Integrated Analysis of Bulk transcriptomics and Single-Cell RNA Sequencing Data identifies Glycolysis-Related Prognostic Gene signature of sepsis

Abstract

Abstract Background It has been widely recognized that the perturbation of the immune system induced by sepsis underlies the pathophysiology of sepsis and determines the patient’s prognosis. Failure of previous studies targeting one single marker highlighted the complexity and heterogeneity of immunopathology in sepsis. This study aimed to develop a glycolysis-related gene model able to predict sepsis prognosis. Methods Univariate Cox regression analysis and least absolute shrinkage and selection operator (LASSO) regression were applied to establish a glycolysis-related risk model. Kaplan-Meier analysis was performed to test the model’s prognostic value, which also was verified in the other cohort. Single-cell RNA-sequencing (scRNA-seq) data were downloaded from Gene Expression Omnibus (GEO) to further explore the cell origin of glycolysis-related signature genes, and Seurat was used for data quality control and analysis. Cell abundances were validated via bulk-gene-expression deconvolution. Results In patients with sepsis, fourteen glycolysis-related genes associated with 28-day survival were finally identified and fitted into a prognostic model. Kaplan-Meier analysis showed that, whether in the training or validation cohort, the mortality of the High-Score group identified by this model was significantly higher than that of the Low-Score group (P values were 1.578e-7 and 4.572e-4 respectively). Enrichment analysis based on Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) suggested that the High-Score group was mainly enriched in neutrophil activation, neutrophil degranulation, and neutrophil-mediated immunity as well as ferroptosis, while negative regulation of monocyte activation and interferon-gamma production and was downregulated. Protein-protein interaction (PPI) analysis demonstrated that Fructose-1,6-Bisphosphatase 1 (FBP1) might play an essential role in the hub gene network. ScRNA-seq analysis showed that majority of these signature genes were expressed in myeloid cells. Monocytes in survived septic patients had much higher FBP1 expression, which was also verified by bulk-gene-expression deconvolution. Conclusions The constructed glycolysis-related prognostic gene signature could effectively predict the 28-day mortality of septic patients. High expression of FBP1 in monocytes may play a protective effect on patients with sepsis, which deserves further research and analysis.