Identification of a Chlorogenic Ester as a Monoamine Oxidase (MAO-B) Inhibitor by Integrating "Traditional and Machine Learning" Virtual Screening and In Vitro as well as In Vivo Validation: A Lead against Neurodegenerative Disorders?

Abstract

Parkinson's disease (PD) is the furthermost motor disorder of adult-onset dementia connected to memory and other cognitive abilities. Monoamine oxidases (MAOs) have gained significant attention in recent years owing to their possible therapeutic use against PD. Expression of MAO-B has been found to be elevated in PD patients for increased uptake of dopamine, producing hydrogen peroxide and finally causing neuronal injury. In this work, two new compounds have been identified as leads against MAO-B, and one of those compounds has been validated in vitro and in vivo. From the Protein Data Bank, MAO-B protein structures complexed with selegiline, 6-hydroxy-N-propargyl-1(R)-aminoindan, or a chromen derivative have been selected as templates for shape-based virtual screening (SB-VS) against the Traditional Chinese Medicinal (TCM) natural database. In parallel, using machine learning, a molecular-descriptor-based support vector model (SVM) was prepared and screened. For this purpose, naïve Bayesian, logistic regression, and random forest strategies were employed with the best specific molecular descriptor, which yielded a model with an overall accuracy (Q) of 0.81. Two common hit compounds lead-1 and lead-2 resulting from both shape and SVM screenings were analyzed through molecular docking and molecular dynamics (MD) simulation (200 ns). Also, from trajectory analysis such as molecular mechanics generalized Born surface area (MMGB/SA) and the residual interaction network (RIN) analyzer, both leads were found to bind at the active site with a favorable correlated motion, including domain movements. Lead-2, which is a chlorogenic ester, was synthesized and found to have no cytotoxic effect up to 50 μg/mL on Neuro-2A cells. The significant reactive oxygen species (ROS) scavenging activity by lead-2 could be correlated to its neuroprotective efficacy. Its capacity to inhibit human MAO-B through a competitive mode could be observed. An experimental zebra fish model confirms the neuroprotection by lead-2 by assessing the locomotor activities under malathion influence and treatment of lead-2. Also, histopathology analysis revealed that lead-2 could slow down degeneration in the brain. The present study emphasizes that integrating machine learning in parallel with traditional virtual screening may be useful to identify effective lead compounds for a given target.