Tracing fucosylation and astrocyte-associated pathogenic pattern in major depressive disorder: evidence from machine learning-based multi-omics and clinical validation.

Background: Major depressive disorder (MDD) remains a debilitating psychiatric condition with substantial global prevalence. Emerging evidence highlights the role of aberrant fucosylation (FUS) and astrocyte dysfunction as key contributors to MDD pathogenesis; however, their reciprocal molecular interactions remain poorly defined. Objective: To identify and validate a fucosylation-astrocyte (FA)-associated molecular signature and its central pathogenic factor in MDD through a multi-omics and machine learning framework. Methods: We employed Limma, xCell, and WGCNA on bulk RNA-seq data from MDD patient brain tissues (GSE54566, GSE54570) to identify FA-associated shared differentially expressed genes (DEGs). Lasso-logistic regression was applied to the GSE53987 (MDD patient bulk training set) to identify the FAA-associated central pathogenic factor which molecular functions was estimated by single-gene GSEA analysis. The diagnostic performance of this gene was validated in GSE53987 and GSE241921 (MDD patient bulk brain independent set). Consensus clustering identified FA-associated molecular subgroups in GSE54564 (MDD patient bulk brain validation set). Single-cell data (GSE336230) of MDD brain patient tissues explored the cellular localization and pathogenic role of the hub gene in astrocytes. Drug screening was performed using the Drug Reflector platform in GSE53987, followed by molecular docking for enrichment therapeutic candidate targeting hub gene for MDD. Finally, clinical validation of hub gene expression was performed on MDD brain postmortem amygdala samples from 10 MDD patients and 10 controls using q-RT-PCR. Results: We identified 8 FA-associated shared DEGs, which can stratify MDD patient into 2 molecular groups. G6PD can be considered as the up-regulated potential diagnostic pathogenic factor, which was mainly distributed in astrocyte. BRD-K97481123 can be a potential therapeutic candidate with G6PD. Conclusion: Our findings unveil a novel FA-associated pathogenic signature can be considered as a potential therapeutic target for MDD. This study provides a new framework for understanding MDD to enhance clinical translation.