Neuropilin-1 facilitates SARS-CoV-2 cell entry and infectivity

1An outbreak of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infections has caused a pandemic associated with a severe acute pulmonary disease named COVID-19 (coronavirus disease 2019). A related coronavirus, SARS-CoV, led to a much smaller outbreak in 2003, possibly due to infection occurring predominantly in the lower respiratory system, whereas SARS-CoV-2 spreads rapidly through active pharyngeal viral shedding. Despite these differences, uptake of both viruses is mediated by the same cellular receptor: angiotensin-converting enzyme 2 (ACE2). One hypothesis to explain the enhanced spreading of SARS-CoV-2 is the presence of a polybasic furin-type cleavage site, RRAR^S, at the S1-S2 junction in the SARS-CoV-2 spike (S) protein that is absent in SARS-CoV. Similar sequences are found in the S proteins of many other pathogenic human viruses, including Ebola, HIV-1, and highly virulent strains of avian influenza. The presence of the polybasic cleavage site in SARS-CoV-2 results in enhanced pathogenicity by priming the fusion activity and could potentially create additional cell surface receptor binding sites. Proteolytic cleavage of RRAR^S by furin exposes a conserved C-terminal motif, RXXROH [where R is arginine and X is any amino acid; R can be substituted by lysine (K)], in the S protein. Such C-terminal sequences that conform to the C-end rule (CendR) are known to bind to and activate neuropilin (NRP1 and NRP2) receptors at the cell surface. Recent cryo–electron microscopy structures of the SARS-CoV-2 S protein demonstrated that the S1-S2 junction is part of a solvent-exposed loop and is therefore accessible for receptor interactions.

2To determine whether SARS-CoV-2 can use NRP1 for virus entry and infectivity, we generated lentiviral particles pseudotyped with the SARS-CoV-2 S protein. Pseudoviruses are well suited for virus entry assays, as they allow viral entry to be distinguished from other virus life-cycle steps. Human embryonic kidney 293 T (HEK-293T) cells, which have almost no detectable ACE2 and NRP1 transcripts (fig. S1), were transfected with plasmids that encode the two established host factors, human ACE2 and the transmembrane protease serine 2 (TMPRSS2), or NRP1. When expressed alone, ACE2 rendered cells susceptible to infection (Fig. 1A). Although NRP1 did not promote infection in HEK-293T cells, its coexpression with ACE2 and TMPRSS2 markedly enhanced infection (Fig. 1, A and B). NRP1 expression increased infection in Caco-2 cells, which endogenously express ACE2 (Fig. 1C and fig. S1D), showing that NRP1 can potentiate infection in the presence of other host factors. To test the specificity of NRP1-dependent virus entry, we developed monoclonal antibodies (mAbs) that were designed to functionally block the extracellular b1b2 domain of NRP1, which is known to mediate binding to CendR peptides. The mAb3 was observed to bind to the recombinant b1b2 domain of wild-type (WT) NRP1 but not to the triple-mutant b1b2 domain (S346A, E348A, and T349A in the CendR binding pocket) (fig. S2A). The potency of the mAbs in preventing cellular binding and internalization of NRP ligands was tested using 80-nm silver nanoparticles (AgNP) coated with the prototypic NRP1-binding CendR peptide RPARPAROH (fig. S2B). mAb3 efficiently blocked AgNP-CendR binding (fig. S2C) and internalization (fig. S2, D and E), whereas another monoclonal antibody, mAb2, had no effect and was used as a control in further experiments. Treatment of HEK-293T with mAb3 significantly reduced infection by SARS-CoV-2 pseudoviruses in cells expressing ACE2, TMPRSS2, and NRP1 (Fig. 1D), but not in cells expressing ACE2 and TMPRSS2 only (fig. S2F). When SARS-CoV-2 pseudovirus was preincubated with recombinant, soluble extracellular b1b2 domain of NRP1, the wild type significantly reduced infection compared with the triple mutant (Fig. 1E and fig. S2G).

3Next, we explored the role of NRP1 using SARS-CoV-2 isolated from COVID-19 patients from the Helsinki University Hospital. We used WT SARS-CoV-2 and a cleavage-impaired SARS-CoV-2 mutant that was isolated from Vero-E6 cells, which rapidly accumulate mutations at the furin cleavage site of the S protein during passaging (Fig. 2, A and B). First, we confirmed that furin cleaved the WT, but not the mutant, SARS-CoV-2 S protein by analyzing S protein processing in Chinese hamster ovary cells with functional (parental) or deficient (FD11) furin enzyme (fig. S3). Next, we validated that exogenous ACE2 expression rendered HEK-293T cells susceptible to infection with SARS-CoV-2 (Fig. 2, C and D). NRP1 expression alone caused lower levels of infection, which were detectable only with increasing virus titer (Fig. 2, C and D). We then compared the ability of WT and mutant SARS-CoV-2 to infect HEK-293T that stably express ACE2; ACE2 and TMPRSS2; or ACE2, TMPRSS2, and NRP1. Infection of these cell lines by the WT, but not the mutant, virus increased in the presence of NRP1, providing further evidence that NRP1 requires a furin-cleaved substrate for its effects (Fig. 2, E and F). We studied the effect of the NRP1-blocking antibody, mAb3, on infection of Caco-2 cells by WT and mutant SARS-CoV-2 and found that preincubation with NRP1-blocking antibody reduced WT virus infection by ~40%, whereas the control mAb2 had no effect (Fig. 2, G and H). NRP1-blocking antibody had no effect on the infection by the mutated virus (Fig. 2, G and H).

4Cleavage of SARS-CoV-2 S protein at the S1-S2 site generates the C-terminal end sequence TQTNSPRRAROH. To determine whether this specific sequence can function as a substrate for NRP1, we used AgNPs coated with TQTNSPRRAROH peptide or different control peptides, including one with a terminal amide group (TQTNSPRRARNH2), which reduces NRP1 binding (Fig. 3A). We found that AgNP-TQTNSPRRAROH, but not control AgNPs, were efficiently taken up by HEK-293T cells expressing NRP1 (Fig. 3, B and C). Next, we determined whether AgNP-TQTNSPRRAROH particles were also internalized into cells in vivo. We chose to study nanoparticle entry in the mouse olfactory epithelium, owing to the known expression of NRP1 in the olfactory system, including olfactory neuronal cells of the epithelium (fig. S4). AgNPs-TQTNSPRRAROH and control AgNP-TQTNSPRRARNH2 were administered into the nose of anesthetized adult mice. Six hours after administration, we observed a significantly larger uptake of AgNP-TQTNSPRRAROH than of AgNP-TQTNSPRRARNH2 into the olfactory epithelium (Fig. 3, D and E) and, unexpectedly, into neurons and blood vessels of the cortex (Fig. 3, F and G). Similar results were obtained for AgNPs coated with the prototypic NRP1-binding CendR peptide RPARPAROH (fig. S5).

5Having obtained evidence for a role of NRP1 in cell entry of SARS-CoV-2, we examined whether NRP1 expression correlated with the detection of virus RNA in single-cell transcriptomes. For these analyses, we used published single-cell RNA sequencing (scRNA-seq) datasets of cultured experimentally infected human bronchial epithelial cells and cells isolated from bronchoalveolar lavage fluid (BALF) of severely affected COVID-19 patients. Among the proposed entry and amplification factors, NRP1, FURIN, and TMPRSS11A were enriched in SARS-CoV-2–infected cells compared with noninfected cells (fig. S6). We also detected increased expression of these proteins after infection (fig. S6). In addition, RNA expression of NRP1 and its homolog NRP2 was elevated in SARS-CoV-2–positive cells compared with adjacent cells in the BALF of severely affected COVID-19 patients (fig. S7).

6Because the availability of virus receptors and entry cofactors on the surface of host cells determines infectivity, we compared the expression patterns of ACE2 and NRP1 in published scRNA-seq datasets of human lung tissue and human olfactory epithelium. Whereas ACE2 was detected at very low levels, both NRP1 and NRP2 were abundantly expressed in almost all pulmonary and olfactory cells, with the highest levels of expression in endothelial cells (figs. S8 and S9). We confirmed these results by examining NRP1 immunoreactivity in human autopsy tissue and detected NRP1 in the epithelial surface layer of the human respiratory and olfactory epithelium (fig. S10A). ACE2 was hardly detectable in these tissues (fig. S10B). Within the olfactory epithelium, NRP1 was also observed in cells positive for oligodendrocyte transcription factor 2 (OLIG2), which is mostly expressed by olfactory neuronal progenitors (fig. S10C).

7In light of the widely reported disturbance of olfaction in a large fraction of COVID-19 patients and the enrichment of NRPs in the olfactory epithelium, we analyzed a series of autopsies from six COVID-19 patients and eight noninfected control patients to determine whether SARS-CoV-2 could infect NRP1-positive cells (Fig. 4 and table S1). Using antibodies against the S protein, we detected infection of the olfactory epithelium in five of six COVID-19 patients. The infected olfactory epithelial cells showed high expression of NRP1 (Fig. 4, A and B). Additional costaining indicated infection of cells positive for OLIG2 (Fig. 4B and fig. S11).

8There is limited knowledge about the virus–host interactions that determine cellular entry of SARS-CoV-2. Viruses display considerable redundancy and flexibility because they can exploit weak multivalent interactions to enhance affinity. To date, studies of SARS-CoV-2 entry have focused almost entirely on ACE2, which is expressed at very low protein levels in respiratory and olfactory epithelial cells. This raises the possibility that cofactors are required to facilitate virus–host cell interactions in cells with low ACE2 expression. NRP1 could represent such an ACE2 potentiating factor by promoting the interaction of the virus with ACE2. The reason a number of viruses use NRPs as entry factors may be their high expression on epithelia facing the external environment and their function in enabling cell, vascular, and tissue penetration.