Model Prediction And Molecular Docking Of Angiotensin-Converting Enzyme Inhibitors

Angiotensin converting enzyme is essential for the regulation of blood pressure,which can catalyze the inactive decapeptide Angiotensin I forming an active vasoconstrictor octapeptide—Angiotensin II and inactivate the antihypertensive bradykinin.At present,the clinical treatment of hypertension is basically to hunt for the appropriate Angiotensin converting enzyme inhibitors to protect the heart,kidneys,brain and reduce the incidence of cardiovascular disease.Both in theoretical research and clinical treatment,ACE inhibitors have been extensively applied to the prevention and treatment of hypertension.In this work,the theoretical methods in computational biology,such as quantitative structure-activity relationship(QSAR),molecular docking were applied to explore the interaction mechanism of ACE inhibitors.The main contents and conclusions are as follows:(?)554 ACE inhibitors derived from Binding Database were characterized by the vsurf_descriptors,which were used to establish the SVM models with only 3 two-dimension descriptors and 4 three-dimension descriptors,respectively.The optimized SVM model with two-dimension descriptors achieved the overall accuracy,sensitivity,specificity,AUC and MCC of 89.637%,91.185%,82.716%,0.938 and 0.686 on 444 training samples,81.818%,84.337%,74.074%,0.909 and 0.549 on 110 test samples.The model was further validated by 114 external test set with accuracy,sensitivity,AUC of 67.544%,70.526% and 0.572.The model with 4 three-dimension descriptors presented the results on accuracy,sensitivity,specificity,AUC and MCC,88.739%,90.91%,80.43%,0.914 and 0.678 with 444 training samples;86.364%,87.10%,82.35%,0.909 and 0.591 with 110 test samples.The 114 external test samples of accuracy,sensitivity,AUC were 71.93%,94% and 0.809.(?)FASGAI and NNAAIndex were applied to characterize the ACE inhibitors with dipeptide and tripeptide in the peptide database(BIOPEP).Comparing the models with two kinds of descriptors,the one with FASGAI descriptors was more excellent than the other one.The dipeptide model only with 2 FASGAI descriptors obtained the overall accuracy,sensitivity,specificity,AUC and MCC of 82.051%,79.167%,83.333%,0.903 and 0.601 on 78 training samples,and 90%,77.778%,100%,0.956 and 0.811 on 20 test samples.The tripeptide model with 3 FASGAI descriptors received the results of accuracy,sensitivity,specificity and AUC of 79.245%,77.907%,85% and 0.846 on 106 training set,76.923%,81.818%,50% and 0.658 on test set.(?)Both SVM and molecular docking methods on all inhibitors(554 molecules,98 dipeptides and 132 tripeptides)were used to establish the new classification models,which was further used to validate the reliability and stability of SVM models.It was so obvious that the optimal binding conformations showed hydrogen bonds of 554 inhibitors;the aromaticity of C-terminal and the hydrophobicity of N-terminal for dipeptide;the hydrophobicity and ?-helix of the N-terminal,the electrostatics of the second aromatic amino for tripeptide.These properties of inhibitors had a significant effect on binding conformations and provided theoretical foundation for the design of drug inhibitors.