Unexpected Receptor Functional Mimicry Elucidates Activation of Coronavirus Fusion.

Protein expression and purification

The SARS-CoV S 2P and MERS-CoV S 2P ectodomains were produced in 500mL HEK293F cells grown in suspension using FreeStyle 293 expression medium (Life technologies) at 37°C in a humidified 8% CO2 incubator rotating at 130 rpm The cultures were transfected using 293fectin (ThermoFisher Scientific) with cells grown to a density of 1 million cells per mL and cultivated for three days. The supernatants were harvested and cells resuspended for another three days, yielding two harvests. Clarified supernatants were purified using a 5mL Cobalt affinity column (Takara). Purified protein was filtered or concentrated and flash frozen in Tris-saline (50 mM Tris pH 8.0, 150 mM NaCl) prior to negative staining and cryo-EM analysis. Wild-type SARS-CoV S was expressed in 293F cells and purified the same way as the SARS-CoV S 2P. Wild-type SARS-CoV S was filtered through 0.2 μm filter and the quality of the protein was assessed by negative stain electron microscopy and SDS-PAGE. Wild-type SARS-CoV S was not frozen and was stored at 4°C and used within 7 days of purification.

Expression and purification of human angiotensin-converting enzyme ectodomain (ACE2, residues 1–614) fused to the Fc region of human IgG (hFc) was performed following the protocol previously described for CEACAM1a(^{Walls et al., 2016a}) and cleavage of the Fc fragment was achieved using trypsin or thrombin. SARS-CoV S₁-Fc was produced in 250mL HEK293F cells grown in suspension using FreeStyle 293 expression medium (Life technologies) at 37°C in a humidified 8% CO2 incubator rotating at 130 rpm The cultures were transfected using 293fectin (ThermoFisher Scientific) with cells grown to a density of 1 million cells per mL and cultivated for six days. The supernatant was harvested, clarified and incubated with 0.25mL protein A beads (0.5 mL slurry) (GenScript) overnight. Beads were washed thoroughly with phosphate buffered saline prior to elution with 0.1M glycine pH3.0 which was immediately neutralized with 1M Tris pH8.5. LCA60 and S230 Fabs were expressed using transient transfection of Expi293F cells, purified by affinity chromatography using CaptureSelect CH1-XL column (ThermoFisher), buffer exchanged to phosphate buffer saline using a HiTrap fast desalting column (GE Healthcare) and sterilized through a 0.2 μm filter.

Pseudovirus production

MLV-based SARS-CoV S-pseudotyped were prepared as previously described (^{Millet and Whittaker, 2016}). HEK293T cells were cotransfected with a SARS-CoV S encoding-plasmid, a MLV Gag-Pol packaging construct and the MLV transfer vector encoding a luciferase reporter using the Lipofectamine 2000 transfection reagent (Life Technologies) according to the manufacturer’s instructions. Cells were incubated for 5 hours at 37°C with 5% CO₂ with transfection medium. Cells were then washed with DMEM 2X and DMEM containing 10% FBS was added for 12 hours. The cells were washed again 2X with DMEM and were kept in DMEM without FBS for 48 hours (^{Wang et al., 2014}). The supernatants were then harvested and filtered through 0.45-μm membranes.

MERS-CoV production

Virions were purified from the supernatant of infected Vero E6 cells. Cells were infected at 90% confluence at multiplicity of infection of 0.01, in Serum Free Medium (VP-SFM; ThermoFisher Scientific). Cultures were maintained at 37°C, 5% (v/v) CO₂ until 3 days post infection when the culture supernatant was removed. Approximately 50% of cells were rounding, but most remained attached to the growth surface. For virion purification, culture supernatant was clarified by centrifugation at 1,400 g for 10 min then overlaid on a cushion of 20% (w/w) Sucrose in Hanks Buffered Saline solution (HBSS: ThermoFisher Scientific).Supernatant virions were centrifuged through the cushion at 100,000 g for 90 min at 4°C. Pelleted virions were resuspended in HBSS and disrupted by heating for 10 min at 95°C in 1% (w/v) sodium dodecyl sulfate. All live virus work was undertaken under UK ACDP Containment level 3 conditions.

MERS-CoV S/LCA60 complex formation

Flash frozen MERS-CoV 2P stabilized S at 5mg/mL was incubated in a 1:1.5 molar ratio with LCA60 at 2.3 mg/mL on ice for three hours. Trypsin at 3 μg/mL was added for twenty minutes at 37°C prior to injection over a gel filtration column (Superose 6 10/300 from GE Life Sciences) equilibrated in 20mM Tris pH8.0 and 150mM NaCl. The complex peak fractions were concentrated to 0.5 mg/mL. 3 μL was applied to a glow discharged (20 s at 20 mA) lacey carbon copper grid with a thin layer of evaporated continuous carbon prior to plunge freezing into liquid ethane with a Mark IV vitrobot (ThermoFisher Scientific) using −1 blot force and 2 s blot time at 95% humidity and 25°C.

SARS-CoV S/S230 complex formation

Lacey carbon copper grids were coated with a thin-layer of continuous carbon using a carbon evaporator and glow discharged for 20 s at 20 mA. Flash frozen SARS-CoV 2P stabilized S was diluted to 0.05 mg/mL and 2 μl was applied to the grid and incubated for 1 min. Subsequently, the grid was washed three times in Tris-saline buffer before 2 μL of S230 Fab at 0.02 mg/mL was applied to the grid and incubated for five minutes. The grid was washed again three times with Tris-saline after the Fab incubation and manually blotted prior to plunge freezing into liquid ethane at room temperature.

Cryo-EM data collection

Data were acquired using the Leginon software (^{Suloway et al., 2005}) to control an FEI Titan Krios transmission electron microscope operated at 300 kV and equipped with a Gatan K2 Summit direct detector and Gatan Quantum GIF energy filter, operated in zero-loss mode with a slit width of 20 eV. Automated data collection was carried out using Leginon at a nominal magnification of 105,000x with a pixel size of 0.685 Å. The dose rate was adjusted to 8 counts/pixel/s, and each movie was acquired in super-resolution mode fractionated in 50 frames of 200 ms. For MERS-CoV S/LCA60, 2,400 micrographs were collected in a single session with a defocus range comprised between 0.6 and 3.0 μm. For SARS-CoV S/S230, 5,900 images were collected in a single session with a defocus range comprised between 0.8 and 3.0 μm.

CryoEM data processing of MERSCoV S/LCA60

Movie frame alignment was carried out with MotionCorr2 (^{Zheng et al., 2017}) using a dose weighting of 40 e^-/Ų. The parameters of the microscope contrast-transfer function were estimated with GCTF (^{Zhang, 2016}). Particles were automatically picked with DoGPicker (^{Voss et al., 2009}) through the Appion interface (^{Lander et al., 2009}). Particle images were extracted and processed with Relion3.0 (^{Kimanius et al., 2016}) with a box size of 768 pixels² binned to 384 pixels² yielding a pixel size of 1.37Å. Reference free 2D classification was used to parse 240,000 particles from the original 500,000 particle images using Relion3.0. An initial model was generated from PDBID 5W9J with all MERS-CoV S domain B closed and removing the G4 Fab (^{Pallesen et al., 2017}). Relion3.0 3D classification with C1 symmetry was utilized to select the ∼126,000 best particles for state 1 and the ∼114,000 best particles for state 2. CTF refinement in Relion3.0 was used to refine per-particle defocus values. Particles images from each state were subjected to the Bayesian polishing procedure implemented in Relion3.0 (^{Zivanov et al., 2018, Zivanov et al., 2019}) and 3D refinement before performing 3D classification, without refining angle and shifts, using soft masks focusing on the B domains and Fabs with a tau factor of 50. 82,000 and 78,000 particles were selected for states 1 and 2, respectively. Particle images from state 1 underwent another round of domain B/LCA60 focused classification to select 54,000 particles used for the final reconstruction. Finally, CryoSPARC non-uniform refinement (^{Punjani et al., 2017}) was used to obtain the final reconstructions at 3.5 and 3.6 Å resolution, respectively, using soft masks excluding the open B domains and bound Fabs. Reported resolutions are based on the gold-standard FSC = 0.143 criterion (^{Rosenthal and Henderson, 2003, Scheres and Chen, 2012}) and Fourier shell correlation curves were corrected for the effects of soft masking by high-resolution noise substitution(^{Chen et al., 2013}). Local resolution estimation, filtering and sharpening was carried out using CryoSPARC. Analysis was carried out using the state 1 structure since the reconstruction is best resolved.

CryoEM data processing of SARS-CoV S/S230

Movie frame alignment was carried out with MotionCorr2 (^{Zheng et al., 2017}) using a dose weighting of 40 e^-/Ų. The parameters of the microscope contrast-transfer function were estimated with GCTF (^{Zhang, 2016}). Particles were automatically picked with DoGPicker (^{Voss et al., 2009}) through the Appion interface (^{Lander et al., 2009}). Particle images were extracted and processed with Relion3.0 (^{Kimanius et al., 2016}) with a box size of 768 pixels² binned to 72 pixels² (for 2D classification), 128 pixels² (for 3D classification) or 384 pixels² (for 3D refinement, yielding a pixel size of 1.37Å). Two rounds of Relion reference-free 2D classification were used to select ∼350,000 particles clustering in well-defined class averages. An initial model was generated from a SARS-CoV S structure with all B domains closed (PDB: 5X58) (^{Yuan et al., 2017}). Relion 3D classification without applying symmetry was utilized to select ∼162,000 particle images for state 1 (including numerous conformations of the two bound Fabs as well as partial occupancy) and ∼126,000 particle images for state 2 (including numerous conformations of the three bound Fabs as well as partial occupancy). Particle images in each subset were then subsequently subjected to a new round of 3D classification without imposing symmetry and using the same initial model as for the previous round. A subset of ∼65,000 particle images from state 1 and ∼71,000 particle images from state 2 were selected based on reduced conformational heterogeneity of the bound Fabs (which exhibit a continuum of conformational states) and independently used to run 3D auto-refinements using Relion. Particle images from each state were subjected to the Bayesian polishing procedure implemented in Relion3.0 (^{Zivanov et al., 2018, Zivanov et al., 2019}) and 3D refinement before performing 3D classification, without refining angle and shifts, using soft masks focusing on the B domains and Fabs with a tau factor of 30. ∼23,000 and 20,000 particles were selected for states 1 and 2, respectively, based on mobility of the bound Fabs which were less well resolved than SARS-CoV S (especially the S₂ subunit) in the maps. Finally, CryoSPARC non-uniform refinement (^{Punjani et al., 2017}) was used to obtain the final reconstructions at 4.2 Å and 4.5 Å resolution for states 1 and 2, respectively. Reported resolutions are based on the gold-standard FSC = 0.143 criterion (^{Rosenthal and Henderson, 2003, Scheres and Chen, 2012}) and Fourier shell correlation curves were corrected for the effects of soft masking by high-resolution noise substitution (^{Chen et al., 2013}). Local resolution estimation, filtering and sharpening was carried out using CryoSPARC.

CryoEM Model building and analysis

UCSF Chimera (^{Goddard et al., 2007}) and Coot were used to fit atomic models (PDB: 5W9J for MERS and PDB: 5X58 for SARS) into the cryoEM maps. The crystal structure of LCA60 or S230 were also fit into density using UCSF Chimera. The models were subsequently manually rebuilt using Coot (^{Brown et al., 2015, Emsley et al., 2010}). N-linked glycans were hand-built into the density where visible and the models were refined and relaxed using Rosetta (^{Wang et al., 2016}). Glycan refinement relied on a dedicated Rosetta protocol, which uses physically realistic geometries based on prior knowledge of saccharide chemical properties (^{Frenz et al., 2019}), and was aided by using both sharpened and unsharpened maps. Models were analyzed using MolProbity (^{Chen et al., 2010}) and privateer (^{Agirre et al., 2015}). Figures were generated using UCSF ChimeraX (^{Goddard et al., 2018}).

Mass spectrometry

In solution: MERS-CoV S or SARS-CoV S were either unaltered or subjected to Endo-F and Endo-H deglycosylation treatment. Two microliters of the deglycosylases were allowed to incubate with 20 μg of S for 4-14 hr in 50mM sodium acetate pH 4.4 at 37°C in a 20 μL reaction. 6 μg of S were then incubated in 100mM Tris pH 8.5, 2% sodium deoxycholate, 10mM tris(2-carboxyethyl)phosphine, and 40mM iodoacetamide at 95°C for five minutes and 25°C for thirty minutes in the dark. Denaturated, reduced, and alkylated S (1.6 μg) was then diluted into fresh 50mM Ammonium bicarbonate and incubated for 14 hr at 37°C either with 0.032 μg of trypsin (Sigma Aldrich), chymotrypsin (Sigma Aldrich), trypsin then sequentially chymotrypsin, alpha lytic protease (Sigma), or AspN protease (Pierce). Formic acid was then added to a final concentration of 2% and the samples spun at 14,000 rpm for 20 min and then again for 5 min to precipitate the sodium deoxycholate and collect the peptides from the supernatants. The SARS S digests with trypsin, chymotrypsin and alpha lytic protease were processed in parallel after cotton-enrichment of glycopeptides. Approximately 0.5 mg of cotton wool was packed into 100 μL pipette tips. Glycopeptides were diluted in 80% acetonitrile with 0.1% formic acid and bound to the cotton plug, followed by washing of the cotton plug with 80% acetonitrile with 0.1% formic acid. Glycopeptides were eluted in 0.1% formic acid in water.

In gel: 120 μL of MERS-CoV England1 and EMC2012 virions and 40 μg of purified MERS2P were run on a 4%–15% gradient Tris-Glycine Gel (BioRad). One lane of each sample was loaded with 40 μL while a second lane was loaded with 40 μL, allowed to run for 2 min, and loaded with another 40 μL. The bands were excised and washed with milliQ water and acetonitrile. 1 mg/mL DTT in 50mM ammonium bicarbonate was incubated with the gel pieces at 60°C for 1 hour. Following an acetonitrile wash, 10 mg/mL iodoacetamide in 50mM ammonium bicarbonate was incubated for 30 min at room temperature in darkness. Following extensive washes, either 30 μg/mL trypsin or chymotrypsin in ammonium bicarbonate was incubated with the gel pieces for 1.5 hours on ice. The solution was replaced with ammonium bicarbonate and incubated overnight at 37°C. The supernatant was collected and dried down in a speedvac for 1 hr. The peptides were reconstituted in 40 μL 10% formic acid, 5% DMSO.

Peptide supernatant was collected and between 4 and 8 μL were run on a Orbitrap Fusion Tribrid (ThermoFisher Scientific) mass spectrometer. A 35 cm analytical column and a 3cm trap column filled with ReproSil-Pur C18AQ 5 μm (Dr. Maisch GMBH) beads was used. Nanospray LC-MS/MS was used to separate peptides over a 110-min gradient from 5% to 30% acetonitrile. A positive spray voltage of 2100 was used with an ion transfer tube temperature of 350°C. An electron-transfer/higher energy collision dissociation ion fragmentation scheme was used (^{Frese et al., 2013}). The supplemental higher energy collision dissociation energy was 0.15 for the runs. A resolution setting of 120,000 with an AGC target of 2^∗10⁵ was used for MS1, and a resolution setting of 30,000 with an AGC target of 1^∗10⁵ was used for MS2. The data was then analyzed using Byonic (^{Bern et al., 2012}) against a custom database of recombinant coronavirus spike proteins, searching for glycan modifications with 12/24 ppm search windows for MS1/MS2, respectively. The in-solution digestions of the recombinant ectodomain were searched against a custom database of recombinant coronavirus spike proteins, proteases, and common contaminants. The in-gel digestion experiments were first searched against a database comprised of the full MERS-CoV proteome and the full Chlorocebus sabaeus (virus produced in VeroE6 cells) or human proteome (ectodomain produced in HEK293F cells) retrieved from UniProt, with variable N-linked glycosylations omitted from the search to identify background contaminants. The top 16 contaminants were then assembled together with the S sequence into a focused database for the glycopeptide identifications. In all experiments, up to three missed cleavages were permitted using C-terminal cleavage at R/K for trypsin, F/Y/W/M/L for chymotrypsin, R/K/F/Y/W/M/L for trypsin-chymotrypsin, T/A/S/V for alpha lytic protease, and N-terminal cleavage at D for AspN. Carbamidomethylation of cysteine was set as fixed modification, methionine oxidation and carbamidomethylation of lysine as variable common 1, glycan modifications as variable common 2, with a variable common max. parameter of 3. The N-linked glycan databases of Byonic were merged into a single non-redundant set of glycan compositions for the searches. All reported glycopeptides in the Byonic results files were manually inspected for quality of fragment assignments. All glycopeptide identifications were merged into a single non-redundant list per sequon. Glycans were classified based on HexNAc content as high-mannose (2 HexNAc), hybrid (3 HexNAc) or complex (> 3 HexNAc). Byonic search results were exported to mzIdentML format to build a spectral library in Skyline (^{MacLean et al., 2010}) and extract peak areas for individual glycoforms from MS1 scans. The full database of variable N-linked glycan modifications from Byonic was manually added to the Skyline project file in XML format. Byonic search results from in-gel digestions of both virion samples and ectodomain were merged into a single spectral library and manually inspected for quantitation. Reported peak areas were pooled based on the number of HexNAc, Fuc or NeuAc residues to distinguish high-mannose/hybrid/complex glycosylation, fucosylation and sialylation, respectively.

SEC-MALS

30 μl of each sample at 2.5 mg/mL were loaded onto a Vanquish column (ThermoFisher Scientific) with Acquity UPLC protein BEH SEC, 450Å, 2.5 μm, 4.6 mm x 150mm (Waters) and passed through a Wyatt μDAWN coupled to a Wyatt Optilab UT-rEX differential refractive index detector. Data were analyzed using Astra 6 software (Wyatt Technology Corp) to quantify the protein and glycan weight.

Author Contributions

A.C.W., X.X., J.S., Y.-J.P., and D.V. designed the experiments. A.C.W., X.X., M.A.T., and E.C. designed, expressed, and purified the proteins. A.C.W., X.X., J.Q., and D.V. performed cryo-EM sample preparation, data collection, processing, and built and refined the atomic models. A.C.W. and Y.-J.P. crystallized the Fabs. Y.-J.P. and D.V. processed the diffraction datasets and built and refined the atomic models. A.C.W., X.X., and J.S. performed the MS experiments. A.C.W. and X.X. performed all binding assays. R.G. produced MERS-CoV isolates. M.D., A.L., M.Z., and D.C. provided key reagents. A.C.W., X.X., Y.-J.P. M.A.T., J.S., F.A.R., D.C., and D.V. analyzed the data. A.C.W., X.X., and D.V. prepared the manuscript with input from all authors.