Integrated machine learning, single-cell RNA sequencing, and experimental validation for exploratory investigation of Fructus Aurantii in treating major depressive disorder

Fructus Aurantii (FA), a Chinese herbal medicine, has been indicated to have antidepressant effects in our previous study. The potential mechanism of FA in the therapy of major depressive disorder (MDD) was explored through machine learning, single-cell RNA sequencing (scRNA-seq), and experimental validation. We conducted differential analysis and WGCNA on the expression cohorts of MDD to determine differentially expressed genes and key module genes. FA-related targets were retrieved from multiple databases. Through the intersection of the aforementioned genes, candidate core targets were screened using the cytoHubba plug-in and machine learning methods. Subsequent analyses of these targets included clinical diagnostic evaluation, enrichment analysis, immune infiltration analysis, and molecular docking. scRNA-seq was used to explore the potential cellular mechanisms of MDD. Finally, RT-qPCR was employed to verify the gene expression level of the candidate core targets in the hippocampus. Four candidate core targets (ALOX5, PIK3CG, CHUK, and LRRK2) were determined using the cytoHubba plug-in and machine learning methods (LASSO regression, SVM-RFE, and Boruta). These targets exhibited notably elevated expression levels in the MDD group, and this finding was validated via RT-qPCR. The nomogram showed favorable value in clinical decision-making and diagnosis. Integrated analysis of signaling pathways and immune cells indicated that immune inflammation exerts an impact on MDD. Moreover, eosinophils and macrophages exhibited strong correlations with these targets and played crucial roles in immune inflammation. The active components of FA (limonin, auraptene, isosinensetin, nomilin, and luteolin) showed high binding affinity to these candidate core targets. Additionally, the expression of LRRK2 changed remarkably during the developmental stage of excitatory neurons. This research suggests that FA may exert its therapeutic effects through candidate core targets including ALOX5, PIK3CG, CHUK, and LRRK2, and key pathways associated with inflammation, immune dysregulation, and neuronal damage, providing a new perspective and important reference for subsequent MDD research and the development of therapeutic strategies.