[Key genes of intervertebral disc degeneration with ferroptosis and prediction of potential traditional Chinese medicine based on bioinformatics and machine learning].

This study aimed to identify key ferroptosis-related genes associated with intervertebral disc degeneration(IDD) through bioinformatics and machine learning algorithms, predict potential TCM, and conduct experimental validation. IDD datasets(GSE124272 and GSE150408) were obtained from the GEO database, while ferroptosis-related genes were retrieved from the FerrDb and GeneCards databases. Data preprocessing was performed using R. Weighted gene co-expression network analysis(WGCNA) was employed to identify differentially expressed genes(DEGs), which were intersected with ferroptosis-related genes to construct a protein-protein interaction(PPI) network. Gene Ontology(GO) and Kyoto Encyclopedia of Genes and Genomes(KEGG) enrichment analyses were subsequently conducted. Three machine learning algorithms, i.e., LASSO regression, support vector machine-recursive feature elimination(SVM-RFE), and random forest, were applied to screen feature genes and construct prediction models. Potential TCM was predicted through the Coremine Medical database and validated through molecular docking. For experimental validation, mouse nucleus pulposus cells were divided into a normal control group, an erastin(1 μmol·L~(-1))-induced model group, and an icariin(10 μmol·L~(-1)) intervention group. Cell viability, oxidative stress indicators [reactive oxygen species(ROS), lipid peroxides(LPO), reduced glutathione(GSH), oxidized glutathione(GSSG), and malondialdehyde(MDA)], and mitochondrial function were assessed. Immunofluorescence staining was used to detect the levels of aggrecan and type Ⅱ collagen(collagen Ⅱ), and Western blot was performed to evaluate the expression of key proteins, including glutathione peroxidase 4(GPX4), signal transducer and activator of transcription 3(STAT3), and acyl-CoA synthetase long-chain family member 1(ACSL1). The results identified 14 ferroptosis-related DEGs associated with IDD. GO analysis revealed enrichment in biological processes such as interleukin(IL)-8 and IL-17 regulation, radial glial cell differentiation, vascular endothelial growth factor regulation, and inflammatory responses; cellular components related to lysosomes; and molecular functions involving protein kinase binding. KEGG pathway analysis showed significant enrichment in the human cytomegalovirus infection pathway. Machine learning algorithms identified four key genes, including perilipin 4(PLIN4), STAT3, lysine demethylase 6B(KDM6B), and ACSL1, with the constructed diagnostic model achieving an area under the curve(AUC) of 0.890. Potential TCM, including Epimedii Folium, Tripterygii Radix, and Cuscutae Semen, were predicted, among which icariin showed high binding affinity to ACSL1 and STAT3. Experimental validation demonstrated that compared with the normal control group, the model group showed significantly decreased cell viability, elevated levels of ROS, LPO, GSSG, and MDA, and increased expression of STAT3 and ACSL1 proteins, along with impaired mitochondrial membrane potential and activity. In contrast, levels of GSH, aggrecan, collagen Ⅱ, and GPX4 protein expression were markedly reduced. Compared with the model group, icariin intervention significantly improved the oxidative stress status of the cells, reduced levels of ROS, LPO, GSSG, MDA, and expression of STAT3 and ACSL1 proteins, improved mitochondrial membrane potential and activity, and increased levels of GSH, aggrecan, collagen Ⅱ, and GPX4 protein expression. In conclusion, the diagnostic model based on PLIN4, STAT3, KDM6B, and ACSL1 provides a new strategy for the early diagnosis and treatment of IDD. Icariin may influence IDD progression by modulating ferroptosis-related genes and holds potential therapeutic value.