Unveiling Cuproptosis-Driven Molecular Clusters and Immune Dysregulation in Ankylosing Spondylitis.

Abstract

Background: Ankylosing spondylitis (AS) is a chronic autoimmune disease characterized by inflammation of the sacroiliac joints and spine. Cuproptosis is a newly recognized copper-induced cell death mechanism. Our study explored the novel role of cuproptosis-related genes (CRGs) in AS, focusing on immune cell infiltration and molecular clustering.

Methods: By analyzing the peripheral blood gene expression datasets obtained from GSE73754, GSE25101, and GSE11886, we identified the expression patterns of cellular factors and immune infiltration cell related to cuproptosis. Subsequently, we employed weighted gene co-expression network analysis (WGCNA) to identify differentially expressed genes (DEGs) within each cluster and utilized the "GSVA" and "GSEABase" software packages to examine variations in gene sets enriched across various CRG clusters. Finally, we selected the best-performing machine learning model to predict genes associated with AS. Datasets (GSE25101 and GSE73754) and ELISA to assess the expression levels of the five genes and their corresponding proteins.

Results: Seven cuproptosis-related DEGs and four immune cell types were identified, revealing significant immune heterogeneity in the immune cell infiltration between the two cuproptosis-related molecular clusters in AS. The eXtreme Gradient Boosting (XGB) model showed the highest predictive accuracy, achieving an area under the receiver operating characteristic curve (AUC) of 0.725, and 5-gene prediction models were established. It showed satisfactory performance in the GSE25101 dataset (AUC = 0.812). According to the blood serum samples of AS patients and controls, PELI1 had a higher expression level (AUC = 0.703, p = 0.07), while ICAM2 and RANGAP1 had lower expression levels (AUC = 0.724, 0.745, and p = 0.011, 0.000, respectively) in AS patients.

Conclusion: We explored the correlation of cuproptosis in AS, and developed the optimal machine learning model to identify high-risk genes associated with AS. We also explored the pathogenesis and treatment strategies of AS, targeting PELI1, ICAM2, and RANGAP1.