key: cord-287172-h8zoplkm
authors: Ghobrial, Moheb; Charish, Jason; Takada, Shigeki; Valiante, Taufik; Monnier, Philippe P.; Radovanovic, Ivan; Bader, Gary D.; Wälchli, Thomas
title: The human brain vasculature shows a distinct expression pattern of SARS-CoV-2 entry factors
date: 2020-10-21
journal: bioRxiv
DOI: 10.1101/2020.10.10.334664
sha: 
doc_id: 287172
cord_uid: h8zoplkm

A large number of hospitalized COVID-19 patients show neurological symptoms such as ischemic- and hemorrhagic stroke as well as encephalitis, and SARS-CoV-2 can directly infect endothelial cells leading to endotheliitis across multiple vascular beds. These findings suggest an involvement of the brain- and peripheral vasculature in COVID-19, but the underlying molecular mechanisms remain obscure. To understand the potential mechanisms underlying SARS-CoV-2 tropism for brain vasculature, we constructed a molecular atlas of the expression patterns of SARS-CoV-2 viral entry-associated genes (receptors and proteases) and SARS-CoV-2 interaction partners in human (and mouse) adult and fetal brain as well as in multiple non-CNS tissues in single-cell RNA-sequencing data across various datasets. We observed a distinct expression pattern of the cathepsins B (CTSB) and -L (CTSL) - which are able to substitute for the ACE2 co-receptor TMPRSS2 - in the human vasculature with CTSB being mainly expressed in the brain vasculature and CTSL predominantly in the peripheral vasculature, and these observations were confirmed at the protein level in the Human Protein Atlas and using immunofluorescence stainings. This expression pattern of SARS-CoV-2 viral-entry associated proteases and SARS-CoV-2 interaction partners was also present in endothelial cells and microglia in the fetal brain, suggesting a developmentally established SARS-CoV-2 entry machinery in the human vasculature. At both the adult and fetal stages, we detected a distinct pattern of SARS-CoV-2 entry associated genes’ transcripts in brain vascular endothelial cells and microglia, providing a potential explanation for an inflammatory response in the brain endothelium upon SARS-CoV-2 infection. Moreover, CTSB was co-expressed in adult and fetal brain endothelial cells with genes and pathways involved in innate immunity and inflammation, angiogenesis, blood-brain-barrier permeability, vascular metabolism, and coagulation, providing a potential explanation for the role of brain endothelial cells in clinically observed (neuro)vascular symptoms in COVID-19 patients. Our study serves as a publicly available single-cell atlas of SARS-CoV-2 related entry factors and interaction partners in human and mouse brain endothelial- and perivascular cells, which can be employed for future studies in clinical samples of COVID-19 patients.

ACE2 is expressed in vascular endothelial-and smooth muscle cells of the brain, as previously shown at the protein level using immunohistochemistry 61 We visualized the data using t-distributed stochastic neighbor embedding (t-SNE) and

computed differential gene expression analyses for each specific organ in adult (26 9 organs/tissues, ca. 283'355 cells) and fetal (16 organs/tissues, ca. 188'587 cells) human organs/tissues (Figure 1a ,b).

In the adult, the expression of ACE2 was generally low in all tissues, in agreement with the previously reported low expression level of ACE2 in scRNA seq datasets 64 , demonstrating the highest expression in the jejunum, duodenum, kidney, and ileum ( Figure 1c) . TMPRSS2 expression displayed a somewhat broader distribution than ACE2 (as reported in Sungnak et al., Nat Med, 2020 64 ) with the highest expression in the prostate, followed by trachea, colon, stomach, pancreas, kidney, jejunum, duodenum, and lung ( Figure 1c ). In the temporal lobe and the cerebellum, both ACE2 and TMPRSS2 demonstrated low expression ( Figure 1c ). As mentioned above, CTSB and CTSL can substitute TMPRSS2 protease activity. Therefore, to address potential alternative mechanisms for protein S protease priming and thus SARS-CoV-2 entry into cells, we hereafter mainly focus on CTSB and CTSL. CTSB and CTSL were highly expressed in various organs, and CTSB showed the highest expression in the thyroid, heart, liver, and temporal lobe, followed by the lung, artery, and cerebellum ( Figure 1c ).

CTSL, in contrast, showed low expression in the temporal lobe and cerebellum (Figure 1c ), and CTSB showed higher expression than CTSL in both the temporal lobe and the cerebellum ( Figure 1c ). In the fetus, the expression of both ACE2 and TMPRSS2 was low in most tissues, displaying, however highest expression values in the intestine and adrenal gland ( Figure 1d ); ACE2 and TMPRSS2 showed very similar expression patterns across all organs, as revealed by dot plots (Figure 1d ). Similar to the adult, CTSB and CTSL were variably but overall highly expressed in all tested organs, and CTSB showed higher expression than CTSL in both the spinal cord and brain (Figure 1d ).

To examine differential expression patterns of the brain/CNS and the periphery (non-CNS),

we compared the brain/CNS (pool of temporal lobe and cerebellum in the adult, and pool of brain and spinal cord in the fetus) to a pool of all peripheral organs (Figure 1e ,f).

Interestingly, ACE2 and TMPRSS2 showed a higher expression in the periphery in both the adult and the fetus (Figure 1e ,f). In the adult, CTSB was higher expressed in the brain/CNS, both in intensity and percent of expression, whereas CTSL showed higher expression in the periphery (Figure 1e ). In the fetus, both CTSB and CTSL showed a higher expression in the periphery (Figure 1f ).

CTSB also displayed a higher expression than CTSL in the adult and fetal mouse brain/CNS as compared to the periphery, thereby confirming the observations made in humans (Extended Interestingly, CTSB was the highest expressed SARS-CoV-2 entry associated gene in adult and fetal brain endothelial whereas CTSL showed the highest expression of all SARS-CoV-2 entry-associated genes in adult and fetal peripheral endothelial (Figure 2g ,h).

Together, these data suggest that the distinct expression pattern of SARS-CoV-2 proteases and associated enzymes in the CNS is developmentally established. .

The brain vasculature is comprised of endothelial cells and perivascular cells of the neurovascular unit including neurons, pericytes, astrocytes, and immune cells such as microglia and macrophages 38,39,67,68,69 (Figure 3a ). To examine the expression patterns of SARS-CoV-2 entry factor transcripts in the neurovascular unit and thereby address potential entry mechanisms of SARS-CoV-2, we next addressed brain endothelial-and perivascular cells ( Figure 3a) . First, looking at all brain cells (endothelial-and perivascular cells pooled 13 together), we observed that CTSB showed the highest transcript expression of SARS-CoV-2 entry associated genes in both the adult and fetal brain, and that the SARS-CoV-2 entry associated genes were more highly expressed in the adult than in the fetal stage while displaying certain similarity ( Figure 3b ). Next, we compared the SARS-CoV-2 entryassociated gene expression patterns between endothelial-and perivascular cells at the adult and fetal stages (Figure 3c ,d,f,h,j). Next, we wondered whether these observed CTSB/CTSL-and SARS-CoV-2 entry associated factor expression patterns were retained in in vitro brain organoid models which were recently employed to study SARS-CoV-2 invasiveness to the brain 76 . In a recent single cell RNA-seq datasets of brain organoids 77 , CTSB showed the highest expression in cells undergoing epithelial-to-mesenchymal transition (EMT), followed by endothelial-like and endothelial-like progenitor cells and neuronal-like progenitor cells (Extended Data Figure   6a -c). When comparing the expression patterns in endothelial-like cells to the pool of perivascular cells, CTSB was is higher in endothelial-like cells and CTSL was higher in 15 perivascular cells (Extended Data Figure 6c ). These expression patterns were similar to the ones in the adult and fetal human brain datasets, indicating that brain organoids might constitute a model system to further explore the brain vascular-related observations made here Together, these data indicate that endothelial cells and perivascular cells (mainly microglia) of the fetal and adult CNS neurovascular unit express CTSB but not TMPRSS2, suggesting a potential alternative of how SARS-CoV-2 might enter the brain endothelium and neurovascular unit. Importantly, additional validation experiments need to be carried out to validate these findings at the protein level.

To further characterize the SARS-CoV-2 entry factor transcript expression patterns in brain endothelial cells, we examined the different brain endothelial cell clusters according to their arterio-venous specification, as previously described in mouse brain 78, 79, 80 (Figure 4a ). We found that CTSB was expressed in multiple brain endothelial cell types including veins, capillaries, endothelial cells 1 (EC1), antigen presenting-and vein (antigen presenting) endothelial cells (Figure 4b ). In all adult and fetal brain endothelial cell clusters, CTSB was amongst the highest expressed SARS-CoV-2 entry-associated transcripts (Figure 4a ,b and Extended Data Figure 7a ,b). In the adult brain vasculature, the veins cluster showed the highest average expression of CTSB, followed by EC1, capillaries, antigen-presenting veins, and antigen-presenting endothelial cells (Figure 4a,b) , whereas CTSB showed the highest expression in fetal veins > capillaries > arteries > EC1 (Extended Data Figure 7a,b) .

Notably, recent studies using single-cell transcriptomics in human and mouse tissues revealed endothelial clusters in a variety of tissues including the human bladder, and kidney (and others) as well as the mouse and human lung expressing MCH class II genes such as HLA-DPA1 and HLA-DRA involved in MHC class II-mediated antigen processing, loading and presentation 80 .

These observations of professional antigen-presenting signatures of endothelial cells suggested potential immune functions of endothelial cells, and CTSB might be presented at the cell-surface upon SARS-CoV-2-infection of brain endothelial cells.

These results suggest that CTSB expression in different endothelial clusters including antigenpresenting endothelial cells is developmentally established and might be reactivated in pathological conditions, for instance upon endothelial cell infection with SARS-CoV-2.

CTSB correlates with genes and pathways involved in inflammation, angiogenesis, coagulation, and blood-brain-barrier permeability in brain endothelial cells Next, we wondered which genes were associated with CTSB in brain endothelial cells. We performed Spearman's correlation analysis of CTSB in brain endothelial cells within the datasets. The observed correlation coefficients of the top 50 CTSB-correlated genes were relatively low, which might be due to dropout effects and technical noise in the scRNA seq datasets, and the top 50 CTSB correlated genes showed a slight predominance for the capillary-venous side, consistent with that of CTSB ( Figure 4c ). Notably, we observed various genes associated with immune functions/inflammation including innate and antiviral immune Table 1 ).

Expression of most of these inflammatory genes is highest in venous endothelial cells, and for some in vein antigen-presenting endothelial cells, antigen-presenting endothelial cells, and EC1 (Figure 4c ), indicating immunity functions for these endothelial cell clusters. The relatively high expression of inflammatory-related genes in antigen-presenting (venous) endothelial cells is plausible whereas the relative preference for the venous side remains to be further explored.

Moreover, genes involved in angiogenesis and cell-extracellular matrix interactions including SPRED2, FSCN1, SLC9A3R2, GPR124, and others, and were also abundantly present in the rank list and were highest in capillaries, veins, and EC1 (Figure 4c, green and Supplementary   Table 1 ). We also observed several genes associated with endothelial glucose-and fatty acid metabolism, namely MRPL45 and TXN2, which were highest in EC1 and veins (Figure 4c, blue and Supplementary Table 1) .

We observed a similar expression pattern of the top 50 CTSB-correlated genes in the fetal brain vascular endothelial cells (Extended Data Figure 7c and Supplementary Genes involved in angiogenesis and cell-extracellular matrix interactions including SPARC, RHOA, TMSB10, VIM, CLDN5, and others, as well as genes associated with endothelial glucose-and fatty acid metabolism, namely SEPW1 and POMP were also abundantly present and were all highest in veins, followed by capillaries, arteries, and EC1 (Extended Data Table 2) .

We next performed pathway analysis using GSEA and Cytoscape on the CTSB-correlated genes. Notably, the top regulated pathways included inflammation, angiogenesis, coagulation, cell-extracellular matrix interaction, viral-host interaction, vascular metabolism, blood-brain-barrier permeability, and reactive oxygen species (ROS) in both the adult and the fetal brain endothelium ( Together, these data reveal that CTSB is highly expressed in various endothelial cell clusters of the fetal and adult human brain and that pathways downstream of CTSB might provide a suggestive explanation of some of the neurovascular symptoms observed in COVID-19

patients.

Upon entry of the SARS-CoV-2 virus into the host cell, the virus interacts with multiple intracellular proteins 62 . Therefore, we next addressed the expression of the recently described 332 SARS-CoV-2 interaction partners (= protein-protein interactions (PPIs) = proteins that are physically associated with SARS-CoV-2 proteins) in the different adult and fetal organs (Extended Data Figure 8a -d). In the adult, SARS-CoV-2 interactions partners were highly expressed in trachea, cerebellum, artery, kidney, and temporal lobe (Extended Data Figure   8e ), and the pooled analysis showed a higher expression in CNS versus non-CNS organs (Extended Data Figure 8f ). In the fetus, SARS-CoV-2 interactions partners were highly expressed in fetal male gonad, adrenal gland, brain conjointly with heart, intestine and kidney (Extended Data Figure 8g ). In the pooled analysis, however, due to the relatively low expression in the fetal spinal cord the expression was higher in non-CNS-as compared to CNS organs (Extended Data Figure 12h ).

Next, we focused on the endothelial cells in the different adult and fetal organs (Extended Data Figure 8i -l). In the adult, endothelial cells of the temporal lobe and the cerebellum were among the organs/tissue displaying high expression of SARS-CoV-2 interaction partners, together with endothelial cells from artery, cervix, trachea, uterus, kidney, muscle and spleen (Extended Data Figure 8m ). The relatively high expression in CNS versus non CNS endothelial cells was further revealed in the pooled analysis (Extended Data Figure 8n ). At the fetal stage, brain endothelial cells showed the highest expression of SARS-CoV-2 interaction partners among all organs/tissues, followed by endothelial cells from male gonad, adrenal gland, pancreas, heart, and muscle (Extended Data Figure 8o ). Again, due to the relatively 

When addressing the different brain endothelial cell clusters according to their arterio-venous specification, we found that in both adult and fetal brain endothelial cell clusters, SARS-CoV-2 interaction partners showed a tendency of higher expression in the veins and capillaries as compared to the arteries (Extended Data Figure 9y -z). In the adult endothelial cells, the cluster endothelial cell 1 EC1 showed the highest expression of SARS-CoV-2 interaction partners, followed by antigen-presenting veins, veins, capillaries, EndoMT, and arteries (Extended Data Figure 9y ), whereas SARS-CoV-2 interaction partners showed the highest expression in fetal capillaries > arteries > veins > EC1 (Extended Data Figure 9z ).

With CTSB being a SARS-CoV-2 entry associated gene and SARS-CoV-2 interaction partners being intracellular virus-interaction partners post viral cellular entry, we next wondered whether there was a correlation between the top CTSB correlated pathways and the SARS-CoV-2 interaction partners. Notably, the top CTSB-correlated pathways including inflammation, angiogenesis, coagulation, cell-extracellular matrix interaction, vascular metabolism, blood-brain-barrier permeability, and reactive oxygen species (ROS) all displayed a high correlation with SARS-CoV-2 interaction partners in both the adult and fetal brain endothelium (Extended Data Figure 10a -zvi, Extended Data Figure 11a -zvi). Thus, all these pathways that were recently linked to COVID-19 patients with neurological-and vascular symptoms 30,49,81,82 showed a strong correlation with the intracellular SARS-CoV-2 interaction partners, therefore suggesting a link between CTSB-mediated cellular entry and intracellular signaling post viral entry into the host cell.

CTSB -but not TMPRSS2 -protein is expressed in the human brain vasculature whereas CTSL-but not TMPRSS2 -protein is expressed in the peripheral vasculature   Finally, to examine the observed mRNA expression patterns of CTSB, CTSL, and TMPRSS2 at the protein level, we took advantage of the publicly available Human Protein Atlas dataset (https://www.proteinatlas.org/).

TMPRSS2 did not show protein expression in the brain, but was expressed in the prostate, jejunum, kidney, and others, again similar to the observed mRNA expression patterns ( Figure   5a ). Interestingly, and in line with the observations made in the scRNA-seq data, immunohistochemical analysis revealed that TMPRSS2 was neither expressed in the brain endothelium ( In contrast to CTSB, CTSL did not show protein expression in the brain, but was expressed in numerous other organs known to be affected by systemic SARS-CoV-2, such as the prostate, lung, gastrointestinal tract, liver, kidney, jejunum, and colon, among others (Extended Data Immunohistochemical analysis using the validated antibody CAB000459 83, 84, 85 revealed that CTSL was expressed in endothelial cells of various organs but not in the brain vasculature, displaying high cytoplasmic/membranous expression in peripheral vascular endothelial cells across various peripheral organs including the liver, placenta, lung, spleen, colon, kidney, pancreas, and heart (Extended Data Figure 15a -m).

To further validate these findings, we performed double immunostainings for an endothelial cell marker (CD31) and CTSB, CTSL, and TMPRSS2 on human temporal lobe specimens from our neurosurgical operating theatre. CTSB showed high expression in CD31 + endothelial cells as well as in CD31cells of the neuropil and intermediate expression in

NeuN + neurons, GFAP + astrocytes, IB1 + microglia (Figure 6a -zii), in agreement with the observations made in the HPA and the scRNA seq datasets described above. TMPRSS2 and CTSL either were absent in brain endothelial-and perivascular cells or showed very low expression levels (Figure 6a -l), again confirming the scRNA-seq-and human protein atlas data.

Together, these data indicate that whereas TMPRSS2 is absent in brain-and peripheral vascular endothelial cells, CTSB is expressed in the brain vasculature and CTSL in the peripheral vasculature at both the mRNA and protein levels, suggesting potential alternative mechanism for SARS-CoV-2 cellular entry into the brain and peripheral vasculature.

In this study, analysing multiple scRNA-seq datasets as well as referring to the Human Protein Atlas and performing immunofluorescent stainings, we found that the human brain vasculature expresses a distinct profile of SARS-CoV-2 entry associated genes. Most interestingly, the viral entry-associated protease CTSB but not TMPRSS2 is highly expressed in brain vascular endothelial cells whereas CTSL but not TMPRSS2 is highly expressed in vascular endothelial cells of peripheral organs. These observations suggest a potential mechanism of SARS-CoV-2 viral entry into brain-and peripheral endothelial cells that might underlie the neurovascular-and vascular symptoms observed in some COVID-19 patients (Extended Data Figure 16 ).

The expression patterns we found can provide insight about the susceptibility of the human vasculature to SARS-CoV-2 infections and couldin a next stepbe followed by similar studies on currently still limited clinical samples from COVID-19 patients. Towards this end we examined gene expression of SARS-CoV-2 entry-associated genes and interaction partners in multiple scRNA-seq datasets from different tissues, including the brain, spinal cord and various peripheral tissues. We are aware of the limitations of these analyses, as for instance sparse cell types might be lacking or underrepresented/under-detected due to their low abundance, technical limitations related to isolation protocols, or bioinformatic analyses including technical/computational dropout effects. Thus, the specificity is high (meaning positive results are highly reliable) whereas the sensitivity is limited and thus negative results should be interpreted with care.

SARS-CoV-2 shows systemic effects affecting multiple organ systems including the brain, liver, kidney, heart, and others, and increasing evidence suggests that blood vessel endothelial cells exert crucial roles in the underlying pathogenesis 30,49,82,81 . Whereas cellular entry of SARS-CoV-1 exclusively depends on CTSL and TMPRSS1 50,53 , cellular entry of SARS-

CoV-2 can occur either via TMPRSS2 or CTSB/CTSL 53,59 (Extended Data Figure 16 ). Thus, the widespread expression of CTSB in the brain (vasculature) and the expression of CTSL in multiple other SARS-CoV-2 affected organs/tissues suggest that CTSB and CTSL might be involved in alternative entry mechanisms and transmission routes that could be responsible for the neurovascular-and vascular phenotypes observed in COVID-19 patients 30,32 .

Furthermore, our findings that CTSB isin addition to the brain endotheliumalso highly expressed in microglia (and that CTSL shows high expression in peripheral endothelial cells and macrophages) suggests a potential mechanism involving these two cell types e.g. via an endothelial-to-microglia crosstalk 86, 87 , that could explain endotheliitis/endothelialitis in peripheral organs 49,81 and in the brain 30 . It has been suggested that the SARS-CoV-2 affected endothelial cell is characterized by increased vascular leakage, pro-coagulative and proinflammatory states in the lung 82 , and that a combination of those might account for the 30% of COVID-19 patients with severe lung damage in part owed to an overreacting inflammatory response 82 and multi-organ failure 82 . Interestingly, our endothelial cluster analysis revealed a relative preference of CTSB expression for the venous endothelial side while the correlation analysis showed a high expression of inflammatory-related genes in antigen-presenting (venous) endothelial cells, suggesting a preference of SARS-CoV-2 entry factors and downstream pathways for the venous endothelium. In light of the findings that SARS-CoV-1 caused venous vasculitis in small veins of the brain and lung 88 , and that SARS-CoV-2 causes severe endothelialitis particularly in venous vascular bed 89consistent with our findingsit will be exciting to further investigate the potentially different susceptibility of endothelial cells along the arteriovenous specification.

Moreover, CTSB is known to exert pivotal roles in cancer including brain tumors and brain inflammation/inflammatory brain disease via Interleukin-1β (IL1-β) and tumor necrosis factor-α (TNF-α) 90 , and is capable of crossing the blood-brain-barrier 91 . Thus, we speculate that the co-expression of CTSB in microglia and endothelial cells could account for vascular leakage and opening of the blood-brain-barrier with subsequent increased leukocyte migration across the BBB and infection of the brain vascular endothelium 30,82 which might explain neurological symptoms such as stroke, epilepsy, necrotizing hemorrhages, and encephalopathy 30,32 . The results from our correlation-and pathway analyses showing that CTSB is tightly linked to genes and pathways involved in viral entry, inflammation, angiogenesis, coagulation, vascular metabolism, and blood-brain-barrier leakage, further supports these speculations. For instance, we observed clusters of antigen (MHC class II)presenting brain endothelial cells expressing CTSB, possibly explaining cellular crosstalk (for instance with microglia) required for inflammatory responses 82 ; as endothelial cells cannot act as antigen-presenting cells on their own. Along these lines, the other cathepsinsin addition to CTSB -showing higher expression in the adult-(CTSA, CTSD) and fetal (CTSC, CTSF)

brain endothelium were all linked to inflammatory processes. For instance, CTSB is known to interact with CTSA and CTSD 52,53,92-94 and to be involved in brain inflammation and other immune functions such as brain inflammation 90 and immune response 95 . Moreover, the viral receptor/receptor-associated enzyme ST6GAL1 encoding for the human Beta-galactoside alpha-2,6-sialyltransferase 1 96 , is a protein that is involved in the generation of the cell-surface carbohydrate determinants and differentiation antigens HB-6, CDw75, and CD76 97 and that is found in mouse high endothelial cells of mesenteric lymph node and Peyer's patches, where its suggested function is to be involved in the B cell homing to Peyer's patches 98 .

Interestingly, our correlation analysis indeed showed a high correlation of CTSB with inflammatory pathways, angiogenesis, and viral-host-interaction, therefore further suggesting a direct link of CTSB with inflammatory, angiogenic, and coagulative responses in the brain endothelium, pathways that were all recently shown to exert pivotal roles in SARS-CoV-2 pathogenesis and COVID-19 patients in peripheral-and CNS organs 30, 81, 82 . Moreover, 26 inflammatory-, angiogenic-, coagulative-, cell-ECM interaction, vascular/BBB permeability-, metabolism, and oxidative stress pathways all correlated with both the SARS-CoV-2 entryassociated gene CTSB as well with the SARS-CoV-2 intracellular interaction partners, suggesting a link between SARS-CoV-2 entry mechanisms and intracellular signaling. Thus, the role of CTSB and the other SARS-CoV-2 entry-associated genes in inflammatory, angiogenic, and the other aforementioned responses within the brain endothelium deserves further investigation.

For instance, we observed a high correlation of with endothelial glucose-and fatty acid metabolism, known to be key for viral replication and propagation 82, 99 , indicating that endothelial cells represent an attractive metabolic target for SARS-CoV-2 infection.

To that regard, it was recently suggested that risk factors for COVID-19 such as old age, obesity, hypertension, and diabetes mellitus are all characterized by pre-existing vascular dysfunction with altered vascular endothelial metabolism 99 . Thus, whether certain endothelial clusters in specific organs and certain pathological conditions display a metabolic signature that is more prone to SARS-CoV-2 infection remains to be explored.

Interestingly, CTSB and CTSL have been shown to substitute TMPRSS2 for viral entry of the Ebola virus, which affects the vasculature in the brain and in peripheral organs 100 , but their role in SARS-CoV-2 remains unknown. Whereas the expression of CTSB in the brain and in multiple peripheral tissues was previously reported 101-105,106,107 , the high expression of CTSB in the brain vasculature endothelium and of CTSL in the peripheral vascular endothelium at both the mRNA and protein levels was not reported to our knowledge, nor was the absence of TMPRSS2. Our own experiments revealed CTSB expression in endothelial cells, neurons, astrocytes, pericytes and microglia within the human brain neurovascular unit and might indeed provide the basis of a potential explanation for some of the putative SARS-CoV-2mediated effects on the human blood-brain-barrier 30 as well as some of the observed neurovascular symptoms in COVID-19 patients 30, 81 . We thus propose a working model in which SARS-CoV-2 (and to a much lower extent SARS-CoV-1) can infect brain endothelial cells via ACE2 and mainly CTSB and peripheral endothelial cells via ACE2 and mainly CTSL (Extended Data Figure 16 ). However, regarding the periphery, further validation is needed addressing the CTSL expression patterns within the perivascular niche in various peripheral vascular beds taking into account their tissue-specific properties 108 . We are also well aware of the limitations of our study needing further confirmation using functional assays in vivo and in vitro.

Taken together, our findings may have important implications for understanding SARS-CoV-2 cellular entry and viral transmissibility in the brain, the brain vasculature, and in peripheral vascular beds. Targeting CTSB (and CTSL) using already approved drugs/known inhibitors (for instance E-64d, ammonium chloride) 53 could result in inhibiting angiogenesis, vascular metabolism, vascular leakage, and the inflammatory response, resulting in vascular normalization 82, 53 . As CTSB is expressed in the brain endothelium, and as SARS-CoV-2 invades the brain (putatively via CTSB) and might affect blood-brain-barrier integrity 109,30 , our discoveries might have important translational implications for both neurovascular-and vascular symptoms observed in COVID-19 patients.

In summary, our work further illustrates the opportunities emerging from integrative analyses of publicly available datasets including scRNA sequencing and the Human Protein Atlas. Tissue sections (40 µm) of human adult brain (derived from temporal lobectomies) were stained for TMPRSS2, CTSB, CTSL (green), the vascular endothelial cell markers CD31

(pan-endothelial marker, red), the microglial marker IBA1 (red), the neuronal marker NeuN (red), the pericyte marker PDGFRB (red), the astrocytic marker GFAP (red), and DAPI nuclear counterstaining (blue). a-l Expression of TMPRSS2, CTSB, and CTSL in blood vessel endothelial cells in the adult human brain. 

Datasets were retrieved from published datasets of multiple human and mouse tissues of the human and mouse cell atlas 63, 111 . Adult brain datasets were retrieved from publically available sources including (Jäkel et al. 70 

Benjamini -Hochberg correction false-discovery rate (FDR) q-value that ranges from 0 (highly significant) to 1 (not significant). The resulting pathways are ranked using NES, and 

Adult ECs Fetal ECs Adult

Average Expression Percent Expressed 0 10 20 30

Non-CNS ECs Fetal endothelial cells 

Average Expression Percent Expressed Figure 3 antigen presenting antigen presenting A C E 2 Mast cell Mast cell T S B  C  T S L  C  T S A  C  T S C  C  T S D  C  T S E  C  T S F  C  T S G  C  T S H  C  T S K  C  T S O  C  T S S  C  T S V  C  T S W  C  T T S B  C  T S L  C  T S A  C  T S C  C  T S D  C  T S E  C  T S F  C  T S G  C  T S H  C  T S K  C  T S O  C  T S S  C  T S V  C  T S W  C  T S Macrophage Epithelial cell Mural cells Extended Data Figure 7 

PVCs ECs A C E 2 T M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P E P D P P S T 6 G A L 1 S T 3 G A L PercentM A P K 1 S L C 1 6 A 1 M A V S C D 6 3 S P R E D 2 R P S 7 P 1 0 F S C N 1 M X 1 A L A D H I G D 1 B M R P L 4 5 S L C 6 A 1 P R T G S L C 9 A 3 R 2 A R L 6 I P 5 E I F 5 B C 9 R P S A G P R 1 2 4 K E A P 1 D N A J C 1 3 C Y B 5 R 3 T X N 2 R A S G R P 2 P R E L P C G N L 1 S P O C K 2 P 2 R Y 1 4 F A M 1 3 5 B R P S 5 C 2 0 o r f 2 7 T M E M 8 7 A S F T 2 D 3 P R E P R B M 1 2 B G R M 3 D N A J C 8 C A 4 A C 0 0 4 4 5 3 . 8 M Y L K F A M 3 5 A X R N 1 E L T D 1 R P L 3 6 A L C Y B A T R A D D U B E 2 L 6 E S D B N I P LA C E 2 T M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P E

A C E 2 T M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P E P D P P 4 S T 6 G A L 1 S T 3 G A L 4

A C E 2 T M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P E P D P P 4 S T 6 G A L 1 S T 3 G AA C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P EA C E 2 T M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P EP R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P EP R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z AT M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C TA C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P ET M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P EA C E 2 T M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P EA C E 2 T M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z Ah j A C E 2 T M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z A N P ET M P R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z AP R S S 2 C T S B C T S L C T S A C T S C C T S D C T S E C T S F C T S G C T S H C T S K C T S O C T S S C T S V C T S W C T S Z AO A T M S B 1 0 G N G 1 1 S L C 3 A 2 A C T B P E C A M 1 Z F P 3 6 L 1 P R O C R C D 8 1 K L F 6 G N A I 2 S 1 0 0 A 1 0 V I M H 3 F 3 B T M S B 4 X R H O B V A C C N L 1 T S C 2 2 D 3 S E R P I N H 1 P K M C S R P 2 M Y L 1 2 A R N A S E 1 E F N A 1 C A V 1 P S A P F A M 2 1 3 A R H O C N D U F B 2 Y W H A Z A 2 M B S G D S T N S E L K M A G E D 2 D A D 1 B 2 M C D C

SARS-CoV-2 interaction partners FA metabolism Extended Data Figure 16 

The neuroinvasive potential of SARS-CoV2 may play a role in the respiratory failure of COVID-19 patients

Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study

A Novel Coronavirus from Patients with Pneumonia in China

The socio-economic implications of the coronavirus pandemic (COVID-19): A review

Coronavirus Patients Are Reporting Neurological Symptoms. Here's What You Need to Know

Sars-Cov-2: Underestimated damage to nervous system

Guillain-Barre Syndrome Associated with SARS-CoV-2 Infection in a Pediatric Patient

Guillain-Barre Syndrome Associated with SARS-CoV-2 Detection and a COVID-19 Infection in a Child

SARS-CoV-2-associated Guillain-Barre syndrome with dysautonomia

Guillain-Barre Syndrome Associated with SARS-CoV-2

Guillain-Barre syndrome associated with SARS-CoV-2 infection

Guillain-Barre Syndrome associated with SARS-CoV-2 infection

Guillain-Barre syndrome associated with SARS-CoV-2 infection: causality or coincidence?

The neuroinvasive potential of SARS-CoV2 may play a role in the respiratory failure of COVID-19 patients

CoV-2: Olfaction, Brain Infection, and the Urgent Need for Clinical Samples Allowing Earlier Virus Detection

Neurological involvement of coronavirus disease 2019: a systematic review

Neurological associations of COVID-19

Neurological and neuropsychiatric complications of COVID-19 in 153 patients: a UK-wide surveillance study

Inhibitors of cathepsin L prevent severe acute respiratory syndrome coronavirus entry

Evidence that TMPRSS2 Activates the Severe Acute Respiratory Syndrome Coronavirus Spike Protein for Membrane Fusion and Reduces Viral Control by the Humoral Immune Response

Efficient Activation of the Severe Acute Respiratory Syndrome Coronavirus Spike Protein by the Transmembrane Protease TMPRSS2

A Transmembrane Serine Protease Is Linked to the Severe Acute Respiratory Syndrome Coronavirus Receptor and Activates Virus Entry

Proteolytic Cleavage of the SARS-CoV-2 Spike Protein and the Role of the Novel S1/S2 Site

Effect of angiotensin-converting enzyme inhibition and angiotensin II receptor blockers on cardiac angiotensin-converting enzyme 2

Tissue distribution of ACE2 protein, the functional receptor for SARS coronavirus. A first step in understanding SARS pathogenesis

A SARS-CoV-2 protein interaction map reveals targets for drug repurposing

Construction of a human cell landscape at single-cell level

SARS-CoV-2 entry factors are highly expressed in nasal epithelial cells together with innate immune genes

Single-cell RNA-seq data analysis on the receptor ACE2 expression reveals the potential risk of different human organs vulnerable to 2019-nCoV infection

Single cell RNA sequencing of 13 human tissues identify cell types and receptors of human coronaviruses

The neurovascular unit -concept review

The blood-brain barrier/neurovascular unit in health and disease

The Neurovascular Unit Coming of Age: A Journey through Neurovascular Coupling in Health and Disease

Altered human oligodendrocyte heterogeneity in multiple sclerosis

Single-Cell Multi-omic Integration Compares and Contrasts Features of

Integrative single-cell analysis of transcriptional and epigenetic states in the human adult brain

Massively parallel single-nucleus RNA-seq with DroNc-seq

Single-cell genomics identifies cell type-specific molecular changes in autism

Molecular Diversity of Midbrain Development in Mouse, Human, and Stem Cells

Neuroinvasion of SARS-CoV-2 in human and mouse brain. bioRxiv

Engineering of human brain organoids with a functional vascular-like system

A molecular atlas of cell types and zonation in the brain vasculature

Single-Cell Transcriptome Atlas of Murine Endothelial Cells

An Integrated Gene Expression Landscape Profiling Approach to Identify Lung Tumor Endothelial Cell Heterogeneity and Angiogenic Candidates

Pulmonary Vascular Endothelialitis, Thrombosis, and Angiogenesis in Covid-19

COVID-19: the vasculature unleashed

Profiling of Atherosclerotic Lesions by Gene and Tissue Microarrays Reveals PCSK6 as a Novel Protease in Unstable Carotid Atherosclerosis

Matrix Metalloproteinase-2 Cleavage of the beta 1 Integrin Ectodomain Facilitates Colon Cancer Cell Motility

Heterogeneity in signaling pathways of gastroenteropancreatic neuroendocrine tumors: a critical look at notch signaling pathway

Microglial interactions with the neurovascular system in physiology and pathology

Microglia-blood vessel interactions: a double-edged sword in brain pathologies

The clinical pathology of severe acute respiratory syndrome (SARS): a report from China

Direct evidence of SARS-CoV-2 in gut endothelium

The NLRP3 Inflammasome Contributes to Brain Injury in Pneumococcal Meningitis and Is Activated through ATP-Dependent Lysosomal Cathepsin B Release

The authors declare no competing financial interests.