In Silico Classification Modeling Of Blood-brain Barrier Permeability Of Traditional Chinese Medicine Permeability And Experimental Validation

The ability of brain penetration is one of critical pharmacokinetics properties of drugs.Therefore,the estimation of blood brain barrier(BBB)permeability has become an important research topic in central nervous system drugs discovery.Traditional Chinese medicine(TCM)has played a key role in the treatment of central nervous system(CNS)disease and become an important resource for CNS drug discovery.However,few studies are available on the blood-brain barrier penetration of the traditional Chinese medicine-derived compounds.Among various methods for estimating BBB permeability,in silico modeling of BBB permeability plays an important role in early discovery of CNS drugs due to its high-throughput and cost-effectiveness.In addition,significant difference in thechemical space distribution of chemical drugs and TCM makes the applicability of available drug-based in silico models to predict BBB permeability of TCM questionable,therefore the development of a TCM oriented machine learning model would be needed to automate the screening the large TCM compounds dataset for the prioritization of potential CNS active molecules.Herein,based on the investigation of the applicability of chemical drug-based in silico models to predict BBB permeability of TCM,we constructed and validated a TCM-oriented in silico BBB permeability classification model by using machine learning methods.The main contents of this research are as follows:(1)In Silico Classification Modeling of Blood-Brain Barrier Permeability of Traditional Chinese MedicineFirst,applicability of available drug-based in silico models to predict BBB permeability of TCM were investigated.Our results suggested poor prediction performance of chemical drug based BBB penetration models on TCM BBB permeability,thus it is not suitable for TCM database screening.Using four machine learning methods including SVM,RF,Naive Bayes and PNN,the TCM-specific BBB penetration models were successfully constructed and applied to screen large scale TCM database.The results showed that the model performance is better,as the accuracy higher than 75%and the consensus model’s accuracy is as high as 85%.In the screening of traditional Chinese medicine database with 11,032 molecules,4,249 was predicted to be BBB+ and 6,783 compounds as BBB-.(2)Evaluation of BBB permeability of the TCM-derived compounds by PAMPA-BBB analysisBBB permeation classification results were confirmed on 32 selected TCM molecules by in vitro parallel artificial membrane permeability assay.The results showed that the prediction results of 26 of 32 tested ingredients of traditional Chinese medicine are consistent with the experimental results,and overall accuracy of the consensus model is around 81%,which indicated that the better performance of the TCM-based BBB penetration models in prediction of large scale TCM database.(3)Evaluation of BBB permeability of the TCM-derived compounds using in vitro BBB modelsThe in vitro BBB models were constructed by bEnd.3 cells culture and used for the evaluation of BBB permeability of the TCM-derived compounds which defined as BBB+ in PAMPA-BBB and without relevant literature reports.Evaluation results showed that Nuciferin has good BBB penetrationability,and it is worth further study.