key: cord-0964178-exn0zctm
authors: Rowland, Raymond; Brandariz-Nuñez, Alberto
title: Analysis of the role of N-linked glycosylation in cell-surface expression, function and binding properties of SARS-CoV-2 receptor ACE2
date: 2021-05-11
journal: bioRxiv
DOI: 10.1101/2021.05.10.443532
sha: 1da5e42162c1a4b06183faa67017d2e7a97da4ad
doc_id: 964178
cord_uid: exn0zctm

Human angiotensin I-converting enzyme 2 (hACE2) is a type-I transmembrane glycoprotein that serves as the major cell entry receptor for SARS-CoV and SARS-CoV-2. The viral spike (S) protein is required for attachment to ACE2 and subsequent virus-host cell membrane fusion. Previous work has demonstrated the presence of N-linked glycans in ACE2. N-glycosylation is implicated in many biological activities, including protein folding, protein activity, and cell surface expression of biomolecules. However, the contribution of N-glycosylation to ACE2 function is poorly understood. Here, we examined the role of N-glycosylation in the activity and localization of two species with different susceptibility to SARS-CoV-2 infection, porcine ACE2 (pACE2) and hACE2. The elimination of N-glycosylation by tunicamycin (TM) treatment or mutagenesis, showed that N-glycosylation is critical for the proper cell surface expression of ACE2 but not for its carboxiprotease activity. Furthermore, nonglycosylable ACE2 localized predominantly in the endoplasmic reticulum (ER) and not at the cell surface. Our data also revealed that binding of SARS-CoV and SARS-CoV-2 S protein to porcine or human ACE2 was not affected by deglycosylation of ACE2 or S proteins, suggesting that N-glycosylation plays no role in the interaction between SARS coronaviruses and the ACE2 receptor. Impairment of hACE2 N-glycosylation decreased cell to cell fusion mediated by SARS-CoV S protein but not SARS-CoV-2 S protein. Finally, we found that hACE2 N-glycosylation is required for an efficient viral entry of SARS-CoV/SARS-CoV-2 S pseudotyped viruses, which could be the result of low cell surface expression of the deglycosylated ACE2 receptor. Importance Elucidating the role of glycosylation in the virus-receptor interaction is important for the development of approaches that disrupt infection. In this study, we show that deglycosylation of both ACE2 and S had a minimal effect on the Spike-ACE2 interaction. In addition, we found that removal of N-glycans of ACE2 impaired its ability to support an efficient transduction of SARS-CoV and SARS-CoV-2 S pseudotyped viruses. Our data suggest that the role of deglycosylation of ACE2 on reducing infection is likely due to a reduced expression of the viral receptor on the cell surface. These findings offer insight into the glycan structure and function of ACE2, and potentially suggest that future antiviral therapies against coronaviruses and other coronavirus-related illnesses involving inhibition of ACE2 recruitment to the cell membrane could be developed.

expression of ACE2 but not for its carboxiprotease activity. Furthermore, 25 nonglycosylable ACE2 localized predominantly in the endoplasmic reticulum 26 (ER) and not at the cell surface. Our data also revealed that binding of CoV and SARS-CoV-2 S protein to porcine or human ACE2 was not affected 28 by deglycosylation of ACE2 or S proteins, suggesting that N-glycosylation plays 29 no role in the interaction between SARS coronaviruses and the ACE2 receptor. for the development of approaches that disrupt infection. In this study, we show 39 that deglycosylation of both ACE2 and S had a minimal effect on the Spike-40 ACE2 interaction. In addition, we found that removal of N-glycans of ACE2 41 impaired its ability to support an efficient transduction of SARS-CoV and SARS-42

CoV-2 S pseudotyped viruses. Our data suggest that the role of deglycosylation 43 of ACE2 on reducing infection is likely due to a reduced expression of the viral 44 receptor on the cell surface. These findings offer insight into the glycan 45 structure and function of ACE2, and potentially suggest that future antiviral 46 therapies against coronaviruses and other coronavirus-related illnesses 47 Introduction: 50 Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) is a highly 51 transmissible betacoronavirus that emerged in 2019 and is responsible for the 52 current pandemic (1-7). Outcomes of human infection range from 53 asymptomatic infection to severe clinical disease (8, 9) . Infection is frequently 54 associated with severe acute respiratory syndrome (SARS) but may also trigger 55 other responses leading to multiorgan failure and death (5-7, 10-12). Several surface and showed no accumulation in the ER ( Fig. 2A) . Similarly, incubation 170 of the cells with the mannosidase-I inhibitor, KIF, had little or no effect on the 171 cell-surface localization of ACE2 (Fig. 2B) . These observations indicate that the the presence of hACE2* mutant on the cell surface was decreased compared 223 with wild type hACE2 (Fig. 4C) . Additionally, KIF treatment did not reduce the 224 cell surface expression of both hACE2 and pACE2 (Fig. 4D) , which is consistent 225 with the immunofluorescence data shown above (Fig. 2C ). In agreement with 226 the findings shown in the previous section, these data demonstrate that ACE2 227 N-deglycosylation results in a pronounced decrease of cell surface expression. To test the impact of N-glycosylation in the folding of ACE2 receptor, we 232 determined the carbopeptidase activity of both nonglycosylated hACE2 and 233 pACE2, generated by either TM treatment of cells or by mutagenesis. To 234 directly analyze the carbopeptidase activity of ACE2 proteins, we tested the 235 ability of immunoprecipitated ACE2 variants ( Fig. 5A and 5B) to hydrolyze a 236 synthetic peptide substrate. The ACE2 variants described in the western blot to 237 the right of each graph were incubated with a fluorophore-labeled substrate. 238

As shown in Fig. 5A and 5B, the constructs lacking glycosylation did not lose 239 carboxippetidase activity when compared to wild type. These results show that 240 N-glycosylation is not required for ACE2 protease activity and suggest that the 241 N-glycosylation-deficient ACE2 variants are in a native conformational state. 242 243

The ACE2 receptor is extensively glycosylated bearing high-mannose, hybrid, confirmed the interaction of both hACE2 and pACE2 with SARS-CoV-2 S 271 protein (Fig. 6B ). Next, we tested the ability of N-glycosylation-deficient ACE2 variants, generated either by TM treatment or by mutagenesis, to bind S by 273 performing similar methods. As shown in Fig. 7A showed association with S protein (Fig. 7C) . These results indicated that N-277 glycosylation is not required for the ability of ACE2 to bind SARS-CoV-2 S 278 protein. Interestingly, a nonglycosylated variant of S generated after TM 279 treatment, was able to interact with ACE2, suggesting that N-glycosylation of 280 SARS-CoV-2 S protein is not required for its capacity to bind ACE2 receptor 281 containing lysates were digested with PNGase F. As expected, both treatments 287 produced S protein bands of similar size (Fig. 7D) . Consistently, N-288 deglycosylation of both ACE2 and S protein by mutagenesis or TM treatment, 289 respectivally, did not disrupt their interaction (Fig. 7E ). Additional colocalization 290 data further confirmed the association between the N-glycosylation-deficient 291 ACE2 variants and the Spike protein ( Fig. 7F and 7G ). Finally, we explored the 292 ability of SARS-CoV S protein to bind nonglycosylated ACE2 variants. As 293 shown in Fig. S2A and S2D, N-glysosylation-defective ACE2 variants from 294 human and pig were capable to bind SARS-CoV S protein, which agrees with 295 a previous work that showed that modifications of the N-linked glycan structure 296 of hACE2 did not affect to its binding to the S protein (57). Colocalization results confirmed the S-ACE2 association ( Fig. S2B and S2F ). Moreover, a 298 nonglycosylated S variant, generated by TM treatment, was able to interact with 299 ACE2 receptors ( Fig. S2A and S2D ). Similar to SARS-CoV-2 S protein, the 300 expression of the nonglycosylated S variant was only detected in the pull-down 301 assay when using ACE2 as bait ( Figure S2A and S2D), suggesting that the 302 blocking of N-glycosylation might affect the stability or expression of the S 303 protein. In addition, digestion of untreated S-containing lysates with PNGase 304 F confirmed that TM treatment generated a N-glycosylation-defective S variant 305 ( Fig. S2C ). Consistently, the N-deglycosylation of both ACE2 and S protein by 306 mutagenesis or TM treatment of cells, respectivally, did not disrupt their 307 interaction (Fig. S2E) . Overall, these findings indicated that N-glycosylation is 308 not required for the ability of ACE2 to bind either SARS-CoV S or SARS-CoV-309 2 S protein. This observation is consistent with previous findings that demonstrated that 374 hACE2 is the receptor of SARS-CoV-2 (4, 16, 17, 64). In contrast, the viral entry 375 of SARS-CoV-2 S pseudotyped viruses in N-nonglycosylable ACE2-expressing 376 cells was reduced compared to hACE2-expressing cells ( Fig. 9B and 9C) , 377

suggesting that ACE2 needs to be N-glycosylated to support viral entry. 378

Similarly, viral entry of SARS-CoV S pseudotyped viruses was also reduced in 379 cells expressing a N-glycosylation-deficient variant ( Fig. 9B and 9C ). These 380 results indicated that hACE2 N-glycosylation is required to allow an efficient 381 viral entry of both SARS-CoV-2 and SASRS-CoV-2, which is in agreement with 382 our previous results that showed that N-glycosylation is a prerequisite for the 383 proper cell surface expression of hACE2. ACE2, (2) ACE2 N-glycosylation is not required for its carboxipeptidase activity, 394

(3) Association of SARS-CoV-2/SARS-CoV-2 S protein with ACE2 is not 395 disrupted by N-glycosylation inhibition of ACE2 receptors or the viral proteins, 396

(4), Impairment of hACE2 N-glycosylation affects cell-cell fusion mediated by SARS-CoV S protein but not the membrane fusion induced by SARS-CoV-2 S 398 protein, and (5), hACE2 N-glycosylation is indirectly required for an efficient 399 viral entry of the SARS-CoV/SARS-CoV-2 S pseudotyped viruses. 400

In agreement with previous studies (32-36), we verified, by digestion 401 with glycosidases and by treatment with inhibitors that interfere at different 402 stages of N-glycosylation biosynthesis, that hACE2 is N-linked glycosylated. 403

Digestion with Endo H confirmed a small presence of high mannose/hybrid 404 type glycans at hACE2 receptor. This finding is consistent with previous 405 glycomic and glycoproteomic analysis that found that complex-type glycans 406

were much more abundant than high mannose/hybrid type glycans across all 407 N-glycosylation sites of hACE2 (34, 35). By using the same methods, we also 408 found that pACE2, that showed four similar N-glycosyaltion sites with human 409 (Fig. 1A) , displays a similar N-glycosylation pattern as its homologous hACE2. previous reports that showed that the higher ability to mediate cell-cell fusion of 464 the SARS-CoV-2 S protein is likely due to the presence of a furin cleavage site 465 in its sequence (27, 59, 63). In agreement with this notion, insertion of the S1/S2 466 furin-cleavage site significantly potenciated the capacity of SARS-CoV S 467 protein to mediate cell-cell fusion but did not affect virion entry (78). In general, 468 the S1/S2 furin-recognition site is missing in most of β-B coronaviruses, and 469 their S proteins are uncleaved in normal conditions. For instance, in the case 470 of SARS-CoV, that mainly uses the endosomal membrane fusion pathway to activated (79-81). The specific role of the S1/S2 cleavage site in the viral life-473 cycle of SARS-CoV-2 remains to be further investigated. A recent study 474

showed that furin promotes both SARS-CoV-2 infectivity and cell-cell spread 475 but it is not absolutely essential for SARS-CoV-2 infection and replication 476 occurs in its absence, suggesting that furin-targeting drugs may reduce but not induced membrane fusion (57). It remains to be determinated whether the reduced fusogenic activity is due to the aberrant glycan structure of ACE2 or to 497 misfolding of the glycoprotein. 498

By analyzing the effect of blocking complex N-glycans formation of 499 hACE2 receptor, it has been proposed that ACE2 glycans may not regulate 500 viral entry of SARS-CoV-2 (38). In contrast to a full inhibition of N-glycosylation 501 biosynthesis, our data showed that inhibition of complex N-glycosylation (by 502 using KIF) did not alter the cell surface expression of ACE2. Moreover, a whole 503 N-glycosylation depletion of the ACE2 receptor did not disrupt the S-ACE2 504 interaction, which would explain why the ACE2 N-glycans do not have a role in 505 the viral entry of SARS-CoV-2. These obsevations support that idea that the 506 reduction of the viral entry in cells expressing a non-glycosylated ACE2 is 507 probably due to the lack of available cell-surface ACE2. However, the N-linked 508 sugars present on the S protein were critical for the virus to enter the host cells 509 since inhibition of complex N-glycan biosynthesis enhanced S protein 510 proteolysis, suggesting that N-glycosylation might play a role regulating SARS-511

CoV-2 S protein stability (38, 39). In agreement with these findings, we 512 observed that the expression of a nonglycosylated S variant was very low and 513 only was detected by Western blotting when it was coprecipiated with ACE2, 514

suggesting that the bloking of N-glycosylation might affect the stability or 515 expression of the S protein. Our data showed that the levels of wild-type hACE2 protein were higher 517 than the N-glycosylation-deficient ACE2 mutant, suggesting that N-518 glycosylation inhbition might affect the stability or expression of the ACE2 519 receptor. Since N-glycosylation plays an important role in protein stability by stability is affected by the presence of its N-linked glycan motifs. Cycloheximide 522 (CHX), an inhibitor of protein synthesis, was used to analyze the turnover rates 523 of the hACE2 variants. We found that after up to 10h of drug treatment, hACE2 524 and hACE* levels did not change, suggesting tha N-glycosylation is not 525 implicated in hACE2 stability and is not absolutely required for correct folding 526 (Fig. S3) . At this moment, our findings indicated that N-glycosylation is required the two fluorescent signals (See Fig. 2 and 3 ). Statistical analysis of the data 638 was performed by using Student's t test. p<0.05 was considered to be 639 statistically significant. 

World Health Organization coronavirus disease 776 2019 (COVID-19) situation report

A novel coronavirus from patients with pneumonia in China

A new coronavirus associated with human respiratory 782 disease in China

A pneumonia outbreak associated with a new 785 coronavirus of probable bat origin

Clinical and immunological features of severe and moderate 788 coronavirus disease 2019

Clinical features of patients infected with 2019 novel coronavirus in

Role of N-glycosylation in cell surface 990 expression and protection against proteolysis of the intestinal anion 991 exchanger SLC26A3

Identification and functional 995 characterisation of N-linked glycosylation of the orphan G protein-coupled 996 receptor Gpr176

Effect of N-glycosylation on ligand binding affinity of rat V1a 999 vasopressin receptor

N -linked glycosylation of 1002 platelet P2Y12 ADP receptor is essential for signal transduction but not for 1003 ligand binding or cell surface expression

N-glycosylation requirements for the AT1a angiotensin II 1006 receptor delivery to the plasma membrane

Structural and functional analysis 1010 of the canine histamine H2 receptor by site-directed mutagenesis: N-1011 glycosylation is not vital for its action

Structural basis of receptor recognition by SARS-CoV-2

Identification of sialic acid-1017 binding function for the Middle East respiratory syndrome coronavirus 1018 spike glycoprotein

Canine and feline parvoviruses 1021 preferentially recognize the non-human cell surface sialic acid N-1022 glycolylneuraminic acid

The sweet spot: defining virus-sialic acid interactions

Exploitation of 1027 glycosylation in enveloped virus pathobiology

Furin cleavage of the SARS coronavirus 1030 spike glycoprotein enhances cell-cell fusion but does not affect virion 1031 entry

Activation of the SARS 1033 coronavirus spike protein via sequential proteolytic cleavage at two 1034 distinct sites

SARS coronavirus, but not human 1038 coronavirus NL63, utilizes cathepsin L to infect ACE2-expressing cells

Inhibitors of cathepsin L prevent severe acute respiratory syndrome 1042 coronavirus entry

The Polybasic Cleavage Site in SARS

Effects of N-glycosylation on protein 1049 conformation and dynamics: Protein Data Bank analysis and molecular 1050 dynamics simulation study

Structure and function in 1056 rhodopsin: the role of asparagine-linked glycosylation

N-glycosylation and expression in human 1063 tissues of the orphan GPR61 receptor

Glycosylation of the 1065 mammalian alpha 1-adrenergic receptor by complex type N-linked 1066 oligosaccharides

The six N-linked 1079 carbohydrates of the lutropin/choriogonadotropin receptor are not 1080 absolutely required for correct folding, cell surface expression, hormone 1081 binding, or signal transduction

-conjugated goat anti-mouse IgG (red). ER was visualized by staining cells 1137 with calnexin antibody, followed by Alexa 488-conjugated goat anti-rabbit IgG 1138 (green)

Other cells were 1165 transfected with a S-expressing vector. Cells expressing ACE2 variants or S 1166 protein were lysed at 24h post-transfection. The cell lysates were mixed in a 1167 1:1 ratio, and incubated with anti-FLAG beads. To control for background 1168 binding of S protein to anti-FLAG beads, we performed similar experiments with 1169 HEK293T cells that were independently transfected with a S-expressing 1170 plasmid or an empty pCDNA3.1 vector. The amount of untagged and FLAG-1171 tagged proteins in the lysates (Input) and immunoprecipitates (IP) was 1172 analyzed by Western blotting with anti-S and anti-FLAG antibodies. WB, 1173 Western blot; IP, Immunoprecipitation. Similar results were obtained in three 1174 independent experiments and representative data is shown

After 24 h, cells were 1177 fixed and immunostained using anti-FLAG antibody followed by Alexa 488-1178 conjugated goat anti-rabbit IgG (green). The S protein was visualized using 1179 anti-S monoclonal antibody

Similar results were obtained in 1181 three separate experiments and representative data is shown

) 293T cells were transfected with a plasmid 1187 expressing SARS-CoV-2 S protein and treated with TM (1 μg/ml). 16 h post-1188 treatment, the S protein was precipitated by using ACE2 as a bait (lane 3), as 1189 indicated in Figure 6

1192 and G) Cells were co-transfected with ACE2-expressing plasmids and a S-1193 expressing vector. Cells were either untreated (+DMSO) or incubated with TM 1194 (1μg/ml) (+TM) dissolved in DMSO for 16h. All cells were fixed and 1195 immunostained using anti-FLAG antibody and then with Alexa 488-goat anti-1196 rabbit IgG (green). S protein was visualized by immunostaining it with anti-S 1197 monoclonal antibody, followed by Alexa 594-goat anti-mouse IgG (red)

Effect of hACE2 N-glycosylation on SARS-CoV S and SARS

CoV-2 S protein fusion activity 1203

HEK 293T effector cells were co-transfected with a GFP-expressing 1204 plasmid along with one of the following plasmids: a plasmid expressing SARS

At 24 h post transfection, cells were detached and co-1207 cultured with HEK293T target cells co-expressing either hACE2 or hACE2* and 1208 TMPRRSS2 for 24 h. Target cells transfected with an empty plasmid were 1209 included as negative control. Representative results are shown. A Western blot GAPDH was used as a loading control. (B) Quantitative representation of 1212 syncytia shown in (A). Values are means ± SD from one representative 1213 experiment performed in triplicate

Student's t-test) with respect to the negative controls and hACE2*. NS, not 1215 significant

Contribution of ACE2 N-glycosylation to viral entry of SARS

CoV/SARS-CoV-2 S pseudotyped viruses

A) HEK293T cells were transiently transfected with plasmids expressing 1220 hACE2 or hACE2* proteins, and their expression levels were analyzed by 1221 Western blotting using anti-FLAG and anti-GAPDH antibodies

Representative fluorescence images of HEK293T cells expressing wild-type 1223 hACE2 or hACE2* after infection with equalised amounts of GFP-expressing

SARS-CoV-2 or VSVG pseudotyped viruses. (C) The percentage 1225 of GFP-positive cells was measured at 72h post-infection by FACS

was used as a positive control. Values are means ± SD from one 1227 representative experiment performed in triplicate

as determined by two-tailed Student's t-test) with respect to wild-type 1229 hACE2 and Mock cells. NS, not significant