| Hyperlipidemia is abnormally elevated levels of any or all lipids and/or lipoproteins in the blood,which is an important pathogenic factor of atherosclerosis or other cardiovascular diseases.The researches of blood lipid are clear now,including metabolic pathways,the key targets of marketed drugs,and so on.In addition,lipid-lowing drugs in tradition Chinese medicine(TCM)have an exact effect.Thus,regarding the lipid-lowing targets as the starting point of this research,the identifying of active compounds from TCM and the evaluation of ADMET property were performed.The above researched will be benefit to clarify the material basis and mechanism action of TCM.The modernization of TCM will be further enhanced.Furthermore,new drug R&D of TCM basing on monomeric compounds cannot match well with the theory of TCM,thus,component compatibility is one of the major directions in the field of TCM.In this paper,7 anti-hyperlipidemia targets,including 3-Hydroxy-3-Methylglutaryl Coenzyme A Reductase(HMG-CoA Reductase),Peroxisome Proliferator-Activated Receptor-?(PPAR-?),Niemann-Pick Cl-like 1(NPC1L1),Nicotinic Acid Receptor(GPR109A),Cholesteryl Ester Transfer Protein(CETP),Microsomal Triglyceride Transfer Protein(MTP),Diacylglycerol Acyltransferase 1(DGAT1)were defined as carrier,corresponding identification models were built by using molecular simulation technologies.Based on these models,virtual screening was performed to identify potential active compounds from TCM.Compound toxicity discrimination models and the permeability prediction models of three body barriers were built,which can provide the basis for the prediction of drug-like properties of the potential active compounds.In addition,literature retrieval,data mining method and molecular simulation technology were combined together to discuss the methods of component compatibility of TCM on clinical application and action target level.This paper included the following contents:identification models generation and virtual screening,the construction of drug-like prediction models,and the discussion of component compatibility methods.1.Identification models generation and virtual screening.Firstly,the methodology of identification models generation was established.Because the structural characteristics of the active compounds are obvious,the 3D structures of the active pockets are similar but not different,thus,the allosteric active site of the mGluR1 and mGluR5 was selected as carrier to clarify the principle of method selection during this research.Compared with previous research,the methodology in this paper has the following improvement:?For the construction of pharmacophore models,different method was selected based on the structural characteristics of the active compounds.For example,the structures of mGluR1 negative allosteric modulators(NAM)can be divided into three types,and the compound structure within one type is similar with each other.Thus,GALAHAD models regarding to the three types rather than a total model were built,and the results showed that each model achieved high specificity.While,the mGluR5 NAMs match the construction principle of HypoGen,thus quantitative model was built for mGluR5 NAMs.?For molecular docking,the active pockets were fully analyzed,the RMSD value of initial compound and docking results of active compounds were fully considered.Based on these,the appropriate method was selected.For example,there is a water molecular in the active pocket of mGluR5,which makes positive contribution to form hydrogen bond interaction.Cross-over analysis were performed in multi-conditions and showed that this water molecular must be reserved during the computational process.?The results of pharmacophore model and molecular docking were analyzed comprehensively.The prediction results based on ligand-based models and receptor-based models were consistent.In conclusion,the above methodology embodied the full workflow of virtual screening and highlighted the significance of method selection,which could be used for drug discovering of TCM.Secondly,the seven identification models of anti-hyperlipidemia targets were built to discover potential active compounds from TCM.The appropriate methods were selected based on the different data.For nicotinic acid receptor,because of the small number and small structure of the active compounds,bioisosterism and fragment search methods rather than pharmacophore modeling were used to perform the ligand-based drug discovery;For MTP,there is no crystal structure,the homology modeling was carried out to perform molecular docking study;For DGAT1,there is no crystal structure and no appropriate template structure for homology modeling,only the pharmacophore modeling was carried out.During the process,the test sets and several evaluators were used to evaluate the pharmacophore models,and showed rationality and reliability.The molecular docking models of each target,which can reproduce the interaction modes between the active compound and receptor,were used to evaluate the compounds hit by pharmacophore models.Combining the pharmacophore models and molecular docking models together,the approved positive drugs of each target included in DrugBank were searched.The results were as follows:The identification model of HMG-CoA reductase inhibitor can map 7 positive drugs(9 were included);identification model of PPAR-? agonist can map 3 positive drugs(5 were included);the identification models of NPC1L1 inhibitor,CETP inhibitor,MTP inhibitor can map the only recorded positive drug——ezetimibe,torcetrapib,lomitapide,respectively.Moreover,the hits can match well with models.Therefore,our models possess certain reliability,which could be used for discovering active compounds from TCM.Thirdly,the optimal models of each target were used to search TCMD.A hit list of 571 compounds was obtained based on the model of HMG-CoA reductase inhibitor,HMG-CoA Reductase-Comp.1 and HMG-CoA Reductase-Comp.2 have been proved to have anti-hyperlipidemia effect;A hit list of 182 compounds was obtained based on the model of PPAR-? agonist,among them,PPAR-?-Comp.1,PPAR-?-Comp.2,and PPAR-?-Comp.3 have been proved to have anti-hyperlipidemia effect;A hit list of 285 compounds was obtained based on the model of NPC1 L1 inhibitor,among them,NPC1L1-Comp.1,TCM-Comp.1,and NPC1L1-Comp.2 have been proved to have anti-hyperlipidemia effect;A hit list of 6 compounds was obtained based on the model of nicotinic acid receptor agonist,the source plants of some compounds have been proved to have anti-hyperlipidemia effect,such as valerian,betel nut,and boschniakia rossica;A hit list of 298 compounds was obtained based on the model of CETP inhibitor,among them,CETP-Comp.1 and CETP-Comp.2 have been proved to have the effect to treat cardiovascular disease;A hit list of 58 compounds was obtained based on the model of MTP inhibitor,among them,the source plants of TCM-Comp.1 and MTP-Comp.1 have been proved to have the effect to treat cardiovascular disease;A hit list of 434 compounds was obtained based on the model of DGAT1 inhibitor,the source plants of DGAT1-Comp.1 and DGAT1-Comp.2 have been proved to have anti-hyperlipidemia effect.The hit compounds or their source plants have the anti-hyperlipidemia effect or have the effect to treat cardiovascular disease,but the exact action targets are still not clear now.The results of this study indicated that our hits obtained by using molecular simulation technologies have strong targeting effect and can be further evaluated by biological experiments2.Construction of drug-like prediction modelsFive toxicity discrimination models were constructed by using SVM methods,including hepatotoxicity,nephrotoxicity,neurotoxicity,cardiotoxicity,and reproductive toxicity;and permeability prediction models of three body barriers were built,including blood brain barrier(BBB),placental barrier,and skin barrier.During the process of experiments,different data processing methods and parameters selection methods were combined together to build the prediction models.The results indicated that the cross-over method achieved high sensitivity and specificity.The above models all obtained high accuracy for the prediction of drug-like properties of compounds in TCM.In addition,the prediction models of the main mechanisms of nephrotoxicity,cardiotoxicity,and BBB permeability were built,including renal tubular necrosis,hERG potassium channel inhibitor,BBB passive diffusion and efflux transport prediction models.The results indicated that the S VM method can be used to expound the mechanism of drug-like property.Moreover,the BBB permeability and CNS activity model,the placental barrier permeability and reproductive toxicity model were combined together,respectively,which can provide suggestions for the system construction of the drug discovery and drug-like property prediction.The toxicity of all potential active compounds were predicted by the constructed toxicity prediction models,which showed only tripterygic acid has toxicity.The potential active compounds of the three subtypes of mGluRs were predicted by BBB permeability model and CNS activity model.The results indicated that most compounds have CNS activity and can cross BBB.The above experiments suggested that the models built in our study can be used to predict the drug-like properties of the compounds in TCMs.They are especially suitable for the compound which is hard to extract and separate from source plants.3.Discussion of component compatibility methods.Based on the multiple targets combination and multiple active site interaction level,component compatibility methods were discussed in this paper.The discovery of multiple targets system included methods which based on western medicines used together in clinical and based on herb pairs in TCM.Then,combining with the virtual screening results,this study provided methods for the discovery of multi-component TCMs.In addition,for the allosteric regulation,multi-components which can target on different active site in one target were discussed and verified,which can provide new idea for the interpretation of TCM mechanism and the discovery of multi-component TCMs.This study was based on molecular simulation and SVM method to discover potential drug-like active compounds from TCMs for anti-hyperlipidemia.Furthermore,component compatibility methods were discussed.This study provides new ideas for new drug research. |
PDF Full Text Request