and pro-inflammatory cytokines(TNF-α,IL-1β,IL-6),thereby increasing the body weight,increasing the length of colon,and relieving the symptoms of hematochezia in the mice with ulcerative colitis.Combined with the analysis of genes related to Nrf2 signaling pathway in colon tissue,it was speculated that the protective effect of ISO on mice with colitis might be related to the activation of Keap1-Nrf2-ARE pathway.In conclusion,this study established and optimized a positive/negative effect prediction model of food-derived electrophilic compounds based a semi-supervised KNN method,and screened 4 electrophilic compounds with cytoprotective potential through affinity and covalent docking.Then,based on cell and animal injury models,it was verified that the electrophilic compound ISO could activate the Nrf2 signaling pathway and exert a cytoprotective effect. |