Integrated network toxicology, machine learning, and multi-omics characterization of 2,4-DTBP neurotoxicity

2,4-Di-tert-butylphenol (2,4-DTBP) is an emerging alkylphenol antioxidant with potential neurotoxicity concerns, but its molecular toxicological basis remains poorly defined. Here, we applied an integrated framework combining network toxicology, machine learning-assisted target prioritization, molecular docking, cellular validation, and integrated transcriptomic-metabolomic profiling to delineate 2,4-DTBP-induced neurotoxic mechanisms. Network toxicology implicated PI3K/AKT, MAPK, calcium signaling, cytokine-cytokine receptor interaction, and ATP-binding cassette (ABC) transporter pathways as major 2,4-DTBP-associated axes. By integrating three machine learning algorithms including least absolute shrinkage and selection operator (LASSO) regression, support vector machine-recursive feature elimination (SVM-RFE), and random forest (RF), we convergently prioritized six candidate genes potentially associated with 2,4-DTBP-related neurotoxicity: Erb-B2 receptor tyrosine kinase 2 (ERBB2), colony-stimulating factor 1 receptor (CSF1R), RET proto-oncogene (RET), C-C motif chemokine ligand 22 (CCL22), ATP-binding cassette subfamily C member 5 (ABCC5), and ATP-binding cassette subfamily C member 1 (ABCC1). SHapley Additive exPlanations (SHAP) quantified their contributions and improved model interpretability, while molecular docking provided preliminary structural insights into the potential interactions between 2,4-DTBP and its corresponding target proteins. In SH-SY5Y cells, 100 μM 2,4-DTBP exposure elicited pronounced cytotoxicity characterized by oxidative stress, mitochondrial alterations, lipid peroxidation, intracellular iron accumulation, apoptosis, ferroptosis-related perturbations, and pronounced inflammatory responses. Western blotting further confirmed alterations in the phosphorylation levels of AKT, ERK, NF-κB p65, and CaMKII. Ferrostatin-1 mitigated the 2,4-DTBP-induced reduction in cell viability, oxidative stress, mitochondrial dysfunction, lipid peroxidation, Fe²⁺ accumulation, and cell death, while partially restoring the transcriptional changes in GPX4, SLC7A11, GCH1, and ACSL4. Multi-omics integration further revealed coordinated disruption of inflammatory programs, synapse-related pathways, amino acid and energy metabolism, and redox homeostasis. Collectively, these results support a multi-level mechanistic framework for understanding 2,4-DTBP-induced neurotoxicity and identify candidate molecular targets and potential biomarkers for assessing the environmental neurotoxicity risks of emerging phenolic contaminants.