利用机器学习鉴定并验证骨关节炎中潜在的双硫死亡相关基因

Identification and Validation of Disulfidptosis-Related Genes in Osteoarthritis by Machine Learning

骨关节炎(Osteoarthritis, OA)是一种普遍影响中老年人群的慢性炎性疾病,其典型表现为软骨退行性变、关节炎症以及功能受损。面对目前缺乏有效治疗手段的现状,迫切需要更为全面的理论指导。双硫死亡,作为一种新近发现的细胞死亡方式,在多种疾病中均有涉及,其与骨关节炎的关联尚未明确。本研究基于已报道的双硫死亡相关基因,运用GEO数据库中的三组数据集(GSE55235, GSE55457, GSE82107),采用多元机器学习技术将骨关节炎分为两种亚型。研究识别出九个潜在的双硫死亡相关关键差异表达基因(hubDEGs),其中APOD、EDNRB、FOXC2、JUN、LRCH1和MAFF在骨关节炎患者中呈下调趋势,而NUDT1、PNMAL1和ZNF668则呈上调趋势。深入探究发现,这些hubDEGs与骨关节炎中免疫细胞浸润的变化密切相关,主要表现为嗜酸细胞和Th2细胞浸润的减少,以及iDC、巨噬细胞、NK细胞、Treg细胞、Tfh细胞和Th1细胞浸润的增加。利用GSE89408作为独立验证数据集,结合qRT-PCR实验验证了这九个基因在双硫死亡与骨关节炎关联中的关键作用。此外,本研究还构建了一个TF-mRNA-miRNA网络,鉴定出可能的干预靶点,包括CTCF、SP1、hsa-miR-29和hsa-miR-424。这项研究首次揭示了骨关节炎与双硫死亡之间的潜在联系,并识别出参与骨关节炎中双硫死亡过程的关键基因和靶点,为骨关节炎的临床诊断和治疗提供了新颖视角。

Osteoarthritis (OA) is a chronic inflammatory disease that afflicts a large population of middle-aged people, manifesting as cartilage degeneration, joint inflammation, and functional impairment. Currently, effective therapeutic interventions are lacking, which calls for more comprehensive theoretical guidance. Disulfidptosis, a newly discovered form of cell death implicated in various diseases, may be related to OA, but this has not been explored before. Based on previously reported disulfidptosis-related genes, this study used three datasets from the GEO database (GSE55235, GSE55457, GSE82107) and applied various machine learning methods to classify OA into two subtypes. Nine potential disulfidptosis-related hub of Differentially Expressed Genes (hubDEGs) were identified, among which APOD, EDNRB, FOXC2, JUN, LRCH1, and MAFF were down-regulated, and NUDT1, PNMAL1, and ZNF668 were up-regulated. Further exploration revealed that hubDEGs were associated with altered immune cell infiltration in OA, marked by decreased infiltration of eosinophils and Th2 cells, and increased infiltration of iDC, macrophages, NK cells, Treg cells, Tfh cells, and Th1 cells. GSE89408, as an independent dataset, and qRT-PCR confirmed the critical role of these nine genes in the association between disulfidptosis and OA. Moreover, a TF-mRNA-miRNA network was constructed, identifying potential intervention targets, such as CTCF, SP1, hsa-miR-29, and hsa-miR-424. This study is the first to reveal a possible link between OA and disulfidptosis, and to uncover key genes and targets involved in disulfidptosis in OA, providing novel insights for the clinical diagnosis and treatment of OA.