目的 综合分析脓毒症相关单细胞转录组学和转录组学数据集，探讨脓毒症相关态靶病理机制，构建脓毒症态靶预后模型，并挖掘潜在靶向中药及其活性成分。方法 运用差异表达基因分析和韦恩图鉴定出脓毒症态靶相关基因。通过蛋白质-蛋白相互作用（protein-protein interaction，PPI）网络连接度使用随机游走算法得到关键模块基因，采用基因本体论（gene ontology，GO）的生物过程（biological process，BP）和京都基因与基因组百科全书（Kyoto encyclopedia of genes and genomes，KEGG）深入分析关键模块基因的生物学功能；单细胞转录组分析探讨脓毒症态靶相关基因在其中的作用。通过脓毒症患者的基因表达数据及临床生存数据构建脓毒症态靶预后模型。采用脓毒症小鼠模型和实时定量聚合酶链反应（real-time quantitative polymerase chain reaction，RT-PCR）验证脓毒症态靶预后相关基因的表达水平。通过逆向网络药理学筛选脓毒症态靶相关基因的潜在靶向中药及其入血活性成分，并采用分子对接技术验证活性成分与预后相关靶点之间的结合性能。结果 得到70个脓毒症毒态相关基因、67个脓毒症瘀态相关基因、54个脓毒症虚态相关基因，毒态关键模块基因与炎症密切相关，瘀态关键模块基因与凝血密切相关，虚态关键模块基因与细胞稳态密切相关。AUCell评分显示，单核细胞对脓毒症毒态和虚态作用评分最高，血小板对脓毒症瘀态作用评分最高。毒态高评分单核细胞中代谢与炎症相关基因及通路的活性明显上调；瘀态高评分血小板中血小板活化与凝血/血栓形成相关基因及通路的活性明显上调；相对于正常组，脓毒症患者中单核细胞存在大量表达下调的基因，涉及抗原呈递、细胞能量代谢、细胞周期、蛋白质合成等。基于8个靶点［白细胞介素-1受体2型（interleukin-1 receptor type 2，IL1R2）、ADP-核糖基化因子样蛋白4C（ADP ribosylation factor like GTPase 4C，ARL4C）、细胞色素C氧化酶亚基7B（cytochrome C oxidase subunit 7B，COX7B）、真核翻译起始因子2亚基γ（eukaryotic translation initiation factor 2 subunit gamma，EIF2S3）、立即早期反应3（immediate early response 3，IER3）、LSM1同源物（LSM1 homolog，LSM1）、弗林蛋白酶（paired basic amino acid cleaving enzyme，FURIN）和锚蛋白重复结构域9（ankyrin repeat domain-containing protein 9，ANKRD9）］构建的脓毒症态靶预后模型具有良好的生存预后预测能力，动物实验及RT-PCR验证了其表达水平。基于中药分子机制生物信息学分析工具数据库（bioinformatics analysis tool for molecular mechanism of traditional Chinese medicine，BATMAN-TCM）和中药血液吸收成分及代谢物数据库（database of constituents absorbed into the blood and metabolites of traditional Chinese medicine，DCABM-TCM）的逆向网络药理学分析筛选得到多种具有清热解毒、活血化瘀、补益扶正功效的中药及其入血活性成分，分子对接表明代表性入血活性成分与预后相关靶点具有良好的结合性能。结论 单核细胞代谢重编程驱动下的炎症反应激活可能是脓毒症毒态的主要病理机制；血小板活化-凝血/血栓形成可能是脓毒症瘀态的主要病理机制；单核细胞功能紊乱-免疫抑制可能是脓毒症虚态的主要病理机制。基于8个基因的模型可作为脓毒症态靶预后的风险预测模型。

Objective To explore “state and target” pathological mechanism of sepsis by comprehensively analyzing the single-cell transcriptomics and transcriptomic datasets related to sepsis, construct the prognostic model for “state and target” of sepsis, and explore potential target Chinese medicines and active ingredients. Methods Differential expression gene analysis and venn diagram analysis were used to identify “state and target” of sepsis related genes. Key module genes were obtained using the random walk algorithm through the connectivity of the protein-protein interaction (PPI) network. Gene ontology (GO) biological process (BP) and Kyoto encyclopedia of genes and genomes (KEGG) were used to analyze the biological functions of key module genes. To explore the role of “state and target” of sepsis related genes by single-cell transcriptome analysis. A prognostic model for “state and target” of sepsis was constructed using gene expression data and clinical survival data from sepsis patients. The expression levels of prognostic genes for “state and target” of sepsis were validated by constructing sepsis mouse model and real-time quantitative polymerase chain reaction (RT-PCR). Through reverse network pharmacology, potential targeted traditional Chinese medicines and their active components were screened, and molecular docking was used to verify the binding performance between active ingredients and prognostic targets. Results A total of 70 genes related to “toxic state” of sepsis, 67 genes related to “stasis state” of sepsis and 54 genes related to “deficiency state” of sepsis were obtained. The key module genes of “toxic state” of sepsis were closely related to inflammation, the key module genes of “stasis state” of sepsis were closely related to coagulation, and the key module genes of “deficiency state” of sepsis were closely related to cell homeostasis. AUCell score showed that monocyte had the highest score for “toxic state” and “deficiency state” of sepsis. Platelet had the highest score for “stasis state” of sepsis. The activities of metabolic and inflammatory related genes and pathways were significantly up-regulated in monocytes with high “toxic state” score. Platelet activation and coagulation/thrombosis related genes and pathways were significantly up-regulated in platelet with high “stasis state” score. Compared with control group, there were a large number of down-regulated genes in monocytes of sepsis patients, which were involved in antigen presentation, cell energy metabolism, cell cycle, protein synthesis, etc. The prognostic model for “state and target” of sepsis constructed based on eight targets [interleukin-1 receptor type 2 (IL1R2), ADP ribosylation factor like GTPase 4C (ARL4C), cytochrome C oxidase subunit 7B (COX7B), eukaryotic translation initiation factor 2 subunit gamma (EIF2S3), immediate early response 3 (IER3), LSM1 homolog (LSM1), paired basic amino acid cleaving enzyme (FURIN) and ankyrin repeat domain-containing protein 9 (ANKRD9)] had good prognostic predictive ability, and the expression levels of these genes were validated using animal experiments and RT-PCR. Based on the bioinformatics analysis tool for molecular mechanism of traditional Chinese medicine (BATMAN-TCM) database and database of constituents absorbed into the blood and metabolites of traditional Chinese medicine (DCABM-TCM), reverse network pharmacology analysis had screened multiple Chinese herbal medicines with effects of clearing heat and detoxifying, activating blood and resolving stasis, and reinforcing healthy qi and their active ingredients entering the blood. Molecular docking showed that representative active ingredients have good binding properties with prognostic targets. Conclusion Activation of inflammatory response driven by monocyte metabolic reprogramming may be the main pathological mechanism of “toxic state” of sepsis. Platelet activation-coagulation/thrombosis may be the main pathological mechanism of “stasis state” of sepsis. Monocyte dysfunction-immunosuppression may be the main pathological mechanism of “deficiency state” of sepsis. A model based on eight genes can serve as a risk prediction model for the prognosis of “state and target” of sepsis.

R285

国家中医药管理局中医药创新团队及人才支持计划项目(ZYYCXTD-D-202203);广州市科技局市校院联合资助项目-广州市中西医结合防治新发突发传染病重点实验室(202201020382);广东省科技计划项目(2023B1212060062);广东省中医急症研究重点实验室开放项目(KF2023JZ06)