Dual-role epitope on SARS-CoV-2 spike enhances and neutralizes viral entry across different variants

Grasping the roles of epitopes in viral glycoproteins is essential for unraveling the structure and function of these proteins. Up to now, all identified epitopes have been found to either neutralize, have no effect on, or enhance viral entry into cells. Here, we used nanobodies (single-domain antibodies) as probes to investigate a unique epitope on the SARS-CoV-2 spike protein, located outside the protein’s receptor-binding domain. Nanobody binding to this epitope enhances the cell entry of prototypic SARS-CoV-2, while neutralizing the cell entry of SARS-CoV-2 Omicron variant. Moreover, nanobody binding to this epitope promotes both receptor binding activity and post-attachment activity of prototypic spike, explaining the enhanced viral entry. The opposite occurs with Omicron spike, explaining the neutralized viral entry. This study reveals a unique epitope that can both enhance and neutralize viral entry across distinct viral variants, suggesting that epitopes may vary their roles depending on the viral context. Consequently, antibody therapies should be assessed across different viral variants to confirm their efficacy and safety.


Introduction
Antibodies have long been used to investigate the structure and function of viral glycoproteins by targeting specific epitopes [1,2].Typically, antibody binding to these epitopes either neutralizes the virus or has no impact on its ability to enter cells.However, recent research indicates that antibodies can enhance a virus's entry into cells when they bind to certain epitopes, sparking interest in how they modulate the function of viral glycoproteins [3][4][5][6][7].No epitopes have been identified that can both neutralize and enhance entry for different variants of the same virus.Discovering such dual-function epitopes would significantly advance our understanding of viral entry and evolution, underscoring the need for targeted antiviral therapies that consider epitope variability across viral variants.
The spike protein of SARS-CoV-2 mediates viral entry into host cells [8][9][10].The spike on the surface of mature virus particles is in the "pre-fusion" state.It is a homotrimer, consisting of three receptor-binding S1 subunits on top of a trimeric membrane-fusion S2 stalk.S1 contains a receptor-binding domain (RBD), an N-terminal domain (NTD), and two subdomains SD1 and SD2 (Fig 1A).The RBD exists in two conformations within the spike: standing up for receptor binding and lying down for immune evasion [11,12].The RBD includes a core subdomain and a receptor-binding motif (RBM).During viral entry, the RBD binds to its host receptor ACE2, and the spike is cleaved at the S1/S2 boundary by a host protease (e.g., proprotein convertase furin, cell-surface TMPRSS2, or lysosomal cathepsins) [12][13][14][15].Then S1 dissociates and S2 undergoes a dramatic structural change to fuse viral and host membranes [8][9][10].After membrane fusion, S2 is in the "post-fusion" state.While the post-fusion spike is the most stable and lowest-energy state, the pre-fusion spike must overcome an initial energy barrier to transition into the post-fusion state [10].
The spike protein of SARS-CoV-2 is key to triggering antibody responses; it is also the sole antigen in many COVID-19 vaccines and the main target for antibody-based treatments [10,16].Antibodies work by latching onto specific epitopes on the spike protein.Neutralizing antibodies block SARS-CoV-2 from entering cells in two main ways.First, by binding to the RBM epitopes on the spike, they prevent the virus from attaching to the ACE2 receptor [17][18][19][20].Second, they can attach to non-RBD epitopes on the spike, keeping the spike in its prefusion state and preventing it from fusing membranes [21][22][23].The ways that antibodies could enhance viral entry are less understood, but they may promote the RBD/ACE2 interaction by maintaining the RBD in the standing up position [3].There is ongoing debate about whether such antibodies actually worsen SARS-CoV-2 infections in vivo [3].Since neutralizing antibodies and entry-enhancing antibodies act in fundamentally different ways with completely opposite outcomes, it is unconceivable that a single antibody could both neutralize the virus and facilitate its entry into cells across various viral variants.
Nanobodies are the antigen-binding domain of heavy-chain-only antibodies produced by camelid animals [24][25][26][27].Here we used nanobodies as probes to investigate the functions of epitopes on the SARS-CoV-2 spike.We identified an intriguing epitope, located outside the RBD of the spike, that enhances the cell entry of prototypic SARS-CoV-2, but neutralizes the

PLOS PATHOGENS
Epitope's dual role in SARS-CoV-2 entry cell entry of Omicron variant.We further examined the molecular mechanisms for such dual functions of this epitope.Our study has important implications for the structure, function and evolution of SARS-CoV-2 spike and for the efficacy and safety of antibody therapeutics.

Discovery of three non-RBD epitopes on prototypic spike
In a recent study, we described the discovery of three RBD-targeting nanobodies, named Nanosota-2, Nanosota-3 and Nanosota-4, from an alpaca immunized with prototypic SARS-CoV-2 spike [28].Here prototypic spike refers to the spike protein from the original SARS-CoV-2 variant plus an additional D614G mutation.Each of the three nanobodies blocks ACE2 from binding to the RBD and thereby neutralizes prototypic SARS-CoV-2 entry.In the current study, we discovered three more nanobodies, named Nanosota-5, Nanosota-6 and Nanosota-7, from the same alpaca immunized with prototypic spike.We determined the cryo-EM structures of prototypic spike complexed with either Nanosota-5 or Nanosota-6 (Figs 1B, 1C, 1D, 1E, S1 and S2 and S1 Table ).The structures revealed that Nanosota-5 binds to the NTD and SD2 from the same spike protomer (Figs 1B, 1C and S1), whereas Nanosota-6 binds to the NTD and SD1 from two different spike protomer (Figs 1D, 1E and S2).The epitopes of Nanosota-5 and -6 are both located outside the RBD and do not overlap with the ACE2-binding region.The location of the Nanosota-7 epitope could not be identified using cryo-EM probably due to its flexibility, but ELISA indicated that the Nanosota-7 epitope is located outside the RBD but within S1 (S3 Fig) .In sum, Nanosota-5, Nanosota-6 and Nanosota-7 all target non-RBD epitopes.
We selected Nanosota-5 for further investigation because it was the most effective at enhancing prototypic SARS-CoV-2 pseudovirus entry among the non-RBD-targeting

PLOS PATHOGENS
Epitope's dual role in SARS-CoV-2 entry nanobodies.Additional pseudovirus entry assays revealed that even at very high concentrations (e.g., 0.4 mg/ml), Nanosota-5-Fc continued to enhance the cell entry of prototypic SARS-CoV-2 pseudoviruses (S4A Fig) .Moreover, Nanosota-5-Fc enhanced the cell entry of both the alpha and delta variants of SARS-CoV-2 pseudoviruses (S4B Fig).Thus, Nanosota-5 enhances the cell entry of prototypic and other pre-Omicron SARS-CoV-2 variants.Additionally, the distance between two Nanosota-5 epitopes on the prototypic trimeric spike is 143 Å, suggesting that Nanosota-5-Fc cannot bind two spike protomers in the same trimeric spike simultaneously.These results provided additional insights into Nanosota-5's impact on the cell entry of pre-Omicron SARS-CoV-2 variants.

A non-RBD epitope enhances cell infection of live prototypic SARS-CoV-2
We further explored whether non-RBD epitopes can enhance the infection of live prototypic SARS-CoV-2 in cultured cells.Specifically, we examined the infection efficiency of live SARS-CoV-2 in ACE2-expressing Vero cells in the presence or absence of Nanosota-5-Fc.Two different virus titers and three different nanobody concentrations were tested.At all nanobody concentrations, significantly enhanced viral infection was observed at the high virus titer (Fig 3A ) and became even more prominent at the low virus titer (Fig 3B).As a comparison, Nanosota-3-Fc, which targets the RBD, potently neutralized the infection of live prototypic SARS-CoV-2 (Fig 3C).Thus, the Nanosota-5 epitope not only enhances the cell entry of prototypic SARS-CoV-2 pseudoviruses, but also enhances the cell infection of live prototypic SARS-CoV-2.

The same non-RBD epitope neutralizes SARS-CoV-2 Omicron variant
After studying the function of non-RBD epitopes in the cell entry of prototypic SARS-CoV-2, we further investigated their function in the cell entry of SARS-CoV-2 Omicron variant.To this end, SARS-CoV-2 Omicron pseudoviruses were used to enter ACE2-expressing HEK293T cells in the presence of Nanosota-5-Fc (Fig 4A).The pseudoviruses of three different Omicron subvariants were tested, including the early subvariant BA.1, the later subvariant BA.5 and the recent subvariant XBB.1.5.Surprisingly, Nanosota-5-Fc effectively neutralized the pseudovirus entry of all three Omicron subvariants (Fig 4A).To validate the above result, we performed a cell-cell fusion assay where cells expressing the spike protein (from either prototypic or XBB.1.5SARS-CoV-2) and cells expressing ACE2 were incubated together for fusion in the presence of Nanosota-5-Fc.The result revealed that while Nanosota-5 enhanced the cell-cell fusion for prototypic spike, it inhibited the cell-cell fusion for XBB.1.5spike (Fig 4B).We also repeated the live SARS-CoV-2 infection assay in the presence of Nanosota-5-Fc, this time using live BA.1, BA.5, and XBB.1.5instead of prototypic SARS-CoV-2.The result demonstrated that Nanosota-5-Fc neutralized all three live Omicron subvariants in cultured cells (Fig 4C).Together, all these lines of evidence reveal that the Nanosota-5 epitope has opposing effects on the functions of the spike proteins from prototypic SARS-CoV-2 and Omicron variant: entry-enhancing for the former and neutralizing for the latter.
To understand how Nanosota-5 binds to XBB.1.5spike, we determined the cryo-EM structure of XBB.1.5spike complexed with Nanosota-5 (Figs 5A and S5 and S1 Table ).The result showed that Nanosota-5 binds to the same epitope on XBB.1.5spike as it does prototypic spike.Again, this epitope is located at the junction of the NTD and SD2 from the same spike .Surprisingly, all of the Nanosota-5-contacting residues are conserved between prototypic and Omicron spikes, including a glycan N-linked to Asn61 in prototypic spike (corresponding to Asn58 in XBB.1.5spike) (Figs 5B, 5C and S6).Additionally, we examined the residues outside but near the Nanosota-5-binding site and identified a few residue differences between prototypic and XBB.1.5spikes.We found that these differences did not significantly affect the surface electrostatic potential in the neighboring regions of the Nanosota-5-binding site (S7 Fig) .Indeed, ELISA results showed that Nanosota-5-His exhibited similar binding affinities for the recombinant spike ectodomains from prototypic SARS-CoV-2 and three Omicron subvariants (BA.1, BA.5, and XBB.1.5)(Fig 5D).Thus, the Nanosota-5 epitope on the spikes from different SARS-CoV-2 variants is conserved, suggesting that its varying effects on the functions of spikes from different viral variants are due to structural differences between prototypic and Omicron spikes outside of this epitope.

Molecular mechanism for the dual function of the non-RBD epitope
We investigated the molecular mechanism behind the non-RBD epitope having opposing effects on the functions of spikes from different SARS-CoV-2 variants.We examined how Nanosota-5-Fc affects the binding interactions between the SARS-CoV-2 spike and ACE2.To do this, recombinant ACE2 was incubated with cell-surface spikes in the presence or absence of Nanosota-5-Fc, and the interaction between the spikes and ACE2 was measured using flow cytometry.Both prototypic and XBB.1.5spikes were tested.The results showed that Nanosota-5-Fc increased the binding affinity between prototypic spike and ACE2 (Fig 6A ), while decreasing the binding affinity between XBB.1.5spike and ACE2 (Fig 6B).Since only the standing-up RBD can bind to ACE2, these results suggest that Nanosota-5-Fc promotes more RBDs to stand up in prototypic spike, while having the opposite effect on XBB.1.5spike.To confirm this conclusion, we further investigated how Nanosota-5-Fc affects the binding interaction between prototypic spike and Nanosota-2, which, like ACE2, is only accessible to the standing-up RBD [28].Flow cytometry results confirmed that Nanosota-5-Fc increased the binding affinity between prototypic spike and Nanosota-2 (Fig 6C).Therefore, the Nanosota-5 epitope has opposing effects on the ACE2-binding affinity of prototypic and XBB.1.5spikes, likely by promoting the RBD to stand up.
We also analyzed whether Nanosota-5 affects post-attachment events during the viral entry process.To do this, we incubated pseudoviruses with cells before adding Nanosota-5-Fc and allowing pseudovirus entry to occur.The results showed that even after viral attachment to cells had occurred, Nanosota-5-Fc still enhanced the cell entry of prototypic pseudoviruses, while neutralizing the cell entry of XBB.1.5pseudoviruses (Fig 6D).Therefore, Nanosota-5 has opposing effects not only on ACE2 binding between prototypic and XBB.1.5spikes but also on post-attachment steps between prototypic and XBB.1.5pseudoviruses.
Since the Nanosota-5 epitope is near the furin cleavage site (Fig 1B ), we examined whether the furin cleavage site impacts Nanosota-5's binding to the SARS-CoV-2 spike.ELISA results showed that Nanosota-5-His had similar binding affinities for prototypic spike, regardless of the presence of the furin cleavage site (S8A Fig) .Next, we investigated whether Nanosota-5 affects the furin cleavage of the spikes.To do this, we co-expressed Nanosota-5-Fc with either prototypic or XBB.1.5spike in cells.The results suggested that Nanosota-5-Fc inhibited furin cleavage of both prototypic and XBB.1.5spikes (S8B Fig) .It is important to note that during molecular maturation, the majority of the spike molecules had already been cleaved (S8B Fig) .Therefore, even though Nanosota-5 inhibits furin cleavage of the spikes, it may not have a significant impact on viral entry through furin cleavage inhibition.Additionally, since Nanosota-5 inhibits furin cleavage of both prototypic and XBB.1.5spikes, and this study focuses on the opposing effects of Nanosota-5 on these spikes, Nanosota-5's inhibition of furin cleavage is not a primary focus of the current study.
Since pre-Omicron and Omicron spikes differ in their sensitivity to TMPRSS2 [29], we analyzed whether TMPRSS2 affects Nanosota-5's impact on SARS-CoV-2 entry.Pseudovirus entry assay results showed that in the presence of TMPRSS2, Nanosota-5-Fc still enhanced the entry of prototypic pseudoviruses while neutralizing the entry of Omicron pseudoviruses into cells overexpressing TMPRSS2 (S8C Fig) .Therefore, the opposing effects of Nanosota-5 on prototypic and Omicron spikes are not due to the different TMPRSS2 sensitivities between the two spikes.

Discussion
Epitopes on viral glycoproteins are not just tools for deciphering the structure and function of these molecules; they are also fundamental to developing antiviral vaccines and antibodybased treatments.Typically, epitopes are classified as either neutralizing, non-neutralizing, or

PLOS PATHOGENS
Epitope's dual role in SARS-CoV-2 entry infection-enhancing.However, by using nanobodies in our research, we have identified a SARS-CoV-2 spike epitope that has a dual role: it neutralizes one variant of SARS-CoV-2 and enhances infection in another.This discovery challenges the traditional categorization of epitopes, underscores the complex evolution of SARS-CoV-2 spike, and offers new insights into antiviral antibody therapies.
Infection-enhancing epitopes in coronavirus spikes have attracted great scientific interest.The classical pathway for coronavirus entry into cells relies on host receptors, where the viruses attach to their receptor, undergo endocytosis, and subsequently fuse viral and host membranes [10].Previously, we discovered a unique molecular mechanism for ADE of coronavirus entry that is dependent on FcR but independent of host receptors.Specifically, neutralizing antibodies that target the RBD simultaneously bind to the RBD on the virus and FcR on the cell surfaces.This interaction guides coronaviruses into cells expressing FcR, such as macrophages (S9 Fig) [4].This mechanism has later been validated in vivo [30][31][32][33].More recently, a group of human antibodies, targeting a non-RBD epitope, was isolated from COVID-19 patients and found to enhance SARS-CoV-2 infection by promoting the spike's binding to its ACE2 receptor [3].While resembling ADE, this mechanism has not yet been confirmed in vivo.Additionally, the infection-enhancing activity of these human antibodies requires Fc, although the reason for this remains unclear [3].In our current study, we identified two entryenhancing epitopes outside the RBD of prototypic SARS-CoV-2 spike and delved deeper into one of them, recognized by Nanosota-5.This epitope is situated at the junction between the NTD and SD2.It partially overlaps with the entry-enhancing epitope mentioned earlier, which is recognized by human antibodies (S10 Fig) .In contrast to the FcR-dependent ADE observed with the RBD epitope, the Nanosota-5 epitope enhances entry through an FcR-independent mechanism.It is not clear whether the Nanosota-5 epitope causes ADE in vivo, which will be investigated in future studies.Overall, our research offers new perspectives on how non-RBD epitopes can increase prototypic SARS-CoV-2's ability to enter cells in vitro.
Remarkably, the Nanosota-5 epitope exhibits dual functions: it enhances the cell entry of prototypic SARS-CoV-2 but neutralizes the cell entry of Omicron variant, including BA.1, BA.5, and XBB.1.5subvariants.This discovery challenges conventional epitope paradigms.To decipher the molecular mechanisms underpinning how the same non-RBD epitope can exert opposing effects on different SARS-CoV-2 variants, we conducted comparative studies between prototypic and Omicron spikes.Our investigations revealed that for prototypic spike, Nanosota-5-Fc promotes its ACE2 binding, thereby enhancing viral entry, and also increases its post-attachment activity, likely facilitating its transition to the post-fusion structure.Surprisingly, although our flow cytometry data showed that Nanosota-5 promotes the prototypic spike to bind ACE2, our cryo-EM study of the prototypic spike in complex with Nanosota-5 revealed that all Nanosota-5-bound prototypic spike molecules were in the three-RBD down conformation, which disfavors ACE2 binding.This contrasts with the cryo-EM studies on unliganded prototypic spike, which typically have 40-50% of the molecules with at least one RBD in the standing up position [11,12].The reason for the discrepancy between our flow cytometry and cryo-EM results is unclear but could be due to the cryo-EM procedure (for instance, Nanosota-5-bound open spike might be too unstable to be characterized by cryo-EM).The open and closed populations of Nanosota-5-bound prototypic spike will be further investigated in future cryo-EM studies.Nevertheless, using flow cytometry, we also demonstrated that Nanosota-5 promotes the prototypic spike to bind Nanosota-2, which, like ACE2, only binds to the standing up RBD of the spike [28].Overall, our data suggest that Nanosota-5 induces the RBD to stand up in prototypic spike, facilitating ACE2 or Nanosota-2 binding.It has been shown that the standing up RBD destabilizes the pre-fusion spike, allowing it to transition more easily to its post-fusion structure [10,34], which aligns with our post-attachment

PLOS PATHOGENS
Epitope's dual role in SARS-CoV-2 entry entry data for prototypic pseudoviruses.In contrast, Nanosota-5 reduces ACE2 binding by Omicron spike and inhibits the post-attachment entry of Omicron pseudoviruses, likely by inducing the RBDs to adopt the lying down conformation in Omicron spike.More detailed mechanistic analysis of the opposing effects observed with the Nanosota-5 epitope across different SARS-CoV-2 variants will be conducted in future studies.
Our study sheds light on the potential for more targeted antiviral antibody therapies.We have found that certain epitopes on viral glycoproteins can have dual roles: they might help one virus variant enter cells but block another variant.This discovery highlights the complexity of viral evolution; accordingly, our approaches to antiviral antibody therapies need to consider the versatile nature of epitopes on viral glycoproteins.
Genes encoding spike ectodomain (residues 14-1211), RBD (residues 319-529), and S1 (residues 14-685) from prototypic SARS-CoV-2 were each subcloned into Lenti-CMV vector (Vigene Biosciences) with an N-terminal tissue plasminogen activator (tPA) signal peptide and a C-terminal His tag.For the spike ectodomain construct, a D614G mutation, two mutations in the furin cleavage site (from RRAR to AGAR), and six proline mutations were introduced to the S2 subunit region to stabilize the spike protein in its prefusion state [35,36].Plasmids encoding Fc-tagged Nanosota-5, -6, and -7 were constructed in the same way as above except that a C-terminal human IgG 1 Fc tag replaced the His tag.
Genes encoding monomeric Nanosota-5, -6, and -7 were each cloned into PADL22c vector (Antibody Design Labs) with an N-terminal PelB leader sequence and C-terminal His tag and HA tag.

Screening of nanobody phage display library
The same nanobody phage display library that was recently used to screen for RBD-targeting nanobodies was also used in the current study to screen for non-RBD-targeting nanobodies [28].Briefly, 5 μg purified SARS-CoV-2 spike ectodomain coated on an ELISA plate was used for one round of bio-panning.100 μl phages from the phage library were added to the coated spike and incubated for 1 hour.After washing, the retained phages were eluted and then used to infect ss320 E. coli.The infected ss320 E. coli were spread onto 2YT AG plates, and single colonies were picked and induced by 1 mM IPTG to express individual nanobodies.The supernatants were subjected to ELISA for identification of strong binders against prototypic SARS-CoV-2 spike.

Protein expression and purification
Nanosota-5, -6 and -7 (with a His tag) were purified as previously described [39].Briefly, nanobody expression was induced by 1 mM IPTG and purified from the periplasm of ss320 E. coli.The pellets were collected and re-suspended in 15 ml TES buffer (0.2 M Tris pH 8, 0.5 mM EDTA, 0.5 M sucrose), shaken on ice for 1 hour, diluted with 40 ml ¼ TES buffer, and then shaken on ice for another hour.The proteins in the supernatant were sequentially purified using an Ni-NTA column and a Superdex200 gel filtration column (Cytiva).
SARS-CoV-2 spike ectodomain (with a His tag), RBD (with a His tag) and individual Fctagged nanobodies were prepared from 293F cells as previously described [40].Briefly, lentiviral particles were packaged using the plasmid encoding one of the above proteins and then used to infect 293F cells for selection of stable cell lines in the presence of Puromycin (Gibco).The proteins were harvested from the supernatant of respective cell culture medium, purified on Ni-NTA column for His-tagged proteins or on Protein A column for Fc-tagged proteins, and purified further on Superdex200 gel filtration column (Cytiva).
To prepare the complexes of the spike and individual nanobodies, 2 mg SARS-CoV-2 spike and each of the His-tagged nanobodies (with the nanobody in excess) were incubated at room temperature for 30 minutes.The above samples were then subjected to gel filtration using a Superose 6 increase 10/300 GL column (Cytiva).

ELISA
To detect the binding between SARS-CoV-2 spike ectodomain and HA-tagged nanobodies from the supernatant of ss320 E. coli, ELISA was conducted as previously described [39].Briefly, ELISA plates were coated with purified SARS-CoV-2 spike ectodomain and were then incubated sequentially with the supernatant of ss320 E. coli (containing nanobodies) and HRP-conjugated anti-HA antibody (1:5,000) (Sigma).ELISA substrate (Invitrogen) was added and the reactions were stopped using 1N H 2 SO 4 .The absorbance at 450 nm (A 450 ) was measured using a Synergy LX Multi-Mode Reader (BioTek).
To detect the binding between SARS-CoV-2 spike domains and purified HA-tagged nanobodies, ELISA plates were coated with SARS-CoV-2 spike domains and were then incubated sequentially with the HA-tagged nanobodies and HRP-conjugated anti-HA antibody (1:5,000) (Sigma).The remaining procedure was the same as above.

Pseudovirus entry assay
The activities of nanobodies in SARS-CoV-2 entry were evaluated using pseudovirus entry assay as previously described [40].Briefly, to prepare the pseudoviruses, HEK293T cells were co-transfected with a pcDNA3.1(+)plasmid encoding SARS-CoV-2 spike, a helper plasmid psPAX2 and a reporter plasmid plenti-CMV-luc.Pseudoviruses were collected 72 hours posttransfection, incubated with individual nanobody at different concentrations at 37˚C for 1 hour, and then used to enter HEK293T cells stably expressing human ACE2 (HEK293T/ hACE2 cells).After another 60 hours, cells were lysed.Aliquots of cell lysates were transferred to new plates, a luciferase substrate was added, and relative light units (RLUs) were measured using an EnSpire plate reader (PerkinElmer).The neutralization potency of each nanobody was calculated and expressed as the concentration of the nanobody capable of inhibiting pseudovirus entry by 50% (IC 50 ).
To assess the effects of Nanosota-5-Fc on pseudovirus entry into TMPRSS2-expressing cells, HEK293T cells transiently expressing human ACE2 and TMPRSS2 were used instead of HEK293T/hACE2 cells.All other procedures remained the same as described above.
For the post-attachment pseudovirus entry assay, pseudoviruses were adsorbed onto HEK293T/hACE2 cells at 4˚C for 1 hour.Unbound pseudoviruses were removed, and the cells were washed three times with cold DMEM.Serially diluted Nanosota-5-Fc was then added to the cells and incubated at 4˚C for 1 hour.The plates were subsequently transferred to 37˚C to allow viral entry.Pseudovirus entry results were measured in the same manner as described above.
Confluent monolayers of Vero E6 cells were transfected with 2.0 μg per well of rSARS--CoV-2-Venus BAC cDNA using Lipofectamine 3000.At 72 hours post transfection, cell supernatants were harvested, labeled as P0, and stored at -80˚C.The P0 virus was used to infect fresh Calu3 cells to generate P1 stocks.P1 viral stocks were aliquoted, titered and stored at −80˚C until use.

Live SARS-CoV-2 infection in vitro
Cell infection of live prototypic SARS-CoV-2 in cultured cells by Nanosota-5-Fc was detected using flow cytometry.Briefly, rSARS-CoV-2-Venus was incubated with different concentrations of Nanosota-5-Fc in 200 ml DMEM at 37˚C for 1 hour.Vero E6 cells (106 cells per well, 12-well plate, triplicates) were either mock inoculated or inoculated with the virus/nanobody mixture.After incubation at 37˚C for 1 hour with gentle rocking every 15 minutes, the inocula were removed and the plates were overlaid with 10% FBS.24 hours post-inoculation, cells were treated with 100 ml of trypsin for 2 minutes.Following this, 500 ml 10% FBS was added to terminate the trypsinization, and cell suspension was centrifuged with 1000 rpm/min for 5 minutes.Cell pellets were resuspended using Cytofix (BD Bioscience) and fixed for 30 minutes.Expression of Venus was evaluated via flow cytometry.All flow cytometry data were acquired using a BD FACSVerse and analysed with FlowJo software.A previously discovered RBD-targeting nanobody Nanosota-3-Fc was used as a comparison [28].

PLOS PATHOGENS
Epitope's dual role in SARS-CoV-2 entry The neutralization potency of Nanosota-5-Fc against live Omicron infections was carried out as previously described [42].Briefly, Nanosota-5-Fc was 10-times serially diluted in 50 μl virus growth medium and then was mixed with 50 μl of one of the live Omicron viruses (subvariants BA.1, BA.5 or XBB.1.5)(3000 TCID 50 /ml) at 37˚C for an hour.The mixture was added to Vero E6 cells, which over express TMPRSS2, in 96-well plates and incubated at 37˚C in 5% CO 2 for 4 days.Cell viability was measured using a neutral red assay (Sigma-Aldrich).The efficacy of Nanosota-5-Fc against the Omicron viruses was calculated and expressed as the concentration capable of maintaining the cell viability by 50% (i.e., IC 50 ) compared to the control virus.

Cell-cell fusion assay
Cell-cell fusion was performed as described previously [43].Briefly, HEK293T cells were cotransfected with the plasmid pFR-Luc (containing a luciferase gene whose expression is controlled by a synthetic promoter) and the plasmid expressing SARS-CoV-2 spike.Besides, HEK293T cells expressing ACE2 were transfected with the plasmid pBD-NF-κB (encoding a fusion protein that can activate the luciferase gene expression of the pFR-Luc).After the cells were cultured for 36 hours, Nanosota-5-Fc at various concentrations were added to the spikeexpressing cells and incubated for 30 minutes.Subsequently, ACE2-expressing cells were overlaid onto spike-expressing cells.The luciferase gene expression was activated when cell-cell fusion occurred.After incubation for 6 hours, the cells were lysed, and relative luciferase units were measured using an EnSpire plate reader (PerkinElmer Life Sciences).

Flow cytometry assay
To assess the effects of Nanosota-5-Fc on the binding between cell-surface spike and recombinant ACE2 ectodomain, various concentrations of Nanosota-5-Fc were incubated with spikeexpressing HEK293T cells for 20 minutes.Then, 2 μg/ml human ACE2 ectodomain (containing a C-terminal His tag) was added and incubated with the cell and Nanosota-5-Fc mixture for another 20 minutes.Spike-bound ACE2 molecules were stained with a PE-conjugated anti-His-tag antibody and analyzed using flow cytometry.The results were analyzed using FlowJo software (version 10).
To assess the effects of Nanosota-5-Fc on the binding between cell-surface spike and recombinant Nanosota-2-His (which only binds to the standing up RBD in the spike), 0.5 μg/ml Nanosota-2-His replaced 2 μg/ml human ACE2 ectodomain in the above experiment.

Furin cleavage inhibition assay
HEK293T cells were seeded into a 6-well plate and transfected with 1 μg of prototypic or XBB.1.5spike-expressing plasmid.Different amounts (0, 0.2, 1, and 5 μg) of plasmids expressing Nanosota-5-Fc were co-transfected with the spike-expressing plasmids.Twenty-four hours post-transfection, cells were collected, and the cell surface spike (containing a C-terminal C9 tag) was detected using an anti-C9 antibody by Western blot.

PLOS PATHOGENS
Epitope's dual role in SARS-CoV-2 entry humidity and plunge-frozen in liquid ethane using a Vitrobot Mark IV (FEI).Cryo-EM data were collected using Latitude-S (Gatan) equipped with a K3 direct electron detector and with a Biocontinuum energy filter (Gatan).For the prototypic spike/Nanosota-5 and prototypic spike/Nanosota-6 complexes, the movies were collected at a nominal magnification of 130,000x (corresponding to 0.664 Å per pixel).For the XBB.1.5spike/ Nanosota-5 complex, the movies were collected at a nominal magnification of 81,000x (corresponding to 1.1 Å per pixel).Statistics of cryo-EM data collection are summarized in S1 Table.

Cryo-EM data processing, model building and refinement
Cryo-EM data were processed using cryoSPARC v3.3.2 [44], and the procedure is outlined in S1, S2 and S5 Figs.Briefly, dose-fractionated movies were subjected to Patch motion correction with MotionCor2 [45] and Patch CTF estimation with CTFFIND-4.1.13[46].Particles were then picked using both Blob picker and Template picker in cryoSPARC v3.3.2 and subjected to the Remove Duplicate Particles Tool.Junk particles were removed through multiple rounds of 2D classifications.Particles from the good 2D classes were used for Ab-initio Reconstruction of three or four maps.The initial models were set as the starting references for heterogeneous refinement (3D classification).The good 3D classes were then subjected to further homogeneous, non-uniform and CTF refinements to generate the final maps with applied C3 symmetry.Map resolutions were determined by gold-standard Fourier shell correlation (FSC) at 0.143 between the two half-maps.Local resolution variation was estimated from the two half-maps in cryoSPARC v3.3.2 or v4.0.3.
Initial model building of the spike/nanobody complexes was performed in Coot-0.8.9 [47] using PDB 7TGY and 8IOS as the starting model for prototypic and XBB.1.5spikes, respectively.The initial model of each nanobody was predicted using SWISS-MODEL (https:// swissmodel.expasy.org/),and then fitted into the density map.Several rounds of refinement in Phenix-1.16[48] and manual building in Coot-0.8.9 were performed until the final reliable models were obtained.Model and map statistics are summarized in S1 Table .Figures were generated using UCSF Chimera X v0.93 [49] and PyMol v2.5.2 [50].