High accuracy prediction of dipeptide angiotensin-converting enzyme (ACE) inhibitory activity by improving the credibility of the 3D-quantitative structure-activity relationship (3D-QSAR) model database and investigating inhibition mechanism

Abstract

The 3D-QSAR model is one of the most effective techniques for predicting the peptide bioactivity based on their structure properties. The accuracy of model prediction is highly dependent on the bio-activity of the peptides in its dataset. However, the same peptide used for constructing the database was reported to have obviously different activities in different published articles, which may significantly decrease the prediction accuracy. Based on this, we chose ACE inhibitory (ACE-I) dipeptides that can be absorbed directly into the human small intestine as research object. Ten dipeptides were randomly selected from the literatures for molecular docking with ACE and synthesized to test their activity. Ultimately, a linear equation was developed between bio-activities and docking scores, which was used to revalidate the reliable ACE-I activities of all reported dipeptides for constructing a credible dataset of 3D-QSAR model. The prediction accuracy was significantly enhanced and the model we established had a high Q2 = 0.840 and R2 = 0.998. Furthermore, dipeptide CW was predicted and tested in vitro to achieve the highest ACE-I activity (IC50 = 0.16 μM), and the inhibition mechanism was also investigated. This study introduced a new method for predicting bio-activity with high precision using the 3D-QSAR model.