Immune mechanisms and signatures in lethal and non-lethal sepsis revealed by single-cell transcriptomics

Abstract Sepsis is a life-threatening inflammatory syndrome caused by systemic infections, potentially leading to multi-organ failure and death. Due to patient heterogeneity, early and accurate diagnosis of sepsis, as well as stratification of patients by severity and potential outcomes, remains a huge challenge. In this study, we leveraged the genetic homogeneity of inbred mice to develop Escherichia coli-induced lethal and non-lethal sepsis models. Using single-cell RNA sequencing, we analyzed dynamic transcriptomes of peripheral blood mononuclear cells and found a marked expansion of monocytes during sepsis. A specific monocyte subset, designated Mono2, exhibited heightened expression of cytokines, chemokines, and pro-inflammatory pathways compared to other subsets. Notably, we observed a clear increasing trend in the abundance of Mono2 population in the lethal groups, and a decreasing trend in the non-lethal groups over time. Genes enriched in pro-inflammatory pathways (eg TNF and NF-κB signaling pathways) showed opposing trends in the 2 models: decreasing as mice recovered in the non-lethal model and increasing as mice succumbed to infection in the lethal model. Furthermore, we identified gene signatures corresponding to different stages of infection in the mouse models. By utilizing publicly available patient data, we validated the diagnostic potential of several genes and gene sets, including ANXA1+FPR1 and BCL2A1 for distinguishing septic patients from healthy individuals, and GBP2 for differentiating survivors from non-survivors. Collectively, this study highlights critical cellular and molecular signatures while providing insights into host immune mechanisms underlying sepsis progression and outcomes, offering valuable resources for advancing sepsis diagnosis and therapeutic strategies.