Identification of differentially expressed genes (DEGs)
The logarithmically processed data were subjected to differential gene analysis by the online analysis software idep.951 (http://bioinformatics.sdstate.edu/idep/), setting the parameters to FC > 4 and p < 0.01. 1569 DEGs were obtained under this filtering condition, and based on the logFC values to distinguish them into 232 up-regulated genes and 1337 down-regulated genes (Fig. 2a), and visualize them using volcano plots (Fig. 2b).
Protein interaction (PPI) networks
To identify core genes from multiple perspectives, we then used the STRING database (https://cn.string-db.org/) to construct protein-protein interaction (PPI) networks and further identify the interactions between genes at the protein level. Identification of important nodes in the network graph was valuable in screening out the most critical genes. A total of 1569 DEGs were submitted to construct the PPI network and 11 genes located at the core of the network were identified, including regulation of cytokine production associated with inflammatory responses (CD6,CEACAM1), cell activation (LCK,NFATC2,CD3E,ZAP70,CARD11,CD6,CEACAM1), loss of response to human ( CEACAM1,LCK,PLCG1), regulation of immune system processes (CD3E,CARD11,CD247, NFATC2 ,PLCG1,LCK,CD6,CEACAM1,IL18R1), T cell receptor signalling pathways (LCK,ZAP70,CD3E.PLCG1,CD247,IL18R1, NFATC2), NK cell-mediated cytotoxicity (ZAP70,LCK,CD247,PLCG1,IL18R1,CARD11) (Fig. 2c).
Enrichment analysis of septic shock
Gene enrichment analysis is a method of analysing gene expression information. Enrichment is the process of classifying genes according to a priori knowledge, i.e. genome annotation information. When genes are classified, it can help to know whether the genes found have certain aspects in common (e.g. function, composition, etc.). We have used the DAVID online software database (https://david.ncifcrf.gov/) and assembled GO analysis to annotate differentially expressed gene functions and visualised them using the OmicShare online tool to screen for biomarkers related to the pathogenesis of septic shock (Fig. 3a). The Kyoto Encyclopedia of Genes and Genomes KEGG pathway analysis was used to screen the signalling pathways in which differentially expressed genes are involved. These DEGs are mainly involved in the T-cell receptor signalling pathway, the MAPK signalling pathway, the PD-L1 and PD1 checkpoint pathways in cancer, the Rap1 signalling pathway, the Ras signalling pathway (Fig. 3b).
Survival curve analysis
The survival curves of the 11 DEGs were queried through the BIDOS online platform (http://www.swmubidos.com/), and four DEGs that were highly correlated with the prognosis of septic shock patients were further screened based on the survival curve analysis, namely CD6, CD247, LCK and CD3E,and these four genes were positively correlated with septic shock patients' prognosis, which may indicate that these genes may be protective genes, and their survival curves are shown in Fig. 4.
Single cell sequencing analysis
We obtained a high quality cell count distribution of 4000–10000 cells by quantitative quality control of each sample Cell Ranger, and a final cell count distribution of 3108–8509 cells after elimination of bicellular, multicellular and apoptotic cells, which were divided into 9 groups of cells by dimensional clustering. The cell types were B-cells, NK-cells, T-cells, Platelets and Monocyte (Fig. 5a). Single cell sequencing analysis showed that CD6, CD247, LCK and CD3E were mainly located in T-cells (Fig. 5b).
Determining the moulding dose
We randomly assigned 60 C57BL/6 male mice into 6 groups with 10h fasting and unrestricted water intake. The mice were divided into control group and different doses of LPS injection group, and the doses of LPS were set at 0 mg/kg, 10 mg/kg, 15 mg/kg, 20 mg/kg, 30 mg/kg and 40 mg/kg for different dose groups. The mice were observed at 0 d, 0.25 d, 0.5 d, 1 d, 2 d, 3 d, 4 d, 5 d, 6 d and 7 d, and their body temperature was measured and the number of dead and surviving mice was recorded. When the mice showed different degrees of reduced activity, unresponsiveness, abnormal hair and body temperature, the intraperitoneal injection of LPS indicated that the mice could not survive.[8, 9] The results indicated that the intraperitoneal injection of LPS was successful in constructing a sepsis model. The results showed that the 7-day mortality rate of the mice increased sequentially with increasing doses of LPS, with a 30% mortality rate at 15 mg/kg of LPS. It was also observed that the mice did not show any new mortality after 72h of LPS injection, and their activity and feeding gradually improved (Fig. 6). Therefore, the final dose of 15 mg/kg LPS was chosen as the survival modeling dose for sepsis, and was taken 72 hours after modeling.
Constructing a model of sepsis
Twenty experimental mice were randomly divided into two groups, namely the blank control group and the sepsis group, 10 mice in each group were weighed and injected intraperitoneally with LPS at a dose of 15 mg/kg to construct a mouse sepsis model. The activity status of the mice was recorded at 0d, 0.25d, 0.5d, 1d, 2d, 3d, 4d, 5d, 6d and 7d. The survival group was taken when the clinical manifestations associated with sepsis reached their peak and no further deaths occurred, and when the mice were in an irreversible state of near death (generalized shivering, no obvious response to painful stimuli, loss of the righting reflex and body temperature below 23°C).[10] 。 Samples were taken after pentobarbital sodium was administered intraperitoneally.
Extraction of mouse spleen RNA
The mice were disinfected with 75% alcohol, placed on a sterile table, the abdominal wall was cut open layer by layer, and the fascial tissue on the surface of the spleen was separated. The quality and concentration of the RNA was determined using an Agilen Bioanalysert (2100) analyser and a NanoDrop ND-2000 (qualifying criteria: OD260/280 = 1.18–2.2, OD260/230 ≥ 2.0, RIN ≥ 6.5, 28S:18S ≥ 1.0). High quality RNA samples that are undegraded and not contaminated with proteins and other impurities are used to construct sequencing libraries.
Acquisition of raw sequencing data
The Illumina platform performs base sequence identification by CASJWA and converts the identified image signal into a storable text signal and stores it in fastq format. The data were quality controlled by the software fastx toolkit 0.0.14 to obtain high quality data to ensure the accuracy of the results, resulting in clean raw transcriptome data.
Validation of differential genes
Bioinformatics analysis of the genomic data from the septic shock group and the normal control group showed that the genes that had an impact on the prognosis of the disease were CD6, CD247, LCK and CD3E, respectively. To verify the authenticity of the results, we constructed a mouse model of sepsis and isolated spleen tissue and extracted RNA from spleen cells to obtain spleen gene expression data from sepsis model mice and normal mice. We determined the expression levels of CD6 and other 4 genes in the experimental mice and visualized the data by the Sento Academic online analysis tool (https://www.xiantao.love/products) to obtain the expression box plots of the above 4 genes in the mouse model (Fig. 7). Therefore, we can conclude that the four differential genes obtained from the bioinformatics analysis of the septic shock group and the normal group are real and reliable and can predict the prognosis of patients with septic shock.