Generic placeholder image

Letters in Drug Design & Discovery

Editor-in-Chief

ISSN (Print): 1570-1808
ISSN (Online): 1875-628X

Letter Article

Prediction and Validation of Proline-containing Tripeptides with Angiotensin I-converting Enzyme Inhibitory Activity Using Machine Learning Models

Author(s): Toshiya Hatakenaka, Yuki Fujimoto, Kouji Okamoto and Tamaki Kato*

Volume 21, Issue 15, 2024

Published on: 28 November, 2023

Page: [3069 - 3075] Pages: 7

DOI: 10.2174/0115701808274195231113053944

Price: $65

Abstract

Background: Numerous inhibitory peptides against angiotensin I-converting enzyme, a target for hypertension treatment, have been found in previous studies. Recently, machine learning screening has been employed to predict unidentified inhibitory peptides using a database of known inhibitory peptides and descriptor data from docking simulations.

Objective: The aim of this study is to focus on angiotensin I-converting enzyme inhibitory tripeptides containing proline, to predict novel inhibitory peptides using the machine learning algorithm PyCaret based on their IC50 and descriptors from docking simulations, and to validate the screening method by machine learning by comparing the results with in vitro inhibitory activity studies.

Methods: IC50 of known inhibitory peptides were collected from an online database, and descriptor data were summarized by docking simulations. Candidate inhibitory peptides were predicted from these data using the PyCaret. Candidate tripeptides were synthesized by solid-phase synthesis and their inhibitory activity was measured in vitro.

Results: Seven novel tripeptides were found from the peptides predicted to have high inhibitory activity by machine learning, and these peptides were synthesized and evaluated for inhibitory activity in vitro. As a result, the proline-containing tripeptide MPA showed high inhibitory activity, with an IC50 value of 8.6 μM.

Conclusion: In this study, we identified a proline-containing tripeptide with high ACE inhibitory activity among the candidates predicted by machine learning. This finding indicates that the method of predicting by machine learning is promising for future inhibitory peptide screening efforts.

Next »

© 2024 Bentham Science Publishers | Privacy Policy