Identification of mitochondrial dynamics-related biomarkers in psoriasis using bioinformatics and Mendelian randomization

Mitochondrial dynamics (MD) are crucial in various inflammatory disorders, yet the specific mechanisms involved in psoriasis remain inadequately understood. Thus, this study aims to discover potential biomarkers and explore the mechanisms related to MD in psoriasis by employing bioinformatics methods in conjunction with the Mendelian randomization (MR) approach. In this investigation, datasets associated with psoriasis, specifically (GSE14905, GSE13355, and ukb-a-100), alongside genes pertinent to MD (MDRGs), were employed. The initial step involved the identification of significant module genes associated with MD through weighted gene co-expression network analysis. Subsequently, the identified module genes were cross-referenced with differentially expressed genes discerned between psoriasis and control groups to extract differentially expressed MDRGs. Additionally, MR analysis was conducted to identify potential candidate genes. The definitive potential biomarkers were determined through protein–protein interaction (PPI) networks, machine learning methodologies, receiver operating characteristic analysis, and expression profiling. Finally, gene set enrichment analysis, alongside immune infiltration and immune response assessments, was executed to elucidate the underlying mechanisms by which the potential biomarkers function in the context of psoriasis. There were 3136 key module genes through weighted gene co-expression network analysis and 643 differentially expressed MDRGs by crossing key module genes and 4310 differentially expressed genes. Afterward, 56 candidate genes with causal relationship to psoriasis were selected by MR analysis. Then 19 hub genes from PPI network were used to further screen 6 feature genes by machine learning, and they had a better ability to distinguish psoriasis (area under the curve > 0. 7). C1orf43, SNF8, STOML2, and MRPS16 were identified as potential biomarkers in psoriasis, and were co-enriched in pyrimidine metabolism, DNAᵣeplication, and proteasome. Eventually, there were 11 differential immune cells (memory B cells, activated dendritic cells, etc) and 13 differential immune responses (antigen processing and presentation, antimicrobials, etc) between psoriasis and control samples in psoriasis (P <. 05). C1orf43, SNF8, STOML2, and MRPS16 were identified as potential biomarkers linked to MD in psoriasis, which provide promising leads for further investigation. These biomarkers require experimental validation to confirm their role in the pathogenesis of psoriasis and their potential as therapeutic targets.