N‐glycosylation of the SARS‐CoV‐2 spike protein at Asn331 and Asn343 is involved in spike‐ACE2 binding, virus entry, and regulation of IL‐6

The coronavirus disease 2019 (COVID‐19) pandemic is an ongoing global public health crisis. The causative agent, the SARS‐CoV‐2 virus, enters host cells via molecular interactions between the viral spike protein and the host cell ACE2 surface protein. The SARS‐CoV‐2 spike protein is extensively decorated with up to 66 N‐linked glycans. Glycosylation of viral proteins is known to function in immune evasion strategies but may also function in the molecular events of viral entry into host cells. Here, we show that N‐glycosylation at Asn331 and Asn343 of SARS‐CoV‐2 spike protein is required for it to bind to ACE2 and for the entry of pseudovirus harboring the SARS‐CoV‐2 spike protein into cells. Interestingly, high‐content glycan binding screening data have shown that N‐glycosylation of Asn331 and Asn343 of the RBD is important for binding to the specific glycan molecule G4GN (Galβ−1,4 GlcNAc), which is critical for spike‐RBD‐ACE2 binding. Furthermore, IL‐6 was identified through antibody array analysis of conditioned media of the corresponding pseudovirus assay. Mutation of N‐glycosylation of Asn331 and Asn343 sites of the spike receptor‐binding domain (RBD) significantly reduced the transcriptional upregulation of pro‐inflammatory signaling molecule IL‐6. In addition, IL‐6 levels correlated with spike protein levels in COVID‐19 patients' serum. These findings establish the importance of RBD glycosylation in SARS‐CoV‐2 pathogenesis, which can be exploited for the development of novel therapeutics for COVID‐19.


INTRODUCTION
The coronavirus disease 2019 (COVID-19) pandemic remains an urgent global public health concern, with over 676 million cases reported and over 6 million deaths worldwide as of May 2023. 1,2Despite worldwide vaccination efforts, the number of infections and fatalities will continue to rise for the foreseeable future.Numerous medications have been tested for the treatment of COVID-19, notably Remdesivir, 3 but few therapies have demonstrated robust efficacy in clinical trials.Therefore, hospital care for COVID-19 patients will become commonplace worldwide, and treating complications such as cytokine storm and organ failure in severe cases will increase the burden on clinical care.
The causative agent of COVID-19, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is a member of the Coronaviridae family of viruses.Prior to 2002, Coronaviridae were known only as minor human pathogens, contributing to about 15%-25% of common colds. 4However, the emergence of an outbreak of severe acute respiratory syndrome (SARS) in 2002 caused by the coronavirus SARS-CoV propelled public health vigilance for Coronaviridae. 5 To date, there are seven known coronaviruses of zoonotic origin that can cause human illness, with the coronaviruses MERS-CoV, SARS-CoV, and SARS-CoV-2 identified as being causal of SARS-like diseases. 6ll coronaviruses are lipid-enveloped, positive singlestrand RNA viruses.The lipid envelopes are made of three major components: two glycoproteins known as spike and membrane and a nonglycosylated envelope protein. 7The spike protein is a 1273 amino acid, a transmembrane protein that is essential for virus entry into host cells. 8tructurally, the spike is comprised of two subunits: an Nterminal S1 that contains a receptor-binding domain (RBD) and a C-terminal S2 domain that functions in host membrane fusion.The process of viral entry encompasses multiple stepwise interactions between host cell and viral proteins but is initiated by the attachment of SARS-CoV-2 RBD to the host cell surface protein angiotensin-converting enzyme 2 (ACE2).Thereafter, the S2 domain of the spike protein engages the type II transmembrane protease (TMPRSS2), which catalyzes the cleavage of spike in a process known as priming.Priming frees up the spike protein for fusion with the host cell membrane and subsequent entry into the host cell. 9 notable feature of the spike protein of SARS-CoV-2 is that it is extensively decorated with up to a hundred N-linked glycans, representing posttranslational modifications that are catalyzed by viral hijacking of the host's glycosylation pathways.10,11 Glycosylation of viral structures such as spike proteins contributes to the virus' host immune system evasion strategies by masking antigenic epitopes.10 Indeed, structural data, along with glyco-proteomic analyses, have proposed that extensive glycosylation of the spike protein shields against access by neutralizing antibodies.12,13 Importantly, glycans on the spike protein may also have a hitherto underappreciated role in host cell receptor interactions and cell membrane fusion during viral entry into the cell.It is now well known that both the viral spike protein and human ACE2 receptor are extensively glycosylated, with a majority of the 22 N-glycosylation sites of spike and 7 N-glycosylation sites of ACE2-bearing carbohydrates.[14][15][16] Specific distribution of glycan structures are important factors that are most likely to dictate the viral spike protein and host ACE2 receptor.17,18 However, the gap in the knowledge on the significance of glycosylation (N-terminal) of RBD in RBD-ACE2 binding and virus entry underscores an exigent need for characterizing the relative influence of residue-specific SARS-CoV-2 spike protein glycosylation in molecular recognition events comprising host-pathogen interaction.
Herein, we show that N-glycosylation of SARS-CoV-2 is essential for binding to ACE2 and for viral entry into cells.Specifically, our results reveal that the glycosylated residues Asn331 and Asn343 on RBD play key roles in binding to host cell receptor ACE2 and in viral infectivity.Furthermore, antibody array analysis revealed that IL-6 was significantly upregulated in conditioned media of cells infected with pseudovirus harboring the RBD protein.We also show that N-glycosylation of Asn331 and Asn343 is involved in the regulation of IL-6 expression and that IL-6 and RBD expression levels were correlated in COVID-19 patient sera.Finally, we report that the glycan (Gal-β−1,4-GlcNAc) may bind to N-glycans of Asn331 and Asn343.Collectively, these data show that N-glycosylation of Asn331 and Asn343 is crucial for the RBD/ACE2 interaction, for SARS-CoV-2 infectivity, and for the regulation of IL-6 expression.These data provide new insight into the molecular mechanism of SARS-CoV-2 pathogenesis and may inform the design and development of novel therapeutics against COVID-19.

N-linked glycosylation of RBD protein is required for binding to ACE2
Current research shows that glycosylation is an important component of RBD binding with human receptor ACE2 interaction for virus entry. 17,19However, the mechanistic pathway is not fully understood.Herein, to determine the role of glycosylation in the RBD-ACE2 interaction and to study in detail the role of glycosylation status during RBD-ACE2 binding, we developed a high throughput, in vitro screening method to measure molecular binding between the SARS-CoV-2 spike protein and the human ACE2 protein. 20Microtiter plates were coated with either recombinant SARS-CoV-2 RBD, S2 domain, nucleocapsid (N) protein, or HIV p24 protein as negative control.Systematic incubation of these plates with recombinant human ACE2 confirmed that ACE2 was specifically bound to RBD and not any of the other viral proteins tested (Figure 1a).To test whether the binding affinity of RBD/ ACE2 was impacted by mutations associated with the omicron and delta variants of RBD, the in vitro binding assay was performed with recombinant omicron and delta variant RBD (Supporting Information S3: Figure 1), revealing a modest reduction in affinity only with the omicron variant.Moreover, recombinant SARS-CoV-2 RBD generated by over-expression within human HEK293 cultured cells bound to the ACE2 with significantly stronger affinity compared to recombinant RBD generated by expression within Escherichia coli (Figure 1a), in a dose-dependent manner (Figure 1b).These data indicate that eukaryoticspecific posttranslational modifications may influence ACE2/RBD binding affinity.Glycosylation, specifically N′-terminal glycosylation is a major eukaryotic posttranslational protein modifying event that is hijacked by enveloped viruses. 14,15N′-terminal glycosylation of viral proteins evades the host immune response and aids in the viral life cycle. 19o determine the extent to which glycosylation influences ACE2/RBD binding affinity, RBD generated within HEK293 cultured cells was deglycosylated to remove both N and O-linked glycans, which resulted in a detectable shift in protein size (Figure 1c).We observed that untreated RBD bound to ACE2 in a dose-dependent manner, while deglycosylation of RBD abolished its binding to ACE2 (Figure 1d).Importantly, deglycosylation of ACE2 had a minimal impact on the binding interaction, suggesting that glycosylation of RBD, rather than of ACE2, is critical for the protein-protein interaction.To further interrogate the glycan linkages that function in ACE2/RBD binding, RBD was treated with multiple deglycosylation enzymes (PNGase F, O-glycosidase, α2-3,6,8,9 neuraminidase A, β1-4 galactosidase S, and β-N-acetylhexosaminidase f ) that target specific glycosylation links.Of the enzymes tested, only treatment with PNGase, which cleaves N-linked oligosaccharides, significantly lowered ACE2/RBD binding affinity (Figure 1e).These data suggest that N-linked glycosylation of RBD protein is required for binding to ACE2.

Glycosylation of RBD at Asn343 is essential for interaction with ACE2
The SARS-CoV-2 spike protein has 22 putative glycosites as determined by the presence of the Asn-Xaa-Ser/Thr, Xaa≠Pro motif sequence. 16Of these sites, N331 and N343, are located on the RBD and have been shown to be N-glycosylated, and when mutated to Gln, were shown to reduce viral infectivity. 16,17To determine whether these residues participate in the RBD/ACE2 molecular interaction during binding, we generated three mutated recombinants of RBD, including N331Q, N343Q, and a N331Q/N343Q double mutation.The binding activity of these mutants to ACE2 was measured using the in vitro binding assay, which revealed that binding was abolished in both N343Q and N331Q/N343Q and was significantly lower in the N331Q mutant version of RBD (Figure 2a).To measure the influence of the N343 and N331 glycosites on infectivity, we employed a pseudovirus system based on a method developed by Whitt et al. 18 Briefly, we created a pseudovirus that expresses the wild-type SARS-CoV-2 spike protein, or mutants thereof, on a viral particle surface, while a plasmid encoding for luciferase is contained inside the viral particle.Upon viral entry into host cells, Luciferase (Luc) is expressed and quantified as a faithful readout of the extent of viral entry as a relative luciferase unit (RLU).In addition to wild type S, we generated pseudovirus expressing mutated versions, namely N331Q, N343Q, and a N331Q/ N343Q double mutation on the viral surface.We confirmed the expression of viral wild-type and mutated derivatives using immunoblot analysis (Figure 2b).Measurement of luciferase activity within two different infected cell types with high ACE2 expression, namely A549 and Vero, revealed that double deletions at N331 and N343 resulted in a substantial reduction in viral infectivity (>80% inhibition).The single mutation N331Q caused a modest decrease of less than 20%, while N343Q exhibited a 50% decrease (Figure 2c,d).In accordance with the results of the in vitro binding assay, the data show that mutation of these specific glycosylation sites on the SARS-CoV-2 spike protein significantly reduces pseudoviral infectivity.ACE2 protein expression was tested in different cell lines, namely NIH-3T3, A431, A549, and Vero (Figure 2e).We confirmed that A549 and Vero cells expressed significant levels of ACE2, while NIH-3T3 and A431 cells did not.The specificity of RBD/ACE2 binding was then assayed with blocking antibodies, including anti-RBD (1H9; 4A9; 1F9), anti-S2, and anti-N antibody (Figure 2f).

Identification of glycan epitopes preferred by RBD protein
To explore the pertinent glycan structures on human receptors that interact with the S protein, 21,22 and to determine the extent to which these interactions are dependent on N331 and N343, a glycan array was hybridized with the wild-type RBD protein or its mutated derivatives.The glycan array data showed that the binding of RBD with terminal galactose-bearing glycan epitopes was reduced significantly when both N331 and N343 were mutated.The candidate glycan molecule was identified as Glycan array data analysis showed that each single mutation, either at N331 or N343Q, did not reduce the binding with Gal ; however, the double mutant (N331Q & N343Q), showed a dramatic reduction in binding (~80%) (Figure 3a).This is consistent with our results with the S-ACE2 binding assay and pseudovirus assay, which indicated that single mutations blocked RBD-ACE2 binding and viral entry to a lesser degree compared to the double mutant.
Based on our results, we hypothesized a model for the interaction of the double N-glycosylation mutant with host cell surface glycans (Figure 3b).According to our model, the binding of RBD with the terminal galactose epitope is diminished when N331 and N343 glycan sites are mutated.This implies that glycans at these sites are critical for the interaction of RBD with terminal galactose structures on the cell surface receptor.Other studies suggest that residues N90 and N322 on ACE2 are predicted to interact with the RBD 22 and are proximal to the RBD. 23In a previous study, we evaluated the presence of this structure on human ACE2 by highresolution LC-MS/MS and found that such terminal Gal structures are indeed present on both N90 and N322 of ACE2 (Figure 3c) and that N90 glycosylation on ACE2 prevents RBD binding while glycosylation at N322 favors RBD binding.Terminal galactose structures are displayed more at N322, which indicates that the terminal galactose structures without sialic acids may be critical in RBD binding with receptors (ACE2).We further confirmed our glycan array results by treating the receptor (ACE2) with neuraminidase to remove sialic acid residues, exposing the terminal galactose residue, and followed with the RBD-ACE2 binding assay.We observed a dosedependent increase in binding.Neuraminidase-treated ACE2 binds with higher affinity (~25%) compared to wild-type RBD glycoprotein in the binding assay (Figure 3d).Since our earlier experiment showed that the glycans on ACE2 had a minimal impact on RBD binding (Figure 1d), we hypothesize that the terminal galactose glycan structures with which RBD interacts, may be displayed on the cell surface, and not necessarily on the ACE2 receptors.Cell surface glycans can enhance the viral binding with its receptors. 25Thus, our observations indicate that glycans with terminal galactose expressed on the cell surfaces could facilitate binding of RBD, and such binding interactions are dependent on N-glycosylation at sites N331 and N343 of RBD.

Identification of IL-6 expression by antibody array analysis
Next, we wanted to understand the pathological relevance of the wild-type (WT) spike-RBD in comparison with the N′ terminal spike-RBD glycosylation mutants.][28] Importantly, IL-6 has been suggested as one of the prime mediators of sustained proinflammatory response in COVID-19 patients.Therefore, we collected serum samples from subjects enrolled in the study and quantified IL-6 and spike protein levels by sandwich ELISA.Patients were  23 The glycan structure shown is drawn using the SNFG nomenclature, where the blue squares represent N-acetylglucosamine, the yellow circles represent galactose, the green circles represent mannose, and the red triangle represents fucose. 23,24(d) RBD-ACE2 binding increased in the presence of ACE2 treated with neuraminidase for 1 h at 37°C in a dose-dependent manner (n = 2 and duplicate for each set).
confirmed as COVID-19 positive or negative using an FDA-approved RT-PCR test.We also measured both IL-6 and RBD levels in COVID-19-infected patient's sera.Results from this study reveal a correlation between IL-6 and RBD levels (R 2 :0.802; p < 0.0001) (Figure 4a).Next, to gain a better understanding of this finding, we performed an antibody array analysis of conditioned media from A549 cells infected with WT and mutant RBD viruses to identify the most robust changes in the host response signaling molecules compared to the mock-infected control.Our antibody array results revealed notable changes in IL-6 levels (Figure 4b).IL-6, a pro-inflammatory immune mediator, was downregulated by >50% when cells were transfected with double mutant RBD plasmid compared to WT RBD.To confirm this finding with a higher sensitivity assay, we performed an IL-6 ELISA (LOD 3 pg/mL) on supernatants collected from WT and RBD-infected A549 cells.We observed a similar downregulation of IL-6 in the presence of the double mutant RBD (Figure 4c).In both immunoassays, the single mutants exhibited an equivocal change in the induction of IL-6, suggesting that the combined effect of the two mutations is more robust than either alone.Taken together, these findings suggest that the two RBD residues N331 and N343 contribute at least partially to host IL-6 expression during SARS-CoV-2 infection.

Mutations at N331 and N343 sites of RBD exhibit blunted ability to elicit IL-6 expression
4][35][36] Hence, we investigated whether the exposure of host cells to wild-type RBD and N-glycan mutant RBD proteins influenced the expression of IL-6.We measured human IL-6 gene expression by RT-PCR in human epithelial and endothelial cells (i.e., HEK-293, A549, and HUVEC cell lines).All three cell lines were treated with either wild-type RBD or mutant RBD proteins for 24 or 48 h at varying concentrations in accordance with previous studies. 37Recombinant IL-6 was used as a positive control protein for this study.Our results revealed minimal upregulation of IL-6 mRNA when cells were treated with either of the RBD single mutants (at N331Q and N343Q) when compared to wild-type RBD protein or recombinant IL-6 treatment.Interestingly, the expression level of IL-6 mRNA in cells treated with the N-glycan double mutant (N331Q & N343Q) was similar to vehicle treatments (Figure 5a-c).These results further support the involvement of the RBD in N-terminal glycosylations at N331 & N343 in the activation of pro-inflammatory immune response (i.e., IL-6 mRNA expression).
9][40][41] Therefore, we next assessed the cytotoxic effects of the wild-type RBD in comparison with the N-terminal RBD glycosylation mutants.We measured cytotoxicity by measuring LDH activity from the lysates of cells treated with varying concentrations of WT and mutant proteins.The LDH assay revealed that, in all three cell lines, wild-type RBD protein was cytotoxic, whereas the RBD mutants had no significant cytotoxicity.(Supporting Information S3: Figure 2a-c).Moreover, the double mutant of RBD protein elicited still less cytotoxicity compared to either single mutant of RBD protein.These trends in LDH activity were consistent with that of the IL-6 expression data.

DISCUSSION
Glycosylation of SARS-CoV-2 spike protein has been widely recognized to have a significant role in COVID-19 pathogenesis. 2,10,42,43The RBD of spike protein is of particular interest because a substantial number of neutralizing antibodies with therapeutic benefits have been identified that directly bind to the RBD of the spike glycoprotein trimer. 44Given the critical importance of spike protein in the viral fusion machinery, structural and functional aspects of putative glycosites on the RBD have gained considerable attention.
To study the role of glycosylation on SARS-CoV-2, we used an in vitro binding assay to show that glycosylation of the RBD protein, and not of ACE2, is critical for the S/ACE2 interaction.We further report that specific glycosylation sites on the SARS-CoV-2 spike protein, namely N331 and N343, are required for spike protein/ACE2 binding and viral infectivity.We, therefore, identify site-specific, N-linked, posttranslational modifications to the SARS-CoV-2 spike protein that are required for its pathogenesis.This finding is consistent with a previous report in which a double mutation at N331 and N343 of the SARS-CoV-2 spike protein markedly reduced infectivity, while single mutations at these residues exhibited less reduction in viral infectivity. 17upporting the concept, prior research has reported on the function of specific-residue glycosylation in host cell and pathogen interactions, innate immune responses, and inflammation. 45,46During viral pathogenesis, viruses co-opt the host's cellular glycosylation pathways to modify their own component proteins, which results in enhanced viral protein stability or can serve to occlude the binding of host antibodies to the virus.Furthermore, the decoration of viral proteins by glycans may hamper the host's ability to generate antibodies to epitopes containing the glycan by limiting antigen presentation by the HLA complex.While glycosylation of N331 and N343 of the SARS-CoV-2 spike protein may function in any of these processes, we show that mutating N331 and N343 lowers viral entry into host cells, which may point to these glycosites participating in maintaining the conformation of the S1 subunit during viral binding and entry.
Because glycosylation at N343 of the SARS-CoV-2 spike protein is essential to viral entry into the host cell, mutations of this residue, or defects in its glycosylation, would likely adversely affect virus fitness.Therefore, this glycosite represents a vulnerability that may be exploited in spike protein-based immunogen design for vaccine development.Indeed, neutralizing antibodies have been shown to directly contact RBD epitopes containing N343.It has been suggested that the identification of crucial glycosites such as those reported here would enable vaccine development strategies that leverage the glycosylation-dependent mechanisms used by antigen-presenting cells to process viral glycoproteins. 30n this study, we identified the glycan epitope (Gal-β−1,4-GlcNAc-β−1,3-(Gal-β−1,4-GlcNAc-β−1,6-) Gal-β−1,4-Glc-Sp5) of spike protein, the binding of which is significantly reduced in the double mutant (N331Q and N343Q) compared to either the single (N331Q and N343Q) mutant or WT RBD protein.This implies that the glycans on the host receptors (i.e., human ACE2) with terminal galactose epitopes are critical in their interaction with the RBD.The N-glycosylations at N331 and N343 of spike protein might have roles in the interaction with F I G U R E 5 N-Glycan mutations at N331 & N343 of RBD protein did not induce IL-6 expression (a-c).Gene expression data for IL-6 mRNA was normalized with respect to 18sRNA (housekeeping control).Cell cytotoxicity was assessed in each cell line of the corresponding experiment in parallel using LDH measurements.LDH fold changes were normalized with corresponding protein amounts quantified in the cellular lysates using the BCA assay method (mg/mL).Each data represents the mean and SE from the n = 3 independent experiments performed in duplicate.Data were analyzed using one-way ANOVA with Tukey's test **p < 0.005, ***p < 0.001, and ****p < 0.0005 for statistical significance.
the receptors, and this interaction could be mediated through glycan structures with terminal galactose epitopes.Our findings are consistent with recently published molecular simulation studies that propose the critical roles of N-glycosylation in the RBD conformation changes and receptor binding. 47he residues N90 and N322 on the ACE2 are proximal to the RBD binding region.Thus, we evaluated the presence of terminal galactose structures on human ACE2 and found that these structures are indeed present on both residues (depicted in Figure 3c). 23N90 glycosylation of ACE2 prevents RBD binding while glycosylation at N322 favors RBD binding. 23,48Interestingly, the glycans with terminal galactose epitopes are relatively more abundant on N322 compared to other glycans.This suggests that these terminal galactose structures without sialic acid end capping on the ACE2 receptor may be critical in RBD binding.If this were true, we would expect ACE2 treated with neuraminidase to bind more tightly to RBD compared neuraminidase untreated ACE2.Indeed, our results showed that RBD binds more strongly with neuraminidase-treated ACE2 in a dosedependent manner (Figure 3d).We, therefore, speculate that this glycan interaction plays an important role in the S-ACE2 binding and potentially in viral docking and entry.
By contrast, the role of the N90 glycosylation in the interaction with the RBD is less clear.Interestingly, several studies suggest that removal of the glycosylation motif at N90 enhances the RBD binding with ACE2, 49,50 while other studies have pointed to a limited role for the glycans on ACE2. 51Further study is required to fully understand the precise role of N90 glycosylation in RBD binding.
It is well known that cytokine storm, and particularly IL-6, are major components of the pathology of SARS-CoV2 infection; however, the mechanism for initiation of a hyperinflammatory response and multiorgan damage from viral infection is poorly understood.Possible roles of IL-6 in COVID-19 pathogenesis have been discussed in several recent findings. 52,53Importantly, the functional significance of N-terminal glycosylation of RBD on cellular inflammation and cytotoxicity remains unclear.To address this knowledge gap, we first profiled cytokine expression and identified that IL-6, a pro-inflammatory signaling molecule, is directly modulated by SARS-CoV-2 in the pseudovirus model system.Our findings are in line with recent reports in which SARS-CoV-2 infection or spike protein treatment of epithelial cells significantly induced IL-6 trans-signaling by activation of the angiotensin II type 1 receptor (AT1) axis to initiate coordination of a hyper-inflammatory response. 54t is evident that SARS-CoV-2 infection or spike protein induces AT1-mediated signaling cascade and thus activates transcriptional regulatory molecules such as NF-κB and AP-1/c-Fos via MAPK activation, resulting in increased IL-6 release.4][35][36] Stimulation of TLRs by bacterial or viral pathogen components or by proinflammatory cytokines such as IL-1, IL-6, and TNF-α can activate cis-regulatory elements, leading to IL-6 synthesis.Recent studies have also shown that S protein treatment robustly induces IL-6 mRNA levels in mammalian cells. 54iven the pathological relevance of the spike protein, we speculated that the mutations at N331 & N343 of RBD protein would affect IL-6 activation and thus abolish IL-6 mediated cytokine storm.Interestingly, our qPCR results (Figure 5a-c) showed that both single and double mutant RBD protein significantly reduced IL-6 activation at the transcriptional level in human epithelial and endothelial cells.
Although both single mutations at N331Q and N343Q RBD protein modestly reduced IL-6 expression, the level of IL-6 in cells treated with the double mutant (N331Q & N343Q) RBD protein was similar to that of control.This suggests that both glycosylation sites contribute to the RBDinduced inflammatory response (i.e., IL-6 production), whose mechanisms remain unclear in our present investigation.This observation was consistent with the finding that IL-6 should be very minimal either in normal sera or in COVID-19-recovered patients. 55In addition, our findings revealed that the double mutant did not induce cytotoxicity as observed in the LDH assay experiments.Given the consistency in our findings from multiple independent mammalian cell lines, it can be thus concluded that the glycosylation of RBD plays an important role in RBD induced cytotoxicity and cellular inflammation.Our IL-6 qPCR results were consistent with the binding assay and cell-based pseudovirus assay data (Figures 1 and 2).Furthermore, the dependence of the viral cell entry on the integrity of the RBD/ACE2 interaction suggests that a vaccine targeting this epitope may retain efficacy through seasonal antigenic drift. 2 The spread of COVID-19 is continuing worldwide with little or no sign of slowing down.Although multiple types of vaccines and therapeutics (including antibodies and small molecules) are being developed, a major concern is that fast mutations of SARS-CoV-2 may restrict the efficacy of vaccines or treatments. 32Therefore, identifying the molecular underpinnings of viral entry into the host cells, including posttranslational modifications, is of crucial value for the development of new therapies to treat COVID-19.We contribute to these endeavors by showing that glycosylation modifications, likely by molecule (Gal-β−1,4-GlcNAc-β −1,3-(Gal-β−1,4-GlcNAc-β−1,6-) Gal-β−1,4-Glc-Sp5) moieties at specific residues to the SARS-CoV-2 spike protein, is required for viral binding and entry.Because there is considerable existing evidence that human neutralizing antibodies target the RBD of spike protein, and some of which have shown therapeutic promise, 29,31,35,36 characterization of glycosylation modifications at N331 and N343 is critical knowledge to inform the development of neutralizing antibodies against COVID-19.
In this study, we show the glycosylation of N331 and N343 residues of RBD protein to be critical for the induction of pro-inflammatory signaling (i.e., IL-6 expression) and cytotoxicity (LDH release).The glycosylation of these 2 residues is, therefore, a key aspect of the molecular mechanism of the host-pathogen interaction as well as its downstream inflammatory pathways in the host cell.Our proposed model (Figure 6) provides a molecular target for the design of vaccines and therapies to block viral infectivity and cytokine storms.

Spike/ACE2 binding assays
We employed an ELISA methodology described previously 20 to analyze the interaction of the RBD with ACE2.In brief, 96-well microplates were coated with recombinant RBD.Recombinant human ACE2 protein was added in varying concentrations, followed by washing, before incubation with goat anti-ACE2 antibody.After the addition of HRP-conjugated antigoat IgG, signals were generated with 3,3',5,5'tetramethylbenzidine substrate and plates read by colorimetry at 450 nm.Recombinant proteins used for the binding assay are listed in Table 1.Deglycosylation of the proteins was performed under native or reducing conditions using Protein Deglycosylation Mix II (New England Biolabs, cat.P6044) or the individual deglycosylases, PNGase F, O-glycosidase, α2-3,6,8,9 neuraminidase A, β1-4 galactosidase S, and β-Nacetylhexosaminidase f (New England Biolabs) following the manufacturer's instructions.SDS-PAGE electrophoresis and immunoblot were performed using standard protocols.Mouse anti-SARS-CoV-2 S1 protein antibody (RayBiotech, cat.130-10864) was used at 1 µg/mL.

Cell culture and treatments
Two cell lines with high ACE2 expression (Vero and A549) and were purchased from the ATCC.Vero C1008 is an African monkey kidney cell line (ATCC® CRL-586™).A549 is an adenocarcinomatoid human alveolar basal epithelial cell line (ATCC® CCL-185™).BHK21/WI-2 is a baby hamster kidney cell line (Kerafast catalog # EH1011).BHK21 cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) containing 5% FBS (Atlanta Biologicals) and penicillin-streptomycin (Corning) at 37°C in 5% CO 2 .Vero and A549 cells were maintained in Dulbecco's Modified Eagle Medium (DMEM) containing 10% FBS and 1x penicillin-streptomycin at 37°C with 5% CO 2 .ΔG-luciferase (G*ΔG-luciferase) VSV pseudotypes carrying SARS-CoV-2 spike or VSV-G-kan were produced from BHK21 cells.The pseudoviral particle titration and neutralization assays were conducted using Vero and A549 cells. 18,34I G U R E 6 Proposed model: Mutations in N-glycosylation sites at N331 and N343 residues of RBD protein inhibit S-ACE2 binding, IL-6 expression and cytotoxicity.I confirm and agree with the possibility that my figure images might be selected and used as a journal cover image at issue publication.using Mapix software.Array-specific data analysis software was used for data computation.The RayBio® Human IL-6 ELISA kit (cat# ELH-IL6) is an in vitro ELISA for the quantitative measurement of human IL-6 and was used to determine the IL-6 level in conditioned media from the pseudovirus assay in parallel with the same experiment.

Real-time PCR analysis
For measuring IL-6 mRNA expression, total RNA was isolated from WT-RBD, Mutant proteins, and human recombinant IL-6 treated HEK-293T and HUVEC cells, respectively, using RNAeasy Mini kit (QIAGEN), and cDNA synthesis was conducted using iScript cDNA synthesis kit (Bio-Rad).qPCR assay was performed by subjecting 500 ng of cDNA using power TRACK Sybr mix (Thermo Fisher Scientific) using primers specific for human IL-6 (forward: 5'-TCGGTCCAGTTGCCTTCT-3' and reverse: 5'-TGAGATGCCGTCGAGGAT-3') and 18srRNA (forward: 5'-TTGGTGGAGCGATTTGTCTG-3' and reverse: 5'-ATCTCGGGTGGCTGAACG-3') as per manufacturer's instructions.The expression levels of IL-6 mRNA were normalized to that of 18srRNA levels.Relative expression of IL-6 mRNA in untreated control and treated samples is expressed as 2 C −Δ t values as described previously, and fold changes are calculated by comparing the 2 C −Δ t values of the treated sample with that of untreated control.

Lactate dehydrogenase (LDH) assay
To determine cytotoxicity, LDH release was measured with the LDH Cytotoxicity Assay Kit (Pierce) according to the manufacturer's protocol.HEK-293, A549, and HUVEC cells were plated, respectively, into 12-well plates and maintained overnight before treatment, following which the cells were treated with WT-RBD, Mutant proteins, and human recombinant IL-6 at 1000 ng/mL for 24 h.After treatment, the cell supernatants were collected by centrifugation.Supernatants were assayed for LDH in triplicates as per the manufacturer's protocol.Absorbances were measured at 490 and 680 nm using a spectrophotometer (Biotek).Absorbance values were normalized to the protein amounts in the analyzed lysates by BCA quantification (Pierce; Thermo Fisher).Fold change in cytotoxicity was calculated based on the difference compared with the LDH-positive control provided with the kit.

Patient's serum collection and IL-6 and RBD quantification
This is a single-center prospective cohort study performed at PanoHealth LLC.Blood samples were collected from adult patients and were selected based on COVID-19 diagnosis by FDA-approved RT-PCR test.Blood samples were processed according to Sterling IRB: 8291-BZhang.Serum was subsequently used for IL-6 and RBD measurement.

Statistical analysis
Data were expressed as mean ± SEM obtained from an appropriate number of independent experiments conducted in either at least duplicate or triplicates, as stated.Statistical significance was assessed using GraphPad Prism software, version 8.2.1 (GraphPad Prism Software).Unpaired or paired two-tailed t tests were used, as stated, to analyze data involving a direct comparison of an experimental group with a control group.One-or two-way ANOVA was used with repeated measures for appropriate experimental groups as stated, followed by Tukey's and Bonferroni's multiple comparison correction.For in vitro quantitative-PCR and LDH release experiments, we performed one-way ANOVA to determine the significance of target gene expression in control versus treatment groups.The reported p values were adjusted to account for multiple comparisons.For all statistical tests, a two-sided confidence level of p < 0.05 (95% CI) was accepted for statistical significance.

F
I G U R E 1 N-linked glycosylation of RBD protein is required for binding to ACE2.(a) Binding activity of SARS-CoV-2 RBD and S2 proteins derived from HEK293 cells or Escherichia coli.Binding was measured using an ACE2 binding assay, and nucleocapsid protein (N) and HIV p24 (P24) were used as negative controls.(b) Binding activity of RBD expressed either in HEK293 cells or in E. coli to titrated concentrations of ACE2.(c) Detection of purified and deglycosylated recombinant RBD and ACE2 proteins by SDS-PAGE.From left to right: untreated proteins, proteins deglycosylated (dg) under native conditions, proteins deglycosylated under reducing conditions.(d) Detection of RBD and ACE2 binding using the ACE2 binding assay under conditions where either or both RBD and ACE2 were deglycosylated (dg).(e) RBD was deglycosylated with individual enzymes: PNGase F, O-glycosidase, α2-3,6,8,9 neuraminidase A, β1-4 galactosidase S, or β-N-acetylhexosaminidase f , prior to the ACE2 binding assay.Statistical significance was tested by T test for all experiments.****p < 0.0001.n ≥ 3 for all experiments.

F I G U R E 2
Glycosylation of RBD at Asn343 is essential for interaction with ACE2.(a) Detection of the binding of RBD wild type, or its N331Q, N343Q, and N331Q/N343Q mutated derivatives to ACE2 using the ACE2 binding assay.40 ng/mL ACE2 was used in all assays.(b) Immunoblot analysis of recombinant SARS-CoV-2 RBD wild type or N331Q, N343Q, and N331Q/N343Q double mutated derivatives expressed within purified pseudovirus particles propagated in BHK21 cultured cells.(c) Detection of luciferase activity in A549 cultured cells infected with pseudovirus harboring S1 wild type, or its N331Q, N343Q, and N331Q/N343Q mutated derivatives.(d) Detection of luciferase activity in Vero cultured cells infected with pseudovirus harboring S1 wild type, or its N331Q, N343Q, and N331Q/N343Q mutated derivatives.(e) ACE2 expression levels (protein) in four cell lines, NIH3T3, A431, A549, and Vero, were detected by immunoblot.(f) The ACE2 binding was measured in the presence of blocking antibodies, including anti-RBD (1H9; 4A9; 1F9), anti-S2, and anti-N, to assess the specificity of ACE2 binding to RBD.Statistical significance was tested by T test for all experiments.****p < 0.001.n = 4/5 for all experiments.

F
I G U R E 3 Identification of glycan binding epitope in the N-Glycan mutant site of RBD protein.(a) Identification of glycan recognition domains in the N331Q and N343Q mutants of RBD protein using a glycan array.Data represents two independent experiments.Data were analyzed using unpaired Tukey's test *p < 0.02 for statistical significance.(b) Schematic model of interaction inhibition in between double mutant (N331Q and N343Q) RBD mutant and WT RBD protein.Adapted from BioRender.com(2021).(c) N-Glycan distribution on sites N90 and N322 of human ACE2 receptor, which are proximal to the RBD binding region of S1 protein (red arrows show the glycans with terminal galactose).Modified from Shajahan and colleagues, with permission.

F
I G U R E 4 Glycosylation mutants of RBD do not induce IL-6 expression.(a) Correlation between IL-6 and RBD in COVID-19-positive sera.Serum samples from COVID-positive patients were analyzed by ELISA for IL-6 and RBD (n = 21).IL-6 was also measured in conditioned media from the pseudovirus assay (A549 cells) by (b) antibody array and (c) sandwich ELISA; all runs were performed in at least duplicate.Data were analyzed using one-way ANOVA (****p < 0.0001) and unpaired Tukey's test **p < 0.0075 for statistical significance.SPIKE MUTATIONS REDUCE ACE2 BINDING | 171 Recombinant proteins used in binding assays.
T A B L E 1