key: cord-0990486-jghu52qi
authors: Cantuti-Castelvetri, Ludovico; Ojha, Ravi; Pedro, Liliana D.; Djannatian, Minou; Franz, Jonas; Kuivanen, Suvi; Kallio, Katri; Kaya, Tuğberk; Anastasina, Maria; Smura, Teemu; Levanov, Lev; Szirovicza, Leonora; Tobi, Allan; Kallio-Kokko, Hannimari; Österlund, Pamela; Joensuu, Merja; Meunier, Frédéric A.; Butcher, Sarah; Winkler, Martin Sebastian; Mollenhauer, Brit; Helenius, Ari; Gokce, Ozgun; Teesalu, Tambet; Hepojoki, Jussi; Vapalahti, Olli; Stadelmann, Christine; Balistreri, Giuseppe; Simons, Mikael
title: Neuropilin-1 facilitates SARS-CoV-2 cell entry and provides a possible pathway into the central nervous system
date: 2020-07-15
journal: bioRxiv
DOI: 10.1101/2020.06.07.137802
sha: 4f7b0222cd263f1a6fd0feb00d31468b4915583f
doc_id: 990486
cord_uid: jghu52qi

The causative agent of the current pandemic and coronavirus disease 2019 (COVID-19) is the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)1. Understanding how SARS-CoV-2 enters and spreads within human organs is crucial for developing strategies to prevent viral dissemination. For many viruses, tissue tropism is determined by the availability of virus receptors on the surface of host cells2. Both SARS-CoV and SARS-CoV-2 use angiotensin-converting enzyme 2 (ACE2) as a host receptor, yet, their tropisms differ3-5. Here, we found that the cellular receptor neuropilin-1 (NRP1), known to bind furin-cleaved substrates, significantly potentiates SARS-CoV-2 infectivity, which was inhibited by a monoclonal blocking antibody against the extracellular b1b2 domain of NRP1. NRP1 is abundantly expressed in the respiratory and olfactory epithelium, with highest expression in endothelial cells and in the epithelial cells facing the nasal cavity. Neuropathological analysis of human COVID-19 autopsies revealed SARS-CoV-2 infected NRP1-positive cells in the olfactory epithelium and bulb. In the olfactory bulb infection was detected particularly within NRP1-positive endothelial cells of small capillaries and medium-sized vessels. Studies in mice demonstrated, after intranasal application, NRP1-mediated transport of virus-sized particles into the central nervous system. Thus, NRP1 could explain the enhanced tropism and spreading of SARS-CoV-2.

pathogenicity by priming of fusion activity 9 and create potentially additional cell surface receptor binding sites. In fact, proteolytic cleavage by furin, exposes a conserved carboxyterminal (C-terminal) motif RXXROH (where R is arginine and X is any amino acid; R can be substituted by lysine, K) on the substrate protein. Such C-terminal sequences that conforms to the 'C-end rule' (CendR) are known to bind to and activate neuropilin receptors (NRP1 and NRP2) at the cell surface 10, 11 . Physiological ligands of NRPs, such as vascular endothelial growth factor A (VEGFA), and class-3 semaphorins (SEMA-3), are cleaved by furin and proprotein convertases, with the resulting C-terminal exposure triggering the receptor interaction 12 . These ligands are responsible for angiogenesis, and for mediating neuronal axon guidance, respectively 12 . Importantly, recent cryo-electron microscopy structure of the SARS-CoV-2 spike demonstrated the S1/S2 junction in a solvent-exposed loop, therefore accessible for interactions 13, 14 .

To determine whether SARS-CoV-2 uses NRP1 for virus entry, we generated replicationdeficient lentiviruses pseudotyped with SARS-CoV-2 spike protein (S) that drive expression of green fluorescent protein (GFP) upon infection. Such SARS-CoV-2 pseudoviruses are ideally suited for virus entry assays, as they allow separating viral entry from other steps in the virus life cycle, such as replication and assembly. Lentiviruses pseudotyped with the vesicular stomatitis virus (VSV) spike G were used as controls. For this assay, we used HEK-293T cells, as they do not express ACE2 and were not infected by lentiviral particles pseudotyped with the SARS-CoV-2 S protein. To test the potential role of NRP1 in virus infection, cells were transfected with plasmids encoding the main attachment receptors, either ACE2 or the transmembrane protease serine 2 (TMPRSS2), and NRP1. Both ACE2 and TMPRSS2 are required for cleavage of the S protein, and thus necessary for the fusion of viral and cellular membranes 9 . When expressed alone, ACE2 rendered cells susceptible to infection (Fig. 1a) .

NRP1 alone allowed lower, yet detectable levels of infection, both in HEK-293T and in Caco-2 cells (Fig. 1a,b) , while cells transfected with plasmids encoding only TMPRSS2 were not infected (Fig. 1a) . The co-expression of TMPRSS2 with either ACE2 or NRP1 potentiated the infection, with ACE2 together with TMPRSS2 being twice as efficient as NRP1 with TMPRSS2 (Fig. 1c) . Maximal levels of infection were achieved when all three plasmids, driving expression of ACE2, NRP1 and TMPRSS2, were used for co-transfection (Fig. 1d) . To further test the specificity of NRP1-dependent virus entry, we developed a series of monoclonal antibodies (mAbs), including function-blocking antibodies, against the extracellular b1b2 domain of NRP1, known to mediate the binding of CendR peptides 12 . The potency of these mAbs in preventing cellular binding and internalization of NRP ligands was tested using 80 nm silver nanoparticles (AgNP) decorated with the prototypic NRP1-binding CendR peptide RPARPAROH 10 . These synthetic AgNP-CendR, but not the control particles without peptides, bind to and are internalized efficiently into PPC1 cells, which express high levels of NRP1 10 .

One of the three antibodies, mAb3, efficiently blocked AgNP-CendR binding (Extended Data Figure 1a ) and internalization (Extended Data Fig. 1b) , while mAb1 had no effect and was used as a control in further experiments. mAb3 antibody recognizes the CendR binding pocket on the b1 domain of NRP1, and its binding to NRP1 with mutated binding pocket is compromised.

Similar results were obtained in NRP1-expressing HEK-293T cells (Extended data Figure   2a ,b). Importantly, upon incubation with HEK-293T cells expressing ACE2, NRP1 and TMPRSS2, mAb3 significantly reduced infection by SARS-CoV-2 pseudoviruses (Fig. 1e) . To provide supporting evidence for the direct binding of virus particles to NRP1, SARS-CoV-2 pseudoviruses were pre-incubated with recombinant, soluble extracellular b1b2 domain of NRP1. The rationale of this approach was that the soluble form of NRP1 b1b2 would bind the virus particles and compete for the binding of NRP1 at the cell surface. As a negative control, we used b1b2 mutant with triple mutation (S346A, E348A and T349A in the CendR binding pocket), known to abrogate the binding of CendR peptides. We validated the effect of the recombinant NRP1 proteins, by showing that b1b2, but not the mutant protein, blocked uptake of AgNP-CendR in HEK-293T cells (Extended Data Fig. 2c) . Importantly, addition of the soluble b1b2 domain of NRP-1 significantly reduced SARS-CoV-2 pseudovirus infection, whereas triple b1b2 mutant had no effect (Fig. 1f) . Thus, these data demonstrate that NRP1 specifically potentiates SARS-CoV-2 pseudovirus infection.

Next, we explored SARS-CoV-2 isolated from COVID-19 patients from the Helsinki University Hospital. RNA viruses such as SARS-CoV-2 have a remarkable ability to adapt to their host environment by generating adaptive mutations in a short period. Confirming recent reports 15 , we found that SARS-CoV-2 viruses, passaged in VeroE6 cells rapidly accumulated mutations around the furin cleavage site of the S protein that impaired its furin cleavage (Fig.   2a,b) . We compared the effect of the blocking antibodies on infection of Caco-2 cells with wildtype and mutated SARS-CoV-2 virus, and found that Caco-2 cells pre-incubated with NRP1 blocking antibody reduced infection with wild-type virus by ~40%, while the control antibody had no effect (Fig. 2c,d) . In contrast, NRP1 blocking antibodies did not reduce the infection of Caco-2 cells with the mutated virus.

Having obtained evidence that NRP1 facilitates SARS-CoV-2 entry, we examined whether NRPs are expressed in infected cells. For these analyses, we used published scRNA-seq datasets of cultured experimentally infected human bronchial epithelial cell (HBECs) and cells isolated from bronchoalveolar lavage fluid (BALF) of severely affected COVID-19 patients 16,17 . Using these datasets, we analyzed which of the proposed SARS-CoV-2 cell entry and amplification factors were correlated with the detection of virus RNA in single cell transcriptomes. Of all the proposed factors only three, NRP1, FURIN and TMPRSS11A, were enriched in SARS-CoV-2 infected compared to non-infected cells (Extended Data Fig. 3 ). In addition, RNA expression of NRP1 and NRP2 was elevated in SARS-CoV-2-positive compared to bystander cells isolated from bronchoalveolar lavage fluids of severely affected COVID-19 patients (Extended Data   Fig. 4) . Severe COVID-19 disproportionately affects patients with diabetes 18 . We analyzed a cryopreserved human diabetic kidney single-nucleus RNA sequencing dataset 19 , and found that among 14 proposed SARS-CoV-2 cell-entry and amplification factors, only NRP1 was significantly upregulated (Extended Data Fig. 5 ). Next, we analyzed the expression pattern of NRP1 by examining the Human Protein Atlas (https://www.proteinatlas.org) 20 , which revealed particularly high levels of NRP1 expression in the epithelial surface layer, outlining the respiratory and gastrointestinal tracts (Extended Data Fig. 6 ). To further resolve expression of SARS-CoV-2 cell-entry receptors, we took advantage of published scRNA-seq datasets of human lung tissue 21 and human olfactory epithelium 22 . In the lung, both NRP1 and NRP2 were abundantly expressed with highest expression in endothelial cells and detectable levels in all pulmonary cells (Extended Data Fig. 7 ). In the adult human olfactory epithelium, we observed abundant expression of both NRP1 and NRP2 in almost all detected cell types of the olfactory epithelium, in sharp contrast to ACE2, which was sparsely expressed in these datasets (Extended Data Fig. 8 ).

In light of the widely reported disturbance of sense of olfaction in a large fraction of COVID-19 patients 23 , and the enrichment of NRPs in the olfactory epithelium, we studied whether SARS-CoV-2 could infect the NRP1-positive cells in the olfactory epithelium in humans. Thus, we analyzed a series of autopsies from six COVID-19 patients and seven non-infected control autopsies for the presence of SARS-CoV-2 infection in the olfactory system (Fig. 3 , Extended Data Table 1 ). Using antibodies against the spike protein, we detected infection in the olfactory epithelium of five out of six COVID-19 patients. Positive signal was particularly well visible in the cells bordering the nasal cavity, whereas no S antigen signal was detectable in the control autopsies (Fig. 3) . The infected olfactory epithelial cells showed high expression of NRP1 (Fig.   3a ). In the underlying tissue, we detected spike protein in NRP1-positive endothelial cells.

Additional co-staining of the transcription factor, OLIG2, and spike protein indicated infection of late olfactory neuronal progenitors and/or newly differentiated olfactory neurons (Fig. 3b) .

Strikingly, within the brain, the olfactory bulb and tracts displayed immunoreactivity for the spike protein especially within NRP1-positive endothelial cells in small capillaries and medium-sized vessels (Fig. 3c) .

These results provide evidence that SARS-CoV-2 infect brain tissue, consistent with its multiorgan involvement 24 , and suggest that viral entry into the brain may occur through the olfactory epithelium. To determine whether NRPs could provide a pathway for virus entry into the brain through the olfactory system, we performed experiments in mice. First, we confirmed the high expression of NRPs in the mouse olfactory epithelium. Virtually all sensory olfactory neurons expressed NRP1 (Extended Data Fig. 9 ). To study whether NRPs could provide a transport pathway into the nervous system, we used the 80 nm diameter AgNPs-CendR, which were validated for their specific interaction with NRPs (Extended Data Fig. 2 ). AgNPs-CendR and control AgNPs lacking the peptide, both of similar size as many viruses, were administered into the nose of anesthetized adult mice (Fig. 4) . Mice were sacrificed 30 min and six hours after intranasal administration and analysis of the olfactory epithelia revealed much larger uptake of AgNP-CendR particles compared to control particles into the epithelia (Fig. 4a-c) . Next, we studied whether AgNP-CendR could be transported to the central nervous system. Strikingly, AgNP-CendR, but not control particles, were readily detected in brain tissue (Fig. 4a,c) .

Whereas AgNP-CendR were found particularly enriched in neurons located in the cortex, they were also found in the olfactory bulb, mainly in neuronal cells (Fig. 4b) , and, to a lesser extent, in endothelial cells (Extended Data Fig. 10 ). These data provide evidence for the existence of a NRP-dependent intranasal brain entry pathway.

There is limited knowledge about the virus-host interactions that determine cellular entry and tissue spreading of SARS-CoV-2. So far, the focus has been entirely on ACE2 as a receptor, but it is clear that viruses display considerable redundancy and flexibility in receptor usage, in particular, because viruses can exploit weak multivalent interactions to enhance affinity 2 . Here, we provide evidence that NRP1 can serve as an entry factor, but more work is required to understand whether the engagement with NRP1 provides attachment, enhances proteolytic cleavage of S, induces receptor-mediated endocytosis, or promotes signaling. The reason why a number of viruses, such as the human T-cell lymphotropic virus type 1 (HTLV-1) 25 , cytomegalovirus (CMV) 26 , Epstein-Barr virus (EBV) 27 and Lujo virus (LUJV) 28 , use NRPs as an entry factor could be due to its localization on epithelia facing the external environment, and enabling cell, vascular and tissue penetration 10 . The functions of NRPs are complex, and it is conceivable that some viruses do not use NRPs as bona fide endocytic receptors, but rather as signaling platforms to facilitate their spread. NRP1 ligation can increase vascular permeability and induce nutrient-sensitive macropinocytosis 11, 29 . In this context, vascular endothelialitis, thrombosis, and angiogenesis that has been observed in COVID-19 patients, together with the upregulation of NRPs in SARS-CoV-2 infected blood vessels, is noteworthy 30 . Mechanistic details of how NRPs promote tissue penetration is under investigation, but the rapid infiltration of CendR peptides across tumor tissue implies active transport from cell to cell 10 . Such transcytotic pathways could be relevant for the transport of viruses from the olfactory sensory nerve endings along the axons into the brain, but paracellular pathways induced by enhanced vascular permeability are also conceivable. Proteolytic cleavage of viral spike proteins exposing CendR sequences are common to many neurotropic viruses (e.g. CMV, HTLV-2, measles, tickborn encephalitis, among others); thus, clarifying whether NRPs are involved in the neuroinvasive potential of furin-activated viruses is an important topic for future research. Lipofectamine solution and the mix was left for 20 minutes at room temperature. The mix was then added to the cells for 6 hours. Cell medium was then changed to normal growth medium (without antibiotics). 24 and 48 hours after transfection, the medium was collected, centrifuged at 1000g for 5 minutes, and filtered through a 0.45 µm filter. The medium was centrifuged at 65000g with a SW28 rotor for 3 hours over a cushion of 10% sucrose prepared in PBS. The pellet was finally resuspended in TBS buffer containing 5% bovine serum albumin overnight at 4°C, aliquoted and frozen at -80°C.

Wild type and triple mutant human NRP1 b1b2 domain (residues 274-584) were expressed in NaCl.

Female BALB/c and C57BL/6 mice, 8-9 weeks old, were immunized intraperitoneally with 17 μg of recombinant NRP1 b1b2 mixed with an equal volume of complete Freund's adjuvant (Sigma-Aldrich Chemie, Steinheim, Germany), followed by a booster immunization four weeks later of the same dose mixed with incomplete Freund's adjuvant (Sigma-Aldrich). Mice received three boosts of the same amount of antigen in PBS on days −3, −2, and −1 prior to fusion. Spleens were excised and the splenocytes were fused with myeloma cells (P3X63Ag8.653) according to a previously described protocol 32 . Beginning on day 10 after fusion, hybridoma supernatants were screened for specific antibodies. Before experiments, the hybridoma supernatants were centrifuged at 300g for 5 min at room temperature and 500 µl dialyzed against 2 L of PBS over night at 4°C prior use.

The silver nanoparticles (AgNPs) were synthesized and functionalized as described previously 33 The particles were washed 3 times by centrifugation at 3,500 × g for 10 min at 4°C, followed by resuspension of the particles in PBST by sonication. Next, biotinylated RPARPAR peptide were coupled to the particles by adding peptide (10 µl, 2 mM in MQ water) to AgNPs (500 µl), followed by incubation at RT for 30 min. The AgNPs were washed, 0.2 µm filtered and stored at 4°C in dark.

The mouse olfactory epithelium, the olfactory bulbs, and the brain were isolated and left in 30% were used for immunohistochemistry. Images were taken using a fluorescence microscope (Olympus BX-63 with a colour camera (Olympus DP80). Image visualization was performed using Omero Server software 5.6, Omero figure 4.0.2 (https://github.com/ome/omero- figure) and InkScape 0.92 34 . The human NRP1 protein immunohistochemistry data on 12 tissue types corresponds to images on the HPA database http://www.proteinatlas.org 20 .

Intranasal delivery was performed as previously described 35 were stained with the specific antibodies. For each tissue a representative area was selected and the area occupied by AgNP particles with a size higher than 1.8 µm 2 was measured. The same threshold value was applied to the images of each area. 

Analyses of the scRNA-seq datasets including filtering, normalization and clustering were conducted using Seurat 3.1 42 . Human lung data from Han et al. 21 , was downloaded from https://figshare.com/articles/HCL_DGE_Data/7235471, in the form of batch-corrected digital gene expression matrices and cell annotation csv files. Cell annotation included cell types, tissue of origin and age. Gene expressions were log-normalized with a scale factor of 10 000 using the NormalizeData function. Next, data was scaled using the ScaleData function and the number of UMI and the percentage of mitochondrial gene content were regressed out as described by the authors 21 Extended Data Fig. 6│ Protein localization of NRP1 in human tissues based on immunohistochemistry. NRP1 immunohistochemistry of 12 human tissues were done using NRP1 antibodies HPA030278 (Atlas Antibodies AB) or CAB004511 (Santa Cruz Biotechnology) (brown), and counterstained with hematoxylin (blue). Particularly high expression of NRP1 was observed in many organs in the epithelial cells facing the external environment. For details, see https://www.proteinatlas.org/ENSG00000099250-NRP1/tissue/primary+data. Fig. 7│ RNA expression of NRP1, NRP2 and ACE2 in adult human lung cells. a, Dot plot visualization of the expression of NRP1, NRP2 and ACE2 in lung cell subtypes. The size of the dots indicates the proportion of cells in the respective cell type having greater-than-zero expression of NRP1 (first column), NRP2 (second column) and ACE2 (third column) while the color indicates the mean expression of aforementioned genes. Average expression is the scaled value. b, UMAPs are showing the lung cell subtypes and expression levels of NRP1, NRP2 and ACE2. Each dot represents a single cell (cell number n = 17,438). The cell cluster identity is noted on the color key legend and labels based on Han et al. 21 . Fig. 8│ RNA expression of NRP1, NRP2 and ACE2 in the adult human olfactory neuroepithelium. a, Dot plot visualization of the expression of NRP1, NRP2 and ACE2 in olfactory neuroepithelia. The size of the dots indicates the proportion of cells in the respective cell type having greater-than-zero expression of NRP1 (first column), NRP2 (second column) and ACE2 (third column) while the color indicates the mean expression of aforementioned genes. Average expression is the scaled value. b, UMAPs depicting the olfactory neuroepithelial cell types (cell number n = 28,622) and expression levels of NRP1, NRP2 and ACE2. The cell cluster identity is noted on the color key legend and labels based on Durante et al. 22 .

Neuropilin regulation of angiogenesis, arteriogenesis, and vascular permeability

Cryo-EM structure of the 2019-nCoV spike in the prefusion conformation

Structure, Function, and Antigenicity of the SARS-CoV-2 Spike Glycoprotein

Characterisation of the transcriptome and proteome of SARS-CoV-2 using direct RNA sequencing and tandem mass spectrometry reveals evidence for a cell passage induced in-frame deletion in the spike glycoprotein that removes the furin-like cleavage site

Single-cell longitudinal analysis of SARS-CoV-2 infection in human bronchial epithelial cells

Single-cell landscape of bronchoalveolar immune cells in patients with COVID-19

Are patients with hypertension and diabetes mellitus at increased risk for COVID-19 infection?

The single-cell transcriptomic landscape of early human diabetic nephropathy

A subcellular map of the human proteome

Construction of a human cell landscape at single-cell level

Single-cell analysis of olfactory neurogenesis and differentiation in adult humans

Neurologic Manifestations of Hospitalized Patients With Coronavirus Disease

Multiorgan and Renal Tropism of SARS-CoV-2

Neuropilin-1 is involved in human T-cell lymphotropic virus type 1 entry

An Unbiased Screen for Human Cytomegalovirus Identifies Neuropilin-2 as a Central Viral Receptor

Neuropilin 1 is an entry factor that promotes EBV infection of nasopharyngeal epithelial cells

NRP2 and CD63 Are Host Factors for Lujo Virus Cell Entry

VEGF-A and the induction of pathological angiogenesis

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

All data generated and/or analysed during this study are either included in this article (and its Supplementary Information) or are available from the corresponding author on reasonable request. Source Data for Figs. 1, 2 and 4 and Extended Data Fig

Functional analysis of potential cleavage sites in the MERS-coronavirus spike protein

Single-cell analysis of the murine chemokines MIP-1alpha, MIP-1beta, RANTES and ATAC/lymphotactin by flow cytometry

Etchable plasmonic nanoparticle probes to image and quantify cellular internalization

The Open Microscopy Environment (OME) Data Model and XML file: open tools for informatics and quantitative analysis in biological imaging

A Positioning Device for the Placement of Mice During Intranasal siRNA Delivery to the Central Nervous System

ImageJ2: ImageJ for the next generation of scientific image data

Detection of 2019 novel coronavirus (2019-nCoV) by real-time RT-PCR

Multiplex PCR method for MinION and Illumina sequencing of Zika and other virus genomes directly from clinical samples

Trimmomatic: a flexible trimmer for Illumina sequence data

The Sequence Alignment/Map format and SAMtools

LoFreq: a sequence-quality aware, ultra-sensitive variant caller for uncovering cell-population heterogeneity from high-throughput sequencing datasets

Comprehensive Integration of Single-Cell Data

FAM is supported by an Australian National Health and Medical Research Council Senior Research Fellowship (GNT1060075). TT and AT are supported by the European Regional Development Fund

V. analyzed the data or supervised data acquisition

Cells were pre-treated with a vehicle control (PBS) or the indicated mAb against NRP1 showing the binding and internalizing of the AgNP-CendR nanoparticles. b, PPC1 cells treated as in (a) were incubated with an etching solution 5 min before fixation to solubilize the non

Representative images and quantification showing the uptake of CF647-labeled AgNP conjugated with biotin-X-RPARPAR (AgNP-CendR, lower row) or biotin alone (control AgNP, upper row) after 30 minutes in HEK-293T cells expressing NRP1. AgNPs (magenta) were visible inside the cells, counterstained with Alexa-Fluor488-phalloidin (yellow) and Hoechst (cyan). b, Representative images and quantification of NRP1-expressing HEK-293T cells treated for 30 minutes with the blocking mAb3 antibody (lower row) or control Ab (upper row). c, Representative images and quantification of NRP1-expressing HEK-293T cells incubated for 30 minutes with AgNPs, together with wild-type b1b2 domain (wt b1b2, lower row) or the mutant b1b2 domain of NRP-1 (mutant b1b2, upper row). n = 3 mice

Schematic representation of the mouse head showing the position of the main olfactory epithelium (MOE), the glomerular (OBglom) and granular (OBgran) layers of the olfactory bulb (OB), and the anterior olfactory nucleus (AON). b, Schematic representation of the MOE, depicting olfactory sensory neurons (OSN) within the epithelium, their apical cilia, and olfactory nerves (olfN) within the basal cell layer (BCL). c, Quantification of NRP1 + neurons in the mouse MOE, OB and AON. n = 3 mice. Data are means ± s.d. One-way ANOVA with Tukey's correction for multiple comparisons. d,e, Confocal images of NRP1 protein (upper panels) and NRP1 RNA (lower panels) in the mouse olfactory system

Extended Data Fig. 10│ The C-end terminal peptide (CendR) mediates the NRP1-dependent uptake to the blood vessels. a, Representative images of main olfactory epithelium (MOE) of animals treated for 6 hours with CF647-labeled AgNPs coated with X-RPARPAR (AgNP-CendR) showing the co-localization between blood vessels (Collagen V, yellow) and the nanoparticles (magenta). b, Representative images (upper row) and magnification (lower row) of the olfactory bulb (OB) from AgNP-CendR treated animals showing the co-localization between AgNP-CendR (magenta) and endothelial cells (CD31, yellow). c, Representative images of the mouse cortex from AgNP-CendR treated animals showing that the majority of the particles were uptaken by the surrounding cells and can no longer be observed in the blood vessels (collagen V, yellow)

The work in Munich and Göttingen was supported by grants from the German Research 

The authors have no competing interests. Table. 1│Patient characteristics of COVID-19 (1 to 6) and control patients (7 to 13). Abbreviations: Afib = atrial fibrillation, aHT = arterial hypertension, AP = acute pancreatitis, BAO = basilar artery occlusion, CHD = chronic heart disease, CKD = chronic kidney disease, COPD = chronic obstructive pulmonary disease, GPA = granulomatosis with polyangiitis, HCC=hepatocellular carcinoma, MI = myocardial infarction, MT = mechanical thrombectomy, Ob = olfactory bulb, OE = olfactory epithelium, s/p = status post, T2DM= type 2 diabetes mellitus.