key: cord-0702049-gcpz5faw
authors: Wang, Xiaofei; Chen, Xiangyu; Tan, Jiaxing; Yue, Shuai; Zhou, Runhong; Xu, Yan; Lin, Yao; Yang, Yang; Zhou, Yan; Deng, Kai; Chen, Zhiwei; Ye, Lilin; Zhu, Yongqun
title: 35B5 antibody potently neutralizes SARS-CoV-2 Omicron by disrupting the N-glycan switch via a conserved Spike epitope
date: 2022-03-29
journal: Cell Host Microbe
DOI: 10.1016/j.chom.2022.03.035
sha: f000e0e4a3c2691b77591857f392de014ccee96c
doc_id: 702049
cord_uid: gcpz5faw

The SARS-CoV-2 Omicron variant harbors more than 30 mutations in the Spike protein, leading to immune evasion from many therapeutic neutralizing antibodies. We reveal that a receptor-binding domain (RBD)-targeting monoclonal antibody 35B5 exhibits potent neutralizing efficacy to Omicron. Cryo-electron microscopy structures of the extracellular domain trimer of Omicron Spike with 35B5 Fab reveal that Omicron Spike exhibits tighter trimeric packing and higher thermostability as well as significant antigenic shifts and structural changes within the RBD, N-terminal domain (NTD), subdomains 1 and 2. However, these changes do not impact targeting of the invariant 35B5 epitope. 35B5 potently neutralizes SARS-CoV-2 Omicron and other variants by causing significant conformational changes within a conserved N-glycan switch that controls the transition of RBD from down to up state, which allows recognition of the host entry receptor ACE2. This mode of action and potent neutralizing capacity of 35B5 indicate its potential therapeutic application for SARS-CoV-2.

The major concern of the COVID-19 pandemic is the emerging antigenic shifted SARS-CoV-2 variants of concern (VOCs) (Williams and Burgers, 2021; Yuan et al., 2021) . Previously SARS-CoV-2 VOCs, especially the Delta (B.1.617.2) variant identified in the early 2021 (Cherian et al., 2021) , harbor enhanced transmission and pathogenicity and show resistance to a variety of 60 therapeutic neutralizing antibodies as well as COVID-19 vaccines (Corti et al., 2021) . This concern is further amplified by a novel SARS-CoV-2 VOC Omicron (B.1.1.529), which was first described in South Africa in November 2021 and rapidly become dominant worldwide (He et al., 2021; Pulliam et al., 2021; Scott et al., 2021) .

The Omicron variant is characterized by 37 Spike (S) amino acid mutations, including 15 mutations in RBD (Cameroni et al., 2021; Dejnirattisai et al., 2022; Hoffmann et al., 2022) , the primary target for neutralizing antibodies. Very recent studies indicated that the Omicron variant was markedly resistant to the majority of therapeutic neutralizing monoclonal antibodies (mAbs) approved for clinical use (Cameroni et al., 2021; Cao et al., 2022; Dejnirattisai et al., 2022; Hoffmann et al., 2022; Liu et al., 2021) . Furthermore, the Omicron variant also escapes current 70 COVID-19 vaccines (Cameroni et al., 2021; Hoffmann et al., 2022; Liu et al., 2021) , including BNT162b2 (Pfizer) and mRNA-1273 (Moderna) (Baden et al., 2021; Polack et al., 2020; van Doremalen et al., 2020) . Neutralizing antibodies with potent efficacy to the Omicron variant are urgently demanded.

RBD-targeting neutralizing antibodies are classified into four classes or into six subgroups, each 75 of which covers a specific epitope. Neutralization of SARS-CoV-2 by these antibodies is carried out through the mechanisms including ACE2 competition, ACE2 molecular mimicry and Fc receptor-mediate neutralization (Barnes et al., 2020; Park et al., 2022) . N-linked glycosylation has J o u r n a l P r e -p r o o f 4 important roles in viral pathology, including mediating protein folding and stability and shaping viral tropism (Choi et al., 2021; Li et al., 2020; Watanabe et al., 2020) . Glycosylation shields 80 specific epitopes to facilitate viral immune evasion (Grant et al., 2020; Reis et al., 2021; Wintjens et al., 2020) . Beyond the shield function, the glycans at N165 and N234 from the N-terminal domain (NTD) in SARS-CoV-2 act a molecular switch to control the conformational transition of RBD from the down state to the up state, which is required for binding of the receptor ACE2 (Casalino et al., 2020; Henderson et al., 2020) . N165 and N234 are conserved in SARS-CoV-1 85 and MERS-CoV, highlighting the common mechanism of RBD conformational transition in S proteins. In this study, we found that despite the significantly structural changes in Omicron S protein, 35B5, an RBD-targeting monoclonal antibody (mAb) cloned from memory B cells of a convalescent COVID-19 patient, shows potent neutralization activities against the SARS-CoV-2

Omicron variant and other VOCs via a distinctive glycan-displacement mechanism, which 90 discriminates 35B5 from the previous identified neutralizing mAbs against SARS-CoV-2.

We tested the neutralization activity of 35B5, which shows broad neutralization to the SARS-CoV-95 2 wild-type (WT), Alpha, Beta and Delta variants in vitro and in vivo (Wang et al., 2021) , to the Omicron variant. 35B5 exhibited high binding capacity to the Omicron S protein (EC50=0.0618 μg/ml), comparable to that to WT (EC50=0.0195 μg/ml) and Delta (EC50=0.0207 μg/ml) S proteins ( Figure 1A ), in enzyme-linked immunosorbent assays (ELISA). In vitro incubation with 35B5 led to complete dissociation of the Omicron S trimer ( Figure 1B ). SARS-CoV-2 pseudovirus-based 100 inhibition assays revealed that 35B5 exhibits the potent neutralizing efficacy to the Omicron J o u r n a l P r e -p r o o f 5 variant with (IC50=0.0147 μg/ml), as well as to SARS-CoV-2 WT (IC50=0.0024 μg/ml) and Delta variant (IC50=0.0069 μg/ml) ( Figure 1C ). Echoing the pseudovirus-based inhibition assays, 35B5 potently neutralizes the authentic SARS-CoV-2 Omicron variant infection with an IC50 value of 0.0499 μg/ml ( Figure 1D ). Thus, 35B5 harbors the nanomolar neutralizing efficacy to the Omicron 105 variant.

We employed the cryo-EM single particle method to determine the complex structure of the Omicron S ectodomain trimer (Omicron S-ECD) with 35B5 Fab. 35B5 Fab and Omicron S-ECD 110 trimer were mixed and incubated with a stoichiometric ratio of 3 to 1 for 20 seconds, and then flash frozen into liquid ethane for sample preparation. Two cryo-EM structures of the Omicron S-ECD-35B5 Fab complexes were successfully determined to the overall resolutions of 3.0 Å and 3.4 Å, respectively. The density map for the up-RBD-35B5 Fab region was obtained at a resolution of 3.35 Å by further local refinements (Figures S1 and S2A-S2F; Table S1 ). The Omicron S-ECD 115 trimer in both the two complexes contains two up-RBDs and one down-RBD (Figures 2A-2C ).

Each up-RBD is bound by a 35B5 Fab. One up-RBD domain in the S-ECD trimer of the complex at the 3.0 Å resolution lacks clear densities, probably due to the 35B5 Fab-caused conformational dynamics of the up-RBD.

Structural superposition of the Omicron and G614 S trimers via alignment by the S2 region 120 revealed that the Omicron S-ECD trimer exhibits tighter structural packing. The NTD and RBD domains of Omicron notably move inward to the central S2 helical regions (Zhang et al., 2020) ( Figure 2D ). The mutation H655Y in the Omicron subdomain-2 (SD2) domain interacts with F643 and thereby increases the stability of the 630 loop (residues 617-644). The highly structured 630 J o u r n a l P r e -p r o o f 6 loop further induce a more ordered structure of the fusion peptide proximal region (FPPR, residues 125 823-862) in the adjacent protomer to stabilize the down state of RBD (Wang et al., 2021) (Figures 2E and 2F) . The tighter trimeric packing and the more ordered FPPR region suggest that the Omicron S trimer is more stable and resistant to premature dissociation. Consistently, differential scanning calorimetry assays revealed that the Omicron S-ECD trimer has higher thermostability than the G614 S-ECD protein ( Figure S2G ). The SARS-CoV-2 Omicron variant also exhibited 130 higher environmental stability than the Alpha, Beta and Delta variants on plastic and skin surfaces (Hirose et al., 2022) .

The Omicron variant is characterized by 15 mutation sites in RBD ( Figure 3A Figure S2H ). The G339D and N440K mutations are located in the epitopes 140 for Class 3 mAbs, including C135 and S309, while S371L, S373P and S375F are located at the RBD interface with Class 4 mAbs that generally harbor cross-species neutralizing activities to SARS-CoV and MERS-CoV ( Figure S2H ). In addition to structural changes of the mutant residues, the surface electrostatic distribution of Omicron RBD is also notably changed, including increased positive charges in the Class 1 and 2 mAb epitopes and enhanced hydrophobic surface 145 at the Class 4 mAb interface ( Figure 3B ). J o u r n a l P r e -p r o o f 7 NTD of S protein is another important antigen for many neutralizing antibodies (Cerutti et al., 2021; Chi et al., 2020) , and plays a role in the S stability through binding RBD of the adjacent protomer (Shang et al., 2020) . Compared to WT NTD, Omicron NTD has 8 mutation sites, among which T95I and A67V increase the inner hydrophobic interactions in the central core region of 150 NTD ( Figures S3A-S3C ). The del69-70, del143-145 and ins214EPE mutations lead to the more disordered loop regions that are invisible in the density map ( Figure S3A ), while G142D and L212I are located on the surface, thereby generating antigenic shifts of NTD and impairing the neutralization activities of the mAbs that targets NTD ( Figure S3A ).

Omicron S2 contains 6 mutations in the central helical region ( Figure 3C ). In the structure of 

Despite the significant antigenic shifts and structural changes of Omicron S-ECD, 35B5 still binds to Omicron RBD at an interface covering the area of 1006 Å 2 ( Figures 3F and S3G Further sequence analysis revealed that the epitopic residues for 35B5 in RBD is invariant in 180 SARS-CoV-2 WT, Alpha, Beta, Delta, Lambda and Omicron variants ( Figure S4A ). Among the 15 mutation sites in Omicron RBD, G339D, G446S and E484A are located at the edge of the 35B5-RBD interface ( Figure 3G ), but have no any structural collision with 35B5 Fab ( Figure 3G ).

The other 12 mutations are located in long distances to the epitope for 35B5 ( Figure 3G ). Thus, the invariant epitopic residues for 35B5 provide the molecular basis for efficient targeting of 185 Omicron RBD by 35B5. In addition, the Omicron S trimer maintains the down-RBD-NTD interface as the same as that in the other variants. Correspondingly, the epitopic residues E340, T345, R346, K444, Y449 and N450 for 35B5 are solvent-exposed in the Omicron down-RBD, which allows initial recognition by 35B5 ( Figures S4B and S4C ).

The L2 loop in the 35B5 epitope is essential for RBD structural integrity and ACE2 binding J o u r n a l P r e -p r o o f 9 The roles of the 35B5 invariant epitope in the S structure were then investigated. The ACE2binding surface on RBD (Figures 4A and 4I) is constituted by the β5-β6 antiparallel sheet with linking loops and α-helices, suggesting that the stability of the β5-β6 sheet are determinative for ACE2 binding during SARS-CoV-2 infection. In the structure of RBD, the 35B5 epitopic residues 195 R346, S349 and Y351 are located in the L2 loop (residues 344-354) and interact with Y451 and L452 in β5, and L492 in β6, respectively, to stabilize the conformations of the two β-strands ( Figure 4A ). Moreover, the residue V350 at the turn in the L2 loop inserts into the hydrophobic pocket under β5, which provides a firm basis to support the strand ( Figure 4A ). These extensive interactions suggest that L2 and its containing 35B5 epitopic residues are crucial for the stability 200 maintenance of β5 and β6, thereby determining ACE2 binding during the viral invasion (Wang et al., 2020).

RBD transits from the "down" state to the "up" state that is required for ACE2 recognition. The the function of the N165-glycan ( Figures 4C and 4D ). In addition, the 35B5-interacting residues 215 R466 and I468 bind to NTD below the down-RBD interface with the N165-glycans ( Figure 4D ).

Thus, the 35B5 epitopic residues control the conformational dynamics of RBD via binding the glycans at N165 from NTD. 

Omicron S protein harbors significant antigenic shifts and structural changes, which leads to immune escaping of the Omicron variant from most mAbs. Our previous study had shown that 235 35B5 had the neutralizing activities against the wild-type, beta and delta variants of SARS-CoV-2. 35B5 dissociates the spike trimer and neutralizes the SARS-CoV-2 through four steps via an J o u r n a l P r e -p r o o f 11 epitope that avoids the prevailing mutation sites on RBD in the beta and delta variants. In this study, we found that 35B5 could neutralize the Omicron variant with a potent neutralizing efficacy, which is much higher than many other neutralization antibodies. The actual reason resulting in 240 dissociation of the spike protein is that 35B5 displaces the conserved glycan switch from RBD, which lead to the unstable "up" states of RBD and eventually causes the shedding of S1 from the S trimer. The invariant epitope in RBD for 35B5 is distinct from those of the previously identified four classes of RBD-targeting mAbs. The glycan displacement action of 35B5 represents a unprecedented neutralization action of mAbs against SARS-CoV-2, which is different from the 

The patent of 35B5 had been licensed. K. D. and Y. Zhu are listed as inventors of the patent. 

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Yongqun Zhu (zhuyongqun@zju.edu.cn).

Materials generated in this study will be made available on request and may require a material transfer agreement. • This paper does not report original code.

• Any additional information required to reanalyze the data reported in this work paper is available from the Lead Contact upon request. Human ACE2-expressing HEK-293T (293T/ACE2) cells were seeded into 96-well plates (Corning, 3599) at a density of 20,000 cells per well. The next day, mAbs were serially diluted in complete media, mixed with WT pseudoviruses (SinoBiological, PSV001) or Delta pseudoviruses (SinoBiological, PSV011) or Omicron pseudoviruses (SinoBiological, PSV016) and incubated for 485 1 hour at 37 ℃. Then, culture media of 293T/ACE2 cells was replaced by pre-incubated mAb/pseudovirus mixture and the cells were cultured for another 16 hours. Next, 293T/ACE2 cells were cultured with fresh complete media for an additional 24 hours and the luciferase activity of SARS-CoV-2 pseudovirus-infected 293T/ACE2 cells were measured by a luciferase reporter assay kit (Promega, E1910). The IC50 (50% inhibitory concentration) values were calculated by 490 fitting a non-linear four-parameter dose-response curve in GraphPad version 6.0.

The Omicron variant strain HKU691 was used for authentic viral neutralization assay. The virus titration was done by plaque assay using TMPRSS2-expressing Vero E6 cells. Authentic SARS-495 CoV-2 neutralization assay was performed according to previous studies (Wang et al., 2021; Zhou et al., 2021) . Briefly, Vero E6 cells were seeded in a 24-well culture plates at a density of 2×10 4 cells per well at 37 ℃ for 24 hours. Then, authentic SARS-CoV-2 Omicron (MOI=0.005) and 5fold serially diluted 35B5 mAbs (from 25 μg/ml to 0.00032 μg/ml) were mixed in the medium with 2% FBS, and were then added into the Vero E6 cells. The culture supernatant of 48 hours 

The Omicron S-ECD gene were cloned using the overlapping PCR method from the plasmid encoding the ectodomains of the SARS-CoV-2 S-6P (HexoPro) mutant, which was kindly provided by Dr. Junyu Xiao of Peking University. HEK293F cells were cultured in the SMM 293-TI medium (Sino Biological Inc.) at 37 ℃ with 8% CO2. The Omicron S-ECD-containing plasmid 510 was transiently transfected into HEK293F cells using 40-kDa linear polyethyleneimine (PEI) (Polysciences) with the PEI:DNA mass ratio of 3:1 and 1 mg DNA for per liter of culture when the cell density reached 2 × 10 6 cells per mL. At day 4 post-transfection, the supernatants of the cell culture were harvested by centrifugation at 10,000 × g for 30 min followed by buffer-exchange by sartorius VIVAFLOW 200 cassette with a molecular weight cutoff of 30 kDa. The secreted 515 Omicron S-ECD protein was purified using HisPur TM cobalt resins (Thermo Scientific). Further purification was carried out using size-exclusion chromatography with a Superose 6 10/300 column (GE Healthcare) in the buffer containing 20 mM HEPES pH 7.2, 150 mM NaCl and 10%

Trehalose. The Fab region of the 35B5 was obtained after the digestion by papain for 40 min at 37 ℃ in a buffer containing 20 mM HEPES pH 7.2, 150 mM NaCl, 5 mM EDTA and 5 mM L-520 cysteine. The obtained 35B5 Fab was purified with a Desalting column (GE Healthcare Life Sciences) to remove L-cysteine, and then further purified in a HiTrap Q column (GE Healthcare Life Sciences). The purified Fabs were collected and concentrated to 0.6 mg/ml.

The S-ECD HexaPro proteins of Omicron and G614 SARS-CoV-2 were diluted in the dilution buffer of 20 mM HEPES, pH 7.2, and 150 mM NaCl to 0.2 mg/ml. The 5000× SYPRO Orange (Sigma) was diluted to 50× using the above dilution buffer. In a 96-well qPCR white plate (Bio-Rad), 18 µl protein or dilution buffer (negative control) and 2 µl dye were mixed together in each J o u r n a l P r e -p r o o f 21 well and incubated for 20 min in dark at room temperature. In a Bio-Rad CFX96 real-time PCR 530 instrument, fluorescence signals were measured every temperature ramp cycle (0.6 ℃ in 10 s) from 25 ℃ to 95 ℃. Data was plotted as the negative first derivative of the relative fluorescence unit (RFU) as a function of temperature by the Bio-Rad CFX Maestro software. Three parallel measurements were performed for each sample and one representative plot was shown.

For negative-staining assays, Omicron S-ECD and 35B5 Fab proteins were diluted in the buffer of 20 mM HEPES, pH 7.2, and 150 mM NaCl to 0.02 mg/ml and 0.04 mg/ml, respectively. 2 μL of 35B5 Fab was mixed with 2 μL Omicron S-ECD and incubated on ice for 3 min. The samples were loaded in the glow-discharged carbon-coated copper grids and stained with 3% Uranyl 540 Acetate (UA). The prepared grids were examined using a transmission electron microscope (HITACHI) operated at 80 kV. Micrographs were recorded using a GATAN camera with 120,000× nominal magnification. 

Raw movie frames were binned, aligned and averaged into motion-corrected summed images using MotionCor2 (Zheng et al., 2017) . The dose-weighted images were then imported into 560 cryoSPARC for the following image processing, including CTF estimation, particle picking and extraction, 2D classification, ab initio 3D reconstruction, heterogeneous 3D refinement and nonuniform homogeneous refinement (Punjani et al., 2017) . 4 representative particle templates were generated in 2D classification of 74,593 particles auto-picked by the blob picker from 1000 micrographs. Using these templates, 2,749,814 particles were extracted with a box size of 386×386 565 and classified into 150 classes in 2D classification. Among them, 45 classes that included 669,842 particles were selected for ab initio 3D reconstruction and heterogeneous refinement. Finally, 556,976 particles reconstructed an apparent architecture of the Omicron S-ECD-35B5 Fab complex and were subjected to two more rounds of ab initio 3D reconstruction and heterogeneous refinement before non-uniform refinement. Then the particles generated two abundant populations 

The statistical analyses were performed in GraphPad version 6.0 unless otherwise mentioned. In the ELISA assay, the mean values ± SEM for three independent experiments were plotted and the 585 EC50 values were calculated by using sigmoidal dose-response nonlinear regression ( Figure 1A ).

In neutralization assays, the mean values ± SEM for three ( Figure 1C ) or two ( Figure 1D ) independent experiments were plotted and the IC50 values were calculated by fitting a non-linear four-parameter dose-response curve. In the differential scanning fluorimetry assay, the melting temperatures (Tm) are shown as mean ± SD for three independent measurements and one 590 representative plot was drawn using the Origin 9.0 software ( Figure S2G ). No methods were used to determine whether the data met assumptions of the statistical approach as no statistical tests were performed in this study. 

A) Interactions of the L2 loop with β5 and β6 in the core region of Omicron RBD

B) Location of the N165-and N234-glycans from NTD at the interface of NTD with the down-RBD in Omicron S-ECD. The N165-and N234-glycans are colored in red and light blue 650 respectively. The glycans are shown as spheres

Interface of the N165-glycans with the down-RBD in Omicron S-ECD. The down-RBD and N165-glycans are shown as surface and sticks, respectively. (D) Detailed interactions of the down-RBD with the N165-glycans. The interacting residues are shown as sticks and labeled as indicated

of the glycans at N165 and N234 with the up-RBDs in the Omicron S-ECD

35B5-bound RBD and NTD (E) were superimposed with the down-RBD with its bound NTD (B) to illustrate the glycan displacement by 35B5 from down-RBD (G). 35B5-bound RBD and NTD (E) were superimposed with the 660 D614G up-RBD with its bound NTD (F) to illustrate the glycan position changes in the two structures (H)

Wang, Chen, et al describe an RBD-targeting monoclonal antibody, 35B5, that potently neutralizes the SARS-CoV-2 Omicron variant. 35B5 neutralizes with nanomolar efficacy by targeting an invariant epitope, resulting in the displacement of the conserved N-glycan switch from the RBD.