Deep Learning-Based Screening for MRTF-A Nuclear Translocation Agonists and Investigation of Its Neuroprotective Effects on Synapses Following Ischemic Stroke Reperfusion
Description
Reperfusion injury following ischemic stroke treatment significantly affects patient prognosis. Myocardin-related transcription factor A (MRTF-A) has been shown to alleviate this injury, making it a promising therapeutic target. In this study, MRTF-A expression was modulated in rat middle cerebral artery occlusion/reperfusion (MCAO/R) models. MRTF-A mitigated reperfusion injury by enhancing the co-localization of PARD3 and Tiam1, thereby preserving synaptic structure and function. Since MRTF-A activation requires its dissociation from G-actin, an active molecular screening zone was defined based on the binding site of the G-actin/MRTF-A RPEL domain. A dataset was constructed using batch molecular docking to train an AttentiveFP deep learning regression model for predicting the binding affinity of small molecules to the G-actin domain. Through molecular scaffold analysis, ADME/T prediction, and molecular dynamics simulation, the potential agonist Pranlukast was identified. Pranlukast competes with the RPEL domain for G-actin binding, thereby promoting MRTF-A activation. Surface plasmon resonance (SPR) and immunofluorescence confirmed Pranlukast's strong binding affinity to G-actin and its ability to promote MRTF-A nuclear translocation. In vivo experiments demonstrated Pranlukast's efficacy in preventing ischemic stroke reperfusion injury and reversing synaptic structural and functional impairments caused by ischemic injury. This study, which integrates specific disease targets with deep learning-based intelligent screening, reveals for the first time that Pranlukast targets the G-actin/MRTF-A interaction via a non-classical pathway. By promoting MRTF-A nuclear translocation, it provides neurosynaptic protection post-ischemic reperfusion, offering a novel therapeutic strategy for ischemic stroke reperfusion injury.