Molecular probes of spike ectodomain and its subdomains for SARS-CoV-2 variants, Alpha through Omicron

Since the outbreak of the COVID-19 pandemic, widespread infections have allowed SARS-CoV-2 to evolve in human, leading to the emergence of multiple circulating variants. Some of these variants show increased resistance to vaccine-elicited immunity, convalescent plasma, or monoclonal antibodies. In particular, mutations in the SARS-CoV-2 spike have drawn attention. To facilitate the isolation of neutralizing antibodies and the monitoring of vaccine effectiveness against these variants, we designed and produced biotin-labeled molecular probes of variant SARS-CoV-2 spikes and their subdomains, using a structure-based construct design that incorporated an N-terminal purification tag, a specific amino acid sequence for protease cleavage, the variant spike-based region of interest, and a C-terminal sequence targeted by biotin ligase. These probes could be produced by a single step using in-process biotinylation and purification. We characterized the physical properties and antigenicity of these probes, comprising the N-terminal domain (NTD), the receptor-binding domain (RBD), the RBD and subdomain 1 (RBD-SD1), and the prefusion-stabilized spike ectodomain (S2P) with sequences from SARS-CoV-2 variants of concern or of interest, including variants Alpha, Beta, Gamma, Epsilon, Iota, Kappa, Delta, Lambda, Mu, and Omicron. We functionally validated probes by using yeast expressing a panel of nine SARS-CoV-2 spike-binding antibodies and confirmed sorting capabilities of variant probes using yeast displaying libraries of plasma antibodies from COVID-19 convalescent donors. We deposited these constructs to Addgene to enable their dissemination. Overall, this study describes a matrix of SARS-CoV-2 variant molecular probes that allow for assessment of immune responses, identification of serum antibody specificity, and isolation and characterization of neutralizing antibodies.


Introduction
The outbreak of the COVID-19 pandemic caused by the severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has led to more than 500 million confirmed cases and 6 million deaths worldwide since its onset in December 2019 (https://covid19.who.int).With joint efforts by public health authorities and scientists, COVID-19 vaccines have been developed at an unprecedented speed, with several vaccines now licensed or granted emergency use authorizations.Despite the rollout of effective vaccines, widespread infections have led to the emergence of numerous variants, including multiple variants of concern (VOC) that displaced the original strain or early circulating strains around the world.These variants harbor mutations in the spike protein, some of which are associated with increased transmissibility and/or immune escape.For example, the D614G mutation contributes a fitness advantage to SARS--CoV-2 and is hence associated with enhanced infectivity [1][2][3], whereas the L452R, S477N, and E484K mutations may lead to reduced protection from re-infection or increased resistance to vaccine-elicited antibodies [4,5].Though COVID-19 vaccines approved and authorized for use thus far are still effective against these variants in preventing severe disease, there have been reports that the B.1.351variant (Beta), the B.1.617.2 variant (Delta), and the B.1.1.529variant (Omicron) are more resistant to neutralization by convalescent plasma, monoclonal antibody treatments, and/or vaccine-elicited antibodies than the original WA-1 strain or the previously prevalent B.1.1.7 variant (Alpha) [6][7][8][9][10][11][12][13].
Biotin-labeled molecular probes are pivotal to antibody discovery and immune evaluation.To respond to global efforts combating the SARS-CoV-2 virus, we previously designed and produced constructs incorporating an N-terminal purification tag, a site-specific proteasecleavage site, the antigen of interest, and a C-terminal sequence targeted by biotin ligase that allow for tag-based purification and in-process biotinylation.Through this strategy, we have produced wildtype SARS-CoV-2 spike ectodomain and subdomain probes allowing for the identification of potent neutralizing antibodies and nanobodies that target the receptor binding domain (RBD) or the N-terminal domain (NTD) of SARS-CoV-2 [14][15][16][17][18][19], the characterization of antibody binding affinities and specificities, and the quantification of immune responses against spike and its subdomains in nonhuman primates to inform vaccine development [20,21] and to find correlates of elicited responses with neutralization [22].Moreover, with the rise of SARS-CoV-2 VOCs, we found that probes incorporating mutations defined by the variants could be helpful in the characterization of the impact of VOC mutations on vaccine effectiveness [23][24][25].
In this study, we report production and characterization of molecular probes of spike and subdomains for SARS-CoV-2 variants.By using the strategy for wildtype probes described previously [19], we created molecular probes comprising the N-terminal domain (NTD), the receptor-binding domain (RBD), the RBD with spike subdomain 1 (RBD-SD1), and the prefusion-stabilized spike ectodomain (S2P) of diverse variants, including Alpha, Beta, Gamma, Epsilon, Iota, Kappa, Delta, Lambda, Mu, and Omicron.We characterized the physical properties of the variant probes, their antibody-binding specificity, and their capacity to sort yeast cells expressing Fab region of SARS-CoV or SARS-CoV-2 spike-binding antibodies.Finally, we examined binding of human antibodies to RBD and NTD domains from six high-profile variants in comparison to the wildtype through sorting B-cell libraries from two convalescent donors.Overall, this study describes how structure-based design enables advantageous production of SARS-CoV-2 variant probes that provide molecular insight into immunogenicity of SARS-CoV-2 variants when isolating neutralizing antibodies, investigating antibody specificity, and monitoring longitudinal vaccine effectiveness against emerging variants.

Probe construct design for tag-based purification with in-process biotinylation
To make the molecular probes for the SARS-CoV-2 variants readily available, we incorporated variant mutations (as listed in Figs 1 and 2) into the probe construct we reported previously https://doi.org/10.1371/journal.pone.0268767.g001[19].The construct design comprised three segments.The first segment was an N-terminal purification tag denoted as 'scFc3C', which contained a modified 'knob-in-hole' single chain constant portion of an antibody (scFc) followed by a cleavage site (LEVLFQGP) for the highly specific human rhinovirus 3C (HRV3C) protease [26].This 'knob-in-hole' feature avoids formation of intermolecular Fc dimer and is defined by a protruding tyrosine in the N-terminalhalf of the Fc-interface (T366Y), a recessed threonine in the C-terminal-half of the Fc-interface (Y407T), in conjunction with an insert of a 20-residue-linker (GGGGS) 4 (Kabat antibody residue numbering [27]).The second segment was the sequence of interest from SARS-CoV-2 variants.We designed molecular probes of the entire spike ectodomain (residues 14 to 1208) with prefusion-stabilizing mutations (S2P), N-terminal domain (NTD), receptor-binding domain (RBD), and RBD plus subdomain 1 (RBD-SD1).The third segment, denoted as '10lnQQ-AVI', comprised a 10-residue linker followed by a biotin ligase-specific sequence (GLNDIFEAQKIEWHE) at the C terminus.
The scFc tag allows transiently expressed proteins to be efficiently captured by protein A resins.The captured target protein can then be cleaved from the resin and biotinylated concurrently to enable streamlined purification and biotinylation on a single protein A column.This method avoids elution with acidic buffer that might alter the conformation and minimizes the need for buffer exchange and size exclusion chromatography.The plasmids for the probe constructs from this study have been deposited to Addgene (www.addgene.org)with accession numbers listed in S1 Table.

Molecular probes of di-proline-stabilized SARS-CoV-2 variant spikes
To obtain probes of the trimeric spike ectodomain for SARS-CoV-2 variants, we incorporated the prefusion-stabilizing di-proline mutations previously reported into SARS-CoV-2 spike protein residues 14 to 1208 [28] along with a T4-phage fibritin trimerization domain (foldon) at the C-terminus [29,30].Briefly, the RRAR furin cleavage site at S1/S2 was replaced with GSAS, and two residues between HR1 and CH domains were substituted with prolines (K986P and V987P).The spike trimer probe constructs (hereafter referred to as 'S2P') were expressed by transient transfection in FreeStyle 293-F cells.In this study, we report production of S2P probes for variants B. captured by protein A resin through N-terminal scFc tag, followed by HRV3C-protease cleavage and BirA-catalyzed biotinylation, and finally passed through size-exclusion chromatography in PBS to remove reaction mixture and undesirable heterogeneous proteins.The final yield of the purified S2P probes ranged from ~0.3 mg to ~2 mg per liter of cell culture.
The purified variant S2P probes, along with WA-1 S2P and D614G S2P probes, all showed primarily a single peak by size-exclusion chromatography (Fig 3A).Size homogeneity was also confirmed by SDS-PAGE, exhibiting one major band (Fig 3B).Bio-layer interferometry (BLI) with streptavidin biosensors confirmed the biotinylation of the variant S2P proteins and ACE2 receptor binding, indicating that the variant S2P probes were well-biotinylated and recognized by ACE2 receptor (Fig 3C).

Molecular probes of variant SARS-CoV-2 NTD, RBD, RBD-SD1 regions
Recent reports of SARS-CoV-2 antibodies have shown that the receptor binding domain (RBD) and the N-terminal domain (NTD) are immunodominant and contain critical epitopes for neutralizing antibodies [34][35][36][37][38]. Therefore, the domain probes for variants are highly desirable for readout of immunogenicity and for isolation and characterization of neutralizing antibodies.Accordingly, we created separate molecular probes comprising the NTD (residues 14-305), the RBD (residues 330-526), and RBD with subdomain 1 (RBD-SD1, residues 319-591) with corresponding mutations listed in Fig 2 .We designed and prepared 14 NTD probes, 12 RBD probes, and 12 RBD-SD1 probes from various variant sequences alongside their original WA-1 strain counterpart.The expression and purification were similar to those for S2P probes as described above.The overall yields were between 1.5-5.9mg/L for the NTD probes, 3.7-11.3mg/L for the RBD probes, and 3.8-18.3mg/L for the RBD-SD1 probes, except for B.1.1.529variant, which yielded 16.6 mg/L for NTD, 0.95 mg/L for RBD (we also produced a shorter version of B.1.1.529variant RBD probe with residues 335-526, which had a lower yield of 0.18 mg/L but had identical antigenic properties), and 0.88 mg/L for RBD-SD1.
All purified biotinylated NTD probes ran as a single peak on size-exclusion chromatography (Fig 5A   epitope by other domains in the trimer.Among the other three SARS-CoV-2 NTD-directed antibodies, 5-7, which targets a hydrophobic pocket in NTD that is distinct from the NTD supersite, could bind to most of the NTD probes, indicating its epitope is unaffected by mutations in the NTD region of most of the variants.As observed with S2P probes, only NTD probes from B.1.617.2 and AY.1 showed low BLI responses to antibody 5-7.On the other hand, antibodies 2-51 and 5-24 showed a discerning difference in binding NTD probes in comparison with S2P probes.In addition to the variants that 2-51 and 5-24 could not recognize in the S2P probes, these two antibodies could not also recognize the variant B.1.429-W152CNTD probe (Fig 5D ).
Structures of antibodies 5-7, 2-51, and 5-24 in complex with SARS-CoV-2 spike or isolated NTD are available [33,39], and we analyzed their epitopes with respect to the impact of the mutated residues in the SARS-CoV-2 variants on antibody binding (S1 Fig) .As described above, 5-7 targets a hydrophobic pocket in NTD away from the NTD supersite.The only residue in the 5-7 epitope with a mutation in the variants is R190, which is mutated to Ser in P.1 (Fig 2).This residue is located on the edge of the hydrophobic pocket (S1A  7D).The B1.1.529RBD or RBD-SD1 probe could bind most other antibodies tested, unlike B.1.1.529S2P, which could not recognize A19-61.1 nor the Class I RBD antibodies A23-58.1 and B1-182.1.These two Class I antibodies, A23-58.1 and B1-182.1,bind to a very small and conserved footprint on RBD when it is in the up position.Consistent with our previous results, A23-58.1 and B1-182.1 could only be knocked out by mutations on F486/ N487/Y489, which is not seen on any of the variants studied here [18].Therefore, all the variant RBD and RBD-SD1 probes showed high responses to these two antibodies.The reason that B.1.1.529S2P probe lost binding to these Class I antibodies is due to the positioning of the RBD domains rather than the mutations in the B.1.1.529RBD region.Overall, the binding results from variant subdomain probes provide molecular insights into antibody recognition of SARS-CoV-2 variants.

Binding analysis to yeast displaying monoclonal antibodies and convalescent immune libraries
To evaluate the variant molecular probes for their utilities in immunogenic assays and cell sorting, we used flow cytometry to measure probe binding to yeast displaying either antigenspecific antibodies or libraries cloned from B cells isolated from COVID-19 convalescent patients [41].The biotinylated SARS-CoV-2 S2P, NTD, and RBD probes were freshly labeled with fluorochromes and used to sort the yeast cells displaying Fabs of S652-112, S652-118, and S652-109, antibodies isolated from a SARS-CoV-1 convalescent subject that bind SARS-CoV-2 S2, NTD and RBD, respectively [19], and Fabs of LY-555, CB6, REGN10933, REGN10987, A19-46.1, and A23-58.1 isolated from COVID-19 convalescent patients [18,42,43].For S652-112 expressing yeast, S2P variant probes showed less than 2-fold (ranged 59%-167%) difference in binding relative to WA-1 S2P probe (Fig 8A  .We found that S2P binding was 59-167% and NTD was 48-200% of WA-1 binding.We noted that S652-118 had reduced binding to B.1.429S2P trimer (59%) but increased binding (146%) to NTD probe of the same variant.While these differences are subtle, they suggest that the decrease in S2P binding may not simply be due to the W152C substitution, which is located on a surface loop equally accessible between S2P and NTD probes, but rather may be due to the different accessibility of the epitope in the trimer versus the isolated NTD (Figs 2 and 8A; S3 and S4 Figs).Conversely, S652-118 binding to the probes of B.1.526and AY.1 had opposite effects, with NTD being worse (48% and 48%, respectively) and slightly higher in the context of these S2P trimer variants (134% and 167%, respectively) (Fig 8A).This suggests that the S652-118 either has stabilizing contacts with other non-NTD domains or that NTD epitope is stabilized in context of the trimeric S-2P.
We next evaluated the binding of S2P and RBD probes to yeast expressing RBD Class I (e.g., CB6, REGN10933, A23-58.1),Class II (e.g., LY-555, A19-46.1) or Class III (e.g., REGN10987, S652-109) antibody Fab regions.We noted that relative to WA-1, binding of D614G S2P and B. .This indicates that the increased binding is not likely due to the presence of the N501Y mutation in B.1.1.7 but rather reflects a preference for these antibodies to bind RBD up and an increased probability of RBD being present in the up position.In addition to and consistent with previously published results, binding of A23-58.1 yeast was not decreased to any of the S2P or RBD variant probes except for the Omicron variant, and, on the flip side, A19-46.1 yeast showed decreased binding to variant S2P and RBD probes containing L452R mutations but exceptionally maintained its binding to the heavily mutated Omicron variant while most of the RBD-targeting antibodies lost binding/ neutralizing capability to Omicron.In line with previous reports showing significant loss of binding and neutralization by CB6 and REGN10933 to B.1.351and P.1, yeast bearing CB6 and REGN10933 showed decreased binding to S2P and RBD (Fig 8A ; S3 and S5 Figs).Similarly, S2P probes with mutations at E484, except for B.1.621,also showed decreased binding to REGN10933 yeast that was not present in the context of RBD monomer probe.Notably, for the Class II antibody LY-555, binding was reduced in the presence of E484Q/K/A in both S2P  VL antibody genes from B cells isolated from PBMCs were cloned into a yeast display vector to generate a yeast library from each donor, expressing human antibody Fabs on yeast surface.The sorting results showed that both libraries responded to all but B.1.1.529RBD probes, with percentage of antigen-positive ranged between 0.07-0.25%for one library and 0.09-0.16%for the other (Fig 8B).This result is consistent with the understanding that prior exposure to SARS-CoV-2 or inoculation with vaccines designed based on WA-1 strain still protects against most variants but the RBD-directed neutralizing titer for B.1.1.529is impaired.Also, when sorting with variant NTD probes, donor 1 library did not bind B. Overall, these results validate the utilities of the variant probes in identification and characterization of antibodies against SARS-CoV-2 variants of concern and in assessment of immune responses and serum antibody specificity.

Discussion
Our initial interest in developing molecular probes comprising key targets of the SARS-CoV-2 spike related to their use in defining and monitoring elicited responses and in facilitating antibody identification and characterization.Other uses in diagnostics (to assess sera reactivity and to provide sensitive markers of infection) and in pathogenesis (to delineate susceptible cells that virus might infect) were also possible.In these uses, one utility of the probes was in delineating the region of the spike recognized by antibodies, e.g., NTD, RBD, SD1 or elsewhere on the ectodomain.
The emergence of SARS-CoV-2 variants highlights an additional purpose, the tracking and characterization of the recognition by sera and antibodies of viral variants.Here, we have developed a matrix of probe regions (spike ectodomain, NTD, RBD, and RBD-SD1) and their variants (Alpha through Omicron) and demonstrated the utilities of these probes in analysis of sera and antibodies.It will be interesting to see how the field utilizes this matrix of probesindividually, such as with the Omicron spike or RBD, as a series, such as assessing sera with RBDs from Alpha through Omicron, or as a matrix.For instance, we have provided WA-1, Beta, Delta, and Omicron probes for confirming elicitation of cross-reactive B cells by mRNA-1273 or mRNA-Omicron boost and for evaluating their ACE2 binding inhibition [44].Another recent research identified convergent immune responses elicited by Beta-or Gammainfection when compared to ancestral virus infection using S2P, RBD, and NTD probes from WA-1, Beta, and Gamma variants [45].To facilitate usage of these probes in COVID-19 related research, the full matrix of probe regions and variants is being made available at Addgene for dissemination, including plasmids for the recently emerged Omicron sublineage BA.2 (S1 and S2 Tables).

Cell lines
Expi293F and FreeStyle 293-F cells were purchased from Thermo Fisher Scientific.The cells were maintained following manufacturer's suggestions and were used as described in detail below.

Construction of expression plasmids for wildtype and variant SARS-CoV-2 molecular probes
DNA sequences encoding the wildtype or variant SARS-CoV-2 spike or specific domains were cloned into a pVRC8400-based expression vector by GeneImmune.The gene of interest was inserted between the DNA sequences coding for a N-terminus purification component and a C-terminus biotinylation tag.The N-terminus purification component was composed of a single chain human Fc with knob-in-hole mutations, a 'GGSGGGGSGG' linker, and an HRV3C protease cleavage site (scFc3C) for purification with protein A resin and tag cleavage.And the C-terminus comprised a 'GGGLVPQQSG' 10lnQQ linker followed by an AVI tag for biotinylation.For the S2P probe constructs, the insertion included the wildtype spike protein residues 14 to 1208, or the corresponding spike gene from variants, with the S1/S2 RRAR furin cleavage site replaced by GSAS, the two proline stabilization mutations K986P and V987P, a GSG linker, and the T4-phage fibritin trimerization domain (foldon) as described by Wrapp and colleagues [28].For the NTD probe constructs, spike gene coding for residues 14-305 was cloned.For the RBD probe constructs, spike gene coding for residues 330-526 or 335-526 was cloned.For the RBD-SD1 probe constructs, spike gene coding for residues 319-591 was cloned.Plasmids for the wildtype and variant SARS-CoV-2 proteins with AVI tag generated in this study have been deposited with Addgene (www.addgene.org)with accession numbers listed in S1 Table.

Expression and preparation of wildtype and variant SARS-CoV-2 molecular probes
The wildtype and variant SARS-CoV-2 molecular probes were produced by transient transfection of FreeStyle 293-F cells [28] and in-process biotinylation [19] as previously described.Briefly, 1 mg of the plasmid encoding the scFc3C tagged and AVI tagged SARS-CoV-2 target protein and 3 ml of the Turbo293 transfection reagent (Speed BioSystems), each in 20 ml Opti-MEM (Thermo Fisher Scientific), were pre-mixed and transfected into FreeStyle 293-F cells at 1 mg plasmid per 0.8 L cells (2×10 6 cells/ml).Cells were incubated in shaker at 120 rpm, 37˚C, 9% CO2.The next day following transfection, 80 ml HyClone SFM4HEK293 medium (Cytiva) and 80 ml FreeStyle 293 Expression Medium (Thermo Fisher Scientific) were added to each 0.8 liter of cells.The transfected cells were allowed to grow for 6-7 days in total before the supernatant was harvested by centrifugation and filtration.Next, the supernatant was incubated with 5ml of PBS-equilibrated protein A resin for two hours, after which the resin was collected and washed with PBS.The captured SARS-CoV-2 protein was then biotinylated using the BIRA500 kit (~2.5 μg per 10 nmol AVI substrate) (Avidity) and cleaved from the Fc purification tag concurrently with 200 μg of HRV3C prepared as described [46].After incubation at 4˚C overnight, the liberated protein was collected, concentrated, and applied to a Superdex 200 16/600 gel filtration column equilibrated with PBS.Peak fractions were pooled and concentrated to 1 mg/ml.Finally, probes were run on SDS-PAGE to examine molecular size and purity.

Expression and preparation of the ACE2 receptor
The human ACE2 (1-740aa) expression plasmid was constructed with a monomeric human Fc tag and an 8xHisTag at the 3'-end of the ACE2 gene.After transient transfection of Expi293F cells, the cell culture supernatant was harvested 5 days post transfection and loaded onto a protein A affinity column.The Fc-tagged protein was eluted with IgG elution buffer and then dialyzed against PBS.

Bio-layer interferometry
An Octet HTX instrument (Sartorius) was used to analyze biotinylation level and antigenicity of the molecular probes and the receptor recognition of the S2P probes.Assays were performed at 30˚C in tilted black 384-well plates (Geiger Bio-One) in PBS with 1% BSA with agitation set to 1,000 rpm.Before running the assays, the streptavidin biosensors were equilibrated in PBS with 1% BSA for at least 20 minutes.Biotinylated SARS-CoV-2 S2P (10 μg/ml), NTD (5 μg/ml), RBD (2.5 μg/ml), and RBD-SD1 (5 μg/ml) were loaded on to streptavidin biosensors for 60 seconds.Binding to biotinylated probes was measured by dipping immobilized probes into solutions of ACE2 receptor at 500 nM or antibodies at 200 nM for 180 seconds.Measurements were assessed in triplicate.Parallel correction to subtract systematic baseline drift was carried out by subtracting the measurements recorded for a loaded sensor dipped into buffer only control well.The response values at the end of association step were reported.

Negative-stain electron microscopy
Purified SARS-CoV-2 S2P variant probes were diluted to approximately 0.02 mg/ml with buffer containing 10 mM HEPES, pH 7, and 150 mM NaCl.A 4.7-μl drop of the diluted sample was applied to a glow-discharged carbon-coated copper grid.The grid was washed three times with the same buffer, and adsorbed protein molecules were negatively stained by applying consecutively three drops of 0.75% uranyl formate.Micrographs were collected at a nominal magnification of 100,000x (pixel size: 0.22 nm) using SerialEM [48] on an FEI Tecnai T20 electron microscope operated at 200 kV and equipped with an Eagle CCD camera or at a nominal magnification of 57,000x (pixel size: 0.25 nm) on a Thermo Scientific Talos F200C electron microscope operated at 200 kV and equipped with a Ceta camera.Particles were picked automatically using in-house written software (Y.T., unpublished).Reference-free 2D classification was performed with Relion 1.4 [49].

Probe conjugation
Biotinylated SARS-CoV-2 proteins were conjugated using either allophycocyanin (APC) streptavidin (full-length spike proteins, every monomer labeled), Brilliant Violet 421 (BV421) streptavidin (RBD proteins), or Brilliant Violet 711 (BV711) streptavidin (NTD proteins).Reactions were prepared at a 4:1 molecular ratio of protein to streptavidin.Labeled streptavidin was added in ⅕ increments, with incubations at 4˚C (rotating) for 20 minutes in between each addition.When labelled, probes were titrated over monoclonal yeast display to determine an optimal staining concentration in the range of 10 to 0.1 ng per ul of staining solution.

Analysis of probe binding to human antibody repertoires expressed in yeast
PBMCs for experimental sorting of natively paired antibody heavy:light yeast display libraries were obtained from 2 participants enrolled in the COVID-19 Cohort Study at Columbia University Irving Medical Center.Patient samples were collected after obtaining informed written consent and appropriate Institutional Review Board approval (IRB#AAAS9517).
B cells were isolated from cryopreserved PBMCs of SARS-CoV-2 convalescent donors as previously described [16,41,[54][55][56].Briefly, B cells were enriched by CD27+ selection and stimulated in vitro for five days to enhance transcription of antibody genes.Single B cells were captured in emulsion droplets consisting of lysis buffer and oligo(dT)-coated magnetic beads for single-cell mRNA capture.Overlap extension RT-PCR was used to transform separate heavy (VH) and light (VL) variable genes into a single, physically linked VH:VL cDNA amplicons.cDNA amplicons encoding heavy and kappa light chain variable regions were cloned into a yeast display vector for expressing antibody VH:VL genes as Fabs for functional screening via FACS and NGS as previously reported [16,41,55,56].Briefly, AWY101 yeast expressing Fabs were cultured in SGCAA media (Teknova) with 2 g/L dextrose (SGDCAA) for 36 hrs at 20˚C to induce Fab expression for surface display.In the first round of screening, 3×10 7  Fab-expressing yeast cells were stained and washed with ice-cold staining buffer (phosphatebuffered saline (PBS) with 0.5% BSA and 2mM EDTA).Washed yeast libraries were stained in dark and at 4˚C with biotinylated antigens pre-conjugated with SA-PE (Streptavidin, R-Phycoerythrin Conjugate, Thermo Fisher Scientific) or SA-APC (Streptavidin, Allophycocyanin Conjugate, Thermo Fisher Scientific), and an anti-FLAG fluorescein isothiocyanate (FITC) monoclonal antibody (clone M2-FITC, Sigma-Aldrich) was added as a marker for Fab expression.Sorting was performed on a SONY MA900 cell sorter and gates for sorting were drawn as

Fig 1 .
Fig 1. Structural modeling reveals a persistent D614G mutation, shared RBD mutations, and distinctive NTD mutations in five variants of concern.Structural models of a SARS-CoV-2 spike protomer from B.1.1.7,B.1.351,P.1, B.1.617.2, and B.1.1.529variants.NTD, RBD, and SD1 domains are shown in orange, cyan, and green, respectively.Mutations are highlighted in the structural diagrams with a red sphere at Cα. Mutations at positions that are not resolved in the cryo-EM structure are labeled in italics with arrows pointing to the disordered regions (shown with dashed lines).

Fig 3 .
Fig 3. Characterization of purified biotinylated SARS-CoV-2 variant S2P probes confirms their homogeneity and antibody-binding specificity.(A) Size exclusion chromatography of the purified biotinylated SARS-CoV-2 S2P probes, including original WA-1 S2P, single mutated D614G S2P, and 15 other variant S2P, in PBS buffer.(B) SDS-PAGE of SARS-CoV-2 S2P variant probes with and without reducing agent.Molecular weight marker (MWM), the WA-1 S2P probe, and the D614G S2P probe, were run alongside the variant S2P probes.(C) Biotinylation and receptor recognition of the SARS-CoV-2 variant S2P probes.The level of biotinylation was evaluated by capture of the S2P probes at 10 μg/ml onto the streptavidin biosensors using Octet.Receptor binding was assessed with 500 nM of dimeric ACE2.Error bars represent standard deviation of triplicate measurements.(D) Antigenicity assessment of the SARS-CoV-2 variant S2P probes.Responses to RBD-directed, NTD-directed, and S2 subunit-directed antibodies were shown in cyan, orange, and rose, respectively.Bar scale between 0 and 1. Negative values not depicted.https://doi.org/10.1371/journal.pone.0268767.g003 ) and appeared as a single major band on SDS-PAGE (Fig5B).The NTD bands look diffused and the observed molecular weights are higher than the theoretical value (~36kDa) because the NTD is heavily N-link glycosylated.Biotinylation of the probes and their binding to the antibody panel were evaluated by BLI.All purified NTD probes bound well to streptavidin biosensors (Fig 5C).As expected, none of the NTD probes reacted to S2-directed or RBD-directed antibodies (Fig 5D).Unlike the moderate signals observed with the variant S2P probes, all the variant NTD probes clearly showed high responses to antibody S652-118, indicating that the epitope of S652-118 was conserved among SARS-CoV, WA-1 SARS-CoV-2, and all the SARS-CoV-2 variants investigated here.The relatively low reactivity of S652-118 to the S2P trimer probes was likely due to the restricted access to the S652-118

Fig 4 .
Fig 4. Negative-stain EM of the biotinylated SARS-CoV-2 variant S2P probes shows individual trimeric spike to be well folded.The variants are labeled above each set of micrographs.The top panel of each set is the representative micrograph; the bottom panel shows the 2D-class averages.Scale bars in micrographs: 50 nm.Scale bars in 2D class average images: 10 nm.The B.1.1.529S2P probe showed mostly trimeric particles, but appeared smaller at pH 7.5; however, at pH 5.5, most particles exhibited trimeric structure similar to other S2P probes (see S7 Fig).

Fig 5 .
Fig 5. Characterization of biotinylated SARS-CoV-2 variant NTD probes confirms their homogeneity and antibody-binding specificity.(A) Size exclusion chromatography of purified biotinylated SARS-CoV-2 NTD probes, including original WA-1 NTD and 14 other variant NTD in PBS buffer.(B) SDS-PAGE of SARS-CoV-2 NTD variant probes with and without reducing agent.Molecular weight marker (MWM) and the WA-1 NTD probe were run alongside the variant NTD probes.(C) Biotinylation of the SARS-CoV-2 variant NTD probes.The level of biotinylation was evaluated by capture of NTD probes at 5 μg/ml onto streptavidin biosensors using Octet.Error bars represent standard deviation of triplicate measurements.(D) Antigenicity assessment of the SARS-CoV-2 variant NTD probes.Responses to RBD-directed, NTD-directed, and S2 subunit-directed antibodies were shown in cyan, orange, and rose, respectively.Bar scale between 0 and 3. Negative values not depicted.https://doi.org/10.1371/journal.pone.0268767.g005 Fig) and appears to play a minor role in 5-7 binding, as there was only moderate reduction of 5-7 binding to the P.1 NTD probe relative to that of WA-1 (Fig 5D).The two NTD variant probes with substantially reduced binding to 5-7, B.1.617.2NTD and AY.1 NTD, were likely due to their deletions and mutation, ΔE156/ΔF157/R158G, which could potentially impact the structure at the hydrophobic pocket (S1A Fig).Antibodies 2-51 and 5-24, on the other hand, target the NTD supersite, which is heavily mutated among the variants (S1B Fig), and they lost binding to most NTD variant probes (Fig 5D).5-24 has a larger epitope than 2-51, a likely explanation for 5-24 to tolerate deletions at residues 144-146 and retain substantial binding to B.1.1.7NTD and B.1.618NTD, compared to 2-51, which lost binding to these two variant probes in addition to most other NTD probes.For the RBD and RBD-SD1 regions, we produced 12 variant probes each, because the B.1.526-E484K is the same as B.1.618,and the two B.1.617variants are the same as the Kappa variant, B.1.617.1, in their RBD and SD1 sequences.All purified RBD and RBD-SD1 molecular probes appeared as a single peak on size-exclusion chromatography (Figs 6A and 7A) and ran as a single band on non-reducing and reducing SDS-PAGE (Figs 6B and 7B).The RBD and RBD-SD1 probes bound well to streptavidin biosensors, also responded well to ACE2 receptor (Figs 6C and 7C), and almost all showed high responses to every RBD-antibodies tested, except antibodies A19-46.1 and A19-61.1 (Figs 6D and 7D).A19-46.1 could not bind probes with L452R/Q mutation, as seen with S2P probes (Figs 3D, 6D and 7D).Residue 452 is located in the center of the A19-46.1 epitope (S1C Fig) and plays an important role in binding interactions [40].On the other hand, the Class III RBD antibody A19-61.1 substantially reduced binding to the B.1.1.529RBD probe, which has mutations at residues 446 and 493 on the periphery of A19-61.1 epitope (S1C Fig), but retained binding to most other RBD or RBD-SD1 variant probes (Figs 6D and

Fig 6 .
Fig 6.Characterization of biotinylated SARS-CoV-2 variant RBD probes confirms their homogeneity and antibody-binding specificity.(A) Size exclusion chromatography of the purified biotinylated SARS-CoV-2 RBD probes, including WA-1 RBD and 12 other variant RBD probes, in PBS buffer.(B) SDS-PAGE of SARS-CoV-2 RBD variant probes with and without reducing agent.Molecular weight marker (MWM) and the WA-1 RBD probe were run alongside the variant RBD probes.(C) Biotinylation and receptor recognition of the SARS-CoV-2 variant RBD probes.The level of biotinylation was evaluated by capture of the RBD probes at 2.5 μg/ml onto the streptavidin biosensors using Octet.Receptor binding was assessed with 500 nM of dimeric ACE2.Error bars represent standard deviation of triplicate measurements.(D) Antigenicity assessment of the SARS-CoV-2 variant RBD probes.Responses to RBD-directed, NTDdirected, and S2 subunit-directed antibodies were shown in cyan, orange, and rose, respectively.Bar scale between 0 and 4. Variants with the L452R mutation did not binding antibody A19-46.1.https://doi.org/10.1371/journal.pone.0268767.g006 and S2 Fig).The variant with the lowest binding had mutation A701V (B.1.351,59%) in the S2 stem (Figs 1 and 8A).While more work is needed to define the S652-112 epitope, this data suggests it may reside near residue 701 in the S2 stem region (Fig 1).The NTD-reactive S652-118 expressing yeast bound to both S2P and NTD probes as expected (Fig 8A; S3 and S4 Figs) 1.1.7S2P probes to yeast expressing A23-58.1 and A19-46.1 was increased 260-275% and 292-214% respectively (Fig 8A and S3 Fig).A23-58.1 is a Class I antibody that binds to the tip of RBD when it is in the up position and A19-46.1 is a Class II antibody that binds to RBD when it is in the up or down position.Neither antibody shows increased binding to the B.1.1.7RBD probe (Fig 8A; S3 and S5 Figs)

Fig 7 .
Fig 7. Characterization of biotinylated SARS-CoV-2 variant RBD-SD1 confirms their homogeneity and antibody-binding specificity.(A) Size exclusion chromatography of the purified biotinylated SARS-CoV-2 RBD-SD1 probes, including original WA-1 RBD-SD1 and 12 other variant RBD-SD1 in PBS buffer.(B) SDS-PAGE of SARS-CoV-2 RBD-SD1 variant probes with and without reducing agent.Molecular weight marker (MWM) and the WA-1 RBD-SD1 probe were run alongside the variant RBD-SD1 probes.(C) Biotinylation and receptor recognition of the SARS-CoV-2 variant RBD-SD1 probes.The level of biotinylation was evaluated by capture of the RBD-SD1 probes at 5 μg/ml onto the streptavidin biosensors using Octet.Receptor binding was assessed with 500 nM of dimeric ACE2.Error bars represent standard deviation of triplicate measurements.(D) Antigenicity assessment of the SARS-CoV-2 variant RBD-SD1 probes.Responses to RBD-directed, NTD-directed, and S2 subunit-directed antibodies were shown in cyan, orange, and rose, respectively.Bar scale between 0 and 3. Negative values not depicted.Variants with the L452R mutation lost binding to A19-46.1.https://doi.org/10.1371/journal.pone.0268767.g007