Fig. 6

Integration of serum EV proteomics, scRNA-seq of PBMCs, and snRNA-seq of lung tissue. a–c KeyMolnet generated a highly complex network of targets with possible relationships by using the “start points and end-points” network search algorithm. In each molecular networks, the top 10 pathways with the highest involvement are listed in order of HScore. Left panel; examination of the molecular network upstream of MACROH2A1. Right panel; examination of the molecular network downstream of MACROH2A1. a The top 10 pathways in the networks of EV proteins with p < 0.05 and fold change > 1.5 or < 0.67 in comparison of Group 2 and 3 cases. b The top 10 pathways in the networks of genes with significantly upregulated or downregulated in differential expression analysis in scRNA-seq of monocytes from PBMCs in comparison of Group 2 and 3 cases and Group 1 cases. c The top 10 pathways in the networks of genes with significantly upregulated or downregulated in differential expression analysis in snRNA-seq of monocytes from lungs in comparison of fatal COVID-19 and controls. d A graphical abstract of our study. We have identified the protein MACROH2A1, as a potential biomarker for predicting severe COVID-19 infections refractory to anti-inflammatory therapy. First, we have successfully identified several biomarker candidates by performing “next-generation proteomics,” a high-throughput non-targeted quantitative proteomics by data-independent acquisition using serum EVs, and identified MACROH2A1 as the best biomarker molecule among them. Furthermore, scRNA-seq of peripheral blood mononuclear cells and single-nucleus RNA sequencing of lung tissues revealed that this molecule is highly expressed and upregulated in monocytes and macrophages, with pathological pathways which were also reflected in EV proteomics, suggesting its deep involvement in the pathogenesis of severe COVID-19 infections in these cells