Introduction
The main clinical manifestations of coronavirus disease 2019 (COVID-19) resulting from severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection are fever, cough, and dyspnea.[1–5] However, numerous neurological abnormalities have been described in patients with COVID-19; these abnormalities include olfactory and gustatory disorders (anosmia, hyposmia, ageusia, and dysgeusia), headache, epilepsy, encephalitis, and acute myelitis.[4–8] SARS-CoV-2 RNA has also been identified in cerebrospinal fluid via genome sequencing.[9,10] These findings suggest that SARS-CoV-2 can directly or indirectly enter the central nervous system (CNS). However, the underlying mechanism remains unknown.
The CNS can be invaded through the hematogenous route or by axonal transport[11]; these routes involve infections of endothelial cells in the blood–brain barrier (BBB) or epithelial cells in the blood–cerebrospinal fluid barrier, as well as infections of the olfactory bulb through the olfactory nerve.[12] The tropism of coronaviruses is primarily dependent on the presence of attachment and entry receptors, along with protease availability and the activities of internalization-related and trafficking-related pathways.[13,14] SARS-CoV-2 uses angiotensin-converting enzyme 2 (ACE2) as an entry receptor.[15,16] Other entry receptor molecules associated with SARS-CoV-2 may include CD147 and CD209.[17,18] Upon receptor binding, spike protein priming enzymes (eg, cell surface transmembrane serine protease [TMPRSS] and endosomal cathepsins) are necessary for some coronaviruses to achieve membrane fusion.[19,20] Thus, the detection of coexpression involving key genes in the viral entry process may help to identify cells that are susceptible to SARS-CoV-2 infection; this will contribute to further investigations of the mechanisms by which SARS-CoV-2 enters the CNS.
In this study, we used single-cell RNA sequencing (scRNA-seq) of normal brain and nasal epithelium specimens, as well as tracheal, bronchial, and lung specimens, to identify susceptible cell types and potential routes of SARS-CoV-2 entry into the CNS, olfactory system, and respiratory system. We sought to clarify how SARS-CoV-2 invades various tissues in the human body, with a particular interest in elucidating the olfactory and CNS symptoms in patients with COVID-19.
Materials and methods
Data sources
We analyzed single-cell transcriptomes of the brain, nasal olfactory epithelium (OE), nasal respiratory epithelium (RE), lung, tracheal epithelium, and bronchial epithelium (Additional Table 1, https://links.lww.com/JR9/A46). Nasal RE/OE and trachea transcriptome data were obtained from the Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/), using accession codes GSE139522[21] and GSE134355.[22] Brain transcriptome data were obtained from the Sequence Read Archive (SRA; https://www.ncbi.nlm.nih.gov/sra), using accession code SRA667466.[23] Lung transcriptome data were obtained from Cell Atlas Data Portal (https://data.humancellatlas.org), using accession code PRJEB31843.[24] Bronchial transcriptome data were obtained from research data published by Lukassen et al.[25] The single-cell data within each dataset had similar sequencing depth to ensure the quality of consensus sequence. Because the data were obtained from public datasets, approval for our study by the ethics committee and patient consent were unnecessary.
Quality control
Barcodes with outlier gene counts may be indicative of dying cells, cells with broken membranes, or doublets. A high fraction of mitochondria is indicative of leaky cytoplasmic membranes allowing mRNA to escape.[26] We removed low-quality data from cells with <200 or >5000 expressed genes, as well as data from cells in which >10% of unique molecular identifiers mapped to mitochondrial genes.
Data integration, dimension reduction, and cell clustering
The R package Seurat[27] and R package Harmony[28] were used to process scRNA-seq data. Unique molecular identifier counts were normalized using the “NormalizeData” function. The top 1000 highly variable genes were identified using the “FindVariableGenes” function. Principal component analysis was performed on single-cell expression matrices using the “RunPCA” function. Subsequently, “Runharmony” was used for batch-effect correction in each dataset. Cell clustering analysis was performed after the construction of a K-nearest-neighbor graph based on the adjusted principal component analysis embeddings by harmony algorithm.
Cell type identification and gene expression analysis
Clusters were annotated based on the expression patterns of known cellular markers; the clustering information was visualized by uniform manifold approximation and projection (UMAP) plots. Then, lost values in the gene expression matrix were imputed by the “RunALRA” function. The imputed gene expression values are displayed in feature and violin plots. The mean expression levels of gene sets are displayed in box plots. For comparisons of expression levels among datasets, the quantile method was used to correct expression data; this global normalization method is widely used.[29,30] To calculate the endocytosis signaling score, we obtained endocytosis-associated genes from “GOBP_ENDOCYTOSIS” gen set in the MigDB dataset (https://www.gsea-msigdb.org). The average expression level of all the genes in this gene set was considered as the expression level of the endocytosis signaling.
Immunohistochemical staining
Human brain tissue specimens were obtained from five patients who had undergone surgery for treatment of traumatic brain injury, including three males and two females. Human lung tissue specimens were obtained during routine surgical intervention from three lung cancer patients, including two males and one female. Ethical approval was granted by the Ethics Committee of Shanghai Changhai Hospital (Approval No. CHEC2020-164). All patients were fully informed of the research and signed the informed consent forms. Formalin-fixed paraffin-embedded brain tissue sections on glass slides were incubated at 37°C overnight, deparaffinized in xylene, and rehydrated in a descending ethanol gradient. Then, tissue sections were incubated in antigen retrieval solution at 95°C for 30 minutes. Next, tissue sections were incubated in 3% hydrogen peroxide for 10 minutes, followed by serum-free protein blocking solution (Cat# 9048-46-8, Sigma, St. Louis, MO, USA) for 20 minutes; they were then incubated with rabbit anti-human ACE2 primary antibody (Cat# ab108252, Abcam, Cambridge, UK) at a dilution ratio of 1:200 overnight at 4°C temperature and goat anti-rabbit IgG secondary antibody (Cat# ab6721, Abcam) at a dilution ratio of 1:1000 for 1 hour at room temperature. After incubation with primary and secondary antibodies, color development was performed using 3-amino-9-ethylcarbazole as the chromogenic substrate and tissue was counterstained with hematoxylin as described previously.[31] The microscopic observation was performed by the optical microscope (Eclipse 80i, Nikon, Tokyo, Japan).
Results
Single-cell maps of mouse brain, healthy human olfactory system, and healthy human respiratory system
To identify susceptible cell types in the brain, olfactory system, and respiratory system, we collected scRNA-seq data for six types of tissue from published papers (Additional Table 1, https://links.lww.com/JR9/A46). In the brain, 11 cell types were identified based on the analysis of 29,678 cells, after the application of quality control criteria (Fig. 1A). The main cell types were oligodendrocytes, interneurons, astrocytes, microglia, and neurons 1 and 2. The other identified cell types included ependymal cells, endothelial cells, pericytes, dividing cells, and myeloid cells.
;)
Figure 1.: Single-cell maps of healthy human respiratory system and mouse brain tissue. UMAP plots showing the distributions of brain cells (A), nasal olfactory epithelial cells (B), nasal respiratory epithelial cells (C), tracheal epithelial cells (D), bronchial epithelial cells (E), and lung cells (F). Single-cell map of respiratory system (G). All the single-cell data in the human respiratory system were combined for another round of reduction and clustering analysis. UMAP plots show the distributions of the cell clusters in the human respiratory system. DCs=dendritic cells, HBCs=horizontal basal cells, NK=natural killer, UMAP=uniform manifold approximation and projection, VSMC=vascular smooth muscle cells.
The nasal epithelium was divided into two types: OE and RE. In total, 19,903 cells and 14 cell types were identified in the OE (Fig. 1B); 10,657 cells and 12 cell types were identified in the RE (Fig. 1C). Epithelial cell types in the RE were horizontal basal cells (HBCs) and Bowman’s gland cells, as well as sustentacular, ciliated, and endodermal progenitor cells. Non-epithelial cell types in the RE were microvillar and endothelial cells, as well as fibroblasts and immune cells (ie, mast, myeloid, T, and B cells). Epithelial cell types in the OE were HBCs, pericytes, and Bowman’s gland cells, as well as gland progenitor, sustentacular, and ciliated cells. Non-epithelial cell types in the OE were neurons, olfactory ensheathing glia cells (OECs), fibroblasts, vascular smooth muscle cells (VSMCs), and immune cells (ie, mast, myeloid, T, and B cells).
After the application of quality control criteria, 9614 cells and 10 cell types were identified in tracheal specimens (Fig. 1D). The cell types were goblet, brush, ciliated, basal, and granular cells, as well as chondrocytes, VSMCs, endothelial cells, fibroblasts, and immune cells (ie, myeloid and B cells). Additionally, 17,228 cells and seven cell types were identified in bronchial specimens after the application of quality control criteria (Fig. 1E). The main cell types identified were ciliated, basal, and goblet cells. The other identified cell types consisted of brush cells, fibroblasts, dividing cells, and ionocytes.
After the application of quality control criteria, 24,710 cells and 14 cell types were identified in lung specimens (Fig. 1F). The cell types identified were alveolar epithelium type 1 (AT1), alveolar epithelium type 2 (AT2), and ciliated cells, as well as fibroblasts, lymph vessel cells, VSMCs, and endothelial cells. The following immune cell types were identified: mast, T, B, and natural killer (NK) cells, along with monocytes, macrophages, and dendritic cells.
Because of the similar cells present among tissues in the respiratory system, we constructed a map of the respiratory system based on olfactory epithelum data, nasal respiratory epithelium data, trachea data, bronchus data, and lung data from 83,126 cells. In this map, we identified 17 subclusters: HBCs, endothelial cells, Bowman’s gland cells, sustentacular cells, alveolar cells, ciliated cells, neurons, OECs, fibroblasts, microvillar cells, VSMCs, chondrocytes, basal cells, mast cells, myeloid cells, B cells, and T/NK cells (Fig. 1G).
Coexpression patterns of key entry genes in mouse brain tissue
To identify susceptible cell types in brain tissue and elucidate potential routes of SARS-CoV-2 entry, we collected scRNA-seq data from mouse brains. In total, 11 subclusters were identified in mouse brain cells (Fig. 2A); marker genes for each subcluster are shown in Figure 2B. ACE2 was strongly expressed in pericytes and weakly expressed in other subclusters (Fig. 2C). To confirm the expression of ACE2 at the protein level, we performed immunohistochemical analysis of normal human brain tissues; the immunohistochemistry results were consistent with the scRNA-seq findings. ACE2 was strongly expressed in vascular pericytes, as well as endothelial cells (Fig. 2D). Lung alveolar cells also exhibited strong expression of ACE2 at the protein level (Fig. 2E); they are regarded as the main target of SARS-CoV-2 lung tissue.[32,33] Therefore, pericytes may be the main target of SARS-CoV-2 in brain tissue; entry through these cells might be involved in indirect damage to neurons. CD147, a potential entry receptor, was expressed in almost all subclusters (Fig. 2F). The spike protein priming enzyme TMPRSS was nearly absent from all subclusters in brain tissue, whereas cathepsin L (CTSL) was expressed in all subclusters (Fig. 2C and F); these findings suggested that spike protein hydrolysis in brain tissue is mediated by CTSL. Because endocytosis is an important aspect of coronavirus infection, we examined the expression of the endocytosis signaling molecules phosphoinositide kinase, FYVE-type zinc finger containing (PIKfyve) and two pore segment channel 2 (TPCN2). We found that PIKfyve was strongly expressed in ependymal cells, pericytes, and neurons 1 and 2 in brain tissue (Fig. 2F), whereas the expression of TPCN2 was minimal and primarily observed in myeloid cells. Furthermore, methylenetetrahydrofolate dehydrogenase 1 (MTHFD1), which is an essential gene for RNA virus replication,[32] was expressed in all subclusters in brain tissue (Fig. 2F). These results suggested that neurons do not constitute a primary site of infection; however, pericytes are the most susceptible cells in brain tissue because they exhibit strong expression of ACE2 and coexpression of CTSL, which facilitate viral entry and replication.
Figure 2.: Single-cell analysis of mouse brain cells. UMAP plot showing the distribution of mouse brain cells (A). Dot plot showing the marker genes for brain clusters (B). Dot scale indicates the fractions of expressing cells and dot color indicates the expression level of the marker genes. UMAP plots showing the expression patterns of ACE2, TMPRSS, and CTSL (C). Red indicates high expression of ACE2 and blue indicates high expression of TMPRSSs or CTSL. Representative immunohistochemical staining of ACE2 in cerebral vascular pericytes (D) and lung tissue of humans (E). Arrows indicate ACE2-postive cells. Violin and box plots for the expression levels of ACE2, CD147, TMPRSS, CTSL, endocytosis-related genes, PIKFYVE, TPCN2, and MTHFD1 across clusters (F). Gene expression levels are measured as log2 (TP10K+1) values and the expression levels of gene sets are measured as mean log2 (TP10K+1) values. ACE2=angiotensin-converting enzyme 2, CTSL=cathepsin L, TMPRSS=transmembrane serine protease, UMAP=uniform manifold approximation and projection.
Coexpression patterns of key entry genes in healthy human nasal olfactory epithelial tissue
Next, we explored whether SARS-CoV-2 could enter the CNS through the olfactory system. We identified 14 subclusters in the OE (Fig. 3A); the marker genes for each subcluster are shown in Figure 3B. ACE2 was strongly expressed in sustentacular, ciliated, and gland progenitor cells; its expression was weaker in HBCs and Bowman’s gland cells (Fig. 3C and D). CD147 was expressed in almost all subclusters, whereas CD209 was only expressed in myeloid cells (Fig. 3D). TMPRSS was expressed in many subclusters in the OE, including gland progenitor, ciliated, and sustentacular cells, as well as neurons; its expression was minimal in fibroblasts, HBCs, and Bowman’s gland cells (Fig. 3C and D). CTSL was expressed in all non-immune subclusters, but it was only expressed in myeloid cells among the immune subclusters (Fig. 3C and D). Additionally, endocytosis signaling was strongly expressed in non-immune subclusters and weakly expressed in immune subclusters (Fig. 3C and D). PIKFYVE and TPCN2 were expressed in most cell types (Fig. 3D). The viral RNA replication target gene MTHFD1 was detected in all subclusters in the OE, except T and B cells (Fig. 3D). Overall, the data indicated that multiple cell types in the OE could serve as target cells for SARS-CoV-2. The gland progenitor, ciliated, and sustentacular cells may constitute the most susceptible cells in the OE because they exhibited the strongest expression of ACE2 and TMPRSS2 (Fig. 3C and D). Thus, we speculate that non-neuronal cell types, rather than sensory or bulb neurons, are responsible for anosmia and associated odor perception deficits in patients with COVID-19.
Figure 3.: Single-cell analysis of healthy human nasal olfactory epithelial cells. UMAP plot showing the distribution of nasal olfactory epithelial cells (A). Dot plot showing the marker genes for nasal olfactory epithelial clusters (B). Dot scale indicates the fractions of expressing cells and dot color indicates the expression level of the marker genes. UMAP plots showing the expression patterns of ACE2, TMPRSS, and CTSL (C). Red indicates high expression of ACE2 and blue indicates high expression of TMPRSS or CTSL. Violin and box plots for the expression levels of ACE2, CD147, CD209, TMPRSS, CTSL, endocytosis-related genes, PIKFYVE, TPCN2, and MTHFD1 across clusters (D). Gene expression levels are shown as log2 (TP10K+1) values, and the expression levels of gene sets are shown as mean log2 (TP10K+1) values. ACE2=angiotensin-converting enzyme 2, CTSL=cathepsin L, HBCs=horizontal basal cells, OEC=olfactory ensheathing glia cells, TMPRSS=transmembrane serine protease, UMAP=uniform manifold approximation and projection, VSMC=vascular smooth muscle cells.
Coexpression patterns of key entry genes in healthy human nasal respiratory epithelial tissue
RE is a major component of the nasal epithelium. We identified 12 subclusters (Fig. 4A) and the marker genes for each subcluster are shown in Figure 4B. ACE2 was strongly expressed in sustentacular, ciliated, and microvillar cells; it was weakly expressed in endodermal progenitor cells (Fig. 4C and D). CD147 was expressed in almost all subclusters, whereas CD209 was expressed in myeloid cells (Fig. 4D). TMPRSS and CTSL had similar expression patterns among subclusters in the RE; they did not exhibit robust expression in immune cells (Fig. 4C and D). Thus, multiple cell types (eg, microvillar, ciliated, and sustentacular) expressed ACE2, TMPRSS/CTSL, endocytosis signaling, PIKFYVE, TPCN2, and MTHFD1; these cells were identified as potential targets of SARS-CoV-2 (Fig. 4D).
Figure 4.: Single-cell analysis of healthy human nasal respiratory epithelial cells. UMAP plot showing the distribution of nasal respiratory epithelial cells (A). Dot plot showing the marker genes for nasal respiratory epithelial clusters (B). Dot scale indicates the fractions of expressing cells and dot color indicates the expression level of the marker genes. UMAP plots showing the expression patterns of ACE2, TMPRSS, and CTSL (C). Red indicates high expression of ACE2 and blue indicates high expression of TMPRSSs or CTSL. (D) Violin and box plots for the expression levels of ACE2, CD147, CD209, TMPRSS, CTSL, endocytosis-related genes, PIKFYVE, TPCN2, and MTHFD1 across clusters. Gene expression levels are shown as log2 (TP10K+1) values, and the expression levels of gene sets are shown as mean log2 (TP10K+1) values. ACE2=angiotensin-converting enzyme 2, CTSL=cathepsin L, HBCs=horizontal basal cells, TMPRSS=transmembrane serine protease, UMAP=uniform manifold approximation and projection.
Coexpression patterns of key entry genes in healthy human tracheal and bronchial tissues
Among the 11 cell types identified in tracheal tissue (Fig. 5A and B), ACE2 was expressed in fibroblasts, VSMCs, and ciliated cells (Fig. 5C and D). CD147 was expressed in all subclusters, whereas CD209 was only expressed in myeloid cells (Fig. 5D). TMPRSS was mainly expressed in goblet, granule, brush, basal, and ciliated cells; in contrast, CTSL was expressed in chondrocytes, endothelial cells, VSMCs, fibroblasts, myeloid cells, and B cells (Fig. 5C and D). Endocytosis signaling, PIKfyve, TPCN2, and MTHFD1 were strongly expressed in all cell types in tracheal tissue (Fig. 5D). Among the seven subclusters in bronchial tissue (Fig. 6A and B), ACE2 was mainly expressed in goblet and ciliated cells, whereas CD147 was expressed in almost all subclusters (Fig. 6C). TMPRSS, CTSL, endocytosis signaling, PIKfyve, TPCN2, and MTHFD1 were expressed in all cell types in bronchial tissue (Fig. 6C and D). Taken altogether, the results indicated that ciliated cells are the most susceptible cells in both tracheal and bronchial tissue because they generally exhibit strong expression of ACE2 and TMPRSS (Figs. 5D and 6C). Goblet cells also demonstrated susceptibility to SARS-CoV-2 through its coexpression of ACE2, TMPRSS/CTSL, endocytosis signaling, and MTHFD1 (Figs. 5D and 6C).
Figure 5.: Single-cell analysis of healthy human trachea cells. UMAP plot showing the distribution of tracheal cells (A). Dot plot showing the marker genes for tracheal clusters (B). Dot scale indicates the fractions of expressing cells and dot color indicates the expression level of the marker genes. UMAP plots showing the expression patterns of ACE2, TMPRSS, and CTSL (C). Red indicates high expression of ACE2 and blue indicates high expression of TMPRSSs or CTSL. Violin and box plots for the expression levels of ACE2, CD147, CD209, TMPRSS, CTSL, endocytosis-related genes, PIKFYVE, TPCN2, and MTHFD1 across clusters (D). Gene expression levels are shown as log2 (TP10K+1) values, and the expression levels of gene sets are shown as mean log2 (TP10K+1) values. ACE2=angiotensin-converting enzyme 2, CTSL=cathepsin L, TMPRSS=transmembrane serine protease, UMAP=uniform manifold approximation and projection, VSMC=vascular smooth muscle cells.
Figure 6.: Single-cell analysis of healthy human bronchial epithelial cells. UMAP plot showing the distribution of bronchial epithelial cells (A). Dot plot showing the marker genes for bronchial epithelial clusters (B). Dot scale indicates the fractions of expressing cells and dot color indicates the expression level of the marker genes. UMAP plots showing the expression patterns of ACE2, TMPRSS, and CTSL (C). Red indicates high expression of ACE2 and blue indicates high expression of TMPRSSs or CTSL. Violin and box plots for the expression levels of ACE2 CD147, TMPRSS, CTSL, endocytosis-related genes, PIKFYVE, TPCN2, and MTHFD1 across clusters (D). Gene expression levels are shown as log2 (TP10K+1) values, and the expression levels of gene sets are shown as mean log2 (TP10K+1) values. ACE2=angiotensin-converting enzyme 2, CTSL=cathepsin L, TMPRSS=transmembrane serine protease, UMAP=uniform manifold approximation and projection.
Coexpression patterns of key entry genes in healthy human lung tissue
In total, 14 subclusters (Fig. 7A) in lung tissue were identified on the basis of cell type-specific marker genes (Fig. 7B). ACE2 was mainly expressed in AT2 cells; it was also expressed in AT1 and ciliated cells (Fig. 7C and D). CD209 was mainly expressed in monocytes and macrophages, whereas CLEC4M was expressed in endothelial cells. CD147 was expressed in all subclusters in lung tissue (Fig. 7D). TMPRSS and CTSL were expressed in most cell types; they were not expressed in mast, NK, T, or B cells (Fig. 7C and D). Endocytosis signaling, PIKFYVE, TPCN2, and MTHFD1 were strongly expressed in lung cells (Fig. 7D).
Figure 7.: Single-cell analysis of healthy human lung cells. UMAP plot showing the distribution of lung cells (A). Dot plot showing the marker genes for lung clusters (B). Dot scale indicates the fractions of expressing cells and dot color indicates the expression level of the marker genes. UMAP plots showing the expression patterns of ACE2, TMPRSS, and CTSL (C). Red indicates high expression of ACE2 and blue indicates high expression of TMPRSSs or CTSL. Violin and box plots for the expression levels of ACE2, CD147, CD209, CLEC4M, TMPRSS, CTSL, endocytosis-related genes, PIKFYVE, TPCN2, and MTHFD1 across clusters (D). Gene expression levels are shown as log2 (TP10K+1) values, and the expression levels of gene sets are shown as mean log2 (TP10K+1) values. CTSL=cathepsin L, DCs=dendritic cells, NK=natural killer, TMPRSS=transmembrane serine protease, UMAP=uniform manifold approximation and projection, VSMC=vascular smooth muscle cells.
Expression patterns of SARS-CoV-2 and neurotropic virus receptor genes in healthy human respiratory and mouse brain tissues
For comprehensive analysis of susceptible cell types in the respiratory system and comparison of similarities among cell types, we constructed a map of the respiratory system and identified 17 subclusters; the main subclusters comprised myeloid, alveolar, T/NK, endothelial, ciliated, neuronal, and B cells (Additional Fig. 1A and B, https://links.lww.com/JR9/A45). We found that ACE2 was strongly expressed in ciliated cells in the OE, RE, tracheal tissue, and bronchial tissue; sustentacular cells in the OE and RE; gland progenitor cells in the RE; and AT2 cells in lung tissue (Fig. 8A). CD209 was strongly expressed in myeloid cells in the OE, RE, and tracheal tissue. TMPRSS, CTSL, endocytosis signaling, TPCN2, and MTHFD1 were expressed in most cell types in the respiratory system (Fig. 8A). The endocytosis-related gene PIKfyve was expressed mainly in myeloid cells and fibroblasts in the RE, as well as endothelial cells in lung tissue (Fig. 8A). These results showed that factors involved in SARS-CoV-2 entry are strongly expressed in the OE and RE, highlighting the potential role of such cells to facilitate viral infection and spread. The most susceptible cells in the respiratory system were ciliated, gland progenitor, and sustentacular cells, as well as HBCs; AT2 cells are the known vulnerable cells for SARS-COV2,[32,33] their susceptibility could be greater than the susceptibility of AT2 cells, indicating that these cells could serve as the portal of entry for SARS-CoV-2.
Figure 8.: Expression levels of SARS-CoV-2 binding receptors, spike protein hydrolysis, endocytosis-related genes, and virus RNA replication genes across tissues. Violin and box plots for the expression levels of SARS-CoV-2 binding receptors, spike protein hydrolysis, endocytosis-related genes, and virus RNA replication genes across tissues (A). Gene expression levels are shown as log2 (TP10K+1) values, and the expression levels of gene sets are shown as mean log2 (TP10K+1) values. Box plots showing the expression levels of exosome biogenesis and release-related genes in the brain and olfactory epithelium (B). The expression levels of gene sets are shown as mean log2 (TP10K+1) values. Schematic depicting potential mechanisms involved in CNS infection (C). Coexpression of entry receptor ACE2 and spike protein priming enzyme cathepsin L in BBB pericytes enables SARS-CoV-2 entry into the cerebrospinal fluid and brain. SARS-CoV-2 may also infect the telencephalon through the olfactory bulb. ACE2=angiotensin-converting enzyme 2, BBB=blood–brain barrier, CNS=central nervous system, HBCs=horizontal basal cells, OEC=olfactory ensheathing glia cells, SARS-CoV-2=severe acute respiratory syndrome coronavirus 2.
To explore the potential mechanism of SARS-CoV-2 infection in the CNS, we compared the expression patterns of entry receptors for various neurotropic viruses between maps of the respiratory system and brain. We found that some neurotropic virus entry receptors were not expressed in neurons, indicating that other mechanisms (eg, other cells) were involved in CNS infection (Additional Figs. 2 and 3, https://links.lww.com/JR9/A45). Notably, many viral receptors (eg, AXL, ISG15, ITGA4, and PECAM1) are expressed in pericytes, suggesting that pericytes mediate CNS infection (Additional Fig. 3, https://links.lww.com/JR9/A45). Furthermore, because exosome biogenesis and release are also associated with viral transport, we explored the brain-specific and OE-specific expression patterns of genes involved in exosome biogenesis and release. The results showed that neurons, ciliated cells, and OECs had strong capacities for exosome biogenesis and release, which may serve as alternative pathways for CNS infection (Fig. 8B). Thus, we speculated the potential mechanisms involved in CNS infection (Fig. 8C). Coexpression of entry receptor ACE2 and spike protein priming enzymes in BBB pericytes enables SARS-CoV-2 entry into the cerebrospinal fluid and brain. SARS-CoV-2 may also infect the telencephalon through the olfactory bulb.
Discussion
scRNA-seq is a powerful technique that can be used to identify susceptible cells through single-cell-level analyses of the coexpression patterns of viral entry genes.[33,34] Although most analyses are based on transcriptional levels, there is clear consistency at the cellular level: the expression patterns of entry genes at the transcriptional level in a particular cell type can accurately predict the protein expression levels of entry receptors.[35] Our results were consistent with the coexpression patterns (ACE2 and TMPTSS2) previously identified among susceptible cell types in multiple tissues.[33] Furthermore, we obtained novel information concerning tissues such as the OE, RE, and brain; we also identified coexpression patterns involving TMPRSS2 and viral replication genes. These results suggest that neurological symptoms and CNS infection in patients with COVID-19 are secondary to SARS-CoV-2 invasion of the BBB via pericytes. Additionally, SARS-CoV-2–induced olfactory disorders could be the result of localized damage to sustentacular cells and ciliated cells in the nasal RE and OE.
The coronavirus infection process begins with viral binding to the cellular membrane through heparan sulfate proteoglycans, which facilitate the interaction between the spike protein and the entry receptor (ie, ACE2).[16,36] Recently, CD147 and CD209 were also identified as potential receptors for SARS-CoV-2; interactions with these proteins also facilitate viral entry into host cells.[17,18] In our study, CD147 was expressed in most cell types; however, these cells were not consistent with published reports of infected cells.[32] Thus, the role of CD147 in SARS-CoV-2 pathogenicity requires further investigation. In contrast, CD209 was mainly expressed in macrophages and monocytes in most tissues. The specific location of CD209 expression implies a role in SARS-CoV-2 entry into macrophages, which were also the target cells for SARS-CoV-2 infection,[32] with the potential for initiation of a cytokine storm.
Successful conformational changes and activation of spike proteins result in membrane fusion; these processes require receptor binding, as well as appropriate protease activation. Endocytosis and cathepsin-mediated endosomal acidification are also important processes during viral fusion with a target cell.[37,38] In the present study, we found that TMPRSS and CTSL were strongly expressed and exhibited a broad distribution. Either TMPRSS or CTSL was expressed in ACE2-expressing cells, consistent with previous reports that ACE2 constitutes a limiting factor for initial viral entry.[39] Genes that encode proteins involved in endocytosis and viral replication are expressed in various cells. Thus, cells that include both the entry receptor (ACE2) and the spike protein priming enzyme (TMPRSS and/or CTSL) may be most susceptible to infection by SARS-CoV-2.
Headache, impaired consciousness, epilepsy, encephalitis, and acute myelitis have been reported in patients with COVID-19.[7] Notably, a case of SARS-CoV-2 encephalitis was reported by physicians in Beijing Ditan Hospital; genomic sequencing revealed the presence of SARS-CoV-2 in the patient’s cerebrospinal fluid. Subsequently, a similar case was reported by physicians in Japan.[10] These reports indicated that SARS-CoV-2 is able to infect the human CNS in some instances. Intracranial infections occur after damage to the BBB, which allows viral invasion.[12] In the present study, we found that the entry receptor ACE2 and spike protein priming enzyme CTSL were coexpressed in pericytes, which are specialized endothelial cells in the brain.[40] Furthermore, endothelial cells generally expressed lower levels of ACE2, which suggests that they can serve as an alternative route of infection. Notably, some vascular pericytes (eg, in the OE and RE) did not express ACE2, indicating that brain pericytes have a distinct role in blood pressure regulation and BBB maintenance.[41] Endothelial cells in the brain are bound together by tight junctions to ensure the integrity of the BBB. The use of endothelial cells and pericytes as host cells could allow viral proliferation; the eventual cell death would destroy the host cells and BBB, thereby releasing viral particles into the brain. Considering the extensive similarities described above in terms of expression patterns for ACE2, TMPRSS2, and CTSL in mice and humans, neurons presumably do not constitute a primary site of infection; however, vascular pericytes may be sensitive to SARS-CoV-2, thus allowing entry through the BBB and subsequent CNS infection.
Respiratory viruses may enter the CNS through neurons in the olfactory and trigeminal nerves (within the nasal cavity), or through neurons in the vagus nerve (within the trachea and lungs).[42] Human coronavirus OC43 can induce CNS neuropathology through interneuron propagation and axonal transport.[43] We found that the entry receptor for human coronavirus OC43 (HLA class I molecule) was expressed in neural cells, which enables the virus to infect neurons (Additional Fig. 2, https://links.lww.com/JR9/A45). However, the entry receptor ACE2 and the spike protein priming enzymes were not coexpressed in neural cells within the CNS, olfactory system, or respiratory system. An earlier epidemic strain of coronavirus, SARS-CoV, also binds ACE2; SARS-CoV particles have been detected in brain tissue, where they were located almost exclusively in neurons.[44] Considering the strong capacities for exosome biogenesis and release in neurons, ciliated cells, and OECs, exosome biogenesis and release could be involved in SARS-CoV-2 infection of the CNS. However, this hypothesis requires additional investigation.
The OE serves as the main odor detection and conduction tissue; it includes ciliated olfactory sensory neurons, glia-like sustentacular cells, and apical microvilli.[45] Olfactory sensory neurons are surrounded by sustentacular cells and the axons of OECs. Sustentacular cells serve as structural support for sensory neurons, whereas OECs guide olfactory sensory neuron axons that cross the cribriform plate at the skull base and terminate in the olfactory bulb.[46–48] Additionally, olfactory sensory neurons express odor receptors on dendritic cilia.[49] Thus, SARS-CoV-2 invasion into ciliated cells and sustentacular cells may lead to olfactory disorders.
SARS-CoV-2 primarily affects the respiratory system, with serious effects on the lungs.[50] In this study, we observed coexpression of ACE2 and TMPRSS/CTSL in the AT1, AT2, and ciliated cells in lung tissue. In lung alveoli, AT1 cells (ie, epithelial cells) are responsible for gas exchange; AT2 cells (ie, alveolar stem-like cells) are responsible for surfactant biosynthesis and immunoregulation.[51] Thus, SARS-CoV-2 infection may cause damage to alveolar infrastructure; the resulting increase in alveolar surface tension can induce dyspnea and hypoxia. The involvement of AT2 cells can lead to extensive inflammation.[52] The host immune response produces a proinflammatory cytokine storm, which manifests as acute lung injury and acute respiratory distress syndrome.[53,54] We found that CD209 was mainly expressed in macrophages and monocytes in lung tissue. Notably, CD209 may be responsible for viral antigen activation of specific T cells, as well as the secretion of perforin, interferon, and granzyme B; these processes can cause further injury to the lungs and remote organs, such as the brain.[55]
Finally, we compared the expression levels of viral entry factors across olfactory and respiratory systems; we found generally high expression levels in ciliated cells in the nasal mucosa, which were higher than the levels in AT2 cells. This finding indicates that such ciliated cells may be the “portal” for SARS-CoV-2 entry because of their anatomical location; they may also serve as a viral reservoir for persistent SARS-CoV-2 infection. These results have important implications for understanding the symptoms of anosmia and potential mechanisms involved in SARS-CoV-2 infection of the CNS.
Limitations
In this study, the coexpression pattern of virus entry genes only indicates the susceptibility of the cell types. It does not really represent the cell types the SARS-CoV-2 infects. Furthermore, we analyzed susceptible cell types based on only normal samples. It would be more convincing if we used samples from COVID-19 patients to study the viral load in different cell types. So, the results were limited.
Conclusion
We identified cells susceptible to SARS-CoV-2 infection in the CNS, olfactory system, and respiratory system. Our findings provide insights into the mechanisms that underlie neurological symptoms and CNS infection in patients with COVID-19.
Acknowledgments
None.
Author contributions
HX, JL: study design. ZK, JW, HZ, DX, HG, and WZ: data analysis. ZK, DX, and HG: data collection and generation. HZ, JW, ZL, XC, and JX: data interpretation. TM, ZK, AQ, and HX: manuscript drafting. AQ, HZ, TM, HX, and JL: overall supervising and organizing the study. All authors approved the final version for publication.
Financial support
This work was supported by the National Natural Ascience Foundation of China (No. 31821003 to HX) and the China Ministry of Science and Technology (No. 2018AAA0100300 to HX). The funders had no role in manuscript design, data collection and analysis, preparation of the manuscript, or decision to publish.
Institutional review board statement
There is no direct involvement of human subjects in this project. All the data use existing biological samples and data from prior studies. Therefore, ethical oversight and patient consent were not handled in this study.
Data availability statement
The single cell data are available at the Gene Expression Omnibus under the accession code GSE139522 and GSE134355, Sequence ReadArchive dataset (SRA https://www.ncbi.nlm.nih.gov/sra) at accession code SRA667466 and Atlas Data Protal (https://data.humancellatlas.org) at accession code PRJEB31843. Brochus data will be made available Mendeley Data (https://data.mendeley.com/datasets/7r2cwbw44m/1).
Conflicts of interest
There are no conflicts of interest.
References
[1]. Guan WJ, Ni ZY, Hu Y, et al. Clinical characteristics of
coronavirus disease 2019 in China. N Engl J Med 2020;382:1708–1720. doi:10.1056/NEJMoa2002032.
[2]. Chen N, Zhou M, Dong X, et al. Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study. Lancet 2020;395:507–513. doi:10.1016/S0140-6736(20)30211-7.
[3]. Huang C, Wang Y, Li X, et al. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet 2020;395:497–506. doi:10.1016/S0140-6736(20)30183-5.
[4]. Uginet M, Breville G, Assal F, et al. COVID-19 encephalopathy: clinical and neurobiological features. J Med Virol 2021;93:4374–4381. doi:10.1002/jmv.26973.
[5]. Sampaio Rocha-Filho PA, Albuquerque PM, Carvalho LCLS, et al. Headache, anosmia, ageusia and other neurological symptoms in COVID-19: a cross-sectional study. J Headache Pain 2022;23:1–11. doi:10.1186/s10194-021-01367-8.
[6]. Dell’Era V, Farri F, Garzaro G, et al. Smell and taste disorders during COVID-19 outbreak: cross-sectional study on 355 patients. Head Neck 2020;42:1591–1596. doi:10.1002/hed.26288.
[7]. Wu Y, Xu X, Chen Z, et al. Nervous system involvement after infection with COVID-19 and other coronaviruses. Brain Behav Immun 2020;87:18–22. doi:10.1016/j.bbi.2020.03.031.
[8]. Helms J, Kremer S, Merdji H, et al. Neurologic features in severe
SARS-CoV-2 Infection. N Engl J Med 2020;382:2268–2270. doi:10.1056/NEJMc2008597.
[9]. Xu Z, Shi L, Wang Y, et al. Pathological findings of COVID-19 associated with acute respiratory distress syndrome. Lancet Respir Med 2020;8:420–422. doi:10.1016/S2213-2600(20)30076-X.
[10]. Moriguchi T, Harii N, Goto J, et al. A first case of meningitis/encephalitis associated with SARS-Coronavirus-2. Int J Infect Dis 2020;94:55–58. doi:10.1016/j.ijid.2020.03.062.
[11]. Shehata GA, Lord KC, Grudzinski MC, et al. Neurological complications of covid-19: Underlying mechanisms and management. Int J Mol Sci 2021;22:4081. doi:10.3390/ijms22084081.
[12]. Li Z, Liu T, Yang N, et al. Neurological manifestations of patients with COVID-19: potential routes of
SARS-CoV-2 neuroinvasion from the periphery to the brain. Front Med 2020;14:533–541. doi:10.1007/s11684-020-0786-5.
[13]. Shang J, Wan Y, Luo C, et al. Cell entry mechanisms of
SARS-CoV-2. Proc Natl Acad Sci U S A 2020;117:11727–11734. doi:10.1073/pnas.2003138117.
[14]. Chen M, Shen W, Rowan NR, et al. Elevated ACE-2 expression in the olfactory neuroepithelium: implications for anosmia and upper respiratory
SARS-CoV-2 entry and replication. Eur Respir J 2020;56:2001948. doi:10.1183/13993003.01948-2020.
[15]. Zhou P, Yang XL, Wang XG, et al. A pneumonia outbreak associated with a new coronavirus of probable bat origin. Nature 2020;579:270–273. doi:10.1038/s41586-020-2012-7.
[16]. Wrapp D, Wang N, Corbett KS, et al. Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation. Science 2020;367:1260–1263. doi:10.1126/science.abb2507.
[17]. Wang K, Chen W, Zhang Z, et al. CD147-spike protein is a novel route for
SARS-CoV-2 infection to host cells. Signal Transduct Target Ther 2020;5:283. doi:10.1038/s41392-020-00426-x.
[18]. Amraei R, Yin W, Napoleon MA, et al. CD209L/L-SIGN and CD209/DC-SIGN act as receptors for
SARS-CoV-2. ACS Cent Sci 2021;7:1156–1165. doi:10.1021/acscentsci.0c01537.
[19]. Hoffmann M, Kleine-Weber H, Schroeder S, et al.
SARS-CoV-2 cell entry depends on
ACE2 and
TMPRSS2 and is blocked by a clinically proven protease inhibitor. Cell 2020;181:271–280.e8. doi:10.1016/j.cell.2020.02.052.
[20]. Meng T, Cao H, Zhang H, et al. The insert sequence in
SARS-CoV-2 enhances spike protein cleavage by TMPRSS. bioRxiv 2020. doi:10.1101/2020.02.08.926006.
[21]. Durante MA, Kurtenbach S, Sargi ZB, et al. Single-cell analysis of olfactory neurogenesis and differentiation in adult humans. Nat Neurosci 2020;23:323–326. doi:10.1038/s41593-020-0587-9.
[22]. Han X, Zhou Z, Fei L, et al. Construction of a human cell landscape at single-cell level. Nature 2020;581:303–309. doi:10.1038/s41586-020-2157-4.
[23]. Zeisel A, Muñoz-Manchado AB, Codeluppi S, et al. Brain structure. Cell types in the mouse cortex and hippocampus revealed by single-cell RNA-seq. Science 2015;347:1138–1142. doi:10.1126/science.aaa1934.
[24]. Madissoon E, Wilbrey-Clark A, Miragaia RJ, et al. scRNA-seq assessment of the human lung, spleen, and esophagus tissue stability after cold preservation. Genome Biol 2019;21:1. doi:10.1186/s13059-019-1906-x.
[25]. Lukassen S, Chua RL, Trefzer T, et al.
SARS-CoV-2 receptor
ACE2 and
TMPRSS2 are primarily expressed in bronchial transient secretory cells. EMBO J 2020;39:e105114. doi:10.15252/embj.20105114.
[26]. Luecken MD, Theis FJ. Current best practices in single-cell RNA-seq analysis: a tutorial. Mol Syst Biol 2019;15:e8746. doi:10.15252/msb.20188746.
[27]. Stuart T, Butler A, Hoffman P, et al. Comprehensive integration of single-cell data. Cell 2019;177:1888–1902.e21. doi:10.1016/j.cell.2019.05.031.
[28]. Korsunsky I, Millard N, Fan J, et al. Fast, sensitive and accurate integration of single-cell data with Harmony. Nat Methods 2019;16:1289–1296. doi:10.1038/s41592-019-0619-0.
[29]. Zhao Y, Wong L, Goh WWB. How to do quantile normalization correctly for gene expression data analyses. Sci Rep 2020;10:15534. doi:10.1038/s41598-020-72664-6.
[30]. Townes FW, Irizarry RA. Quantile normalization of single-cell RNA-seq read counts without unique molecular identifiers. Genome Biol 2020;21:160. doi:10.1186/s13059-020-02078-0.
[31]. Lukenda A, Dotlic S, Vukojevic N, et al. Expression and prognostic value of putative cancer stem cell markers CD117 and CD15 in choroidal and ciliary body melanoma. J Clin Pathol 2016;69:234–239. doi:10.1136/jclinpath-2015-203130.
[32]. Bost P, Giladi A, Liu Y, et al. Host-viral infection maps reveal signatures of severe COVID-19 patients. Cell 2020;181:1475–1488.e12. doi:10.1016/j.cell.2020.05.006.
[33]. Sungnak W, Huang N, BĂ©cavin C, et al.
SARS-CoV-2 entry factors are highly expressed in nasal epithelial cells together with innate immune genes. Nat Med 2020;26:681–687. doi:10.1038/s41591-020-0868-6.
[34]. Zhang H, Kang Z, Gong H, et al. Digestive system is a potential route of COVID-19: an analysis of single-cell coexpression pattern of key proteins in viral entry process. Gut 2020;69:1010–1018. doi:10.1136/gutjnl-2020-320953.
[35]. Brann DH, Tsukahara T, Weinreb C, et al. Non-neuronal expression of
SARS-CoV-2 entry genes in the olfactory system suggests mechanisms underlying COVID-19-associated anosmia. Sci Adv 2020;6:eabc5801. doi:10.1126/sciadv.abc5801.
[36]. Wan Y, Shang J, Graham R, et al. Receptor recognition by the novel coronavirus from Wuhan: an analysis based on decade-long structural studies of SARS coronavirus. J Virol 2020;94:e00127–e00120. doi:10.1128/JVI.00127-20.
[37]. Knyazev E, Nersisyan S, Tonevitsky A. Endocytosis and transcytosis of
SARS-CoV-2 across the intestinal epithelium and other tissue barriers. Front Immunol 2021;12:636966. doi:10.3389/fimmu.2021.636966.
[38]. Zhao MM, Yang WL, Yang FY, et al. Cathepsin L plays a key role in
SARS-CoV-2 infection in humans and humanized mice and is a promising target for new drug development. Signal Transduct Target Ther 2021;6:134. doi:10.1038/s41392-021-00558-8.
[39]. Ou X, Liu Y, Lei X, et al. Characterization of spike glycoprotein of
SARS-CoV-2 on virus entry and its immune cross-reactivity with SARS-CoV. Nat Commun 2020;11:1620. doi:10.1038/s41467-020-15562-9.
[40]. Campisi M, Shin Y, Osaki T, et al. 3D self-organized microvascular model of the human blood-brain barrier with endothelial cells, pericytes and astrocytes. Biomaterials 2018;180:117–129. doi:10.1016/j.biomaterials.2018.07.014.
[41]. Hirunpattarasilp C, Attwell D, Freitas F. The role of pericytes in brain disorders: from the periphery to the brain. J Neurochem 2019;150:648–665. doi:10.1111/jnc.14725.
[42]. Mori I. Transolfactory neuroinvasion by viruses threatens the human brain. Acta Virol 2015;59:338–349. doi:10.4149/av_2015_04_338.
[43]. Dubé M, Le Coupanec A, Wong AHM, et al. Axonal transport enables neuron-to-neuron propagation of human coronavirus OC43. J Virol 2018;92:e00404–e00418. doi:10.1128/JVI.00404-18.
[44]. Xu J, Zhong S, Liu J, et al. Detection of severe acute respiratory syndrome coronavirus in the brain: Potential role of the chemokine mig in pathogenesis. Clin Infect Dis 2005;41:1089–1096. doi:10.1086/444461.
[45]. Villar PS, Delgado R, Vergara C, et al. Energy requirements of odor transduction in the chemosensory cilia of olfactory sensory neurons rely on oxidative phosphorylation and glycolytic processing of extracellular glucose. J Neurosci 2017;37:5736–5743. doi:10.1523/JNEUROSCI.2640-16.2017.
[46]. Kondoh K, Lu Z, Ye X, et al. A specific area of olfactory cortex involved in stress hormone responses to predator odours. Nature 2016;532:103–106. doi:10.1038/nature17156.
[47]. Le Bourhis M, Rimbaud S, Grebert D, et al. Endothelin uncouples gap junctions in sustentacular cells and olfactory ensheathing cells of the olfactory mucosa. Eur J Neurosci 2014;40:2878–2887. doi:10.1111/ejn.12665.
[48]. Beites CL, Kawauchi S, Crocker CE, et al. Identification and molecular regulation of neural stem cells in the olfactory epithelium. Exp Cell Res 2005;306:309–316. doi:10.1016/j.yexcr.2005.03.027.
[49]. Savya SP, Kunkhyen T, Cheetham CEJ. Low survival rate of young adult-born olfactory sensory neurons in the undamaged mouse olfactory epithelium. J Bioenerg Biomembr 2019;51:41–51. doi:10.1007/s10863-018-9774-8.
[50]. Guan WJ, Liang WH, Zhao Y, et al. Comorbidity and its impact on 1590 patients with COVID-19 in China: a nationwide analysis. Eur Respir J 2020;55:2000547. doi:10.1183/13993003.00547-2020.
[51]. Nabhan AN, Brownfield DG, Harbury PB, et al. Single-cell Wnt signaling niches maintain stemness of alveolar type 2 cells. Science 2018;359:1118–1123. doi:10.1126/science.aam6603.
[52]. Kroetz DN, Allen RM, Schaller MA, et al. Type I interferon induced epigenetic regulation of macrophages suppresses innate and adaptive immunity in acute respiratory viral infection. PLoS Pathog 2015;11:e1005338. doi:10.1371/journal.ppat.1005338.
[53]. Desforges M, Le Coupanec A, Dubeau P, et al. Human coronaviruses and other respiratory viruses: underestimated opportunistic pathogens of the central nervous system? Viruses 2019;12:14. doi:10.3390/v12010014.
[54]. Mehta P, McAuley DF, Brown M, et al. COVID-19: consider cytokine storm syndromes and immunosuppression. Lancet 2020;395:1033–1034. doi:10.1016/S0140-6736(20)30628-0.
[55]. Channappanavar R, Perlman S. Pathogenic human coronavirus infections: causes and consequences of cytokine storm and immunopathology. Semin Immunopathol 2017;39:529–539. doi:10.1007/s00281-017-0629-x.
