Inflammation-related biomarkers and berberine therapy in post-stroke depression: evidence from bioinformatics, machine learning, and experimental validation.

OBJECTIVE

Post-stroke depression (PSD), a common neuropsychiatric complication, significantly hinders stroke recovery and quality of life. Given the established role of inflammation in the pathogenesis of PSD, this study aimed to identify key inflammation-related genes and pathways using bioinformatics and machine learning and further evaluate the protective effects of traditional Chinese medicine (TCM) monomer compounds.

METHODS

PSD-related datasets (GSE16561, GSE98793) were obtained from the Gene Expression Omnibus (GEO). Differentially expressed genes (DEGs) were identified using the limma package, followed by functional enrichment analysis with Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG). Three machine learning algorithms-random forest, support vector machine-recursive feature elimination (SVM-RFE), and least absolute shrinkage and selection operator (LASSO)-were applied to screen inflammation-related hub genes. Immune cell infiltration was analyzed using single-sample gene set enrichment analysis (ssGSEA). Candidate TCM compounds were explored via the Coremine Medical database. A PSD rat model was established to validate hub gene expression and to assess the efficacy of berberine (BBR).

RESULTS

Analysis identified 35 inflammation-related DEGs (IDEGs) significantly enriched in immunological processes, including malaria pathogenesis, NETosis, innate immune deficiencies, Rap1 signaling, and IL-17 cascades. The integration of machine learning pinpointed TLR2 and CYP1B1 as core hub genes, demonstrating robust diagnostic performance in external validation. Molecular docking suggested a strong binding affinity between the TCM compound BBR and TLR2/CYP1B1 proteins. PSD rats exhibited prolonged immobility in forced swim/tail suspension tests and decreased sucrose preference versus controls, alongside neuronal damage, edema, and inflammatory infiltration (HE staining). BBR treatment reversed these behavioral deficits and pathological changes. Western blot analysis confirmed elevated TLR2 and CYP1B1 expression in PSD rats, significantly downregulated by BBR. Enzyme-linked immunosorbent assay (ELISA) showed increased serum IL-1β, IL-6, and TNF-α levels in PSD, which BBR effectively reduced.

CONCLUSION

This study identifies TLR2 and CYP1B1 as core inflammation-related genes in PSD. BBR demonstrates therapeutic efficacy as an active monomer compound against PSD, likely mediated through downregulating TLR2 and CYP1B1 expression, consequently diminishing the concentrations of pro-inflammatory mediators (IL-1β, IL-6, TNF-α) that mediate cerebroprotective actions.