Cell type-specific alterations in fatty acid metabolism in neuronal subpopulations of schizophrenia and construction of a diagnostic model

Objective This study integrates single-cell and bulk RNA-seq to investigate cell type-specific alterations in fatty acid metabolism-related genes in the dorsolateral prefrontal cortex (DLPFC) of schizophrenia (SCZ) patients and to evaluate their potential as diagnostic biomarkers. Methods Integrating single-cell sequencing data (9 SCZ patients and 14 controls) and Bulk RNA-seq data (GSE174407, GSE107638), cell subpopulation identification and annotation were performed using Seurat. Key genes were identified by integrating differential gene screening, transcriptional regulatory network construction (SCENIC), pseudotime analysis (Monocle 2), and functional enrichment. A multi-gene diagnostic model was established using LASSO regression. Model performance was validated using ROC curves, nomograms, and immune cell correlation analysis. Finally, an MK-801-induced mouse SCZ model was used to validate the expression of key genes via qPCR. Results The study found that specific neuronal cell subtypes (e.g., CUX2+ NeuN and OPRM1+ NeuN) were significantly upregulated in SCZ, and the differentially expressed genes (DEGs) in these cells (e.g., HSP90AA1 , HSPA1A , PTPRO ) were significantly enriched in fatty acid metabolism pathways. Further regression analysis identified five key genes associated with SCZ pathogenesis ( ACAA1 , ACAT2 , ACSS1 , PSME1 , and S100A10 ). Subsequent analysis indicated that these genes not only participate in inflammatory responses in neuronal cells, showing significant negative correlations with inflammatory genes ( p 0.05), but are also closely related to disease diagnosis and prognosis. A diagnostic model and nomogram for SCZ were constructed based on these genes. The area under the ROC curve (AUC) for the model was 0.856 in the training cohort and 0.779 in the validation cohort, indicating reliable predictive performance for SCZ diagnosis. The SCZ risk predicted by the nomogram closely matched the actual risk. Furthermore, the Decision Curve Analysis (DCA) curve showed that the central gene curve was above the gray line, indicating a significant net benefit from using the nomogram to predict SCZ risk. Finally, significant differential expression of related genes was also found in SCZ mice ( p 0.001). Conclusion This study reveals cell type-specific dysregulation of fatty acid metabolism in SCZ and provides a robust five-gene diagnostic model with translational potential.