The identification and validation of histone acetylation-related biomarkers in depression disorder based on bioinformatics and machine learning approaches

Yujing Lv ¹ Shang Lei ¹ Yi Man ² Jie Chen ³ Guimei Xing ⁴ Mengqing Ma ¹ Jile Lv ¹ Lu Zhang ^5*

• ¹ Graduate School, Bengbu Medical University, Bengbu, China
• ² Department of Oncology，Anhui Jimin Cancer Hospital, Hefei, China
• ³ Department of Psychiartry, Affiliated Psychological Hospital of Anhui Medical University, Hefei, China
• ⁴ Department of Education，Anhui No.2 Provincial People’s Hospital, Hefei, Anhui Province, China
• ⁵ Department of Neurology，Anhui No.2 Provincial People’s Hospital, Hefei, China

Background: Some studies indicated that histone modification may be involved in depression disorder (DD). The maintenance of the histone acetylation state is the work of histone acetyltransferase (HAT) and histone deacetylase (HDAC), which is thought to be a potential diagnostic biomarker of depression. However, it is still unknown how histone acetylation-related genes (HAC-RGs) contribute to the onset and progression of DD.Methods: GSE76826 and GSE98793were obtained from the Gene Expression Omnibus (GEO) database, HAC-RGs were acquired from the GeneCards database. Initially, the differentially expressed genes (DEGs) in GSE76826 were investigated. We used weighted gene co-expression network analysis (WGCNA) to screen key module genes. Candidate genes were selected by intersecting DEGs, key module genes, and HAC-RGs, followed by functional analysis. Two machine learning algorithms were used to identify hub genes, which were used for drug prediction, immunological infiltration studies, nomogram construction, and regulatory network building. The expression levels were verified using the GSE76826 and GSE98793 datasets. Hub gene expression levels in the clinical samples were verified using reverse transcription quantitative PCR (RT-qPCR).Results: The 23 candidate genes were obtained by intersecting 2,316 DEGs, 1010 HAC-RGs and 2,617 key module genes. Three hub genes (JDP2, ALOX5 and KPNB1) were gained by two machine learning algorithms. The nomogram constructed based on these three hub genes showed high predictive accuracy. Additionally, the three hub genes were enriched in the kegg_ribosome. The 9 different immune cells were identified in GSE76826, which were associated with three hub genes. A hub gene-drug network (98 nodes, 106 edges) and an lncRNA-miRNA-mRNA network (56 nodes, 87 edges), were built using the database. The expression level verification indicated that, with the exception of the KPNB1 gene, the DD group had higher levels of JDP2 and ALOX5 and that the expression patterns in GSE76826 and GSE98793 were consistent, with RT-qPCR confirming higher ALOX5 and JDP2 expression in DD samples.Conclusion: This study identified three hub genes (JDP2, ALOX5, and KPNB1) associated with histone acetylation, offering new insight into the diagnosis and treatment of DD.