Deciphering the Multi-Target Synergistic Mechanisms of Anacardiaceae (Mangifera indica, Anacardium occidentale) and Poaceae (Bambusa vulgaris) Leaf Combinations in Inflammatory and Metabolic Disorders: An Integrated Network Pharmacology, Molecular Dynamics, and Machine Learning Workflow
Abstract
Chronic metabolic inflammation—often called metaflammation—underpins the global rise of type 2 diabetes, non-alcoholic fatty liver disease, and atherosclerotic cardiovascular disease. Single-target pharmaceuticals frequently fall short because these conditions emerge from densely interconnected signaling networks rather than isolated molecular defects. Here we present a rigorously integrated computational pipeline that moves beyond conventional network pharmacology to decipher the multi-target synergistic mechanisms of a traditional African polyherbal combination: leaves of Mangifera indica and Anacardium occidentale (Anacardiaceae) together with Bambusa vulgaris (Poaceae). Starting with literature-curated and LC-MS-validated phytochemical libraries, we applied machine-learning classifiers (SwissADME, pkCSM, ProTox-II) to filter 87 initial compounds down to 23 drug-like candidates satisfying Lipinski and Veber criteria with acceptable toxicity profiles. Target fishing via SwissTargetPrediction intersected with GeneCards-derived disease genes for inflammatory-metabolic disorders yielded 68 high-confidence therapeutic targets. Protein–protein interaction network topology (STRING + Cytoscape/CytoHubba) revealed a tightly clustered interactome whose top hubs—TNF, AKT1, RELA, PPARG, PRKAA1 (AMPKα), mTOR, and NFE2L2 (Nrf2)—sit at the crossroads of insulin signaling, oxidative stress defense, and inflammasome activation. Gene Ontology and KEGG enrichment confirmed convergent modulation of the AMPK signaling pathway, Nrf2-mediated antioxidant response, NF-κB/NLRP3 inflammatory axis, and mTOR-driven autophagy. To move past static docking scores, we performed flexible molecular docking followed by 100-nanosecond explicit-solvent molecular dynamics simulations in GROMACS. Four top-scoring complexes (mangiferin–AMPK, orientin–Nrf2, anacardic acid–NLRP3, vitexin–mTOR) all reached structural equilibrium within 35 ns, with backbone RMSD values stabilizing below 0.25 nm. MM-PBSA free-energy calculations returned highly favorable binding affinities (ΔGbind ranging from −31.6 to −42.1 kcal mol⁻¹), driven predominantly by van der Waals and electrostatic interactions that remained stable across the trajectories. These in silico findings provide the first quantitative, dynamics-validated evidence that the chosen Anacardiaceae–Poaceae leaf combination can simultaneously reinforce metabolic sensing (AMPK), bolster endogenous antioxidant capacity (Nrf2), and dampen sterile inflammation (NLRP3/NF-κB). The workflow itself—ML-ADMET filtering, graph-theoretic hub prioritization, flexible docking, and physics-based MD/MM-PBSA—offers a transferable template for future studies of African polyherbal medicines. Downstream experimental validation using Western blotting, RT-qPCR, and metabolomic profiling in relevant cell and animal models is now warranted to translate these computational predictions into standardized, evidence-based therapeutic formulations.