Identification of Key Genes Governing the Effects of Physical Activity on Ferroptosis in Alzheimer's Disease Patients: A Machine Learning-Based Study

Disrupted brain iron metabolism and activated ferroptosis during ageing constitute significant precursors to neurodegenerative diseases. However, whether exercise intervention can modulate ferroptosis-related genes in the central nervous system remains unsystematically elucidated. This study integrated an aged cohort transcriptome dataset (GSE110298) from the GEO database (comprising 11 low-exercise and 23 high-exercise samples). Differential expression genes were screened using the limma package (|log2FC| > 1, FDR<0.05) using the limma package, intersecting these with the FerrDb V2.0 ferroptosis gene set to identify 42 exercise-responsive ferroptosis-associated genes (DFEGs). Multilevel bioinformatics analyses (KEGG/GO enrichment, STRING protein interaction networks, CytoHubba hub gene screening, GSEA pathway activity assessment, and miRNA-transcription factor regulatory network construction) revealed key molecular mechanisms. Hub gene identification: ACSL3, PPARD and TXN were identified as core targets regulating ferroptosis during exercise. Their altered expression was significantly correlated with lipid peroxidation inhibition (ACSL3), enhanced mitochondrial biogenesis (PPARD), and redox homeostasis restoration (TXN). Pathway Mechanisms: DFEGs exhibited significant enrichment in peroxisome metabolism (p=3.2x10-5, including 7 genes such as PEX3 and ACOX1) and the PPAR signalling pathway (p=1.8x10-{square}, including 5 genes such as PPARD and FABP3). GSEA analysis revealed a 31% reduction in Alzheimers disease-associated pathway activity in the high-exercise group (NES = -1.68, FDR = 0.032). Regulatory network: A multi-level regulatory network was constructed encompassing three hub genes, five transcription factors (FOXA2, HNF4A, etc.) and three exercise-responsive miRNAs (miR-124-3p, miR-182-5p, miR-93-5p). Among these, miR-124-3ps targeted inhibition of ACSL3/TXN (TargetScan score >90) may mediate exercise-induced resistance to ferroptosis. This study first reveals a novel mechanism whereby exercise synergistically regulates brain ferroptosis through multiple targets. It proposes that the ACSL3/PPARD/TXN expression combination could serve as a potential biomarker for assessing exercise-induced neuroprotective efficacy, and provides a theoretical basis for developing intervention strategies for neurodegenerative diseases based on exercise-epigenetic interactions.