PHENOLPROPANE AND MEGASTIGMANE FROM Motandra paniculata (POIR) I.M. TURNER (APOCYNCEAE) LEAVES: ARTIFICIAL INTELLIGENCE PREDICTIVE ANTIINFLAMMATORY ACTIVITY AND MOLECULAR DOCKING STUDIES
JOSEPH OISEMUZEIMEN OISEOGHAEDE1*, AMINAT ASABI OYAWALUJA1, BAMISAYE OLAOFE OYAWALUJA2, ABOLADE DAVID OMIYALE3, PECULIAR FEENNA ONYEKERE 4,5
- Department of Pharmacognosy, Faculty of Pharmacy, University of Lagos, College of Medicine of the University of Lagos campus, Idi-Araba, Surulere, Lagos, Nigeria.
- Department of Pharmaceutical Chemistry, Faculty of Pharmacy, University of Lagos, College of Medicine of the University of Lagos campus, Idi-Araba, Surulere, Lagos, Nigeria.
- Department of Systems Engineering, University of Lagos, Akoka, Yaba, Lagos, Nigeria.
- Department of Pharmaceutical Sciences, Retzky College of Pharmacy, University of Illinois, Chicago, Chicago 60612, Illinois, USA.
- Department of Pharmacognosy and Environmental Medicine, Faculty of Pharmaceutical Sciences, University of Nigeria, Nsukka 410001, Enugu, Nigeria
Afr. J Pharm Res Dev; Volume 18(1): 401-407 ; 2026
ABSTRACT
Motandra paniculata, a plant indigenous to Nigeria, is traditionally used in some African communities for the management of pain. This study was undertaken to investigate the anti-inflammatory activity of a phenolpropane (β-hydroxypropiovanillone) and a megastigmane (isololiolide) isolated from the leaves of M. paniculata. This research sought to evaluate their potential as anti-inflammatory leads using in silico docking studies, alongside predictive assessments from Artificial Intelligence (AI) models, specifically Graph Neural Networks (GNNs) and Chemical Bert Representation Task (ChemBERTa). Targeted and blind in silico molecular docking were performed to assess the suitability of the compounds as inhibitors of cyclooxygenase enzymes (COX-1 and COX-2). Separately, a Simplified Molecular Input Line Entry System (SMILES)-based transformer model (ChemBERTa) and a GNN architecture enabled the dual-modal prediction of the compounds’ bioactivity. The phenolpropane (β-hydroxypropiovanillone) exhibited stronger affinity and more reliable binding in the docking studies than the megastigmane (Isololiolide). However, both compounds received similar, moderate-to-low AI-predictive anti-inflammatory activity scores (0.398–0.489) from the GNN and ChemBERTa models. This suggests a low likelihood of potent in vivo activity in their current scaffolds, despite promising target engagement. The results demonstrate that the phenolpropane (β-hydroxypropiovanillone) and megastigmane (isololiolide) isolated from M. paniculata could serve as valuable anti-inflammatory leads. The conflict between their strong target binding (docking) and moderate predicted bioactivity (AI models) highlights the necessity of structural optimisation to improve their overall pharmacological profile for successful drug development.
Keywords: Phenolpropane, megastigmane, anti-inflammatory, docking, artificial intelligence
Email of correspondence: joiseoghaede@unilag.edu.ng;
https://doi.org/10.59493/ajopred/2026.1.1 ISSN: 0794-800X (print); 1596-2431 (online)