Main

SARS-CoV-2 infects target cells via the spike (S) glycoprotein, which is organized as a homotrimer with each monomer comprising S1 and S2 subunits4. The infection process includes binding to cells, triggering of conformational changes in S and then fusion of the viral envelope with the target cell membrane. The S1 subunit of S comprises the N-terminal domain (NTD) and the receptor binding domain (RBD). The RBD interacts with ACE2 through a region defined as the receptor binding motif (RBM). Antibodies against the RBD contribute the majority of the neutralizing activity of polyclonal serum antibodies5,6, potently neutralize SARS-CoV-2 in vitro7,8 and have shown efficacy in clinical trials for prophylaxis and early therapy of COVID-199,10.

The search for SARS-CoV-2-neutralizing antibodies has been facilitated by the use of target cells overexpressing the ACE2 receptor11. However, ACE2 expression in the lower respiratory tract is limited, with low levels found in only a small number of type II alveolar basal, goblet and mucous cells1,2,3. The paradox of low ACE2 levels in the lung and infection in other tissues leading to extrapulmonary complications12 raises the possibility that additional receptors may contribute to viral infection and dissemination, such as DC-SIGN (also known as CD209), L-SIGN (also known as CD209L or CLEC4M), neuropilin-1 (NRP-1), basigin (also known as CD147) or heparan sulfate13,14,15,16,17. It remains to be established whether these molecules act as alternative primary receptors for viral entry, as co-receptors or as attachment receptors that tether viral particles, enhancing their interaction with ACE2.

In this study, we identify DC-SIGN, L-SIGN and SIGLEC1 (also known as CD169, sialoadhesin or Siglec-1) as attachment receptors that enhance ACE2-dependent infection and demonstrate different mechanisms of neutralization by antibodies targeting RBM and non-RBM sites in the presence or absence of lectins.

Lectins are attachment receptors for SARS-CoV-2

To develop an assay for identification of attachment receptors of SARS-CoV-2 infection, we used HEK 293T cells that express low endogenous levels of ACE2. HEK 293T cells were transfected with vectors encoding ACE2 or one of 13 selected lectins and published receptor candidates before infection with vesicular stomatitis virus (VSV) SARS-CoV-2 pseudovirus. Untransfected HEK 293T cells were only weakly permissive to infection, and ACE2 overexpression led to a marked increase in pseudovirus entry. Increased infectivity was also observed in HEK 293T cells following transfection with C-type lectins DC-SIGN and L-SIGN, which were previously reported to be entry receptors13,14,18, as well as with SIGLEC1, which was not previously shown to mediate SARS-CoV-2 entry (Fig. 1a). NRP-1 and CD147 did not enhance SARS-CoV-2 infection in these conditions, although they have been suggested to act as entry receptors15,16. The infection-enhancing activity of the three lectins was also observed with authentic SARS-CoV-2 on cell lines stably expressing these factors (Fig. 1b, Extended Data Fig. 1a–d). A SIGLEC1-blocking antibody inhibited infection of SIGLEC1-expressing HEK 293T cells, supporting a role of this molecule as a SARS-CoV-2 co-factor (Fig. 1c).

Fig. 1: DC-SIGN, L-SIGN and SIGLEC1 function as attachment receptors for SARS-CoV-2 infection.
figure 1

a, VSV-SARS-CoV-2 infection of HEK 293T cells transfected to express ACE2 or a panel of selected lectins or candidate receptors (n = 4 biologically independent replicates). RLU, relative luminescence units. b, Stable cell lines were infected with SARS-CoV-2-Nluc and luciferase levels were quantified at 24 h (n = 6). c, Inhibition of SARS-CoV-2-Nluc infection with anti-SIGLEC1 monoclonal antibody (clone 7-239) (n = 3). d, Indicated cell lines were transduced to express lectins or ACE2 and infected with VSV-SARS-CoV-2 (n = 8). e, Effect of ACE2 siRNA transfection on infection with VSV-SARS-CoV-2 (n = 8).

The ectopic expression of DC-SIGN, L-SIGN and SIGLEC1 did not support infection of ACE2-negative cells, such as HeLa or MRC-5 cells (Fig. 1d), indicating that these lectins do not act as primary entry receptors. The requirement of ACE2 for viral infection of lectin-expressing cells was also demonstrated using ACE2-blocking antibodies or ACE2 small interfering RNA (siRNA) (Fig. 1e, Extended Data Fig. 1e).

Collectively, these data reveal a lectin-facilitated pathway of infection that is evident in cells expressing low levels of ACE2, supporting the notion that SARS-CoV-2 may use these lectins as attachment receptors to tether viral particles, thereby facilitating interaction with ACE2.

Attachment receptors facilitate trans infection

Interaction with ACE2 could take place in cis or in trans, as reported for HIV-119. To address whether ACE2 and lectins can be found on the same cells (that is, in cis), we interrogated the lung cell atlas20 (Extended Data Fig. 2a). DC-SIGN is expressed most prominently on IGSF21+ dendritic cells, L-SIGN has a limited expression on vascular structures and SIGLEC1 is broadly expressed at the surface of alveolar macrophages, dendritic cells and monocytes. ACE2 expression is limited to subsets of alveolar epithelial type-2, basal and goblet cells. We then mined single-cell transcriptomic data on 3,085 lung epithelial and immune cells obtained from bronchoalveolar lavage (BAL) fluid or sputum of individuals who suffered from severe COVID-1921. The distribution of viral RNA per cell varied across annotated cell types. Specifically, the content of viral RNA in macrophages was greater relative to secretory cells (P < 2.2 × 10−16) (Extended Data Fig. 2b). SIGLEC1 was expressed in 41.4% (459 out of 1,107 cells) of SARS-CoV-2+ macrophages, whereas ACE2 expression was negligible in these cells (Fig. 2a). Conversely, ACE2 expression was found in 10.6% (60 out of 565 cells) of SARS-CoV-2+ secretory cells, whereas SIGLEC1 expression was negligible. In the full dataset (including cells from BAL or sputum without detectable SARS-CoV-2), 1,037 cells were annotated as dendritic cells (DCs), out of which 349 (34.6%) were SIGLEC1+ (34.6%). In total, 19 out of 1,037 DCs (less than 2%) had detectable SARS-CoV-2, of which 47% (9 out of 19) exhibited detectable SIGLEC1 expression. Plotting SIGLEC1, DC-SIGN and L-SIGN expression as a function of SARS-CoV-2 viral load revealed a strong positive correlation for SIGLEC1 in macrophages (Fig. 2a). We confirmed this association in a separate transcriptomic dataset of 1,072 SARS-CoV-2+ BAL cells from individuals with severe COVID-1922. We inspected the available sequenced reads from this dataset to assess the nature of viral RNA in SARS-CoV-2+ BAL cells. Minimal viral replication was occurring in this cell population comprising largely of macrophages and other non-epithelial cell types.

Fig. 2: Expression of attachment receptors in infected tissues and their role in mediating trans infection.
figure 2

a, Left, heat map matrix showing counts of cells with detected transcripts for receptor gene(s) (x-axis) by SARS-CoV-2+ cell type (y-axis) (n = 3,085 cells; total number of each cell type in parentheses). Right, correlation of receptor transcript counts with SARS-CoV-2 RNA counts in macrophages and secretory cells (transcript-positive cells shown as fraction of total number of cells of that type). Two-sided correlation test is based on counts (before log transformation) from Ren et al. 21. NK, natural killer cells. CPM, counts per million transcripts; NA, not applicable. b, HeLa cells (grey) transduced with lectins (red) were incubated with VSV-SARS-CoV-2 (blue), washed and co-cultured with VeroE6-TMPRSS2 target cells. Infection (RLU) was assayed in the presence or absence of target cells (purple) (n = 16 biologically independent replicates). c, Trans infection performed as in b. Viral adsorption was performed in the presence of an anti-SIGLEC1 antibody (n = 10). d, Transmission of replication-competent SARS-CoV-2 by lipopolysaccharide (LPS)-activated DCs to susceptible target cells. Cells were pre-incubated with the indicated monoclonal antibodies, exposed to SARS-CoV-2, washed and incubated with indicated target HEK 293T cells. Trans infection was measured six days later as viral release in the supernatant. GM-CSF, granulocyte-macrophage colony-stimulating factor; hPBMCs, human peripheral blood mononuclear cells. Data are mean ± s.d. from two independent experiments including cells from five donors. Two-sided paired t-test.

The above results suggest limited cooperation of ACE2 and SIGLEC1 in cis, because these receptors are rarely expressed in the same cell, suggesting a role for trans infection from SIGLEC1+ myeloid antigen presenting cells to ACE2+ cells. Indeed, lectin-transduced HeLa cells showed enhanced capacity to promote VSV-SARS-CoV-2 trans infection of susceptible Vero E6-TMPRSS2 target cells (Fig. 2b), and SIGLEC1-mediated trans infection was inhibited by SIGLEC1-blocking antibodies (Fig. 2c, Extended Data Fig. 2c).

Next, we evaluated viral attachment and trans infection in primary myeloid cells using replication-competent SARS-CoV-2. Lectins are expressed mostly on antigen-presenting cells such as macrophages, DCs and monocytes, and their expression can be upregulated by innate inflammatory stimuli such as interferons19. While both macrophages and DCs are able to take up SARS-CoV-2 via SIGLEC123, macrophages mostly release inflammatory cytokines upon viral sensing23,24. Here we demonstrate that SIGLEC1 acts as a key factor in the trans infection of susceptible cells from primary DCs. In particular, we found that primary activated DCs cannot be productively infected but can mediate SARS-CoV-2 infection of target cells expressing ACE2 and TMPRSS2, and that this infection was reduced in the presence of an anti-SIGLEC1 antibody (Fig. 2d). In vivo, the trans infection mediated by SIGLEC1 could be relevant once inflammatory DCs migrate to pulmonary tissues upon SARS-CoV-2 infection and could help to spread infection in the lung and to distant tissues. These results are consistent with a role of lectins in dissemination of SARS-CoV-2.

Overexpression of ACE2 impairs neutralization

To investigate how ACE2 and attachment receptor expression levels influence neutralizing activity, we compared three monoclonal antibodies targeting distinct sites on the S protein: (1) S2E12, targeting the RBM site Ia/class 1 in the RBD7; (2) S309, targeting the conserved N-glycan-containing site IV/class 3 distal from RBM25; and (3) S2X333, targeting site i in the NTD26 (Fig. 3a). These monoclonal antibodies completely neutralize infection of Vero E6 cells with authentic SARS-CoV-2, albeit with different potencies, and their activity was not influenced by the expression of the TMPRSS2 protease (Extended Data Fig. 3a, b). To understand the influence of receptor expression on neutralization, we used cell lines expressing ACE2 and TMPRSS2 (endogenously or upon transduction) at levels varying more than 1,000-fold (Fig. 3b, Extended Data Fig. 3c, d). Whereas the RBM monoclonal antibody S2E12 showed comparable neutralizing activity on all target cells, both S309 and S2X333 showed impaired neutralizing activity when tested on cells overexpressing ACE2, in terms of both maximal neutralization and potency (Fig. 3c, d). Similar results were obtained with both VSV-SARS-CoV-2 and authentic SARS-CoV-2-Nluc, a nanoluciferase-expressing infectious SARS-CoV-2 clone. Overall, a negative correlation was found between ACE2 levels and neutralization potency for non-RBM monoclonal antibodies (Extended Data Fig. 3e).

Fig. 3: ACE2 overexpression influences neutralization by different classes of monoclonal antibodies.
figure 3

a, Surface rendering of a composite model of SARS-CoV-2 S bound to S309 (purple), S2E12 (magenta) and S2X333 (orange)7,25,26. The three SARS-CoV-2 S protomers are coloured light blue, gold and pink with N-linked glycans rendered dark blue. b, SARS-CoV-2 S or RBD binding to the indicated cell lines was quantified by flow cytometry. A, ACE2; T, TMPRSS2. Graph shows mean of two biological replicates. MFI, mean fluorescence intensity. c, d, A panel of seven cell lines was infected with SARS-CoV-2-Nluc (c) or VSV-SARS-CoV-2 pseudovirus (d) in the presence of S309, S2E12 or S2X333. Luciferase signals were quantified 24 h after infection (n = 3 biologically independent replicates).

Given this uncertainty in the most relevant in vitro correlates of protection, we investigated the capacity of hamsterized S309 and S2E12 monoclonal antibodies to prevent SARS-CoV-2 infection in Syrian hamsters, an animal model that relies on endogenous expression of ACE227. In a prophylactic setting, S309 was highly effective at doses as low as 0.4 mg kg−1 in terms of reduction of viral RNA and infectious virus levels and histopathological score in the lungs (Extended Data Fig. 4a). Furthermore, we did not observe substantial increased efficacy by co-administering S309 with an equal amount of the potent RBM S2E12 monoclonal antibody (Extended Data Fig. 4b). An ‘Fc-silenced’ version of hamsterized S309 monoclonal antibody (GH-S309-N297A) (Extended Data Fig. 5) was similarly protective against SARS-CoV-2 challenge of hamsters, underscoring that the neutralizing activity of S309 was the primary mechanism of action in this condition.

Together, these data indicate that neutralization assays using cells overexpressing ACE2 underestimate the neutralizing activity of non-RBM monoclonal antibodies, which are similarly protective in a relevant animal model of infection to RBM monoclonal antibodies28. The importance of this finding is also supported by the efficacy data of VIR-7831 (a derivative monoclonal antibody of S309) in a phase 3 clinical trial demonstrating 85% protection against hospitalization and death due to COVID-1929.

Antibody-mediated membrane fusion

Infection of permissive cells involves both interactions with ACE2 and attachment receptors as well as fusion of the viral membrane to cellular membranes. We investigated how different classes of S-specific antibodies may interfere with viral fusion events that are involved in viral entry, but also in cell-to-cell fusion, leading to the formation of syncytia in vitro30 and multi nucleate giant cells in human lung from infected individuals31. RBM-specific SARS-CoV neutralizing monoclonal antibodies can act as ACE2 mimics, triggering the fusogenic rearrangmement of the S protein32. We evaluated monoclonal antibodies of different epitope specificity (Extended Data Table 1) to induce fusogenic rearrangement of soluble S trimers as measured by negative-stain electron microscopy imaging using Fab fragments of the respective monoclonal antibodies (Extended Data Fig. 6a). Five RBM monoclonal antibodies triggered rearrangement to the post-fusion state of a native SARS-CoV-2 S ectodomain trimer, probably owing to conformational selection of open RBDs. Most of these monoclonal antibodies triggered a rapid rearrangement of S, whereas S2D106 did so more slowly. As expected, S2M11, a RBM monoclonal antibody that locks neighbouring RBDs in a closed state7, did not induce fusogenic S rearrangements. Antibodies to NTD (S2X333), to site Ib on RBD (REGN10987 and LyCoV555) or to the N-glycan-containing site at the base of RBD (S309) also did not trigger rearrangement, owing to the absence of conformational selection for open RBDs.

To investigate whether the antibody-mediated triggering of fusogenic rearrangement could promote membrane fusion, we evaluated a panel of monoclonal antibodies for their capacity to induce cell–cell fusion of CHO cells (which lack ACE2 expression) stably transduced with full-length SARS-CoV-2 S. Syncytia formation was triggered by all monoclonal antibodies recognizing antigenic sites Ia and IIa (Extended Data Table 1), which are accessible only in the open RBD state, with half-maximum effective concentration (EC50) values ranging from 20 ng ml−1 for S2E12 to more than 1 µg ml−1 for S2D106 (Extended Data Fig. 6b–d). Syncytia were also formed by the three clinical-stage monoclonal antibodies: REGN10933 (casirivimab), Ly-CoV016 (etesevimab) and CT-P59 (regdanvimab). By contrast, syncytia were not formed in the presence of monoclonal antibodies binding to the open and closed RBD states (S2M11, S309, Ly-CoV555 (bamlanivimab) and REGN10987 (imdevimab)), to the NTD (S2X333) or to a conserved site in the S2 subunit stem helix (S2P6)33. A notable exception is provided by S2X5834, a monoclonal antibody that was structurally defined in this study as binding to site Ib, which is accessible on open and closed RBDs (Extended Data Fig. 7). Of note, syncytia were also formed when using S2E12 Fab, indicating that cell–cell fusion does not result solely from cross-linking of S expressed on opposing cells (Extended Data Fig. 6g). Regarding the possible interaction between fusogenic and non-fusogenic antibodies, we found that syncytia formation induced by S2E12 could be inhibited by different classes of antibodies comprising S2M11 (which locks RBDs in a closed state), S309 (which targets an N-glycan-containing site at the base of RBD) and S2P6 (which destabilizes the stem helix in S2) (Extended Data Fig. 6e). These results highlight that different combinations of antibodies may interfere with each other by promoting or inhibiting membrane fusion.

To address whether antibodies may promote cell-to-cell spread of the infection, we co-cultured S-positive CHO cells with S-negative fluorescently labelled CHO cells. In these conditions, S2E12 promoted unidirectional fusion of S-positive CHO cells with S-negative CHO cells in the absence of ACE2 (defined here as ‘trans fusion’) (Extended Data Fig. 6f). To address whether this mechanism may also mediate ACE2-independent infection of tethered virus, we infected HeLa-DC-SIGN cells with live SARS-CoV-2-Nluc virus in the presence of fusion-enhancing monoclonal antibodies at different dilutions. In these conditions, S2E12, S2D106 and S2X58 did not promote infection (Extended Data Fig. 8a). Collectively, these findings indicate that in certain conditions of antibody concentration and cell–cell proximity, a subclass of RBM antibodies selective for the open conformation of RBD may promote cell–cell fusion with ACE2-negative cells. However, the fusogenic activity of these monoclonal antibodies may not be sufficient to promote entry of virions tethered to the cell surface in the absence of ACE2. It remains to be established whether RBM monoclonal antibodies may mediate ACE2-independent SARS-CoV-2 entry under other conditions, as previously observed for anti-MERS-CoV neutralizing monoclonal antibodies captured by FcγRIIa-expressing cells in vitro35.

Lectin receptors modulate neutralization

Given the dual function of certain RBM antibodies in inhibiting ACE2 binding and triggering fusion and the dependence on attachment receptor expression of neutralization by specific antibodies, we compared the neutralizing activity of a panel of monoclonal antibodies using authentic SARS-CoV-2 and target cells expressing different levels of ACE2 and lectin receptors. When tested on cells overexpressing ACE2, all anti-RBM monoclonal antibodies potently neutralized infection, whereas the non-RBM monoclonal antibodies S309 and S2X333 did not (Figs. 3, 4a, d). However, when tested on cells expressing low levels of ACE2 together with SIGLEC1, DC-SIGN or L-SIGN, S309 and S2X333 showed enhanced neutralizing activity, with S309 reaching 100% of neutralization. Of note, while all RBM monoclonal antibodies retained neutralizing activity on SIGLEC1+ cells, several RBM monoclonal antibodies (S2D106, S2X58, REGN10987, REGN10933 and LyCoV555) lost neutralizing activity on cells expressing DC-SIGN or L-SIGN, showing only partial neutralization at the highest concentrations tested (Fig. 4b, c, e, f, Extended Data Fig. 8c–e). The loss of neutralizing activity of S2X58 and S2D106 monoclonal antibodies observed on DC-SIGN- and L-SIGN-expressing cells was confirmed with both replication-competent SARS-CoV-2 (wild type), as well as with live SARS-CoV-2-Nluc (Extended Data Fig. 8d). However, all neutralizing monoclonal antibodies blocked trans infection of Vero-E6-TMPRSS2 target cells from HeLa cells expressing either DC-SIGN or SIGLEC1 (Extended Data Fig. 9). Together, these data delineate a complex pattern of neutralization of cis or trans viral infection by different classes of monoclonal antibodies whereby the epitope specificity, valency of binding and the ability to trigger fusogenic rearrangement can result in differential blocking efficiency.

Fig. 4: SIGLEC1, DC-SIGN and L-SIGN modulate neutralization by different classes of antibodies.
figure 4

af, Neutralization of infection with SARS-CoV-2-Nluc pre-incubated with indicated monoclonal antibodies on HEK 293T cell lines stably overexpressing lectins or ACE2. Infection was measured by luciferase signal 24 h after infection (n = 3 biologically independent replicates).

Discussion

We have shown that transmembrane lectins act as attachment receptors rather than entry receptors for SARS-CoV-213,14, thus facilitating infection via the canonical ACE2 pathway. This finding addresses the efficiency of lower respiratory tract infection despite the paradoxically low level of ACE2 expression, even in the presence of interferon36,37. The attachment role of lectins in SARS-CoV-2 infection is in line with the known biology of these adhesion molecules, which bind N-glycans characteristic of cellular membranes and pathogen surfaces to promote trans infection38. SIGLEC1 is of particular relevance because it is prominent in lung myeloid cells in association with viral RNA, thus supporting a model of trans infection, tissue dissemination and the triggering of immune responses by myeloid cells, rather than these cells being a direct target for productive infection39. Animal models also support a role of attachment receptors in viral pathogenesis40,41.

Expression of lectin receptors influences the neutralizing activity of different classes of S-specific monoclonal antibodies. In addition, we have observed that various monoclonal antibodies have the ability to interfere with fusion events. We have expanded our initial observation on SARS-CoV and MERS-CoV32,35 by showing that most RBM monoclonal antibodies can trigger the fusogenic rearrangement of S, albeit with varying efficiency. By stabilizing the RBDs in the open conformations, these antibodies might act as receptor mimics. This finding suggests that premature conformational triggering resulting in loss of the potential of a S protein to engender productive infection—we term this mechanism spike inactivation—may be the prominent mode of viral neutralization for this class of antibodies. However, we have also shown that these antibodies can promote fusion of S-expressing cells with neighbouring cells, even if the neighbouring cells lack ACE2. These data are consistent with a recent study reporting that a subset of RBM monoclonal antibodies can enhance S-mediated membrane fusion and formation of syncytia42. Notably, the formation of syncytia has been observed in authoptic samples from severe cases of COVID-1931,43,44. It is tempting to speculate that fusogenic antibodies, although highly effective9,10, may contribute at a later stage to the spread of infection and inflammation.

Overall, our study highlights the finding that ranking of SARS-CoV-2-neutralizing antibodies is highly dependent on the level of ACE2 expression and on the presence of attachment receptors and identifies a mechanism that could result in the creation of multinucleate viral factories, potentially enhanced by specific antibodies.

Methods

Ethics statement

The institutional review board on biomedical research of the Hospital Germans Trias i Pujol (HUGTiP) approved this study. The biologic biosafety committee of the Research Institute Germans Trias i Pujol approved the execution of SARS-CoV-2 experiments at the BSL3 laboratory of the Center of Bioimaging and comparative imaging (CMCIB).

Generation of stable overexpression cell lines

Lentiviruses were generated by co-transfection of Lenti-X 293T cells (Takara) with lentiviral expression plasmids encoding DC-SIGN (CD209), L-SIGN (CLEC4M), SIGLEC1, TMPRSS2 or ACE2 (all obtained from Genecopoeia) and the respective lentiviral helper plasmids. Forty-eight hours after transfection, lentivirus in the supernatant was collected and concentrated by ultracentrifugation for 2 h at 20,000 rpm. Lenti-X 293T (Takara), Vero E6 (ATCC), MRC5 (Sigma-Aldrich), A549 (ATCC) or HeLa (ATCC) cells were transduced in the presence of 6 µg ml−1 polybrene (Millipore) for 24 h. Cell lines overexpressing two transgenes were transduced subsequently. Selection with puromycin and/or blasticidin (Gibco) was started two days after transduction and selection reagent was kept in the growth medium for all subsequent culturing. Single-cell clones were derived from the A549-ACE2-TMPRSS2 cell line, all other cell lines represent cell pools.

SARS-CoV-2 neutralization

Cells cultured in DMEM supplemented with 10% FBS (VWR) and 1× Penicillin/Streptomycin (Thermo Fisher Scientific) were seeded in black 96-well plates at 20,000 cells per well. Serial 1:4 dilutions of the monoclonal antibodies were incubated with 200 pfu of SARS-CoV-2 (isolate USA-WA1/2020, passage 3, passaged in Vero E6 cells) for 30 min at 37 °C in a BSL-3 facility. Cell supernatant was removed and the virus–antibody mixture was added to the cells. Twenty-four hours after infection, cells were fixed with 4% paraformaldehyde for 30 min, followed by two PBS (pH 7.4) washes and permeabilization with 0.25% Triton X-100 in PBS for 30 min. After blocking in 5% milk powder/PBS for 30 min, cells were incubated with a primary antibody targeting SARS-CoV-2 nucleocapsid protein (Sino Biological, catalogue (cat.) no. 40143-R001) at a 1:2,000 dilution for 1 h. After washing and incubation with a secondary Alexa Fluor 647-labelled antibody mixed with 1 µg ml−1 Hoechst33342 for 1 h, plates were imaged on an automated cell-imaging reader (Cytation 5, Biotek) and nucleocapsid-positive cells were counted using the manufacturer’s supplied software (Gen5 v3.08).

SARS-CoV-2-Nluc neutralization

Neutralization was determined using SARS-CoV-2-Nluc, an infectious clone of SARS-CoV-2 (based on strain 2019-nCoV/USA_WA1/2020) encoding nanoluciferase in place of the viral ORF7, which demonstrates comparable growth kinetics to wild type virus45. Cells were seeded into black-walled, clear-bottom 96-well plates at 20,000 cells per well (293T cells were seeded into poly-l-lysine-coated wells at 35,000 cells per well) and cultured overnight at 37 °C. The next day, 9-point fourfold serial dilutions of antibodies were prepared in infection medium (DMEM + 10% FBS). SARS-CoV-2-Nluc was diluted in infection medium at the indicated multiplicity of infection (MOI), added to the antibody dilutions and incubated for 30 min at 37 °C. Medium was removed from the cells, monoclonal antibody–virus complexes were added, and cells were incubated at 37 °C for 24 h. Medium was removed from the cells, Nano-Glo luciferase substrate (Promega) was added according to the manufacturer’s recommendations, incubated for 10 min at room temperature and luciferase signal was quantified on a VICTOR Nivo plate reader using Nivo v3.0.2 software (Perkin Elmer).

SARS-CoV-2 pseudotyped VSV production and neutralization

To generate SARS-CoV-2 pseudotyped VSV, Lenti-X 293T cells (Takara) were seeded in 10-cm dishes for 80% next day confluency. The next day, cells were transfected with a plasmid encoding for SARS-CoV-2 S-glycoprotein (YP_009724390.1) harbouring a C-terminal 19-amino acid truncation using TransIT-Lenti (Mirus Bio) according to the manufacturer’s instructions. One day after transfection, cells were infected with VSV(G*ΔG-luciferase) (Kerafast) at an MOI of 3 infectious units per cell. Viral inoculum was washed off after 1 h and cells were incubated for another day at 37 °C. The cell supernatant containing SARS-CoV-2 pseudotyped VSV was collected at day 2 after transfection, centrifuged at 1,000g for 5 min to remove cellular debris, aliquoted and frozen at −80 °C.

For viral neutralization, cells were seeded into black-walled, clear-bottom 96-well plates at 20,000 cells per well (293T cells were seeded into poly-l-lysine-coated wells at 35,000 cells per well) and cultured overnight at 37 °C. The next day, 9-point fourfold serial dilutions of antibodies were prepared in medium. SARS-CoV-2 pseudotyped VSV was diluted 1:30 in media in the presence of 100 ng ml−1 anti-VSV-G antibody (clone 8G5F11, Absolute Antibody) and added 1:1 to each antibody dilution. Virus:antibody mixtures were incubated for 1 h at 37 °C. Media was removed from the cells and 50 μl of virus:antibody mixtures were added to the cells. One hour post-infection, 100 μl of medium was added to all wells and incubated for 17–20 h at 37 °C. Medium was removed and 50 μl of Bio-Glo reagent (Promega) was added to each well. The plate was shaken on a plate shaker at 300 rpm at room temperature for 15 min and RLUs were read on an EnSight plate reader using Kaleido v3.0 software (Perkin-Elmer).

Transfection-based attachment receptor screen

Lenti-X 293T cells (Takara) were transfected with plasmids encoding the following receptor candidates (all purchased from Genecopoeia): ACE2 (NM_021804), DC-SIGN (NM_021155), L-SIGN (BC110614), LGALS3 (NM_002306), SIGLEC1 (NM_023068), SIGLEC3 (XM_057602), SIGLEC9 (BC035365), SIGLEC10 (NM_033130), MGL (NM_182906), MINCLE (NM_014358), CD147 (NM_198589), ASGR1 (NM_001671.4), ASGR2 (NM_080913), NRP1 (NM_003873). One day after transfection, cells were infected with SARS-CoV-2 pseudotyped VSV at 1:20 dilution in the presence of 100 ng ml−1 anti-VSV-G antibody (clone 8G5F11, Absolute Antibody) at 37 °C. One hour after infection, 100 μl of medium was added to all wells and incubated for 17–20 h at 37 °C. Medium was removed and 50 μl of Bio-Glo reagent (Promega) was added to each well. The plate was shaken on a plate shaker at 300 rpm at room temperature for 15 min and readings in RLU were taken on an EnSight plate reader (Perkin-Elmer).

siRNA-mediated knockdown of ACE2

For reverse transfection of 293T or HuH7 cells, siRNA pools specific for ACE2 (Dharmacon ON-TARGETplus SMARTpool, L-005755-00-0005, pool of 4 individual siRNAs with target sequences: GACAAGAGCAAACGGUUGA, GCGAGUGGCUAAUUUGAAA, GCCAUUAUAUGAAGAGUAU, GGACAAGUUUAACCACGAA) or non-targeting control pool (Dharmacon ON-TARGETplus Non-targeting Pool, D-001810-10-05) were preincubated at 20 nM with Lipofectamine RNAiMAX (Thermo Fisher Scientific) in black-walled clear-bottom 96-well plates according to the manufacturer’s instructions. Fifteen-thousand 293T cells or 10,000 HuH7 cells were seeded on top and incubated at 37 °C. After four days, cells were infected with SARS-CoV-2 pseudotyped VSV as described above.

Trans infection using Hela cells

Parental HeLa cells or HeLa cells stably expressing DC-SIGN, L-SIGN or SIGLEC1 were seeded at 5,000 cells per well in black-walled clear-bottom 96-well plates. One day later, cells reached about 50% confluency and were inoculated with SARS-CoV-2 pseudotyped VSV at 1:10 dilution in the presence of 100 ng/mL anti-VSV-G antibody (clone 8G5F11, Absolute Antibody) at 37 °C for 2 h. For antibody-mediated inhibition of trans infection, cells were pre-incubated with 10 µg ml−1 anti-SIGLEC1 antibody (Biolegend, clone 7-239) for 30 min. After 2 h inoculation, cells were washed 4 times with complete medium and 10,000 VeroE6-TMPRSS2 cells per well were added and incubated for 17–20 h at 37 °C for trans infection. Medium was removed and 50 μl of Bio-Glo reagent (Promega) was added to each well. The plate was shaken on a plate shaker at 300 rpm at room temperature for 15 min and RLUs were read on an EnSight plate reader using Kaleido v3.0 software (Perkin-Elmer).

Cell–cell fusion of CHO-S cells

CHO cells stably expressing SARS-CoV-2 S-glycoprotein were seeded in 96-well plates for microscopy (Thermo Fisher Scientific) at 12,500 cells per well and the following day, different concentrations of monoclonal antibodies and nuclei marker Hoechst (final dilution 1:1,000) were added to the cells and incubated for additional 24 h. Fusion degree was established using the Cytation 5 Imager (BioTek) and an object detection protocol was used to detect nuclei as objects and measure their size. The nuclei of fused cells (that is, syncytia) are found aggregated at the centre of the syncytia and are recognized as a unique large object that is gated according to its size. The area of the objects in fused cells divided by the total area of all the object multiplied by 100 provides the percentage of fused cells

Negative-stain electron microscopy imaging the fusogenic rearrangement of soluble S trimers

SARS-CoV-2 S ectodomain trimer was engineered as follow and recombinantly expressed. The SARS-CoV-2 S(D614G) has a mu-phosphatase signal peptide beginning at Q14, a mutated S1/S2 cleavage site (SGAR), ends at residue K1211 and is followed by a TEV cleavage, fold-on trimerization motif, and an 8× His tag in the pCMV vector. Ten micromolar S was incubated with 13 µM Fab protein for 1 or 48 h at room temperature. Samples were diluted to 0.01 mg ml−1 immediately before protein was adsorbed to glow-discharged carbon-coated copper grids for ~30 s before 2% uranyl formate staining. Micrographs were recorded using the Leginon software46 on a 100kV FEI Tecnai G2 Spirit with a Gatan Ultrascan 4000 4k × 4k CCD camera at 67,000 nominal magnification. The defocus ranged from 1.0 to 2.0 μm and the pixel size was 1.6 Å.

Cryo-electron microscopy

SARS-CoV-2 HexaPro S47 at 1.2 mg ml−1 was incubated with 1.2 fold molar excess of recombinantly purified S2X58 for 10 s at room temperature before application onto a freshly glow discharged 2.0/2.0 UltrAuFoil grid (200 mesh). Plunge freezing used a vitrobot MarkIV (Thermo Fisher Scientific) using a blot force of 0 and 6.5 s blot time at 100% humidity and 23 °C.

Data were acquired using an FEI Titan Krios transmission electron microscope operated at 300 kV and equipped with a Gatan K2 Summit direct detector and Gatan Quantum GIF energy filter, operated in zero-loss mode with a slit width of 20 eV. Automated data collection was carried out using Leginon46 at a nominal magnification of 130,000× with a pixel size of 0.525 Å and stage tilt angles up to 35°. The dose rate was adjusted to 8 counts per pixel s−1, and each movie was acquired in super-resolution mode fractionated in 50 frames of 200 ms. 4,126 micrographs were collected with a defocus range between −0.5 and −3.0 μm. Movie frame alignment, estimation of the microscope contrast-transfer function parameters, particle picking, and extraction were carried out using Warp48. Particle images were extracted with a box size of 800 binned to 400 pixels2, yielding a pixel size of 1.05 Å.

Two rounds of reference-free 2D classification were performed using CryoSPARC49 to select well-defined particle images. These selected particles were subjected to two rounds of 3D classification with 50 iterations each (angular sampling 7.5° for 25 iterations and 1.8° with local search for 25 iterations), using our previously reported closed SARS-CoV-2 S structure as initial model (PDB 6VXX) in Relion50. 3D refinements were carried out using non-uniform refinement along with per-particle defocus refinement in CryoSPARC51. Selected particle images were subjected to the Bayesian polishing procedure implemented in Relion3.052 before performing another round of non-uniform refinement in CryoSPARC followed by per-particle defocus refinement and again non-uniform refinement. Local resolution estimation, filtering, and sharpening were carried out using CryoSPARC. Reported resolutions are based on the gold-standard Fourier shell correlation (FSC) of 0.143 criterion and Fourier shell correlation curves were corrected for the effects of soft masking by high-resolution noise substitution. UCSF ChimeraX53 and Coot54 were used to fit atomic models into the cryo-EM maps.

Immunofluorescence analysis

HEK 293T-derived cell lines were seeded onto poly-d-lysine-coated 96-well plates (Sigma-Aldrich) and fixed 24 h after seeding with 4% paraformaldehyde for 30 min, followed by 2 PBS (pH 7.4) washes and permeabilization with 0.25% Triton X-100 in PBS for 30 min. Cells were incubated with primary antibodies anti-DC-SIGN/L-SIGN (Biolegend, cat. no. 845002, 1:500 dilution), anti-DC-SIGN (Cell Signaling, cat. no. 13193S, 1:500 dilution), anti-SIGLEC1 (Biolegend, cat. no. 346002, 1:500 dilution) or anti-ACE2 (R&D Systems, cat. no. AF933, 1:200 dilution) diluted in 3% milk powder/PBS for 2 h at room temperature. After washing and incubation with a secondary Alexa Fluor 647-labeled antibody mixed with 1 µg ml−1 Hoechst33342 for 1 h, plates were imaged on an inverted fluorescence microscope (Echo Revolve).

ACE2 and TMPRSS2 RT–qPCR

RNA was extracted from the cells using the NucleoSpin RNA Plus kit (Macherey-Nagel) according to the manufacturer’s protocol. Human airway epithelial (HAE) cells were provided by MatTek Life Sciences (MatTek EpiAirway). RNA was reverse transcribed using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems) according to the manufacturer’s instructions. Intracellular levels of ACE2 (forward primer: CAAGAGCAAACGGTTGAACAC, reverse primer: CCAGAGCCTCTCATTGTAGTCT), HPRT (forward primer: CCTGGCGTCGTGATTAGTG, reverse primer: ACACCCTTTCCAAATCCTCAG) and TMPRSS2 (forward primer: CAAGTGCTCCRACTCTGGGAT, reverse primer: AACACACCGRTTCTCGTCCTC) were quantified using the Luna Universal qPCR Master Mix (New England Biolabs) according to the manufacturer’s protocol. Levels of ACE2 and TMPRSS2 were normalized to HPRT. Hela cells were used as the reference sample. All quantitative PCRs were run on a QuantStudio 3 Real-Time PCR System (Applied Biosystems).

SARS-CoV-2 S(D614G) production and biotinylation

Prefusion-stabilized SARS-CoV-2 S(D614G) (comprising amino acids Q14 to K1211) with a C-terminal TEV cleavage site, T4 bacteriophage fibritin foldon, 8× His, Avi and EPEA tags was transfected into HEK 293 Freestyle cells, using 293 fectin as a transfection reagent. Cells were left to produce protein for 3 days at 37 °C. Afterwards, supernatant was collected by centrifuging cells for 30 min at 500g, followed by another spin for 30 min at 4,000g. Cell culture supernatant was filtered through a 0.2-µm filter and loaded onto a 5-ml C-tag affinity matrix column, pre-equilibrated with 50 mM Tris pH 8 and 200 mM NaCl. SARS-CoV-2 D614G S was eluted, using 10 column volumes of 100 mM Tris, 200 mM NaCl and 3.8 mM SEPEA peptide. Elution peak was concentrated and injected on a Superose 6 increase 10/300 GL gel filtration column, using 50 mM Tris pH 8 and 200 mM NaCl as a running buffer. Size-exclusion fractions corresponding to monodisperse SARS-CoV-2 S(D614G) were collected and flash frozen in liquid nitrogen for storage at −80 °C. Purified SARS-CoV-2 S(D614G) protein was biotinylated using BirA500 biotinylation kit from Avidity. To 50 µg of S protein, 5 µg of BirA, and 11 µl of BiomixA and BiomixB was added. Final S protein concentration during the biotinylation reaction was ~1 µM. The reaction was left to proceed for 16 h at 4 °C. Then, protein was desalted using two Zeba spin columns pre-equilibrated with 1× PBS pH 7.4.

Flow cytometry analysis for DC-SIGN, L-SIGN, SIGLEC1 and ACE-2

HEK 293T cells expressing DC-SIGN, L-SIGN, SIGLEC1 or ACE2 were resuspended at 4 × 106 cells per ml and 100 μl per well were seeded onto V-bottom 96-well plates (Corning, 3894). The plate was centrifuged at 2,000 rpm for 5 min and washed with PBS (pH 7.4). The cells were resuspended in 200 μl of PBS containing Ghost violet 510 viability dye (Cell Signaling, cat. no. 13-0870-T100, 1:1,000 dilution), incubated for 15 min on ice and then washed. The cells were resuspended in 100 μl of fluorescence-activated cell sorting (FACS) buffer prepared with 0.5% BSA (Sigma-Aldrich) in PBS containing the primary antibodies at a 1:100 dilution: mouse anti-DC/L-SIGN (Biolegend, cat. no. 845002), rabbit anti-DC-SIGN (Cell Signaling, cat. no. 13193), mouse anti-SIGLEC1 (Biologend, cat. no. 346002) or goat anti-ACE2 (R&D Systems, cat. no. AF933). After 1 h incubation on ice, the cells were washed two times and resuspended in FACS buffer containing the Alexa Fluor 488-labelled secondary antibodies at a 1:200 dilution: goat anti-mouse (Invitrogen cat. no. A11001), goat anti-rabbit (Invitrogen cat. no. A11008) or donkey anti-goat (Invitrogen cat. no. A11055). After incubation for 45 min on ice, the cells were washed 3 times with 200μl of FACS buffer and fixed with 200 μl of 4% PFA (Alfa Aesar) for 15 min at room temperature. Cells were washed 3 times, resuspended in 200 μl of FACS buffer and analysed by flow cytometry using the CytoFLEX flow cytometer (Beckman Coulter).

Flow cytometry of SARS-CoV-2 S and RBD binding to cells

Biotinylated SARS-CoV-2 S(D614G) protein (Spikebiotin, generated in-house) or the biotinylated SARS-CoV-2 S receptor-binding domain (RBDbiotin, Sino Biological, 40592-V08B) were incubated with Alexa Fluor 647 streptavidin (AF647-strep, Invitrogen, S21374) at a 1:20 ratio by volume for 20 min at room temperature. The labelled proteins were then stored at 4 °C until further use. Cells were dissociated with TrypLE Express (Gibco, 12605-010) and 105 cells were transferred to each well of a 96-well V bottom plate (Corning, 3894). Cells were washed twice in flow cytometry buffer (2% FBS in PBS (without Ca and Mg)) and stained with Spikebiotin-AF647-strep at a final concentration of 20 µg ml−1 or RBDbiotin-AF647-strep at a final concentration of 7.5 µg ml−1 for 1 h on ice. Stained cells were washed twice with flow cytometry buffer, resuspended in 1% PFA (Electron Microscopy Sciences, 15714-S) and analysed with the Cytoflex LX (Beckman Coulter).

Recombinant expression of SARS-CoV-2-specific monoclonal antibodies

Human monoclonal antibodies were isolated from plasma cells or memory B cells of SARS-CoV-2-immune donors, as previously described25,55,56. Recombinant antibodies were expressed in ExpiCHO cells at 37 °C and 8% CO2. Cells were transfected using ExpiFectamine. Transfected cells were supplemented one day after transfection with ExpiCHO Feed and ExpiFectamine CHO Enhancer. Cell culture supernatant was collected 8 days after transfection and filtered through a 0.2-µm filter. Recombinant antibodies were affinity purified on an ÄKTAxpress FPLC device using 5 ml HiTrap MabSelect PrismA columns followed by buffer exchange to Histidine buffer (20 mM histidine, 8% sucrose, pH 6) using HiPrep 26/10 desalting columns.

SARS-CoV-2 trans-infection assay on primary cells

Cell lines used have been described57. Isolation and culture of primary cells was performed as described58. In brief, peripheral blood mononuclear cells were obtained with a Ficoll–Hypaque gradient (Alere Technologies) from blood donors and monocyte populations (>90% CD14+) were isolated with CD14 negative selection magnetic beads (Miltenyi Biotec). Macrophages were obtained by culturing these cells in the presence of 100 ng ml−1 of macrophage colony-stimulating factor (M-CSF) for 7 days and replacing media and cytokines every 2 days. DCs were obtained culturing these cells in the presence of both 1,000 IU ml−1 granulocyte-macrophage colony-stimulating factor (GM-CSF) and interleukin-4 (R&D Systems) for 7 days and replacing medium and cytokines every 2 days. Activated cells were differentiated by culturing myeloid cells at day 5 for 2 more days in the presence of 1,000 IU ml−1 of interferon-α (IFNα) (Sigma-Aldrich) or 100 ng ml−1 LPS (Sigma-Aldrich). The SARS-CoV-2 viral strain used on primary cells was isolated in March 2020 from a nasopharyngeal swab as described57. The virus was propagated for two passages and a virus stock was prepared by collecting the supernatant from Vero E6 cells. The genomic sequence was deposited at the GISAID repository (http://gisaid.org) with accession ID EPI_ISL_510689. For trans infection, cells were pre-incubated for 15 min at room temperature with 10 μg ml−1 anti-SIGLEC1 7–239 (Abcam) or IgG1 isotype control (BD Biosciences), or left untreated before viral exposure. Uptake experiments with SARS-CoV-2 were performed by pulsing 0.1 × 106 DCs with 200 µl SARS-CoV-2 with 106.15 median tissue culture infectious dose (TCID50) per ml for 3 h at 37 °C. After extensive washing, cells were co-cultured at a ratio 3:1 with HEK 293T cells expressing or not expressing ACE2 and TMPRSS-2. Six days later, the amount of viral release to the supernatant was measured with a SARS-CoV-2 nucleocapsid protein High-sensitivity Quantitative ELISA (Immuno Diagnostics).

SIGLEC1 surface expression analysis by FACS

A total of 2.5 × 105 myeloid cells was blocked with 1 mg ml−1 of human IgGs and stained with anti-SIGLEC1–PE or matched isotype–PE control (BioLegend) at 4 °C for 30 min. The mean number of SIGLEC1 monoclonal antibody binding sites per cell was obtained with a Quantibrite kit (Becton Dickinson), subtracting values obtained for isotype control. Samples were analysed with a Canto Flow Cytometer using Flow Jo software to evaluate collected data.

SARS-CoV-2 infection model in hamster

Virus preparation

The SARS-CoV-2 strain used in this study, BetaCov/Belgium/GHB-03021/2020 (EPI ISL 109 407976|2020-02-03), was recovered from a nasopharyngeal swab taken from an asymptomatic patient infected with SARS-CoV-2—confirmed by quantitative PCR with reverse transcription (RT–qPCR)—who returned from Wuhan, China in February 2020. A close relation with the prototypic Wuhan-Hu-1 2019-nCoV (GenBank accession 112 number MN908947.3) strain was confirmed by phylogenetic analysis. Infectious virus was isolated by serial passaging on HuH7 and Vero E6 cells;59 passage 6 virus was used for the study described here. The titre of the virus stock was determined by end-point dilution on Vero E6 cells by the Reed and Muench method60. Live virus-related work was conducted in the high-containment ABSL3 and BSL3+ facilities of the KU Leuven Rega Institute (3CAPS) under licenses AMV 30112018 SBB 219 2018 0892 and AMV 23102017 SBB 219 20170589 according to institutional guidelines.

Cells

Vero E6 cells (African green monkey kidney, ATCC CRL-1586) were cultured in minimal essential medium (Gibco) supplemented with 10% fetal bovine serum (Integro), 1% l- glutamine (Gibco) and 1% bicarbonate (Gibco). End-point titrations were performed with medium containing 2% fetal bovine serum instead of 10%.

SARS-CoV-2 infection model in hamsters. The hamster infection model of SARS-CoV-2 has been described59,61. The specific study design is shown in the schematic below. In brief, wild-type Syrian Golden hamsters (Mesocricetus auratus) were purchased from Janvier Laboratories and were housed per two in ventilated isolator cages (IsoCage N Biocontainment System, Tecniplast) with ad libitum access to food and water and cage enrichment (wood block). The animals were acclimatized for four days before the start of the study. Housing conditions and experimental procedures were approved by the ethics committee of animal experimentation of KU Leuven (license P065- 2020). Sample size was dimensioned in order to have a significant difference of at least 1 log viral RNA (effect size d = 2.004) between control and treated groups, by using a two-tailed t-test with 80% power and an α error of 0.05, calculated with G*Power 3.1 software. Female 6- to 8-week-old hamsters were anaesthetized with ketamine/xylazine/atropine and inoculated intranasally with 50 μl containing 2 × 106 TCID50 SARS-CoV-2 (day 0).

Treatment regimen

Hamsters were allocated into different monoclonal antibody or dosing groups at random. Animals were prophylactically treated 48 h before infection by intraperitoneal administration and monitored for appearance, behaviour and weight. At day 4 after infection, hamsters were euthanized by intraperitoneal injection of 500 μl Dolethal (200 mg ml−1 sodium pentobarbital) (Vétoquinol). Lungs were collected and viral RNA and infectious virus were quantified by RT–qPCR and end-point virus titration, respectively. Blood samples were collected before infection for PK analysis. The circulating antibody levels were measured by mesoscale bridging ELISA using an anti-human LS mutation monoclonal antibody for capture and anti-human CH2 monoclonal antibody for detection34.

SARS-CoV-2 RT–qPCR. Collected lung tissues were homogenized using bead disruption (Precellys) in 350μl RLT buffer (RNeasyMinikit,Qiagen) and centrifuged (10,000 rpm, 5 min) to pellet the cell debris. RNA was extracted according to the manufacturer’s instructions. Of 50 μl eluate, 4 μl was used as a template in RT–qPCR reactions. RT–qPCR was performed on a LightCycler96 platform (Roche) using the iTaq Universal Probes One-Step RT-qPCR kit (BioRad) with N2 primers and probes targeting the nucleocapsid59. Standards of SARS-CoV-2 cDNA (IDT) were used to express viral genome copies per mg tissue or per ml serum.

End-point virus titrations

Lung tissues were homogenized using bead disruption (Precellys) in 350 μL minimal essential medium and centrifuged (10,000 rpm, 5 min, 4 °C) to pellet the cell debris. To quantify infectious SARS-CoV-2 particles, end-point titrations were performed on confluent Vero E6 cells in 96-well plates. Viral titres were calculated by the Reed and Muench method60 using the Lindenbach calculator and were expressed as TCID50 per mg tissue. The samples for RNA and virus titration were run by the technicians in a blinded manner, without knowing the treatment group.

Histology

For histological examination, lungs were fixed overnight in 4% formaldehyde and embedded in paraffin. Tissue sections (5 μm) were analysed after staining with haematoxylin and eosin and scored blindly for lung damage by an expert pathologist. The scored parameters, to which a cumulative score of 1 to 3 was attributed, were the following: congestion, intra-alveolar haemorrhagic, apoptotic bodies in bronchus wall, necrotizing bronchiolitis, perivascular oedema, bronchopneumonia, perivascular inflammation, peribronchial inflammation and vasculitis.

Binding of immunocomplexes to hamster monocytes

Immunocomplexes were generated by complexing S309 monoclonal antibody (hamster IgG, either wt or N297A) with a biotinylated anti-idiotype Fab fragment and Alexa Fluor 488-streptavidin, using a precise molar ratio (4:8:1, respectively). Pre-generated fluorescent immunocomplexes were serially diluted and incubated at 4 °C for 3 h with freshly revitalized hamster splenocytes, obtained from a naive animal. Cellular binding was then evaluated by cytometry upon exclusion of dead cells and physical gating on monocyte population. Results are expressed as Alexa Fluor 488 mean florescent intensity of the entire monocyte population.

Binding of immunocomplexes to hamster FcgRIV and human FcgRIIIA by biolayer interferometry

Antibody immunocomplexes were first generated by cross-linking hamster S309 and S309-N297A with a specific F(ab)2 anti-S309 idiotype monoclonal antibodies (2:1 µg ml−1 ratio) for 40 min at room temperature. Recombinant golden hamster (GH) FcγRIV and biotinylated human FcγRIIIA (Acro Biosystems) were diluted to 0.25 and 0.5 µg ml−1 respectively in kinetic buffer (PBS + BSA 0.01%, pH 7.1) and immobilized on anti-His-Penta or streptavidin SAX Biosensors (FortéBio) for 10 min. FcγRs-coated biosensors were incubated for 5 min with immunocomplexes mixture (4.5 µg ml−1), to allow association, followed by a 5-min dissociation step. The shift in the biolayer interferometry signal generated by change in molecular binding was recorded in real time using an Octet RED96 system (FortéBio). Binding raw data were exported and plotted using GraphPad Prism software (V9).

Bioinformatic analyses

Processed Human Lung Cell Atlas (HLCA) data and cell-type annotations20 were downloaded from Github (https://github.com/krasnowlab/HLCA). Processed single-cell transcriptome data and annotation of lung epithelial and immune cells from individuals infected with SARS-CoV-221,22 were downloaded from the NCBI GEO database (ID: GSE158055) and Github (https://github.com/zhangzlab/covid_balf). Available sequence data from the second single-cell transcriptomics study by Liao et al. 22 were downloaded from NCBI SRA (ID: PRJNA608742) for inspection of reads corresponding to viral RNA using NCBI MagicBlast and SAMtools. Reads that supported a junction between the 5′ leader sequence and the transcription regulatory sequence (TRS) preceding open reading frames for viral genes were counted as evidence of subgenomic mRNA, a surrogate readout for viral replication. Such reads constituted a small fraction of TRS-containing viral reads, ranging from undetectable to 3.4%. The proportion of single guide RNA relative to genomic RNA was estimated by counting TRS-containing reads supporting a leader–TRS junction. Criteria and methods for detection of leader-TRS junction reads were adapted from Alexandersen et al. 62. The viral genome reference and TRS annotation were based on Wuhan-Hu-1 NC_045512.2/MN90894763. Only two samples from individuals with severe COVID-19 had detectable leader–TRS junction reads (SRR11181958 and SRR11181959). Summary visualizations and analyses for the datasets above were performed using R and the following R packages: data.table, ggplot2, cowplot, scales, RColorBrewer, viridis, scater, and SingleCellExperiment.

Statistics and reproducibility

If not stated otherwise, all experiments were performed with at least three biological replicates for each condition. Experiments were independently repeated at least two times, and one representative dataset is shown. All error bars throughout the study represent the s.d. Statistical tests are detailed in the respective figure legends.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.