Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Site-specific glycan analysis of the SARS-CoV-2 spike

Site-specific glycan analysis of the SARS-CoV-2 spike RESEARCH CORONAVIRUS virions by budding into the lumen of endo- plasmic reticulum–Golgi intermediate com- Site-specific glycan analysis of the SARS-CoV-2 spike partments (15, 16). However, observations of complex-type glycans on virally derived mate- 1,2,3 1 4 4 1 Yasunori Watanabe *, Joel D. Allen *, Daniel Wrapp , Jason S. McLellan , Max Crispin † rial suggests that the viral glycoproteins are subjected to Golgi-resident processing en- The emergence of the betacoronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the zymes (13, 17). causative agent of coronavirus 2019 (COVID-19), represents a considerable threat to global human health. High viral glycan density and local protein Vaccine development is focused on the principal target of the humoral immune response, the spike (S) architecture can sterically impair the glycan glycoprotein, which mediates cell entry and membrane fusion. The SARS-CoV-2 S gene encodes 22 N-linked maturation pathway. Impaired glycan matura- glycan sequons per protomer, which likely play a role in protein folding and immune evasion. Here, tion resulting in the presence of oligomannose- using a site-specific mass spectrometric approach, we reveal the glycan structures on a recombinant type glycans can be a sensitive reporter of SARS-CoV-2 S immunogen. This analysis enables mapping of the glycan-processing states across the native-like protein architecture (8), and site- trimeric viral spike. We show how SARS-CoV-2 S glycans differ from typical host glycan processing, which specific glycan analysis can be used to compare may have implications in viral pathobiology and vaccine design. different immunogens and monitor manufac- turing processes (18). Additionally, glycosylation can influence the trafficking of recombinant evereacute respiratorysyndromecorona- spike is dominated by host-derived glycans, immunogen to germinal centers (19). virus 2 (SARS-CoV-2), the causative path- with each trimer displaying 66 N-linked glyco- To resolve the site-specific glycosylation of ogen of coronavirus 2019 (COVID-19) (1, 2), sylation sites. The S protein is a key target in the SARS-CoV-2 S protein and visualize the induces fever, severe respiratory illness, vaccine design efforts (6), and understanding distribution of glycoforms across the protein S and pneumonia. SARS-CoV-2 uses an the glycosylation of recombinant viral spikes surface, we expressed and purified three bio- extensively glycosylated spike (S) protein that can reveal fundamental features of viral biology logical replicates of recombinant soluble mate- protrudes from the viral surface to bind to and guide vaccine design strategies (7, 8). rial in an identical manner to that which was angiotensin-converting enzyme 2 (ACE2) to Viral glycosylation has wide-ranging roles used to obtain the high-resolution cryo–electron mediate host-cell entry (3). The S protein is a in viral pathobiology, including mediating pro- microscopy (cryo-EM) structure, albeit without trimeric class I fusion protein, composed of tein folding and stability and shaping viral a glycan-processing blockade using kifunensine two functional subunits, responsible for recep- tropism (9). Glycosylation sites are under selec- (4). This variant of the S protein contains all tor binding (S1 subunit) and membrane fusion tive pressureasthey facilitateimmuneevasion 22 glycans on the SARS-CoV-2 S protein (Fig. 1A). (S2 subunit) (4, 5). Thesurface of theenvelope by shielding specific epitopes from antibody Stabilization of the trimeric prefusion struc- neutralization. However, we note the low muta- ture was achieved by using the 2P stabilizing tion rate of SARS-CoV-2 and that as yet, there mutations (20)atresidues 986 and 987, a GSAS School of Biological Sciences, University of Southampton, Southampton SO17 1BJ, UK. Oxford Glycobiology Institute, have been no observed mutations to N-linked (Gly-Ser-Ala-Ser) substitution at the furin cleav- Department of Biochemistry, University of Oxford, South glycosylation sites (10). Surfaces with an un- age site (residues 682 to 685), and a C-terminal Parks Road, Oxford OX1 3QU, UK. Division of Structural usually high density of glycans can also enable trimerization motif. This helps to maintain Biology, University of Oxford, Wellcome Centre for Human Genetics, Oxford OX3 7BN, UK. Department of Molecular immune recognition (9, 11, 12). Theroleof quaternary architecture during glycan process- Biosciences, The University of Texas at Austin, Austin, glycosylation in camouflaging immunogenic ing. Before analysis, supernatant containing TX 78712, USA. protein epitopes has been studied for other the recombinant SARS-CoV-2 S was purified *These authors contributed equally to this work. †Corresponding author. Email: max.crispin@soton.ac.uk coronaviruses (10, 13, 14). Coronaviruses form by size exclusion chromatography to ensure Fig. 1. Expression and validation of the SARS-CoV-2 S glycoprotein. (A) Schematic representation of the SARS-CoV-2 S glycoprotein. The positions of N-linked glycosylation sequons (N-X-S/T, where X ≠ P) are shown as branches (N, Asn; X, any residue; S, Ser; T, Thr; P, Pro). Protein domains are illustrated: N-terminal domain (NTD), receptor binding domain (RBD), fusion peptide (FP), heptad repeat 1 (HR1), central helix (CH), connector domain (CD), and transmembrane domain (TM). (B) SDS–polyacrylamide gel electro- phoresis analysis of the SARS-CoV-2 S protein (indicated by the arrowhead) expressed in human embryonic kidney (HEK) 293F cells. Lane 1: filtered supernatant from transfected cells; lane 2: flow- through from StrepTactin resin; lane 3: wash from StrepTactin resin; lane 4: elution from StrepTactin resin. (C) Negative-stain EM 2D class averages of the SARS-CoV-2 S protein. 2D class averages of the SARS-CoV-2 S protein are shown, confirming that the protein adopts the trimeric prefusion conformation matching the material used to determine the structure (4). Watanabe et al., Science 369, 330–333 (2020) 17 July 2020 1of4 RESEARCH | REPORT that only native-like trimeric protein was ana- lyzed(Fig. 1B andfig.S1). The trimeric con- formation of the purified material was validated by using negative-stain EM (Fig. 1C). To determine the site-specific glycosylation of SARS-CoV-2 S, we used trypsin, chymotrypsin, and a-lytic protease to generate three glyco- peptide samples. These proteases were se- lected to generate glycopeptides that contain a single N-linked glycan sequon. The glycopep- tideswereanalyzedbyliquidchromatography– mass spectrometry, and the glycan composi- tions were determined for all 22 N-linked glycan sites (Fig. 2). To convey the main pro- cessing features at each site, the abundances of each glycan are summed into oligomannose- type, hybrid-type, and categories of complex- type glycosylation based on branching and fucosylation. The detailed, expanded graphs showing the diverse range of glycan compo- sitions are presented in table S1 and fig. S2. Two sites on SARS-CoV-2 S are principally oligomannose-type: N234 and N709. The pre- dominant oligomannose-type glycan structure observed across the protein, with the exception of N234, is Man GlcNAc (Man, mannose; 5 2 GlcNAc, N-acetylglucosamine), which demon- strates that these sites are largely accessible to a-1,2-mannosidases but are poor substrates for GlcNAcT-I, which is the gateway enzyme in the formation of hybrid- and complex-type glycans in the Golgi apparatus. The stage at which processing is impeded is a signature related to the density and presentation of gly- cans on the viral spike. For example, the more densely glycosylated spikes of HIV-1 Env and Lassa virus (LASV) GPC exhibit numerous sites dominated by Man GlcNAc (21–24). 9 2 A mixture of oligomannose- and complex- type glycans can be found at sites N61, N122, N603, N717, N801, and N1074 (Fig. 2). Of the 22 sites on the S protein, 8 contain substantial populations of oligomannose-type glycans, high- lighting how the processing of the SARS-CoV-2 S Fig. 2. Site-specific N-linked glycosylation of the SARS-CoV-2 S glycoprotein. The schematic illustrates glycans is divergent from host glycoproteins the color code for the principal glycan types that can arise along the maturation pathway from oligomannose- (25). The remaining 14 sites are dominated by to hybrid- to complex-type glycans. The graphs summarize quantitative mass spectrometric analysis of processed, complex-type glycans. the glycan population present at individual N-linked glycosylation sites simplified into categories of glycans. The Although unoccupied glycosylation sites oligomannose-type glycan series (M9 to M5; Man GlcNAc to Man GlcNAc ) is colored green, afucosylated were detected on SARS-CoV-2 S, when quan- 9 2 5 2 and fucosylated hybrid-type glycans (hybrid and F hybrid) are dashed pink, and complex glycans are grouped tified they were revealed to form a very minor according to the number of antennae and presence of core fucosylation (A1 to FA4) and are colored pink. component of the total peptide pool (table S2). Unoccupancy of an N-linked glycan site is represented in gray. The pie charts summarize the quantification of In HIV-1 immunogen research, the holes gen- these glycans. Glycan sites are colored according to oligomannose-type glycan content, with the glycan sites erated by unoccupied glycan sites have been labeled in green (80 to 100%), orange (30 to 79%), and pink (0 to 29%). An extended version of the site- shown to be immunogenic and potentially specific analysis showing the heterogeneity within each category can be found in table S1 and fig. S2. give rise to distracting epitopes (26). The high The bar graphs represent the mean quantities of three biological replicates, with error bars representing the occupancy of N-linked glycan sequons of SARS- standard error of the mean. CoV-2 S indicates that recombinant immuno- gens will not require further optimization to enhancesiteoccupancy. (3D) structure (Fig. 3). This combined mass tion sites (N165, N234, N343) can be observed, Using the cryo-EM structure of the trimeric spectrometric and cryo-EM analysis reveals how especially when the receptor binding domain SARS-CoV-2 S protein [Protein Data Bank (PDB) the N-linked glycans occlude distinct regions is in the “down” conformation. The shield- ID 6VSB] (4), we mapped the glycosylation across the surface of the SARS-CoV-2 spike. ing of receptor binding sites by glycans is a status of the coronavirus spike mimetic onto the Shielding of the receptor binding sites on common feature of viral glycoproteins, as ob- experimentally determined three-dimensional the SARS-CoV-2 spike by proximal glycosyla- served on SARS-CoV-1 S (10, 13), HIV-1 Env Watanabe et al., Science 369, 330–333 (2020) 17 July 2020 2of4 RESEARCH | REPORT (27), influenza hemagglutinin (28, 29), and glycans to camouflage one of the most conserved is in contrast to other viral glycoproteins; for LASV GPC (24). Given the functional con- and potentially vulnerable areas of their re- example, the dense glycan clusters in several straints of receptor binding sites and the re- spective glycoproteins (30, 31). strains of HIV-1 Env induce oligomannose- sulting low mutation rates of these residues, We note the dispersion of oligomannose-type type glycans that are recognized by antibodies there is likely selective pressure to use N-linked glycans across both the S1 and S2 subunits. This (32, 33). In SARS-CoV-2 S, the oligomannose- type structures arelikelyprotected by thepro- tein component, as exemplified by the N234 glycan, which is partially sandwiched between the N-terminal and receptor binding domains (Fig. 3). We characterized the N-linked glycans on extended flexible loop structures (N74 and N149) and at the membrane-proximal C ter- minus (N1158, N1173, N1194) that were not re- solved in the cryo-EM maps (4). These were determined to be complex-type glycans, con- sistent with steric accessibility of these residues. Whereas the oligomannose-type glycan con- tent (28%) (table S2) is above that observed on typical host glycoproteins, it is lower than other viral glycoproteins. For example, one of the most densely glycosylated viral spike proteins is HIV-1 Env, which exhibits ~60% oligomannose-type glycans (21, 34). This sug- gests that the SARS-CoV-2 S protein is less densely glycosylated and that the glycans form Fig. 3. Structure-based mapping of SARS-CoV-2 S N-linked glycans. Representative glycans are less of a shield compared with other viral glyco- modeled onto the prefusion structure of the trimeric SARS-CoV-2 S glycoprotein (PDB ID 6VSB) (4), with proteins, including HIV-1 Env and LASV GPC, one RBD in the “up” conformation and the other two RBDs in the “down” conformation. The glycans are which may be beneficial for the elicitation of colored according to oligomannose content as defined by the key. ACE2 receptor binding sites are highlighted neutralizing antibodies. in light blue. The S1 and S2 subunits are rendered with translucent surface representation, colored light Additionally, the processing of complex-type and dark gray, respectively. The flexible loops on which the N74 and N149 glycan sites reside are represented glycans is an important consideration in im- as gray dashed lines, with glycan sites on the loops mapped at their approximate regions. munogen engineering, especially considering Fig. 4. Underprocessing of viral glycan shields. From left to right, MERS-CoV S (10), SARS-CoV-1 S (10), SARS-CoV-2 S, LASV GPC (24), and HIV-1 Env (8, 21). Site- specific N-linked glycan oligomannose quantifications are colored according to the key. All glycoproteins were expressed as soluble trimers in HEK 293F cells apart from LASV GPC, which was derived from virus-like particles from Madin-Darby canine kidney II cells. Watanabe et al., Science 369, 330–333 (2020) 17 July 2020 3of 4 RESEARCH | REPORT that epitopes of neutralizing antibodies against nogen integrity and will also be important to 32. G. B. E. Stewart-Jones et al., Cell 165, 813–826 (2016). 33. D. Sok et al., Sci. Transl. Med. 6, 236ra63 (2014). SARS-CoV-2 S can contain fucosylated glycans monitor as manufacturing processes are scaled 34. L. Cao et al., Nat. Commun. 9, 3693 (2018). at N343 (35). Across the 22 N-linked glycosyl- for clinical use. Glycan profiling will therefore 35. D. Pinto et al., Nature 10.1038/s41586-020-2349-y ation sites, 52% are fucosylated and 15% of the also be an important measure of antigen qua- (2020). 36. L. K. Pritchard, D. J. Harvey, C. Bonomelli, M. Crispin, glycans contain at least one sialic acid residue lity in the manufacture of serological testing K. J. Doores, J. Virol. 89, 8932–8944 (2015). (table S2 and fig. S3). Our analysis reveals that kits. Last, with the advent of nucleotide-based 37. M. Crispin, SARS-CoV-2 spike site-specific N-linked glycan analysis. N343 is highly fucosylated with 98% of detected vaccines, it will be important to understand MassIVE Database (2020); https://doi.org/10.25345/C54X4K. glycans bearing fucose residues. Glycan mod- how those delivery mechanisms affect immu- ACKNOWLEDGMENTS ifications can be heavily influenced by the cellu- nogen processing and presentation. We thank M. Dixon and M. Gowland-Pryde for supporting our work lar expression system used. We have previously on this project during the difficulties arising from the pandemic demonstrated for HIV-1 Env glycosylation that and G. Ould for critical reading of the manuscript. Funding: This REFERENCES AND NOTES work was funded by the International AIDS Vaccine Initiative, the processing of complex-type glycans is driv- Bill and Melinda Gates Foundation through the Collaboration for 1. C. Huang et al., Lancet 395, 497–506 (2020). en by the producer cell but that the levels of AIDS Vaccine Discovery (OPP1084519 and 1196345 to M.C.), 2. X. Yang et al., Lancet Respir. Med. 10.1016/S2213-2600(20) oligomannose-type glycans were largely inde- the NIAID (R01-AI127521 to J.S.M.), and the Scripps Consortium for 30079-5 (2020). HIV Vaccine Development (CHAVD) (AI144462 to M.C.). M.C. is pendent of the expression system and are much 3. M. Letko, A. Marzi, V. Munster, Nat. Microbiol. 5,562–569 (2020). a Supernumerary Fellow of Oriel College, Oxford, and professor 4. D. Wrapp et al., Science 367, 1260–1263 (2020). more closely related to the protein structure adjunct at Scripps Research, CA. Author contributions: Y.W. and 5. A. C. Walls et al., Cell 181, 281–292.e6 (2020). and glycan density (36). J.D.A. performed mass spectrometry experiments and analyzed 6. F. Amanat, F. Krammer, Immunity 52, 583–589 (2020). data. Y.W. built glycosylated models. J.S.M. and M.C. supervised Highly dense glycan shields, such as those 7. L. Cao et al., Nat. Commun. 8, 14954 (2017). experiments. Y.W., J.D.A., and D.W. expressed and purified observed on LASV GPC and HIV-1 Env, feature 8. A.-J. Behrens et al., J. Virol. 91, e01894–e16 (2017). proteins. Y.W., J.D.A., and M.C. wrote the manuscript with input so-called mannose clusters (22, 24)onthe pro- 9. Y. Watanabe, T. A. Bowden, I. A. Wilson, M. Crispin, Biochim. from all authors. Competing interests: J.S.M. is an inventor Biophys. Acta 1863, 1480–1497 (2019). on U.S. patent application no. 62/412,703 (“Prefusion Coronavirus tein surface (Fig. 4). Whereas small mannose- 10. Y. Watanabe et al., Nat. Commun. 11, 2688 (2020). Spike Proteins and Their Use”), and D.W. and J.S.M. are inventors type clusters have been characterized on the S1 11. M. Dalziel, M. Crispin, C. N. Scanlan, N. Zitzmann, R. A. Dwek, on U.S. patent application no. 62/972,886 (“2019-nCoV Vaccine”). subunitofMiddleEastrespiratory syndrome Science 343, 1235681 (2014). Data and materials availability: Mass spectrometry raw files 12. C. N. Scanlan, J. Offer, N. Zitzmann, R. A. Dwek, Nature 446, (MERS)–CoV S (10), no such phenomenon has have been deposited in the MassIVE proteomics database (37). 1038–1045 (2007). The plasmid is available from J.S.M. under a material transfer been observed for the SARS-CoV-1 or SARS- 13. A. C. Walls et al., Cell 176, 1026–1039.e15 (2019). agreement with The University of Texas at Austin. This work is CoV-2 S proteins. The site-specific glycosyla- 14. T. J. Yang et al., Proc. Natl. Acad. Sci. U.S.A. 117,1438–1446 (2020). licensed under a Creative Commons Attribution 4.0 International 15. S. Stertz et al., Virology 361, 304–315 (2007). tion analysis reported here suggests that the (CC BY 4.0) license, which permits unrestricted use, distribution, 16. P. Venkatagopalan, S. M. Daskalova, L. A. Lopez, K. A. Dolezal, and reproduction in any medium, provided the original work is glycan shield of SARS-CoV-2 S is consistent B. G. Hogue, Virology 478,75–85 (2015). properly cited. To view a copy of this license, visit https:// with other coronaviruses and similarly exhibits 17. G. Ritchie et al., Virology 399, 257–269 (2010). creativecommons.org/licenses/by/4.0/. This license does not numerous vulnerabilities throughout the gly- 18. A. A. Hargett, M. B. Renfrow, Curr. Opin. Virol. 36,56–66 apply to figures, photos, artwork, or other content included in the (2019). article that is credited to a third party; obtain authorization from can shield (10). Last, we detected trace levels of 19. T. Tokatlian et al., Science 363, 649–654 (2019). 323 325 the rights holder before using such material. O-linked glycosylation at Thr /Ser (T323/ 20. J. Pallesen et al., Proc. Natl. Acad. Sci. U.S.A. 114, S325), with over 99% of these sites unmodified E7348–E7357 (2017). SUPPLEMENTARY MATERIALS 21. W. B. Struwe et al., Cell Rep. 24, 1958–1966.e5 (2018). (fig. S4),suggestingthatO-linkedglycosyla- science.sciencemag.org/content/369/6501/330/suppl/DC1 22. A.-J. Behrens et al., Cell Rep. 14, 2695–2706 (2016). tion of this region is minimal when the struc- Materials and Methods 23. M. Panico et al., Sci. Rep. 6, 32956 (2016). Figs. S1 to S4 ture is native-like. 24. Y. Watanabe et al., Proc. Natl. Acad. Sci. U.S.A. 115, 7320–7325 Tables S1 and S2 (2018). Our glycosylation analysis of SARS-CoV-2 References (38, 39) 25. I. Loke, D. Kolarich, N. H. Packer, M. Thaysen-Andersen, offers a detailed benchmark of site-specific MDAR Reproducibility Checklist Mol. Aspects Med. 51,31–55 (2016). Data S1 glycan signatures characteristic of a natively 26. M. Bianchi et al., Immunity 49, 288–300.e8 (2018). 27. J. Jardine et al., Science 340, 711–716 (2013). View/request a protocol for this paper from Bio-protocol. folded trimeric spike. As an increasing num- 28. C.-J. Wei et al., Sci. Transl. Med. 2, 24ra21 (2010). ber of glycoprotein-based vaccine candidates 29. R. Xu et al., Science 328, 357–360 (2010). 31 March 2020; accepted 29 April 2020 are being developed, their detailed glycan 30. X. Wei et al., Nature 422, 307–312 (2003). Published online 4 May 2020 analysis offers a route for comparing immu- 31. M. Zhang et al., Glycobiology 14, 1229–1246 (2004). 10.1126/science.abb9983 Watanabe et al., Science 369, 330–333 (2020) 17 July 2020 4of4 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Science (New York, N.y.) Pubmed Central

Site-specific glycan analysis of the SARS-CoV-2 spike

Science (New York, N.y.) , Volume 369 (6501) – May 4, 2020

Loading next page...
 
/lp/pubmed-central/site-specific-glycan-analysis-of-the-sars-cov-2-spike-RAg0ErG1F8

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Pubmed Central
Copyright
Copyright © 2020 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. Distributed under a Creative Commons Attribution License 4.0 (CC BY).
ISSN
0036-8075
eISSN
1095-9203
DOI
10.1126/science.abb9983
Publisher site
See Article on Publisher Site

Abstract

RESEARCH CORONAVIRUS virions by budding into the lumen of endo- plasmic reticulum–Golgi intermediate com- Site-specific glycan analysis of the SARS-CoV-2 spike partments (15, 16). However, observations of complex-type glycans on virally derived mate- 1,2,3 1 4 4 1 Yasunori Watanabe *, Joel D. Allen *, Daniel Wrapp , Jason S. McLellan , Max Crispin † rial suggests that the viral glycoproteins are subjected to Golgi-resident processing en- The emergence of the betacoronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the zymes (13, 17). causative agent of coronavirus 2019 (COVID-19), represents a considerable threat to global human health. High viral glycan density and local protein Vaccine development is focused on the principal target of the humoral immune response, the spike (S) architecture can sterically impair the glycan glycoprotein, which mediates cell entry and membrane fusion. The SARS-CoV-2 S gene encodes 22 N-linked maturation pathway. Impaired glycan matura- glycan sequons per protomer, which likely play a role in protein folding and immune evasion. Here, tion resulting in the presence of oligomannose- using a site-specific mass spectrometric approach, we reveal the glycan structures on a recombinant type glycans can be a sensitive reporter of SARS-CoV-2 S immunogen. This analysis enables mapping of the glycan-processing states across the native-like protein architecture (8), and site- trimeric viral spike. We show how SARS-CoV-2 S glycans differ from typical host glycan processing, which specific glycan analysis can be used to compare may have implications in viral pathobiology and vaccine design. different immunogens and monitor manufac- turing processes (18). Additionally, glycosylation can influence the trafficking of recombinant evereacute respiratorysyndromecorona- spike is dominated by host-derived glycans, immunogen to germinal centers (19). virus 2 (SARS-CoV-2), the causative path- with each trimer displaying 66 N-linked glyco- To resolve the site-specific glycosylation of ogen of coronavirus 2019 (COVID-19) (1, 2), sylation sites. The S protein is a key target in the SARS-CoV-2 S protein and visualize the induces fever, severe respiratory illness, vaccine design efforts (6), and understanding distribution of glycoforms across the protein S and pneumonia. SARS-CoV-2 uses an the glycosylation of recombinant viral spikes surface, we expressed and purified three bio- extensively glycosylated spike (S) protein that can reveal fundamental features of viral biology logical replicates of recombinant soluble mate- protrudes from the viral surface to bind to and guide vaccine design strategies (7, 8). rial in an identical manner to that which was angiotensin-converting enzyme 2 (ACE2) to Viral glycosylation has wide-ranging roles used to obtain the high-resolution cryo–electron mediate host-cell entry (3). The S protein is a in viral pathobiology, including mediating pro- microscopy (cryo-EM) structure, albeit without trimeric class I fusion protein, composed of tein folding and stability and shaping viral a glycan-processing blockade using kifunensine two functional subunits, responsible for recep- tropism (9). Glycosylation sites are under selec- (4). This variant of the S protein contains all tor binding (S1 subunit) and membrane fusion tive pressureasthey facilitateimmuneevasion 22 glycans on the SARS-CoV-2 S protein (Fig. 1A). (S2 subunit) (4, 5). Thesurface of theenvelope by shielding specific epitopes from antibody Stabilization of the trimeric prefusion struc- neutralization. However, we note the low muta- ture was achieved by using the 2P stabilizing tion rate of SARS-CoV-2 and that as yet, there mutations (20)atresidues 986 and 987, a GSAS School of Biological Sciences, University of Southampton, Southampton SO17 1BJ, UK. Oxford Glycobiology Institute, have been no observed mutations to N-linked (Gly-Ser-Ala-Ser) substitution at the furin cleav- Department of Biochemistry, University of Oxford, South glycosylation sites (10). Surfaces with an un- age site (residues 682 to 685), and a C-terminal Parks Road, Oxford OX1 3QU, UK. Division of Structural usually high density of glycans can also enable trimerization motif. This helps to maintain Biology, University of Oxford, Wellcome Centre for Human Genetics, Oxford OX3 7BN, UK. Department of Molecular immune recognition (9, 11, 12). Theroleof quaternary architecture during glycan process- Biosciences, The University of Texas at Austin, Austin, glycosylation in camouflaging immunogenic ing. Before analysis, supernatant containing TX 78712, USA. protein epitopes has been studied for other the recombinant SARS-CoV-2 S was purified *These authors contributed equally to this work. †Corresponding author. Email: max.crispin@soton.ac.uk coronaviruses (10, 13, 14). Coronaviruses form by size exclusion chromatography to ensure Fig. 1. Expression and validation of the SARS-CoV-2 S glycoprotein. (A) Schematic representation of the SARS-CoV-2 S glycoprotein. The positions of N-linked glycosylation sequons (N-X-S/T, where X ≠ P) are shown as branches (N, Asn; X, any residue; S, Ser; T, Thr; P, Pro). Protein domains are illustrated: N-terminal domain (NTD), receptor binding domain (RBD), fusion peptide (FP), heptad repeat 1 (HR1), central helix (CH), connector domain (CD), and transmembrane domain (TM). (B) SDS–polyacrylamide gel electro- phoresis analysis of the SARS-CoV-2 S protein (indicated by the arrowhead) expressed in human embryonic kidney (HEK) 293F cells. Lane 1: filtered supernatant from transfected cells; lane 2: flow- through from StrepTactin resin; lane 3: wash from StrepTactin resin; lane 4: elution from StrepTactin resin. (C) Negative-stain EM 2D class averages of the SARS-CoV-2 S protein. 2D class averages of the SARS-CoV-2 S protein are shown, confirming that the protein adopts the trimeric prefusion conformation matching the material used to determine the structure (4). Watanabe et al., Science 369, 330–333 (2020) 17 July 2020 1of4 RESEARCH | REPORT that only native-like trimeric protein was ana- lyzed(Fig. 1B andfig.S1). The trimeric con- formation of the purified material was validated by using negative-stain EM (Fig. 1C). To determine the site-specific glycosylation of SARS-CoV-2 S, we used trypsin, chymotrypsin, and a-lytic protease to generate three glyco- peptide samples. These proteases were se- lected to generate glycopeptides that contain a single N-linked glycan sequon. The glycopep- tideswereanalyzedbyliquidchromatography– mass spectrometry, and the glycan composi- tions were determined for all 22 N-linked glycan sites (Fig. 2). To convey the main pro- cessing features at each site, the abundances of each glycan are summed into oligomannose- type, hybrid-type, and categories of complex- type glycosylation based on branching and fucosylation. The detailed, expanded graphs showing the diverse range of glycan compo- sitions are presented in table S1 and fig. S2. Two sites on SARS-CoV-2 S are principally oligomannose-type: N234 and N709. The pre- dominant oligomannose-type glycan structure observed across the protein, with the exception of N234, is Man GlcNAc (Man, mannose; 5 2 GlcNAc, N-acetylglucosamine), which demon- strates that these sites are largely accessible to a-1,2-mannosidases but are poor substrates for GlcNAcT-I, which is the gateway enzyme in the formation of hybrid- and complex-type glycans in the Golgi apparatus. The stage at which processing is impeded is a signature related to the density and presentation of gly- cans on the viral spike. For example, the more densely glycosylated spikes of HIV-1 Env and Lassa virus (LASV) GPC exhibit numerous sites dominated by Man GlcNAc (21–24). 9 2 A mixture of oligomannose- and complex- type glycans can be found at sites N61, N122, N603, N717, N801, and N1074 (Fig. 2). Of the 22 sites on the S protein, 8 contain substantial populations of oligomannose-type glycans, high- lighting how the processing of the SARS-CoV-2 S Fig. 2. Site-specific N-linked glycosylation of the SARS-CoV-2 S glycoprotein. The schematic illustrates glycans is divergent from host glycoproteins the color code for the principal glycan types that can arise along the maturation pathway from oligomannose- (25). The remaining 14 sites are dominated by to hybrid- to complex-type glycans. The graphs summarize quantitative mass spectrometric analysis of processed, complex-type glycans. the glycan population present at individual N-linked glycosylation sites simplified into categories of glycans. The Although unoccupied glycosylation sites oligomannose-type glycan series (M9 to M5; Man GlcNAc to Man GlcNAc ) is colored green, afucosylated were detected on SARS-CoV-2 S, when quan- 9 2 5 2 and fucosylated hybrid-type glycans (hybrid and F hybrid) are dashed pink, and complex glycans are grouped tified they were revealed to form a very minor according to the number of antennae and presence of core fucosylation (A1 to FA4) and are colored pink. component of the total peptide pool (table S2). Unoccupancy of an N-linked glycan site is represented in gray. The pie charts summarize the quantification of In HIV-1 immunogen research, the holes gen- these glycans. Glycan sites are colored according to oligomannose-type glycan content, with the glycan sites erated by unoccupied glycan sites have been labeled in green (80 to 100%), orange (30 to 79%), and pink (0 to 29%). An extended version of the site- shown to be immunogenic and potentially specific analysis showing the heterogeneity within each category can be found in table S1 and fig. S2. give rise to distracting epitopes (26). The high The bar graphs represent the mean quantities of three biological replicates, with error bars representing the occupancy of N-linked glycan sequons of SARS- standard error of the mean. CoV-2 S indicates that recombinant immuno- gens will not require further optimization to enhancesiteoccupancy. (3D) structure (Fig. 3). This combined mass tion sites (N165, N234, N343) can be observed, Using the cryo-EM structure of the trimeric spectrometric and cryo-EM analysis reveals how especially when the receptor binding domain SARS-CoV-2 S protein [Protein Data Bank (PDB) the N-linked glycans occlude distinct regions is in the “down” conformation. The shield- ID 6VSB] (4), we mapped the glycosylation across the surface of the SARS-CoV-2 spike. ing of receptor binding sites by glycans is a status of the coronavirus spike mimetic onto the Shielding of the receptor binding sites on common feature of viral glycoproteins, as ob- experimentally determined three-dimensional the SARS-CoV-2 spike by proximal glycosyla- served on SARS-CoV-1 S (10, 13), HIV-1 Env Watanabe et al., Science 369, 330–333 (2020) 17 July 2020 2of4 RESEARCH | REPORT (27), influenza hemagglutinin (28, 29), and glycans to camouflage one of the most conserved is in contrast to other viral glycoproteins; for LASV GPC (24). Given the functional con- and potentially vulnerable areas of their re- example, the dense glycan clusters in several straints of receptor binding sites and the re- spective glycoproteins (30, 31). strains of HIV-1 Env induce oligomannose- sulting low mutation rates of these residues, We note the dispersion of oligomannose-type type glycans that are recognized by antibodies there is likely selective pressure to use N-linked glycans across both the S1 and S2 subunits. This (32, 33). In SARS-CoV-2 S, the oligomannose- type structures arelikelyprotected by thepro- tein component, as exemplified by the N234 glycan, which is partially sandwiched between the N-terminal and receptor binding domains (Fig. 3). We characterized the N-linked glycans on extended flexible loop structures (N74 and N149) and at the membrane-proximal C ter- minus (N1158, N1173, N1194) that were not re- solved in the cryo-EM maps (4). These were determined to be complex-type glycans, con- sistent with steric accessibility of these residues. Whereas the oligomannose-type glycan con- tent (28%) (table S2) is above that observed on typical host glycoproteins, it is lower than other viral glycoproteins. For example, one of the most densely glycosylated viral spike proteins is HIV-1 Env, which exhibits ~60% oligomannose-type glycans (21, 34). This sug- gests that the SARS-CoV-2 S protein is less densely glycosylated and that the glycans form Fig. 3. Structure-based mapping of SARS-CoV-2 S N-linked glycans. Representative glycans are less of a shield compared with other viral glyco- modeled onto the prefusion structure of the trimeric SARS-CoV-2 S glycoprotein (PDB ID 6VSB) (4), with proteins, including HIV-1 Env and LASV GPC, one RBD in the “up” conformation and the other two RBDs in the “down” conformation. The glycans are which may be beneficial for the elicitation of colored according to oligomannose content as defined by the key. ACE2 receptor binding sites are highlighted neutralizing antibodies. in light blue. The S1 and S2 subunits are rendered with translucent surface representation, colored light Additionally, the processing of complex-type and dark gray, respectively. The flexible loops on which the N74 and N149 glycan sites reside are represented glycans is an important consideration in im- as gray dashed lines, with glycan sites on the loops mapped at their approximate regions. munogen engineering, especially considering Fig. 4. Underprocessing of viral glycan shields. From left to right, MERS-CoV S (10), SARS-CoV-1 S (10), SARS-CoV-2 S, LASV GPC (24), and HIV-1 Env (8, 21). Site- specific N-linked glycan oligomannose quantifications are colored according to the key. All glycoproteins were expressed as soluble trimers in HEK 293F cells apart from LASV GPC, which was derived from virus-like particles from Madin-Darby canine kidney II cells. Watanabe et al., Science 369, 330–333 (2020) 17 July 2020 3of 4 RESEARCH | REPORT that epitopes of neutralizing antibodies against nogen integrity and will also be important to 32. G. B. E. Stewart-Jones et al., Cell 165, 813–826 (2016). 33. D. Sok et al., Sci. Transl. Med. 6, 236ra63 (2014). SARS-CoV-2 S can contain fucosylated glycans monitor as manufacturing processes are scaled 34. L. Cao et al., Nat. Commun. 9, 3693 (2018). at N343 (35). Across the 22 N-linked glycosyl- for clinical use. Glycan profiling will therefore 35. D. Pinto et al., Nature 10.1038/s41586-020-2349-y ation sites, 52% are fucosylated and 15% of the also be an important measure of antigen qua- (2020). 36. L. K. Pritchard, D. J. Harvey, C. Bonomelli, M. Crispin, glycans contain at least one sialic acid residue lity in the manufacture of serological testing K. J. Doores, J. Virol. 89, 8932–8944 (2015). (table S2 and fig. S3). Our analysis reveals that kits. Last, with the advent of nucleotide-based 37. M. Crispin, SARS-CoV-2 spike site-specific N-linked glycan analysis. N343 is highly fucosylated with 98% of detected vaccines, it will be important to understand MassIVE Database (2020); https://doi.org/10.25345/C54X4K. glycans bearing fucose residues. Glycan mod- how those delivery mechanisms affect immu- ACKNOWLEDGMENTS ifications can be heavily influenced by the cellu- nogen processing and presentation. We thank M. Dixon and M. Gowland-Pryde for supporting our work lar expression system used. We have previously on this project during the difficulties arising from the pandemic demonstrated for HIV-1 Env glycosylation that and G. Ould for critical reading of the manuscript. Funding: This REFERENCES AND NOTES work was funded by the International AIDS Vaccine Initiative, the processing of complex-type glycans is driv- Bill and Melinda Gates Foundation through the Collaboration for 1. C. Huang et al., Lancet 395, 497–506 (2020). en by the producer cell but that the levels of AIDS Vaccine Discovery (OPP1084519 and 1196345 to M.C.), 2. X. Yang et al., Lancet Respir. Med. 10.1016/S2213-2600(20) oligomannose-type glycans were largely inde- the NIAID (R01-AI127521 to J.S.M.), and the Scripps Consortium for 30079-5 (2020). HIV Vaccine Development (CHAVD) (AI144462 to M.C.). M.C. is pendent of the expression system and are much 3. M. Letko, A. Marzi, V. Munster, Nat. Microbiol. 5,562–569 (2020). a Supernumerary Fellow of Oriel College, Oxford, and professor 4. D. Wrapp et al., Science 367, 1260–1263 (2020). more closely related to the protein structure adjunct at Scripps Research, CA. Author contributions: Y.W. and 5. A. C. Walls et al., Cell 181, 281–292.e6 (2020). and glycan density (36). J.D.A. performed mass spectrometry experiments and analyzed 6. F. Amanat, F. Krammer, Immunity 52, 583–589 (2020). data. Y.W. built glycosylated models. J.S.M. and M.C. supervised Highly dense glycan shields, such as those 7. L. Cao et al., Nat. Commun. 8, 14954 (2017). experiments. Y.W., J.D.A., and D.W. expressed and purified observed on LASV GPC and HIV-1 Env, feature 8. A.-J. Behrens et al., J. Virol. 91, e01894–e16 (2017). proteins. Y.W., J.D.A., and M.C. wrote the manuscript with input so-called mannose clusters (22, 24)onthe pro- 9. Y. Watanabe, T. A. Bowden, I. A. Wilson, M. Crispin, Biochim. from all authors. Competing interests: J.S.M. is an inventor Biophys. Acta 1863, 1480–1497 (2019). on U.S. patent application no. 62/412,703 (“Prefusion Coronavirus tein surface (Fig. 4). Whereas small mannose- 10. Y. Watanabe et al., Nat. Commun. 11, 2688 (2020). Spike Proteins and Their Use”), and D.W. and J.S.M. are inventors type clusters have been characterized on the S1 11. M. Dalziel, M. Crispin, C. N. Scanlan, N. Zitzmann, R. A. Dwek, on U.S. patent application no. 62/972,886 (“2019-nCoV Vaccine”). subunitofMiddleEastrespiratory syndrome Science 343, 1235681 (2014). Data and materials availability: Mass spectrometry raw files 12. C. N. Scanlan, J. Offer, N. Zitzmann, R. A. Dwek, Nature 446, (MERS)–CoV S (10), no such phenomenon has have been deposited in the MassIVE proteomics database (37). 1038–1045 (2007). The plasmid is available from J.S.M. under a material transfer been observed for the SARS-CoV-1 or SARS- 13. A. C. Walls et al., Cell 176, 1026–1039.e15 (2019). agreement with The University of Texas at Austin. This work is CoV-2 S proteins. The site-specific glycosyla- 14. T. J. Yang et al., Proc. Natl. Acad. Sci. U.S.A. 117,1438–1446 (2020). licensed under a Creative Commons Attribution 4.0 International 15. S. Stertz et al., Virology 361, 304–315 (2007). tion analysis reported here suggests that the (CC BY 4.0) license, which permits unrestricted use, distribution, 16. P. Venkatagopalan, S. M. Daskalova, L. A. Lopez, K. A. Dolezal, and reproduction in any medium, provided the original work is glycan shield of SARS-CoV-2 S is consistent B. G. Hogue, Virology 478,75–85 (2015). properly cited. To view a copy of this license, visit https:// with other coronaviruses and similarly exhibits 17. G. Ritchie et al., Virology 399, 257–269 (2010). creativecommons.org/licenses/by/4.0/. This license does not numerous vulnerabilities throughout the gly- 18. A. A. Hargett, M. B. Renfrow, Curr. Opin. Virol. 36,56–66 apply to figures, photos, artwork, or other content included in the (2019). article that is credited to a third party; obtain authorization from can shield (10). Last, we detected trace levels of 19. T. Tokatlian et al., Science 363, 649–654 (2019). 323 325 the rights holder before using such material. O-linked glycosylation at Thr /Ser (T323/ 20. J. Pallesen et al., Proc. Natl. Acad. Sci. U.S.A. 114, S325), with over 99% of these sites unmodified E7348–E7357 (2017). SUPPLEMENTARY MATERIALS 21. W. B. Struwe et al., Cell Rep. 24, 1958–1966.e5 (2018). (fig. S4),suggestingthatO-linkedglycosyla- science.sciencemag.org/content/369/6501/330/suppl/DC1 22. A.-J. Behrens et al., Cell Rep. 14, 2695–2706 (2016). tion of this region is minimal when the struc- Materials and Methods 23. M. Panico et al., Sci. Rep. 6, 32956 (2016). Figs. S1 to S4 ture is native-like. 24. Y. Watanabe et al., Proc. Natl. Acad. Sci. U.S.A. 115, 7320–7325 Tables S1 and S2 (2018). Our glycosylation analysis of SARS-CoV-2 References (38, 39) 25. I. Loke, D. Kolarich, N. H. Packer, M. Thaysen-Andersen, offers a detailed benchmark of site-specific MDAR Reproducibility Checklist Mol. Aspects Med. 51,31–55 (2016). Data S1 glycan signatures characteristic of a natively 26. M. Bianchi et al., Immunity 49, 288–300.e8 (2018). 27. J. Jardine et al., Science 340, 711–716 (2013). View/request a protocol for this paper from Bio-protocol. folded trimeric spike. As an increasing num- 28. C.-J. Wei et al., Sci. Transl. Med. 2, 24ra21 (2010). ber of glycoprotein-based vaccine candidates 29. R. Xu et al., Science 328, 357–360 (2010). 31 March 2020; accepted 29 April 2020 are being developed, their detailed glycan 30. X. Wei et al., Nature 422, 307–312 (2003). Published online 4 May 2020 analysis offers a route for comparing immu- 31. M. Zhang et al., Glycobiology 14, 1229–1246 (2004). 10.1126/science.abb9983 Watanabe et al., Science 369, 330–333 (2020) 17 July 2020 4of4

Journal

Science (New York, N.y.)Pubmed Central

Published: May 4, 2020

References