Screening of Core Genes and Mechanistic Exploration of Ketamine-induced Psychiatric Disorders: A Multibrain Region Transcriptomic and Machine Learning-based Bioinformatics Analysis

Background: Ketamine is a non-competitive N-methyl-D-aspartate (NMDA) receptor antagonist. It has attracted considerable attention for its rapid antidepressant effects in recent years, but ketamine-induced psychotic-like symptoms limit its clinical application. The molecular mechanisms and key targets underlying ketamine-induced psychiatric disorders remain unclear. Aims and Objectives: In this study, we utilized multi-brain region transcriptome data and bioinformatics methods to identify the key genes and pathways involved. Materials and Methods: First, we obtained transcriptome data of ketamine-treated and control brain tissues (including frontal cortex, hippocampus, striatum, and amygdala) from public databases (GEO). Simultaneously, we retrieved psychiatric disorder-related gene sets from the GeneCards database. For each brain region sample, we performed single-sample gene set enrichment analysis (ssGSEA) to calculate enrichment scores for the psychiatric disorder gene set and assess differences between groups. We applied Weighted Gene Co-expression Network Analysis (WGCNA) to identify gene modules associated with the high-expression phenotype and conducted Gene Ontology (GO) functional annotation. In each brain region, differentially expressed genes (DEGs) between the high-expression and control groups were identified and intersected with WGCNA modules to obtain candidate key genes. Based on these candidates, we used three machine learning algorithms (least absolute shrinkage and selection operator (LASSO) regression, support vector machine recursive feature elimination (SVM-RFE), and Random Forest) to obtain 12 sets of candidate feature genes, comparing model performance using receiver operating characteristic (ROC) curves and area under the curve (AUC). Results: The results indicated that the LASSO model for the frontal cortex exhibited the best performance, identifying nine feature genes (Galr1, Cbr3, Crem, Fosl2, Mypn, Maff, Rhbg, Tslp, Klra2). Further GO/KEGG enrichment analysis and protein-protein interaction (PPI) network analysis highlighted the close association of Fosl2 and Maff with ketamine-induced psychiatric disorders. Comparison with our prior proteomic data on the prefrontal cortex of a ketamine model revealed a markedly downregulated protein Cbr3. Subsequent quantitative polymerase chain reaction (qPCR) assays in a ketamine-induced psychiatric disorder mouse model confirmed these findings: Cbr3 was significantly downregulated, while Fosl2 and Maff were significantly upregulated in the prefrontal cortex, consistent with our analysis. Thus, Cbr3, Fosl2, and Maff were identified as core genes in ketamine-induced psychiatric disorders. Finally, we evaluated the correlation between these core genes and immune cell infiltration, and analyzed their functions in humans using Genotype-Tissue Expression (GTEx) data and genome-wide association study (GWAS) loci. Conclusion: This study comprehensively applied gene set enrichment, WGCNA, and machine learning to multi-brain region transcriptomes to systematically screen for potential core genes of ketamine-induced psychiatric disorders, with preliminary qPCR validation. These findings provide new insights into molecular markers and mechanisms in this field.