Network analysis between neuron dysfunction and neuroimmune response based on neural single-cell transcriptome of COVID-19 patients

Abstract

Despite global vaccination efforts, COVID-19 breakthrough infections caused by variant virus continue to occur frequently, long-term sequelae of COVID-19 infection like neuronal dysfunction emerge as a noteworthy issue. Neuroimmune disorder induced by Inflammatory factor storm was considered as a possible reason, however, little was known about the functional factors affecting neuroimmune response to this virus. Here, using medial prefrontal cortex single cell data of COVID-19 patients, expression pattern analysis indicated that some immune-related pathway genes expressed specifically, including genes associated with T cell receptor, TNF signaling in microglia and Cytokine-cytokine receptor interaction and HIF-1 signaling pathway genes in astrocytes. Besides the well-known immune-related cell type microglia, we also observed immune-related factors like IL17D, TNFRSF1A and TLR4 expressed in Astrocytes. Based on the ligand-receptor relationship of immune-related factors, crosstalk landscape among cell clusters were analyzed. The findings indicated that astrocytes collaborated with microglia and affect excitatory neurons, participating in the process of immune response and neuronal dysfunction. Moreover, subset of astrocytes specific immune factors (hinged neuroimmune genes) were proved to correlate with Covid-19 infection and ventilator-associated pneumonia using multi-tissue RNA-seq and scRNA-seq data. Function characterization clarified that hinged neuroimmune genes were involved in activation of inflammation and hypoxia signaling pathways, which could lead to hyper-responses related neurological sequelae. Finally, a risk model was constructed and testified in RNA-seq and scRNA data of peripheral blood.

Keywords: COVID-19; Neural sequelae; Neuroimmune characteristics; Risk model; scRNA-seq.

Fig. 1

Analysis of brain single cell. (A) Visualization of major cell clusters using UMAP, which included six cell categories (Excitatory Neurons, Oligodendrocyte, Neural Progenitor Cells, Astrocytes, Microglia and Interneurons through cell annotation). Dots, individual cells, colors, cell clusters. (B) The proportion of the cells number of each cell type. (C) Violin plots displaying the expression of IL17D, TNFRSF1A and TLR4 across the cell clusters identified. The y axis showed the normalized read count. (D) Differential expression of cytokines and the related biological processes in each cell cluster. The color contours of each cluster were consistent with Fig. 1-A.

Fig. 2

Cell crosstalk based on ligand and receptor. (A) The ligand-receptor interaction in each cell cluster, and the crosstalk between classified nerve cells. (B) The crosstalk among Astrocytes, Microglia and Excitatory Neurons consists of ligands and receptors. (C) ANGPTL4 was specifically expressed in Astrocytes, and it regulated receptors ITGB1 and CDH11. (D) Astrocytes expressed specifically factor PRR5, which interacted with ITGB1 and CDH11. (E) TGFB2 was specifically expressed in Astrocytes, and it regulated receptors ACVR1B and ACVR1C. (F) LAMA1 was specifically expressed in Astrocytes, and it regulated receptors SV2B and ITGA2.

Fig. 3

Mechanisms Characterization of cluster for neuronal dysfunction. (A) The crosstalk between astrocytes and microglia based on ligand and receptor interactions. (B) BAG3 was specifically expressed in astrocytes, and it regulated receptors HSPA9 and HSPA4. (C) DLL1 was specifically expressed in microglia, and it regulated receptors NCTCH1 and NCTCH1. (D) SPP1 was specifically expressed in microglia, and it regulated receptors ITGB5 and CD44.

Fig. 4

Recognition module associated with COVID-19 severity based on tracheal aspirate data. (A) Cluster dendrogram showed correlations between clinical features and modules by WGCNA. (B) Functional analysis was implemented basing on genes in ME-turquoise module. Goplot showed that 32 genes were associated with top five biological processes. (C) Expression of 14 genes in ME-turquoise module were screened by univariate Cox regression. (D) Cumulative risk of COVID-19 was analyzed by using expression of PRR5 and GEM.

Fig. 5

Recognition of hinged neuronimmune genes associated with COVID-19 severity and functional characterization. The circos plot showed the genes positions of functional module in the chromosomes (the outer circle). Differential expressions of genes in the peripheral blood RNA-seq data were displayed in the heat map (the middle circle). The inner network demonstrated the interaction between genes, and the darker colors represented stronger interactions. (B) The enrichment plot showed the distribution of hinged genes in the regulation of hypoxia response process. (C)The most hinged gene in (B): the expression of ANGPTL4 was significantly different between COVID-19 (n = 16) and healthy subjects (n = 10) (p < 0.05). Moreover, the transcriptional regulatory network involved ANGPTL4 was constructed, which was shown in the top right corner. (D) The enrichment plot showed the distribution of hinged genes in the regulation of TNFA signaling via NFKB response process. (E) The most hinged gene in (D): the expression of ACKR3 was significantly different between COVID-19 (n = 16) and healthy subjects (n = 10) (p < 0.001). Moreover, the transcriptional regulatory network involved ACKR3 was constructed, which was shown in the top right corner.

Fig. 6

Construction and evaluation of a risk model. (A) 8 genes were screened by multivariate cox regression method to construct a risk model. (B) Cumulative risk of COVID-19 patients was analyzed by using risk module, which had significant classification efficiency with p < 0.01. (C) Risk scores differenced in COVID-19 patients aged less than 60 years (n = 69) and greater than 60 years (n = 29). (D) Risk scores difference was displayed between female (n = 30) and male (n = 68) with COVID-19. (E) Risk scores differenced in COVID-19 patients between the alive (n = 47) and the dead (n = 51). (F) It had the differences between in whether it's Hispanic ethnicity (Yes n = 47, No n = 51). (G) Risk scores were different between COVID-19 patients and normal groups. (H) Proportion differences at high-risk and low-risk group was different between COVID-19 patients (n = 9) and normal groups (n = 7), and 87.5% COVID-19 patients were in the high-risk group, in contrast, 75% normal groups were in the low-risk group.

Fig. 7

ScRNA data validated the model. ScRNA data from GSE158055, which included 171 patients with COVID-19, progression (n = 83)) and 25 healthy individuals. (A) ANGPTL4 expressed in CD8 cells, CD4 cells, B cells and other immunity cells, and expressed highly in COVID-19 patients with severity. (B) BAG3 expressed in immunity cells, and expressed highly in COVID-19 patients with severity. (C) GPC4 expressed in some immunity cells, and had rare expression in healthy subjects. (D) S100A13 expressed in immunity cells, and expressed highly in COVID-19 patients with severity.