类风湿性关节炎（RA）是一种常见的自身免疫性疾病，其发病机制尚不清楚。实验室生物标志物的缺陷导致缺乏临床上可用的 RA 早期诊断和治疗策略。本研究旨在鉴定血清外泌体 lncRNA 作为有前途的生物标志物，并揭示它们影响成纤维样滑膜细胞 (FLS) 特征基因以诱导 RA 恶性特性的潜在机制。有目的地挖掘血清外泌体（GSE271161和PRJNA911001）和FLS（GSE103578、GSE122616、GSE128813、GSE181614和GSE83147）的RNA测序数据集。对差异表达（DE）lncRNA/蛋白质编码基因进行可视化和功能富集，筛选重要的lncRNA，并构建竞争性内源RNA（ceRNA）和蛋白质-蛋白质相互作用（PPI）网络。对验证队列进行实时定量 PCR、受试者工作特征曲线 (ROC) 和相关性分析。结果，我们总共筛选了131个血清外泌体DElncRNA和125个FLS DEmRNA，它们主要富集于增殖、炎症和代谢途径。使用4种机器学习算法（SVM、KNN、RF、Logit）对DElncRNA表达谱进行深度学习，以识别具有更好性能的模型和具有更高重要性分数的lncRNA，从而建立了连接血清外泌体lncRNA和FLS的特征基因。简而言之，我们提出4种RA代表的血清外泌体lncRNA（DLEU2、FAM13A-AS1、MEG3和SNHG15）可能作为实验室检测的有价值的指标，它们介导的细胞间通讯和ceRNA网络可能调控FLS的特征基因，从而在 RA 中产生恶性表型和适应性滑膜微环境。版权所有 © 2024 Elsevier B.V. 保留所有权利。

Rheumatoid arthritis (RA) is a common autoimmune disease whose pathogenesis is poorly understand. Gaps in laboratory biomarkers cause a lack of clinically available strategies for the early diagnosis and treatment of RA. This study aims to identify serum exosomal lncRNAs as promising biomarkers and to unravel potential mechanisms by which they affect characteristic genes of fibroblast-like synoviocytes (FLSs) to induce RA malignant properties. RNA sequencing datasets of serum exosomes (GSE271161 and PRJNA911001) and FLSs (GSE103578, GSE122616, GSE128813, GSE181614 and GSE83147) were purposively mined. Visualization and functional enrichment of differentially expressed (DE) lncRNAs/protein-coding genes, screening of significant lncRNAs, and construction of competing endogenous RNAs (ceRNAs) and protein-protein interaction (PPI) network were carried out. Quantitative real-time PCR, receiver operating characteristic curve (ROC) and correlation analysis were conducted on the validation cohort. As a result, we screened a total of 131 serum exosomal DElncRNAs and 125 FLSs DEmRNAs, which were predominantly enriched in the proliferative, inflammatory and metabolic pathways. In-depth learning of DElncRNAs expression profiles was performed to identify models with better performance and lncRNAs with higher importance scores using 4 machine learning algorithms (SVM, KNN, RF, Logit), which led to the establishment of ceRNAs network linking serum exosomal lncRNAs and characteristic genes of FLSs. In short, we proposed that 4 RA-representative serum exosomal lncRNAs (DLEU2, FAM13A-AS1, MEG3 and SNHG15) may be applied as valuable indicators for laboratory tests, and their-mediated intercellular communication and ceRNAs network may regulate the characteristic genes of FLSs, thereby generating malignant phenotypes and adaptive synovial microenvironment in RA.Copyright © 2024 Elsevier B.V. All rights reserved.