Jin Rao,1 Xuefu Wang,2 Zhinong Wang1 

1Department of Cardiothoracic Surgery, Changzheng Hospital, Naval Medical University, Shanghai, People’s Republic of China; 2School of Health Science and Engineering, University of Shanghai for Science and Technology, Shanghai, People’s Republic of China

Correspondence: Zhinong Wang; Xuefu Wang, Email wangzn007@smmu.edu.cn; wangxf2311@163.com

Background: Cardiac macrophages are a heterogeneous population with high plasticity and adaptability, and their mechanisms in heart failure (HF) remain poorly elucidated.
Methods: We used single-cell and bulk RNA sequencing data to reveal the heterogeneity of non-cardiomyocytes and assess the immunoreactivity of each subpopulation. Additionally, we employed four integrated machine learning algorithms to identify macrophage-related genes with diagnostic value, and in vivo validation was performed. To assess the immune infiltration characteristics in HF, we utilized the CIBERSORT and single sample gene set enrichment analysis (ssGSEA). An unsupervised consensus clustering algorithm was applied to identify the macrophage-related HF subtypes. Furthermore, the scMetabolism was employed to explore the specific metabolic patterns of the macrophage subtypes. Finally, CellChat was used to investigate cell-cell interactions among the identified subtypes.
Results: The immunoreactivity score of macrophages in the HF was higher than that in the other cell types. GSEA of macrophage clusters indicated a significant enrichment of leukocyte-mediated immune processes, antigen processing, and presentation. The intersection of the results from machine learning revealed that SERPINA3, GPAT3, ANPEP, and FCER1G can serve as feature genes and form a diagnostic model with a good predictive capability. Unsupervised consensus clustering algorithms reveal the immune and metabolic subtypes of macrophages. The metabolic heterogeneity of macrophage subpopulations can lead to macrophage polarization into different types, which may be related to the metabolic reprogramming between glycolysis and mitochondrial oxidative phosphorylation. Cellular communication revealed that macrophages form a network of interactions with neutrophils to support each other’s functions and maintenance. The complex efferent and afferent signals are closely associated with myocardial fibrosis.
Conclusion: SERPINA3, GPAT3, ANPEP, and FCER1G can potentially serve as immune therapeutic targets and central biomarkers. The immunological and metabolic heterogeneity of macrophages may offer a more precise direction to explore the mechanisms underlying HF and novel immunotherapies.

Keywords: heart failure, single cell sequencing, machine learning, macrophage, diagnostic models, immune infiltration, metabolic reprogramming