Table_4_Identification and validation of a novel mitochondrion-related gene signature for diagnosis and immune infiltration in sepsis.xlsx

BackgroundOwing to the complex pathophysiological features and heterogeneity of sepsis, current diagnostic methods are not sufficiently precise or timely, causing a delay in treatment. It has been suggested that mitochondrial dysfunction plays a critical role in sepsis. However, the role and mechanism of mitochondria-related genes in the diagnostic and immune microenvironment of sepsis have not been sufficiently investigated.
MethodsMitochondria-related differentially expressed genes (DEGs) were identified between human sepsis and normal samples from GSE65682 dataset. Least absolute shrinkage and selection operator (LASSO) regression and the Support Vector Machine (SVM) analyses were carried out to locate potential diagnostic biomarkers. Gene ontology and gene set enrichment analyses were conducted to identify the key signaling pathways associated with these biomarker genes. Furthermore, correlation of these genes with the proportion of infiltrating immune cells was estimated using CIBERSORT. The expression and diagnostic value of the diagnostic genes were evaluated using GSE9960 and GSE134347 datasets and septic patients. Furthermore, we established an in vitro sepsis model using lipopolysaccharide (1 µg/mL)-stimulated CP-M191 cells. Mitochondrial morphology and function were evaluated in PBMCs from septic patients and CP-M191 cells, respectively.
ResultsIn this study, 647 mitochondrion-related DEGs were obtained. Machine learning confirmed six critical mitochondrion-related DEGs, including PID1, CS, CYP1B1, FLVCR1, IFIT2, and MAPK14. We then developed a diagnostic model using the six genes, and receiver operating characteristic (ROC) curves indicated that the novel diagnostic model based on the above six critical genes screened sepsis samples from normal samples with area under the curve (AUC) = 1.000, which was further demonstrated in the GSE9960 and GSE134347 datasets and our cohort. Importantly, we also found that the expression of these genes was associated with different kinds of immune cells. In addition, mitochondrial dysfunction was mainly manifested by the promotion of mitochondrial fragmentation (p<0.05), impaired mitochondrial respiration (p<0.05), decreased mitochondrial membrane potential (p<0.05), and increased reactive oxygen species (ROS) generation (p<0.05) in human sepsis and LPS-simulated in vitro sepsis models.
ConclusionWe constructed a novel diagnostic model containing six MRGs, which has the potential to be an innovative tool for the early diagnosis of sepsis.

背景：由于脓毒症（sepsis）复杂的病理生理特征与异质性，当前的诊断方法缺乏足够的精准性与时效性，易导致治疗延误。已有研究表明，线粒体功能障碍在脓毒症进程中发挥关键作用，但线粒体相关基因在脓毒症诊断及免疫微环境中的作用与机制尚未得到充分研究。
方法：本研究从GSE65682数据集的人类脓毒症样本与正常对照样本中筛选出线粒体相关差异表达基因（differentially expressed genes, DEGs）。通过最小绝对收缩和选择算子（Least absolute shrinkage and selection operator, LASSO）回归与支持向量机（Support Vector Machine, SVM）分析，挖掘潜在的诊断生物标志物。采用基因本体（Gene Ontology, GO）富集分析与基因集富集分析（Gene Set Enrichment Analysis, GSEA），鉴定与这些生物标志物基因相关的关键信号通路。此外，借助CIBERSORT算法评估这些基因与浸润免疫细胞比例的相关性。利用GSE9960、GSE134347数据集以及脓毒症患者队列，验证诊断基因的表达水平与诊断价值。本研究还采用脂多糖（lipopolysaccharide, LPS）（浓度为1 µg/mL）刺激CP-M191细胞，构建体外脓毒症模型。分别对脓毒症患者外周血单个核细胞（peripheral blood mononuclear cell, PBMC）与CP-M191细胞的线粒体形态与功能进行检测评估。
结果：本研究共筛选得到647个线粒体相关差异表达基因。经机器学习分析，确定6个关键线粒体相关差异表达基因，分别为PID1、CS、CYP1B1、FLVCR1、IFIT2及MAPK14。基于这6个基因构建诊断模型，受试者工作特征（receiver operating characteristic, ROC）曲线结果显示，该模型区分脓毒症样本与正常对照样本的曲线下面积（area under the curve, AUC）达1.000，该结果在GSE9960、GSE134347数据集以及本研究队列中得到验证。值得注意的是，本研究还发现这些基因的表达水平与多种免疫细胞存在相关性。此外，在人类脓毒症样本与脂多糖诱导的体外脓毒症模型中，线粒体功能障碍主要表现为线粒体碎片化程度升高（p<0.05）、线粒体呼吸功能受损（p<0.05）、线粒体膜电位降低（p<0.05）以及活性氧（reactive oxygen species, ROS）生成增多（p<0.05）。
结论：本研究构建了一种包含6个线粒体相关基因（mitochondria-related genes, MRGs）的新型诊断模型，该模型有望成为脓毒症早期诊断的创新性工具。