Identification and Evaluation of Lipocalin-2 in Sepsis-Associated Encephalopathy via Machine Learning Approaches

Jia Hu,1,2,* Ziang Chen,3,4,* Jinyan Wang,1,2 Aoxue Xu,1,2 Jinkai Sun,1,2 Wenyan Xiao,5,6 Min Yang5,6 1Department of Anesthesiology and Perioperative Medicine, The Second Affiliated Hospital of Anhui Medical University, Hefei, Anhui Province, People’s Republic of China; 2Key Laboratory of Anesthesiology and Perioperative Medicine of Anhui Higher Education Institutes, Anhui Medical University, Hefei, Anhui Province, People’s Republic of China; 3Department of Urology, Huashan Hospital, Fudan University, Shanghai, People’s Republic of China; 4Fudan Institute of Urology, Huashan Hospital, Fudan University, Shanghai, People’s Republic of China; 5The Second Department of Critical Care Medicine, The Second Affiliated Hospital of Anhui Medical University, Hefei, Anhui Province, People’s Republic of China; 6The Laboratory of Cardiopulmonary Resuscitation and Critical Care Medicine, The Second Affiliated Hospital of Anhui Medical University, Hefei, Anhui Province, People’s Republic of China*These authors contributed equally to this workCorrespondence: Wenyan Xiao; Min Yang, The Second Department of Critical Care Medicine, The Second Affiliated Hospital of Anhui Medical University, 678 Furong Road, Hefei, Anhui Province, People’s Republic of China, Email [email protected]; [email protected]: Sepsis-associated encephalopathy (SAE) critically contributes to poor prognosis in septic patients. Identifying and screening key genes responsible for SAE, as well as exploring potential targeted therapies, are vital for improving the management of sepsis and advancing precision medicine.Patients and Methods: Single-cell RNA sequencing (scRNA-seq) was administrated to identify cell subpopulations related to poor prognosis in septic patients. Next, hierarchical dynamic weighted gene co-expression network analysis (hdWGCNA) was employed to identify genes associated with specific neutrophil subpopulations. Enrichment analysis revealed the biological functions of these genes. Subsequently, neuroinflammation-related genes were obtained to construct a neuroinflammation-related signature. The AddModuleScore algorithm was used to calculate neuroinflammation scores for each cell subpopulation, whereas the CellCall algorithm was used to assess the crosstalk between neutrophils and other cell subpopulations. To identify key genes accurately, four binary classification machine learning algorithms were utilized. Finally, Western blotting and behavioral tests were used to confirm the role of LCN2-related neuroinflammation in septic mice.Results: This study utilized scRNA-seq to reveal the critical role of peripheral neutrophils during sepsis, identifying these neutrophils as contributors to poor prognosis and associated with neuroinflammation. On the basis of various machine learning algorithms, we discovered that Lipocalin-2 (LCN2) may be the key gene involved in neutrophil-induced SAE. To prove these findings, we conducted in vivo experiments and an animal model. Increased LCN2 expression and cognitive dysfunction occurred in septic mice. Additionally, the levels of markers of astrocytes and microglia and inflammatory factors such as TNF-α and IL-6 were significantly increased. All these phenomena were reversed by the downregulation of LCN2.Conclusion: The upregulation of LCN2 expression on peripheral neutrophils is a critical step that triggers neuroinflammation in the central nervous system during SAE.Keywords: sepsis-associated encephalopathy, cognitive dysfunction, lipocalin-2, single-cell RNA sequencing, machine learning, neuroinflammation