key: cord-0319937-5w2vwsog
authors: Kumar, Sanjeev; Patel, Anamika; Lai, Lilin; Chakravarthy, Chennareddy; Valanparambil, Rajesh; Davis-Gardner, Meredith E.; Edara, Venkata Viswanadh; Linderman, Susanne; Reddy, Elluri Seetharami; Gottimukkala, Kamalvishnu; Nayak, Kaustuv; Bajpai, Prashant; Singh, Vanshika; Frank, Filipp; Cheedarla, Narayanaiah; Verkerke, Hans P.; Neish, Andrew S.; Roback, John D.; Mantus, Grace; Goel, Pawan Kumar; Rahi, Manju; Davis, Carl W.; Wrammert, Jens; Suthar, Mehul S.; Ahmed, Rafi; Ortlund, Eric; Sharma, Amit; Murali-Krishna, Kaja; Chandele, Anmol
title: Structural insights for neutralization of BA.1 and BA.2 Omicron variants by a broadly neutralizing SARS-CoV-2 antibody
date: 2022-05-13
journal: bioRxiv
DOI: 10.1101/2022.05.13.491770
sha: 20eae74ccadba46d8283bf7863b570dcd0619472
doc_id: 319937
cord_uid: 5w2vwsog

The SARS-CoV-2 BA.1 and BA.2 (Omicron) variants contain more than 30 mutations within the spike protein and evade therapeutic monoclonal antibodies (mAbs). Here, we report a receptor-binding domain (RBD) targeting human antibody (002-S21F2) that effectively neutralizes live viral isolates of SARS-CoV-2 variants of concern (VOCs) including Alpha, Beta, Gamma, Delta, and Omicron (BA.1 and BA.2) with IC50 ranging from 0.02 – 0.05 μg/ml. This near germline antibody 002-S21F2 has unique genetic features that are distinct from any reported SARS-CoV-2 mAbs. Structural studies of the full-length IgG in complex with spike trimers (Omicron and WA.1) reveal that 002-S21F2 recognizes an epitope on the outer face of RBD (class-3 surface), outside the ACE2 binding motif and its unique molecular features enable it to overcome mutations found in the Omicron variants. The discovery and comprehensive structural analysis of 002-S21F2 provide valuable insight for broad and potent neutralization of SARS-CoV-2 Omicron variants BA.1 and BA.2.

The antigenic residues targeted by 002-S21F2 broadly neutralizing antibody (bnAb) are 132 highly conserved among current and previous SARS-CoV-2 VOC ( Fig. 2K and S8, S9A) . 133

Our structural data shows that 002-S21F2 continues to maintain potent neutralization 134 against Omicron variants BA.1 and BA.2 which harbor epitope mutations at G339D and 135 N440K (Fig. 1C, 1D, 2G and S7) . We suspect that this neutralization ability will persist 136 with newly listed variants (BA.2.13, BA.2.12.1, BA.3 and BA.4/BA.5), as they are not 137 reported to contain any additional mutations within the 002-S21F2 epitope region (Fig.  138   S8) . Furthermore, sequence alignment of Sarbecovirus RBDs shows 10 out of 19 139 conserved residues in SARS-CoV suggesting potential for cross-reactivity with other 140 Sarbecoviruses ( Fig. 2K and Fig. S9B) . 141

Structural comparison of the 002-S21F2 epitope with other class-3 mAbs, including the 142 two available therapeutic mAbs effective against Omicron -Ly-CoV1404 (Bebtelovimab) 143 and S309 (Sotrovimab), show some similarities between the 002-S21F2 and C135 144 binding sites (Fig. 2J) (16, 17) . However, C135 is unable to neutralize Omicron as a lysine 145 mutation at RBD site N440 position would sterically clash with the C135 heavy chain 146 CDR2 (Fig. 2J) (18) . In support of this, RBD deep mutational scanning shows that an 147 N440K mutation (present in BA.1 and BA.2) disrupts the RBD-C135 interaction (19) . 148

Although 002-S21F2 recognizes an epitope outside the ACE2 binding motif, it may 149 directly block ACE2 interaction through head-to-head inter-spike crosslinking as 150 observed at saturating spike to IgG concentration (Fig. S5C ). This corroborates a recent 151 report that positively correlates high neutralization potency to inter-spike crosslinking 152 ability within the RBD-5/class-3 antibodies (11). Interestingly, we also observed a higher 153 ratio of all RBD "down" conformations (~54% particles) in antibody bound spike data 154 compared to apo spike-6P (which only shows ~35% of all RBD "down" conformation 155 particles). Both putative mechanisms may interfere with the ACE2 binding and 156 contribute to neutralization. 157

Sequence analysis of 002-S21F2 revealed that its heavy chain (HC) variable region is 158 comprised from VH5-51, DH5-24, and JH4 genes; the light chain (LC) gene utilizes VK1-159 33 and JK2 (Fig. S4A) . Of the 5252 SARS-CoV-2 mAb sequences banked in the CoV-AbDab 160 database (20), only 2 others utilized this combination of VH5-51 and VK1-33 (Fig. S4B) . 161

Alignment of the 002-S21F2 mAb sequence to its germline sequence revealed 4 amino 162 acid (AA) mutations in the HC that spanned the FR1 and CDR1 regions, and 3 AA 163 mutations present in the FR3 and CDR3 regions of the LC (Fig. S4C) . This low frequency 164 of somatic hypermutations (SHM), 2.7% in the HC and 1.7% in the LC, suggests that the 165 memory B cell that expressed this mAb had not yet undergone extensive selection in the 166 germinal center. A comparison of 002-S21F2 with SARS-CoV-2 therapeutic mAbs 167 approved for clinical use revealed no obvious genetic similarities, suggesting that 002-168 S21F2 exhibits unique genetic characteristics (Fig. 3A) . 169

Structural studies have reported that SARS-CoV-2 bnAbs target only a few antigenic sites 170 on the RBD which are majorly recognized by class-3 and class-4 mAbs (11, 12, 17, 18) . 171

Omicron and its sublineages can evade natural and vaccine generated immunity and pose 172 a threat to immune-compromised, vaccine-hesitant and unvaccinated adults and 173 children. However, only 2 of the currently approved therapeutic antibodies have shown 174 neutralization potential to Omicron -an S309 derivative (Sotrovimab) and Ly-CoV1404 175 (Bebtelovimab). We show that 002-S21F2 potently neutralizes both Omicron BA.1 and 176 BA.2 and previous VOC without sacrificing potency, similar to a recently reported mAb 177

Bebtelovimab (16). In contrast, Sotrovimab neutralized BA.1 with ~3-fold higher potency 178 (than WA.1) yet poorly neutralizes BA.2 (9). Both of these bnAbs are class-3 antibodies 179 that recognize overlapping epitopes on the outer face of RBD but are distinct from the 180 002-S21F2 epitope (Fig. 3B ). This suggests that the epitopes defined by the SARS-CoV-2 181 class-3 bnAbs target distinct antigenic residues on the outer face of the RBD, and that this 182 surface may potentially form the basis for an effective vaccine. For example, selective 183 steering of B cell immune responses to the RBD class-3 antigenic sites, defined by 002-184 S21F2 and Bebtelovimab bnAbs, may induce potent antibody responses against SARS-185

CoV-2 VOC. A similar successful strategy of epitope-focused vaccine candidates has been 186 previously used to guide induction of HIV-1 bnAbs VRC01 and PGT121 that are currently 187 in clinical trials (21-23). In addition, 002-S21F2 maintains potent neutralization to 188

Omicron variants despite being isolated from a convalescent individual infected in the 189 early months of the pandemic, when only the ancestral SARS-CoV-2 WA.1 strain was 190

reported. Further, our structural analysis shows that the limited number of SHMs 191 observed in this near germline bnAb 002-S21F2 are not involved in recognizing the 192 antigenic sites (Fig. 2) 

Experimental work, data acquisition and analysis of data by S.K., A.P., L.L., C.R.C., R.V., Source data are provided in this paper. 250

Human subjects 252 COVID-19 recovered individuals have been described earlier (14). Of these, five subjects 253 chosen based on the frequency of receptor binding protein-positive memory B cells and 254 the available number of banked peripheral blood mononuclear cells (PBMCs) were 255 included in this study for human monoclonal antibodies generation. 256

The recombinant SARS-CoV-2 RBD gene was cloned, expressed, purified and ELISAs were 258 in the duplicate wells incubated at the highest dilution of the respective specimen). 293

Purified SARS-CoV-2 RBD protein was labelled with Alexa Fluor 488 using a microscale 295 protein labelling kit (Life Technologies, #A30006) as per the manufacturer's protocol. 296

Ten million PBMCs of select COVID-19 recovered donors were stained with RBD-Alexa 297 

The antibody genes were amplified as described earlier (25, 26). Briefly, cDNA was 310 synthesized, and antibody variable gene VDJ segments were amplified by reverse 311 transcription-polymerase chain reaction (RT-PCR) using a template-switching rapid 312 amplification of complementary DNA (cDNA) ends (RACE) approach (Davis et al., 313 manuscript in preparation). Gene segments were cloned into AbVec6W vectors (25). 4 314 colonies from each transformed plate were randomly picked and the insert was checked 315 by performing colony PCR using nested PCR primers. The sequence integrity of the 316 plasmids was verified by Sanger sequencing (Macrogen sequencing, South Korea). 317

The immunogenetic analysis of both heavy chain and light chain germline assignment, 319 framework region annotation, determination of somatic hypermutation (SHM) levels 320 supernatants were tested for their SARS-CoV-2 RBD binding potential by enzyme-linked 330 immunosorbent assay (ELISA). Supernatant with positive RBD binding signals was next 331 purified using Protein A/G beads (Thermo Scientific), concentrated using a 30 kDa or 100 332 kDa cut-off concentrator (Vivaspin, Sartorius) and stored at 4°C for further use. 333

The potential of human ACE2 and SARS-CoV-2 RBD interaction inhibition by RBD-specific 335 mAbs was measured with the cPass SARS-CoV-2 surrogate virus neutralization test 336 (sVNT) kit (Genscript, Singapore) as described previously (28), as per the manufacturer's 337 protocol. Briefly, each mAb at 20 μg/ml concentration was mixed with equal volumes of 338 recombinant HRP-conjugated RBD and incubated for 30 min at 37°C. Next, 100 μl of this 339 mixture was transferred to 96-well plates coated with recombinant hACE2 receptor and 340 further incubated for 15 min at 37°C. The plate was washed four times with 1X PBST 341 buffer followed by the addition of tetramethylbenzidine (TMB) substrate). The plate was 342 incubated for 15 min at room temperature, and the reaction was stopped by adding the 343 stop solution. Absorbance was measured at 450 nm and the percentage of inhibition of 344 each sample was calculated using the following formula: % inhibition = (1-(OD450 345 sample/ OD450 of negative control)) x 100. Controls were included in duplicate; samples 346 were analyzed in the singular. Inhibition >30% was regarded as a positive neutralization. 347

Binding analysis of SARS-CoV-2 mAb to spike protein was performed using an 349 electrochemiluminescence assay as previously described (29) NaH2PO4.H2O and 300mM NaCl in PBS followed by spike protein elution in elution buffer 389 containing 235mM Imidazole, 6.7mM NaH2PO4.H2O and 300mM NaCl in PBS. Eluted 390 protein dialyzed against PBS and concentrated. The concentrated protein ran onto a 391 Superose-6 Increase 10/300 column and protein eluted as trimeric spike collected. The 392 quality of the protein was evaluated by SDS-PAGE and by Negative Stain-EM. 393

Spike protein was diluted to 0.05mg/ml in PBS before grid preparation. A 3µL drop of 395 diluted protein was applied to previously glow-discharged, carbon-coated grids for ~60 396 sec, blotted and washed twice with water, stained with 0.75% uranyl formate, blotted 397 and air-dried. Between 30-and 50 images were collected on a Talos L120C microscope 398 (Thermo Fisher) at 73,000 magnification and 1.97 Å pixel size. Relion-3.1 (30) or 399 Cryosparc v3.3.2 (31) was used for particle picking, 2D classification. Particles belonging to the best IgG bound 3D class were refined in non-uniform 3D 426 refinement with per particle CTF and higher-order aberration correction turned on. To 427 further improve the resolution of the RBD-IgG binding interface a soft mask was created 428 covering one RBD and interacting Fab region of IgG and refined locally in Cryosparc using 429

Local Refinement on signal subtracted particles. All maps were density modified in 430 Phenix (32) using Resolve CryoEM. The combined Focused Map tool in Phenix was used 431 to integrate high resolution locally refined maps into an overall map. Additional data 432 processing details are summarized in Figure S5-S6 . 433

The initial spike models for WA.1 (PDB:7lrt) or Omicron (PDB:7tf8) as well as individual 434 heavy and light chains of the Fab region of an IgG (generated with Alphafold (33)) were 435 docked into cryoEM density maps using UCSF ChimeraX (34). The full Spike-mAb model 436 was refined using rigid body refinement in Phenix, followed by refinement in Isolde (35) . 437

The final model was refined further in Phenix using real-space refinement. Glycans with 438 visible density were modelled in Coot (36 

Etesevimab 002-S11F2 002-02 034-3 034-32 002-S13B6 002-13 036-1F9 002-S22D8 002-S13D5 002-S14D3 002-S14A4 002-S21A10 034-43 002-S14C4 002-S21E9 002-S21F3 036-1A10 002-S14D2 036-1E2 002-S13E1 002-S13F5 002-S13D6 036-1B1 002-S21E11 002-S13A2 002-S13C1 002-S12F5 002-S13A6 002-S13B2 002-S13C4 002-S11F3 036-2D2 036-35 002-4 036-15 002-54 002-S11E4 036-2B10 002-S22C11 002-S21B10 002-43 002-14 002-58 002-S22A4 034-2F10 036-2C2 002-18 036-1A11 002-S11G4 034-2C4 002-37 002-S22D1 002-S21F2 034-2C2 036-1F7 002-48 002-55 002-S13B3 002-S21B12 036-1D8 029-1A4 036-1D9 034-1B2 029-1D4 002-S21C4 036-1E12 034-1D9 002-S22A8 029-1C1 029-1D1 023-1B7 036-1B4 002-S22F10 002-6 036-23 023-2C5 029-1C3 029-2E1 002-S21F4 029-2C7 034-6 002-S12F6 036-1E4 034-1E6 029-1E2 002-S21F7 034-1C6 029-2E11 002-S13A4 002-S21E1 034-2 002-S22E1 CR3022 Isotype Control Neg-Pos+ 0 50 100 % ACE2 inhibition 002-S21F2 002-02 034-32 002-04 002-S13A2 002-13 002-S22D8 002-18 002-S13A6 034-3 034-43 002-S21B10 002-S13E1 002-S21E9 002-S21F3 002-S13D6 002-S13F5 002-S12F5 002-S11F2 002-S13B6 036-1F9 002-S13D5 002-S14D3 002-S14A4 002-S21A10 002-S14C4 036-1A10 002-S14D2 036-1E2 036-1B1 002-S21E11 002-S13C1 002-S13B2 002-S13C4 002-S11F3 036- 

Tackling COVID-19 with neutralizing 445 monoclonal antibodies

Current status of therapeutic monoclonal 447 antibodies against SARS-CoV-2

Escape from neutralizing antibodies by SARS-CoV-2 spike protein 449 variants

Rapid spread of SARS-CoV-2 Omicron subvariant BA.2 in a single-451 source community outbreak

mRNA-1273 and BNT162b2 mRNA vaccines have reduced 453 neutralizing activity against the SARS-CoV-2 omicron variant

An infectious SARS-CoV-2 B.1.1.529 Omicron virus escapes 456 neutralization by therapeutic monoclonal antibodies

Considerable escape of SARS-CoV-2 Omicron to antibody 458 neutralization

Antibody evasion properties of SARS-CoV-2 Omicron sublineages. 460

Structure, Function, and Antigenicity of the SARS-CoV-2 Spike 462 Glycoprotein

Defining variant-resistant epitopes targeted by SARS-CoV-2 464 antibodies: A global consortium study

SARS-CoV-2 neutralizing antibody structures inform therapeutic 466 strategies

The Omicron variant is highly resistant against antibody-mediated 468 neutralization: Implications for control of the COVID-19 pandemic

Characterization of neutralizing versus binding antibodies and memory 471 B cells in COVID-19 recovered individuals from India

Commonality despite exceptional 473 diversity in the baseline human antibody repertoire

LY-CoV1404 (bebtelovimab) potently neutralizes SARS-CoV-2 475 variants. bioRxiv

Structural basis for potent antibody neutralization of SARS-CoV-2 477 variants including B.1.1.529

Structural basis of SARS-CoV-2 Omicron immune evasion and 479 receptor engagement

Mapping mutations to the SARS-CoV-2 RBD that escape binding 481 by different classes of antibodies

CoV-AbDab: the coronavirus 483 antibody database

HIV-1 broadly neutralizing antibody precursor B cells revealed by 485 germline-targeting immunogen

HIV Vaccine Design to Target Germline Precursors of Glycan-487 Dependent Broadly Neutralizing Antibodies

A generalized HIV vaccine design strategy for priming of broadly 489 neutralizing antibody responses

Rapid Generation of Neutralizing Antibody Responses in COVID-19

Longitudinal Analysis of the Human B Cell Response to Ebola Virus 493 Infection

Asymmetric and non-stoichiometric glycoprotein recognition by 495 two distinct antibodies results in broad protection against ebolaviruses

IMGT, the international ImMunoGeneTics database

A SARS-CoV-2 surrogate virus neutralization test based on antibody-500 mediated blockage of ACE2-spike protein-protein interaction

Antibody response to SARS-CoV-2 mRNA vaccine in lung 503 cancer patients: Reactivity to vaccine antigen and variants of concern. medRxiv

RELION: implementation of a Bayesian approach to cryo-EM structure 506 determination

PHENIX: a comprehensive Python-based system for 510 macromolecular structure solution

Highly accurate protein structure prediction with AlphaFold

(left) and locally refined (right) maps and resolution estimation based on 0