Targeting the Spike Receptor Binding Domain Class V Cryptic Epitope by an Antibody with Pan-Sarbecovirus Activity

ABSTRACT

Novel therapeutic monoclonal antibodies (MAbs) must accommodate comprehensive breadth of activity against diverse sarbecoviruses and high neutralization potency to overcome emerging variants. Here, we report the crystal structure of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) receptor binding domain (RBD) in complex with MAb WRAIR-2063, a moderate-potency neutralizing antibody with exceptional sarbecovirus breadth, that targets the highly conserved cryptic class V epitope. This epitope overlaps substantially with the spike protein N-terminal domain (NTD) -interacting region and is exposed only when the spike is in the open conformation, with one or more RBDs accessible. WRAIR-2063 binds the RBD of SARS-CoV-2 WA-1, all variants of concern (VoCs), and clade 1 to 4 sarbecoviruses with high affinity, demonstrating the conservation of this epitope and potential resiliency against variation. We compare structural features of additional class V antibodies with their reported neutralization capacity to further explore the utility of the class V epitope as a pan-sarbecovirus vaccine and therapeutic target.

IMPORTANCE Characterization of MAbs against SARS-CoV-2, elicited through vaccination or natural infection, has provided vital immunotherapeutic options for curbing the COVID-19 pandemic and has supplied critical insights into SARS-CoV-2 escape, transmissibility, and mechanisms of viral inactivation. Neutralizing MAbs that target the RBD but do not block ACE2 binding are of particular interest because the epitopes are well conserved within sarbecoviruses and MAbs targeting this area demonstrate cross-reactivity. The class V RBD-targeted MAbs localize to an invariant site of vulnerability, provide a range of neutralization potency, and exhibit considerable breadth against divergent sarbecoviruses, with implications for vaccine and therapeutic development.

INTRODUCTION

The continued emergence of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants of concern (VoCs) and potential for additional zoonotic spillover amid the COVID-19 pandemic has demonstrated the need for both prophylactic protection and therapeutic options with broad immunity against diverse sarbecoviruses. SARS-CoV-2 infection in seropositive individuals, whether through prior infection or vaccination, are now more common due to recently emergent Omicron variants (1–4) and illustrate the increased resistance to prophylactics and therapeutics, as SARS-CoV-2 infections remain high with significant morbidity and mortality worldwide (5–8). Isolation of neutralizing monoclonal antibodies (MAbs) from the plasma of convalescent donors or immunized individuals has yielded beneficial clinical tools for prevention and treatment (9–11). All neutralizing MAbs currently in clinical development, or those that have received emergency use authorization (EUA), target the receptor binding domain (RBD) of the SARS-CoV-2 spike (S) protein (12, 13). The S protein is the primary target of the host immune response and consists of a large ectodomain that trimerizes into a distinctive shape, with a single-pass transmembrane region that anchors the spike to the coronavirus membrane and a short intravirion tail. The RBD is essential for viral entry and subsequent infection, as it directly interacts with the angiotensin-converting enzyme 2 (ACE2) receptor (14, 15). Viral attachment to ACE2 prompts a large conformational rearrangement in the S protein that results in S2′ cleavage and virus-host membrane fusion through the S2 subunit. Most of the RBD-targeted MAbs discovered early in the pandemic have become obsolete, as VoCs accumulated mutations resulting in neutralization escape (16). Current neutralizing MAbs also lack the breadth to be effective against diverse sarbecoviruses (17). All these factors highlight the necessity of expanding the available repertoire of broad and potent SARS-CoV-2 neutralizing MAbs.

Canonically, isolated RBD-targeted MAbs have been clustered into four distinct classes (I to IV) that each bind a particular epitope on the RBD (18). Regions of the RBD targeted by MAbs focused on class I and II epitopes are particularly prone to viral escape, resulting in variants with increased transmissibility and pathogenicity (19, 20). Thus, immunotherapeutic strategies focused on other epitopes that also demonstrate potency while targeting conserved regions of the RBD are of particular interest. These epitopes highlight vulnerabilities of the virus that complicate viral escape. MAbs designated class V were recently described (21) and represent an attractive target for immunogen design and pan-CoV vaccine development. This region, formerly described as the “E465 patch” (22), is nearly invariant across clade 1b sarbecoviruses, of which SARS-CoV-2 is a member, and remains highly conserved among divergent sarbecoviruses, including members of clades 1a, 2, 3, and 4, such as those ubiquitous within bat populations. Despite the mutational constraint of this antigenic site, class V MAbs exhibit significant variability in neutralization efficiency, ranging from robust neutralization, albeit lower than that of most class I and II MAbs, to absence of viral neutralization (21, 23–25).

In addition to the RBD, receptor-binding subunit 1 (S1) also consists of an N-terminal domain (NTD), which shields the adjacent RBD within the trimer prior to receptor binding when the S protein is in the closed (RBD down) conformation (25, 26). The NTD-RBD interface is highly conserved; VoC mutations and deletions of the NTD are primarily found within the flexible loops exposed at the NTD surface that make up the antigenic supersite, exploited by the majority of NTD-targeted MAbs (27–29). In the transition between receptor-inaccessible to receptor-accessible prefusion states, the RBD unfurls from the core of the S protein, away from the NTD, such that the NTD-RBD interface buried in the closed trimer is now exposed, with minimal interactions between the RBD and NTD from the adjacent protomer.

Here, we report the X-ray crystal structure of the SARS-CoV-2 WA-1 RBD, in complex with the human monoclonal antibody Walter Reed Army Institute of Research 2063 (WRAIR-2063), a class V MAb isolated from a COVID-19 convalescent donor (30). We demonstrate that this antibody binds a cryptic, highly conserved epitope that is resistant to mutations found in current VoCs. We link our structural characterization with MAb kinetics measurements and pseudovirus neutralization experiments across several sarbecoviruses, revealing the neutralization breadth and efficacy of the WRAIR-2063 MAb. In addition, we demonstrate that saturation of the SARS-CoV-2 S protein with WRAIR-2063 allows for other classes of MAbs to bind the RBD and appears to enhance binding of class IV MAb CR3022, as well as MAbs that target cryptic epitopes on the NTD and S2. Finally, we compare our results with available structures of class V RBD-targeted MAb complexes and correlate structural details with neutralization breadth and potency. Our findings provide a framework for the development of potent, multi-targeting antivirals with broad cross-reactivity against diverse sarbecoviruses.

RESULTS

Crystal structure of WRAIR-2063 in complex with SARS-CoV-2 WA-1 RBD.

WRAIR-2063 was previously identified by single B cell sorting from the peripheral blood of a convalescent human donor infected with the original WA-1 strain, using a combination of SARS-CoV-2 probes that included the spike ferritin nanoparticle (SpFN), a vaccine candidate presently in clinical trials (30–32). After single-cell SARS-CoV-2 reactive B cells were sequenced, WRAIR-2063 was produced as a human IgG1 in Expi293F cells and purified for functional and structural characterization. To provide a structural understanding of the class V RBD-targeted antibody WRAIR-2063 (Fig. 1A), we determined the X-ray crystal structure of the SARS-CoV-2 WA-1 RBD in complex with WRAIR-2063 and WRAIR-2151 fragment antigen-binding (Fab) regions, to a resolution of 2.77 Å (Table 1). WRAIR-2151 was included to stabilize the complex and provide better crystal packing, as we were unable to crystallize the RBD with WRAIR-2063 alone. We previously described a structure of the WRAIR-2151 Fab in complex with the SARS-CoV-2 RBD (30). As the root mean square deviation (RMSD) value between the WRAIR-2063-2151–RBD and WRAIR-2151–RBD structures was low (0.78 Å across RBD Cα atoms), we concluded that binding of WRAIR-2151 to the RBD did not subsequently influence association with WRAIR-2063. Thus, for this study, we limited our analysis to the WRAIR-2063–RBD complex (Fig. 1B). The determined structure contained RBD residues 334 to 528 and WRAIR-2063 residues 1 to 226 and 1 to 218 of the heavy and light chains, respectively. Modest electron density corresponding to an N-acetylglucosamine moiety was observed attached to RBD N343. The WRAIR-2063 heavy chain is encoded by IGHV3-33*01 with a 15-residue complementarity-determining region (CDR) H3; the light chain is encoded by IGKV2-30*01, with a 9-residue CDR L3 (30). The heavy and light chains are altered by 1 and 3 somatic hypermutations, respectively. A search of the CoV-AbDab repository (33) did not return any close match for the heavy chain, highlighting the uniqueness of the WRAIR-2063 heavy chain and CDR H3, while similar light chain sequences are found in several MAbs.

FIG 1.

Crystal structure and epitope analysis of the SARS-CoV-2 RBD with WRAIR-2063 Fab. (A) Structure of the SARS-CoV-2 RBD (gray surface), aligned with complexes representative of each canonical RBD-targeted antibody epitope class. Class I, CB6 (pale yellow, PDB 7C01); class II, P2B-2F6 (pale green, PDB 8DCC); class III, S309 (teal, PDB 7SOC); class IV, CR3022 (magenta, PDB 7JN5). The class V epitope (represented by WRAIR-2063) is outlined and shaded with purple. (B) Key antibody contact residues of the RBD are shown as sticks and colored to indicate heavy chain (dark purple) and light chain (light purple). Only residues with a BSA of >20 Ų are highlighted. The RBD is rotated about the z axis in relation to the RBD displayed in panel A. (C) BSA (Ų) contributions of the heavy and light chain CDR loops. (D) Surface representation of the RBD is highlighted with the WRAIR-2063 in dark and light violet. CDR loops are displayed and labeled, with critical interface residues of the CDR L3, CDR L1, CDR H2, and CDR H3 shown in expanded views. Hydrogen bonds are indicated with dashed lines. Electron density mesh surface is contoured at 1 σ. (E) The RBD (ribbon representation with surface overlay) with the WRAIR-2063 footprint shown in purple. Mutations found in VoCs are displayed as red spheres and labeled in red.

TABLE 1.

Crystallographic data collection and refinement statistics^a

Parameter	Value for WA-1 RBD with WRAIR-2063 and WRAIR-2151
Crystal growth conditions	100 mM Bis-Tris propane HCl (pH 8.5), 200 mM Na malonate dibasic, 20% (wt/vol) PEG 3350
Cryoprotectant	Mother liquor + 15% (wt/vol) 2,3-butanediol
Beamline (detector)	24-ID-E (Dectris Eiger 16M pixel)
Search models	7N4M, 5GGU (VH), 6B0E (VL)
Space group	P21
Cell constants
a, b, c (Å)	97.1, 97.4, 141.1
α, β, γ (°)	90, 100, 90
Wavelength (Å)	0.9792
Resolution (Å)	139.1–2.76 (2.98–2.86, 2.85–2.76)
Rmerge	0.139
I/σ(I)	9.3 (1.7, 1.0)
Completeness (%)	98.7 (99.7, 89.9)
Redundancy	6.9 (7.2, 5.95)
CC1/2	(0.981, 0.773)
Rpim	0.057 (0.415, 0.6)
Refinement statistics
Resolution (Å)	19.95–2.77 (2.87–2.77)
No. of unique reflections	65,391 (6,207)
Rwork/Rfree* (%)	21.3/26.0
No. of atoms	16,527
Protein	16,386
Ligand/ion	64
Water	77
B factors (Å2)
Protein	91.7
Ligand/ion	103.2
Water	79.3
RMSD
Bond length (Å)	0.009
Bond angle (°)	1.19
Ramachandran plot (%)
Favored/allowed/outliers	98.2/1.8/0.0
PDB ID	8EOO

^aValues in parentheses represent the highest-resolution shells. Values for up to 2 high-resolution shells are presented. R_free was calculated using ~5% randomly selected reflections.

The WRAIR-2063 epitope centers on the external face of the RBD, distinct from class I to IV RBD-targeted MAbs (Fig. 1A), distal from the ACE2 recognition site, and overlaps that of the newly designated class V MAbs (21). This was confirmed by structural superimposition of the WRAIR-2063–RBD complex with RBD-MAb structures representing classes I to IV (Fig. 1A) (34–37). The WRAIR-2063–RBD interface incorporates a network of hydrogen bonds between the RBD and CDR H3, CDR L1, and CDR L3 of WRAIR-2063 (Fig. 1B and C; see Table S1 in the supplemental material), with minimal contributions from CDR H2, for a total of 13 H-bonds and no discernible salt bridges. This finding was confirmed by shotgun mutagenesis epitope mapping, wherein binding of single alanine RBD mutants is compared to that of wild-type RBD (38). These data demonstrated that residues I468 and S469 are critical components of the interface, both of which participate in H-bonds with WRAIR-2063 LC N27 through their backbone N atoms (Fig. 1D, lower left inset), and are 68.6% and 75.4% buried, respectively, at the WRAIR-2063–RBD interface. The antibody-antigen interaction buries a total of 869.3 Ų, with the heavy and light chains contributing 44% and 56% of the total buried surface area (BSA), respectively (Fig. 1C). The WRAIR-2063 epitope does not overlap any mutations found in the SARS-CoV-2 VoCs or variants of interest (VoIs), namely Alpha (B.1.1.7), Beta (B.1.351), Delta (B.1.617.2), Eta/Iota (B.1.525/6), Epsilon (B.1.427/9), or Omicron BA.1, BA.2, BA.2.12.1, BA.2.75, BA.4/5, BQ.1.1, XBB.1, or XBB.1.5 (Fig. 1E), demonstrating the utility of this class V RBD-targeted MAb in pan-SARS-CoV-2 recognition. As the binding site of WRAIR-2063 is distal to the ACE2 recognition motif, with no overlap in epitope, only modest reduction of the RBD-ACE2 interaction was observed in the presence of WRAIR-2063 (30).

To further evaluate the WRAIR-2063 epitope in the context of full-length spike trimer, the WRAIR-2063–RBD complex was superimposed with one RBD of the spike in the prefusion, closed conformation (all RBD down; PDB 6ZGE) (39) and the RBD in the partially open conformation (one RBD up; PDB 6X2A) (15). In the closed conformation, the NTD of an adjacent protomer shields the class V epitope, which would preclude binding of the WRAIR-2063 MAb (Fig. 2A). The frequency of mutations in the NTD (VoCs in particular) at the RBD interface appears to be quite low, intimating the functional importance of the interfacial residues in maintaining stability of the closed spike trimer (40). In the open conformation, however, there is no interaction of the WRAIR-2063 epitope with the NTD (Fig. 2B and D), implying that the spike must be in the open or transitional state to be accessible to the MAb. In the closed state, the overlap between the NTD and WRAIR-2063 binding sites is extensive and nearly entirely shared (Fig. 2D). In the open state, there are no significant steric hindrances or clashes between WRAIR-2063 and the rest of the spike, with likely minimal interaction between WRAIR-2063 and the NTD in the open spike conformation (Fig. 2C). Binding of the MAb with the NTD yielded a dissociation constant (K_d) of ~200 nM (41), a 200-fold weaker interaction than we report here for the RBD, although affinity was determined through surface plasmon resonance, in contrast with our analysis by biolayer interferometry (BLI). Superimposition of the open spike with the WRAIR-2063–RBD complex suggests that loops 109 to 113 and 131 to 137 of the NTD and loops 57 to 63 of the WRAIR-2063 light chain (LC) may mediate this interaction (Fig. 2C). Neither the NTD–WRAIR-2063 nor the NTD-RBD interface residues fall within antigenic sites recognized by NTD-targeted MAbs (27).

FIG 2.

WRAIR-2063 epitope footprint in the context of the SARS-CoV-2 spike trimer. (A) Structural alignment of the WRAIR-2063–RBD with the RBD of the S-2P spike trimer, in the closed conformation (PDB 6ZGE). The aligned RBDs are shown in dark gray, the WRAIR-2063 Fab in violet, and the NTD in pale blue. In this conformation, the WRAIR-2063 epitope is shielded by the NTD. (Lower panel) Butterfly view of the RBD-NTD interface; interacting residues with a BSA of >20 Ų are indicated. (B) Structural alignment of the WRAIR-2063–RBD with the RBD of the S-2P trimer in the partially open (1 RBD up) conformation (PDB 6X2A). Movement of the RBD allows the NTD to shift, exposing the WRAIR-2063 epitope. (C) Possible interface between the NTD (pale blue; bottom of panel) and the WRAIR-2063 LC (violet; top of panel), based on the superimposition of the WRAIR-2063–RBD complex with the open spike. (D) Footprints of the NTD in the closed spike (left), the overlap (orange) of the NTD and WRAIR-2063 footprints in the closed spike (center), and the NTD and WRAIR-2063 footprints in the open spike. In this orientation, interaction of the RBD with the NTD is minimal, displaying no overlap with the WRAIR-2063 epitope.

Breadth of cross-reactivity and epitope conservation of the WRAIR-2063 MAb.

We determined the affinity of WRAIR-2063 to a panel of purified VoC RBDs using BLI. As expected (30), the association of WRAIR-2063 to VoC RBDs was comparable to that of WA-1 (Fig. 3A and B), as WRAIR-2063 bound to all RBDs tested, including SARS-CoV, with low to subnanomolar affinity and minimal dissociation observed. To further evaluate the interface, we created a series of single or double “class V knockout” mutants by site-directed mutagenesis, based on our structural characterization. We were able to analyze only five mutants (Fig. 3A and B), as mutation at the class V epitope appeared to severely impact protein expression, consistent with prior reports of attempts to mutate certain residues within this epitope (23). Although none of our mutants fully eliminated binding, a 10-fold decrease in affinity of the R357A R466A and I468A S469A double mutants was observed compared to that of WA-1 or VoC RBDs, demonstrating that the residues targeted for mutagenesis are involved in the WRAIR-2063–RBD interaction. Furthermore, a more rapid off-rate (k_off) was detected for the class V epitope mutants, particularly the I468E E471R mutant, indicative of a much weaker interaction with the MAb (Fig. 3B).

FIG 3.

WRAIR-2063 binding to VoCs and knockout mutants. (A) Representative sensorgrams used to determine binding kinetics of WRAIR-2063 with SARS-CoV-2 WA-1, variants of concern, and class V knockout mutants, as measured by BLI. (B) Kinetics constants were determined using a minimum of four dilutions of the RBD (31.3 nM to 500 nM) and fitted using a 1:1 binding model.

To understand the conservation of the WRAIR-2063 structural epitope within sarbecoviruses and across betacoronaviruses more broadly, we performed phylogenetic analysis, which revealed that the WRAIR-2063 epitope is nearly identical between SARS-CoV-2 and SARS-CoV (Fig. 4A and B), with the following residue differences N354E, K462R, and E471V. Alignment of the SARS-CoV-2 RBD–WRAIR-2063 structure with the SARS-CoV spike showed a similar arrangement to that observed with the SARS-CoV-2 spike; in the closed state, the class V epitope is completed shielded by the NTD of the adjacent protomer and then completely exposed in the open conformation. Residues mediating the WRAIR-2063–RBD interaction are likely very similar to those identified through our structural characterization with the SARS-CoV-2 WA-1 RBD, based upon BSA determined from superimposition of the SARS-CoV RBD (42) with the SARS-CoV-2 RBD–WRAIR-2063 complex (Fig. 4B).

FIG 4.

WRAIR-2063 epitope conservation and MAb cross-reactivity. (A) Primary sequence and structural analysis of the conserved WRAIR-2063 epitope across betacoronaviruses. Epitope residues are numbered in relation to the WA-1 (Wuhan-Hu-1) reference. The height and color of the histogram indicate the strength of the interaction between WRAIR-2063 and the RBD. Sequences are ordered according to phylogenetic relationship, based on a maximum likelihood phylogenetic tree derived from RBD primary sequences. The RBD structure shown in surface representation is colored to depict sequence conservation between the SARS-CoV-2 and SARS-CoV-1 RBDs, with the WRAIR-2063 epitope outlined in violet. (B) Heat map of SARS-CoV-2 and SARS-CoV RBD with NTD or WRAIR-2063 interface residues and corresponding percent buried surface area. Key antibody contact residues of the SARS-CoV RBD (gray surface; PDB 2GHV) are shown as sticks and colored by heavy chain (dark purple) and light chain (light purple). Only residues with a BSA of >20 Ų are highlighted. (C) Binding of WRAIR-2063, WRAIR-2057, WRAIR-2134, WRAIR-2151, and WRAIR-2173 to sarbecovirus and merbecovirus RBDs, as measured by BLI. The heat map displays area under the curve values. RBD molecules were immobilized to the probe, with WRAIR-2063, -2057, -2134, -2151, or -2173 IgGs in solution.

Alignment of the RBD from SARS-CoV-2 and related reference genomes with other representative sarbecoviruses showed that the WRAIR-2063 epitope is invariant within clade 1b, partially conserved across clades 1a and 2, poorly conserved in clades 3 and 4, and highly variant in other betacoronaviruses, such as Middle East respiratory syndrome (MERS)-CoV (Fig. 4A). To verify our phylogenetic characterization, we tested cross-reactivity of WRAIR-2063 with RBDs from diverse sarbecoviruses and MERS-CoV (Fig. 4C). We observed modest reduction in binding of WRAIR-2063 to other sarbecovirus RBDs, when normalized to WA-1, with 7 to 21% weaker binding observed to other clade 1b RBDs, 7 to 10% weaker binding to clade 1a RBDs, 3 to 21% weaker binding to clade 2 RBDs, and 3 to 38% weaker binding to clades 3 to 4 RBDs (Fig. 4C). No binding was observed to MERS-CoV Erasmus Medical Center or England1 (ENG1) RBDs, suggesting that coverage is limited to sarbecoviruses and may not extend to the greater betacoronavirus family. For comparison, we included class V MAbs WRAIR-2057 and WRAIR-2134, WRAIR-2151, a class IV RBD-targeted MAb, and WRAIR-2173, a class I RBD-targeted MAb, tested against the same panel of RBDs (30). Both WRAIR-2151 and WRAIR-2173 were previously shown to have potent neutralization activity against SARS-CoV-2 pseudoviruses, including VoCs (30). When tested against more diverse sarbecovirus RBDs, in comparison to WRAIR-2063, WRAIR-2151 and WRAIR-2173 showed less binding overall to the sarbecovirus RBDs. Interestingly, both additional class V MAbs, WRAIR-2057 and WRAIR-2134, with 50% inhibitory concentration (IC₅₀) titers ~10-fold or 100-fold less potent than those of WRAIR-2063, respectively, showed considerable variation in binding across our panel of betacoronavirus RBDs. WRAIR-2134 followed a similar trend as WRAIR-2063, albeit with diminished binding to RBDs from clades 1a, 2, and 3 and minimal interaction with clade 4 RBDs. WRAIR-2057 performed similarly to WRAIR-2173, with little to no interaction observed with RBDs outside clade 1b, despite sharing the class V epitope. Combined, these data demonstrate markedly broad recognition by WRAIR-2063 and strong affinity for divergent sarbecovirus RBDs.

Neutralization of clade 1a and 1b pseudoviruses by WRAIR-2063.

To further explore the breadth and potency of WRAIR-2063, we performed neutralization experiments with spike-expressing pseudotyped lentivirus on 293T-ACE2 cells, utilizing diverse sarbecovirus spike proteins, including WA-1, D614G, Omicron BA.2.12.1, Omicron BA.4/5, Omicron BQ.1.1, Omicron XBB.1, Omicron XBB.1.5, RaTG13, and Pangolin-GX from sarbecovirus clade 1b, and SARS-CoV, Frankfurt1, Civet007-2004, WIV1, and SHC014 from clade 1a (43). We included class V MAb S2H97 (21, 44) within our panel of control MAbs. We observed moderate neutralization from our analyses for SARS-CoV-2 D614G at 1.5 μg mL⁻¹, while WRAIR-2063 showed robust neutralization against SARS-CoV at 35 ng mL⁻¹ (Fig. 5A). This correlates well with our prior lentivirus-based pseudovirus neutralization results, which yielded an IC₅₀ of ~1.6 μg mL⁻¹ for the D614G variant and <100 ng mL⁻¹ for SARS-CoV; WRAIR-2063 neutralized WA-1 at ~100 ng mL⁻¹ (30). While an approximate 2-fold lower in IC₅₀ was determined for S2H97 against SARS-CoV-2 variants, when S2H97 was compared to WRAIR-2063, WRAIR-2063 outperformed S2H97 against more diverse sarbecoviruses in our pseudovirus neutralization assay (Fig. 5A), including Pangolin-GX, SARS-CoV, Frankfurt1, and Civet, suggesting that WRAIR-2063 may be more resistant to SARS-CoV-2 polymorphism. Neutralization by WRAIR-2063 and S2H97 was equally poor against RaTG13, WIV1, and SHC014. This difference is striking, particularly because the WRAIR-2063 epitope is well conserved across clade 1a and 1b RBDs (Fig. 5B); the binding difference is likely related to accessibility of the epitope and prefusion dynamics of the S protein. Other control antibodies, namely CR9114, an influenza hemagglutinin-targeted MAb, and DH1047, which binds both the SARS-CoV and the SARS-CoV-2 RBDs at an epitope between classes I and IV, performed as expected, with no neutralization detected for CR9114 (10, 45, 46) and robust neutralization by DH1047 detected for all except the Omicron subvariants.

FIG 5.

Pseudovirus neutralization of diverse sarbecoviruses. (A) Neutralization curves of WRAIR-2063 MAb from a pseudovirus assay, assessed against multiple sarbeco-pseudoviruses, with S2H97, CR9114, and DH1047 included as comparative controls. Geometric mean IC₅₀ values are presented in the table from experiments outlined in the figure above. (B) Sequence alignment of RBDs from clade 1a and 1b sarbecoviruses, with the WRAIR-2063 epitope highlighted in purple. *, :, . denote identical, similar and less similar residues, respectively.

Correlation of class V structural characteristics with neutralization potency.

With the ultimate goal of identifying SARS-CoV-2 neutralizing antibodies that display superior potency and breadth to minimize treatment resistance and emergence of escape mutants, we analyzed available structures of other class V RBD-targeted MAbs and compared structural features and reported RBD association and neutralization levels. At present, to our knowledge, there are 11 published class V MAbs, including WRAIR-2063, -2057, and -2134. Although the general class V binding site is conserved (Fig. 6A), broad structural differences in antibody-RBD association are noted, such as the angle of approach of the MAb to the RBD and BSA of the antibody-RBD interface (Fig. 6B and C). Despite this observation, the epitopes of all 11 MAbs are strikingly similar at the primary sequence level, with several of the same interface residues targeted by all 11 MAbs (Fig. 7A), including W353, R355, K462, F464, and R466; these residues also play a substantial role in mediating the RBD-NTD interaction. Y396, P426, D428, R457, P463, E465, I468, and L518 are engaged by 7 to 9 of the 10 MAbs analyzed.

FIG 6.

Structural comparative analysis of available class V RBD-targeted MAbs. (A) Alignment of the WRAIR-2063–RBD complex with structures of additional RBD-class V antibodies. All structures are shown in the same orientation with respect to the RBD. PDB identifiers and references for each structure are listed in Table 2. (B) Three views showing the binding angle between RBD and the antibody center of mass for each class V MAb. (C) Heat map of SARS-CoV-2 RBD–class V RBD-targeted MAb interface residues and corresponding percent buried surface area.

FIG 7.

In vitro characterization of class V MAbs. (A) Sequence of the SARS-CoV-2 RBD with NTD interacting and, WRAIR-2063, S2H97, ION-300, 6D6. 7D6, WRAIR-2057, COVOX-45, X17, FD20, WRAIR-2134, and N-612-056 binding sites highlighted. (B) Representative sensorgrams used to determine binding kinetics of class V MAbs with SARS-CoV-2 WA-1, as measured by BLI. Kinetics constants were determined using a minimum of four dilutions of the RBD (31.3 nM to 500 nM), fitted using a 1:1 binding model, and indicated in the table at the bottom of the figure. (C) Neutralization curves of class V RBD-targeted MAbs from a pseudovirus assay, assessed against WA-1 pseudotyped virus. IC₅₀ values for each MAb are indicated.

Neutralization potency, however, quantified through similar pseudotyped lentivirus or vesicular stomatitis virus (VSV) neutralization experiments, appears to vary considerably across the 11 MAbs (Table 2). Studies characterizing WRAIR-2063, S2H97, ION-300, 6D6, and 7D6 all report IC₅₀ values of less than 1 μg mL⁻¹, indicating reasonable neutralization capacity. WRAIR-2057, COVOX-45, X17, and FD20 showed modest neutralization, with IC₅₀ values between 1 and 4 μg mL⁻¹, while WRAIR-2134 and N-612-056 had notably weak neutralization capacity, with IC₅₀s of >7 μg mL⁻¹. Reported binding affinities of class V MAbs to the WA-1 RBD also differ considerably, from low nanomolar to low picomolar values (Table 2). However, significant differences are noted in the methods and reagents used to determine dissociation constants.

TABLE 2.

Collated information for all available class V RBD-targeted MAbs

MAb	PDB ID	IC50 in this work (μg mL−1)	IC50 previously published (μg mL−1)	Kd in this work (nM)	Kd previously published (nM)	Angle of approach (°)	Total BSA (Å2)	Total no. of H-bonds	Total no. of salt bridges	Referencea
WRAIR-2063	8EOO	0.45	0.14	1.1	<0.001	64.9	869.3	13	0	30
S2H97	7M7W	0.31	0.35	0.9	0.03	38.9	808.8	9	6	21
ION-300	7BNV	0.99	0.55	3.65	3.4	46.6	922.1	10	3	25
6D6	7EAN	1.37	0.21	3.62	0.01	67.7	937.1	15	2	47
7D6	7EAM	0.33	0.04	3.41	0.003	62.2	676.1	15	1	47
WRAIR-2057	7N4I	1.85	3.27	6.3	<0.001	63.5	877.8	6	3	30
COVOX-45	7BEL	0.32	2.0	4.43	5.9 (Fab)	42.3	1,090.5	23	6	49
X17	7X7T	2.27	1.67	4.72	Not reported for WA-1	38.1	937.0	15	5	63
FD20	7CYV	0.6	1.7	3.07	5.6	50.6	638.1	15	6	23
WRAIR-2134	8SMT	18.59	16.74	4.26	0.06	44.7	926.1	13	7	Under review; 30
N-612-056	7S0B	4.47	7.0	6.31	2.98	61.3	860.9	14	4	24

^aReferences indicate original publication sources.

To provide a more accurate comparison of neutralization potency and binding affinity across class V RBD-targeted MAbs, we produced all 10 MAbs to test by the same methods used to characterize WRAIR-2063. The variable regions of each MAb’s heavy and light chains were codon optimized for expression in human cells, synthesized, and cloned into expression vectors containing the constant regions of human IgG1, Igκ, or Igλ. The binding affinity to WA-1 RBD by BLI was assessed, as was neutralization of WA-1 pseudotyped virus. BLI yielded consistent subnanomolar affinities across all MAbs (Fig. 7B). Only S2H97, however, exhibited binding kinetics similar to that of WRAIR-2063; all other class V MAbs presented an approximate 10-fold higher rate of RBD dissociation. WA-1 (without the D614G mutation) neutralization by WRAIR-2063 was ~2-fold weaker than that of our prior reports, with an IC₅₀ of 451 ng mL⁻¹ (Fig. 7C). Under our experimental conditions, all class V MAbs generally neutralized as expected, compared to published values (Table 2). A 6- to 8-fold higher IC₅₀ values (reduced potency) was observed for 6D6 and 7D6, while slight increases in neutralization potency were seen for WRAIR-2057, COVOX-45, FD20, and N-612-056.

Competition and binding enhancement by WRAIR-2063 of disparate MAbs to the S protein.

We performed a tandem competition assay by BLI to explore the compatibility of WRAIR-2063 with other neutralizing MAbs and further investigate the behavior of the S protein when exposed to WRAIR-2063 in combination with other MAbs that target different S protein epitopes. Under these assay conditions, stabilized WA-1 S protein (HexaPro) was first saturated with WRAIR-2063 and then exposed to WRAIR-2063 in combination with MAbs that target epitopes other than the class V site (Fig. 8A). WRAIR-2057 was included for comparative purposes and, as expected, was competed by WRAIR-2063, consistent with earlier competition experiments performed on the isolated RBD (28, 30). No change in binding was observed for RBD class I/II (WRAIR-2125) or IV (WRAIR-2151) MAbs or for a MAb targeting the NTD antigenic supersite (WRAIR-2039); the epitopes of these three MAbs are readily accessible on the S trimer and maintained following incubation with WRAIR-2063. Notably, when S protein was bound to WRAIR-2063, binding of CR3022, a class IV RBD MAb that recognizes a cryptic epitope inaccessible in the RBD down conformation, increased by 6-fold, suggesting that the S trimer had transitioned to a more open conformation. Similarly, binding to WRAIR-2028 and WRAIR-2024, which target cryptic sites remote from the RBD, also increased upon WRAIR-2063 binding, albeit more modestly, by 3-fold and 2-fold, respectively. The respective epitopes of WRAIR-2028 and WRAIR-2024 are located in the NTD and S2, recapitulating the implication that changes in the S trimer conformation upon binding to the RBD class V site can have distal effects on other domains of S.

FIG 8.

WRAIR-2063 antibody competition with diverse spike-targeted antibodies. (A) BLI-based binding competition assay in which stabilized, biotinylated S protein (HexaPro) was immobilized to streptavidin biosensors, saturated with WRAIR-2063, and then incubated with a combination of WRAIR-2063 and other S protein-directed MAbs. Values represent the percent residual binding of the indicated second MAb after antigen (S protein) saturation with WRAIR-2063. Shading from dark red to white indicates competition strength, while shading from dark blue to white indicates enhancement of binding in the presence of WRAIR-2063. The BLI sensorgram displays the final association step, after WRAIR-2063 saturation, and is representative of duplicate experiments. The WRAIR-2063 response has been subtracted from that of the MAb combinations response to emphasize binding enhancement. (B) Alignment of the RBDs from the WRAIR-2063–RBD and CR3022-RBD (PDB 6W41) complexes. (C) Superimposition of the WRAIR-2063–RBD and CR3022-RBD complexes onto the 1 RBD up S trimer (left; PDB 6VYB) and the 3 RBD up S protein (right; PDB 7V82). RBDs are displayed in dark gray surface representation, with the NTD, WRAIR-2063, and CR3022 shown in cartoon representation, colored blue, purple, and orange, respectively. Steric clashes of CR3022 with the S2 or RBD are highlighted in red. The NTD is not displayed in the profile view of the 3 RBD up S trimer.

DISCUSSION

In this study, we structurally characterized a novel SARS-CoV-2 RBD-targeted monoclonal antibody, WRAIR-2063, isolated from a convalescent donor, in complex with the WA-1 RBD. WRAIR-2063 binds a cryptic RBD epitope, the class V site, which is hidden by the NTD of an adjacent spike protomer when in the closed spike conformation (3 RBD down). The WRAIR-2063 epitope becomes exposed only when at least 1 RBD is in the up conformation, when the S trimer is open or in a closed-to-open transitional state. Superimposition of the WRAIR-2063–RBD complex with the RBD of the S trimer in the 1, 2, or 3 RBD up (open) conformation shows no major steric clashes with RBDs of the adjacent protomers, suggesting that more than one WRAIR-2063 could bind S trimer. The MAb likely has minimal interaction with the NTD of the adjacent protomer within the spike trimer, and the isolated NTD binds WRAIR-2063 with reasonable affinity. Binding of WRAIR-2063 may ultimately affect the prefusion-to-postfusion spike transition, partly accounting for its neutralizing capability, as observed for FD20. This could result in the rapid destruction of the S trimer (23). Class V antibodies 6D6 and 7D6 also caused dissociation of stabilized S trimer (S-2P), resulting in more potent spike shedding than that caused by class IV MAb CR3022 (47). We demonstrated that S protein binding to CR3022 and other MAbs with cryptic S protein epitopes is significantly enhanced in the presence of WRAIR-2063, suggesting that binding of WRAIR-2063 triggers conformational changes in the S trimer that impact accessibility of these cryptic sites. As previously noted for SARS-CoV (26), when in the closed conformation, the SARS-CoV RBD packs more closely with the NTD, while in the same conformation, the SARS-CoV-2 RBD is positioned slightly away from the NTD and closer to the spike trimer’s central cavity. This may indicate that positioning of the class V MAb when bound to the RBD prompts greater conformational rearrangements in the S trimer, ultimately influencing neutralization potency and breadth of MAbs that target related epitopes. Further structural characterization of WRAIR-2063 with S trimer and additional MAbs may clarify this observation.

The WRAIR-2063 epitope is well conserved across divergent sarbecoviruses, which indicates that this epitope is limited in its plasticity in allowing variation and resulting viral escape, as other studies have reported a direct correlation between viral escape and epitope conservation (21). Mutation of the class V epitope by deep mutational scanning of the RBD showed that mutation of these residues predominantly resulted in decreased protein expression, ACE2 binding, and viral infectivity, denoting the functional and structural constraint of several class V epitope residues, particularly around E465 (22, 23). To date, none of the SARS-CoV-2 VoCs display mutations in the class V antigenic site, and the frequency of natural mutation at the epitope remains negligible (https://cov.lanl.gov/), which suggests that the probability of viral mutation at the WRAIR-2063 epitope is quite low. We demonstrated that broad recognition by WRAIR-2063 was possible, with high-affinity binding to RBDs from all known VoCs and only modest reductions in binding to RBDs from other sarbecoviruses. Additionally, we observed moderate neutralization of VoCs tested in our pseudovirus assay, compared to WA-1, with IC₅₀ values ranging from 35 ng mL⁻¹ to ~5 μg mL⁻¹. This highlights the class V epitope as a vulnerability that should be considered during pan-CoV immunotherapeutic design. For reference, FDA filings report the neutralization activities of therapeutic anti-SARS-CoV-2 MAbs under EUA in the United States as 6.4 ng mL⁻¹, 100.1 ng mL⁻¹, and 20 ng mL⁻¹ against the WA-1 isolate by bebtelovimab, sotrovimab, and bamlanivimab, respectively.

It is unknown what percentage of the antibody response to SARS-CoV-2 infection or vaccination is class V RBD-targeted MAbs; overall immunogenicity appears to be dominated by RBD-specific MAbs, followed by NTD-specific MAbs. Estimates are based on widely employed strategies for sorting SARS-CoV-2 reactive IgG memory B cells from peripheral blood mononuclear cells of donors. However, these methods may be biased by the SARS-CoV-2 protein used as a bait during initial B cell isolation, typically the prefusion S protein. While other class V MAbs have been isolated using only prefusion S as the bait (21, 23, 48, 49), our strategy applied a combination of four components: the multivalent SpFN (which displays 8 S trimers on an assembled ferritin nanoparticle) with S1, S2, and RBD. Continued development of isolation strategies using class V epitope knockout variants may further optimize isolation of this type of antibody.

Identification of monoclonal antibodies with high neutralization capacity against disparate sarbecoviruses has been a priority since the early stages of the COVID-19 pandemic. MAb potency is often the primary criterion for lead optimization, which may bias results toward antibodies that target the ACE2 binding site, a region prone to hypermutation; these epitopes are also poorly conserved, limiting their use to variant-specific treatments instead of an ideal pan-sarbecovirus therapeutic. The WRAIR-2063 MAb exhibits considerable breadth, with a nearly invariant antigenic site, and modest neutralization capacity. Thus, that while neutralization potency of WRAIR-2063 is only moderate across diverse sarbecoviruses, therapeutics incorporating a class V RBD-targeted MAb as one component of a cocktail may provide synergistic protection against SARS-CoV-2 variants and prevent viral escape, representing a more robust therapeutic option than what is presently available.

MATERIALS AND METHODS

Reagent and resource sharing.

Further information and requests for resources and reagents should be directed to and will be fulfilled by the corresponding authors, Shelly J. Krebs (skrebs@hivresearch.org) and M. Gordon Joyce (gjoyce@eidresearch.org).

Antibody production.

WRAIR-2063 antibody was purified by protein A affinity chromatography, using Protein A Sepharose fast flow affinity medium (GE Healthcare/Cytiva) from 1 L of expression supernatant. Protein A resin was equilibrated with 20 column volumes (CV) of phosphate-buffered saline (PBS), followed by supernatant loading at room temperature. Unbound protein was removed by washing with 60 CV of PBS. Bound antibody was eluted with IgG elution buffer (Thermo Scientific) and neutralized with 0.1 M Tris, pH 8.0. Purity was assessed by SDS-PAGE under reducing and nonreducing conditions. WRAIR-2063 was filter sterilized with a 0.22-μm filter, flash-frozen in liquid nitrogen, and stored at −80°C.

Fab production.

Freshly purified WRAIR-2063 IgG in PBS buffer (pH 7.4) was mixed with Lys C protease (New England Biolabs) at a 1:2,000 (wt/wt) ratio. The reaction was allowed to proceed for 2 to 3 h in a 37°C water bath. Digestion was assessed by SDS-PAGE, and upon completion, the reaction mixture was passed through protein A beads (0.5 to 1 mL beads) three times and the final flowthrough was assessed by SDS-PAGE for purity.

Production of recombinant proteins.

All expression vectors were transiently transfected into Expi293F cells (Thermo Fisher Scientific) using Turbo293 transfection reagent (Speed Biosystems). Cells were grown in polycarbonate baffled shaker flasks at 34°C or 37°C and 8% CO₂ at 120 rpm. Cells were harvested 5 to 6 days posttransfection via centrifugation at 3,500 rpm for 30 min. Culture supernatants were filtered with a 0.22-μm filter and stored at 4°C prior to purification.

His-tagged RBDs were purified by Ni-nitrilotriacetic acid (Ni-NTA) affinity chromatography, using Ni-NTA resin (Thermo Scientific), followed by dialysis (10,000 molecular weight cutoff) against PBS. Purity for all the proteins was assessed by SDS-PAGE.

Measurement of WRAIR-2063 binding by BLI.

Real-time interactions between purified SARS-CoV-2 VoC RBDs and WRAIR-2063 were monitored on an Octet RED96 instrument (FortéBio). All biosensors were equilibrated in PBS prior to use. All assays were performed at 30°C with agitation set at 1,000 rpm.

For affinity measurements, biosensors were equilibrated in PBS for 30 s. WRAIR-2063 antibody (15 μg mL⁻¹) was immobilized onto anti-human IgG Fc capture (AHC) biosensors (FortéBio/Sartorius) for 60 s. Baseline was established in PBS for 30 s. Loaded biosensors were dipped into wells containing serial dilutions of SARS-CoV-2 WA-1 RBD or VoC RBDs, starting from ~12 to 25 nM, for 60 s, followed by dissociation in PBS for 300 s. Affinity kinetics constants between WRAIR-2063 antibody and RBDs were determined using 4 or 5 concentrations of RBD. Curves were fit with a 1:1 Langmuir binding model, using Data Analysis software 12.0 (FortéBio).

To assess binding to a panel of diverse sarbecovirus RBDs, anti-penta HIS (HIS1K) biosensors (FortéBio) were equilibrated in assay buffer (PBS) for 30 s before loading of His-tagged RBD (30 μg mL⁻¹ diluted in PBS) for 120 s. After baseline was established in PBS for 30 s, immobilized RBD proteins were then dipped into WRAIR-2063, WRAIR-2057, WRAIR-2134, WRAIR-2151, or WRAIR-2173 antibody (30 μg mL⁻¹ diluted in PBS) for 300 s, followed by dissociation for 60 s in PBS. Binding signal for each antibody at 300 s was normalized to the signal for the WA-1 RBD.

Real-time interactions between purified SARS-CoV-2 WA-1 stabilized S trimer, HexaPro (50), and antibodies were monitored on an Octet RED96 instrument (FortéBio). Avi-tagged recombinant S trimer, biotinylated with the BirA biotinylation kit (Avidity), was diluted to 15 μg mL⁻¹ in kinetics buffer (0.1% [wt/vol] bovine serum albumin, 0.02% [vol/vol] Tween-20 in PBS; FortéBio) and loaded onto streptavidin (SA) sensors (FortéBio) for 300 s. Loaded biosensors were immersed into wells containing WRAIR-2063 antibody at 200 nM for 900 s to saturate all binding sites. Next, biosensors were dipped into wells containing the second antibody, in the presence of the first competing antibody, and binding was measured after 900 s of association. Residual binding signal of the second antibody was expressed as a percentage of the maximum binding signal obtained in the absence of WRAIR-2063, run in parallel. Residual binding signal was corrected for increase in signal obtained with WRAIR-2063 alone. Data were analyzed using FortéBio data analysis software v10.0. All assays were performed at 30°C with agitation set at 1,000 rpm.

Shotgun mutagenesis epitope mapping.

Epitope mapping was performed essentially as described previously (30). Using SARS-CoV-2 (strain Wuhan-Hu-1) RBD, 184 residues of the RBD (S residues 335 to 526) were mutated individually to alanine, or alanine residues were mutated to serine. Mutations were confirmed by DNA sequencing, and clones were arrayed in 384-well plates, one mutant per well. Binding of WRAIR-2063 to each mutant clone in the alanine scanning library was determined, in duplicate, by high-throughput flow cytometry. Each S protein mutant was transfected into HEK-293T cells and allowed to express for 22 h. Cells were fixed in 4% (vol/vol) paraformaldehyde (Electron Microscopy Sciences) and permeabilized with 0.1% (wt/vol) saponin (Sigma-Aldrich) in PBS plus calcium and magnesium (PBS++) before incubation with WRAIR-2063 diluted in PBS++, 10% normal goat serum (Sigma), and 0.1% saponin. WRAIR-2063 screening concentrations were determined using an independent immunofluorescence titration curve against cells expressing wild-type S protein to ensure that signals were within the linear range of detection. WRAIR-2063 was detected using 3.75 μg mL⁻¹ of Alexa Fluor 488-conjugated secondary antibody (Jackson ImmunoResearch Laboratories) in 10% normal goat serum with 0.1% saponin. Cells were washed three times with PBS++ plus 0.1% saponin, followed by two washes in PBS, and the mean cellular fluorescence was detected using a high-throughput IntelliCyt iQue flow cytometer (Sartorius). WRAIR-2063 reactivity against each mutant S protein clone was calculated relative to wild-type S protein reactivity by subtracting the signal from that of mock-transfected controls and normalizing it to the signal from wild-type S-transfected controls. Mutations within clones were identified as critical to the WRAIR-2063 epitope if they did not support reactivity to the test MAb but supported reactivity of other SARS-CoV-2 antibodies. This counterscreen strategy facilitates the exclusion of S mutants that are locally misfolded or have an expression defect.

Full-length spike constructs for sarbecovirus neutralization.

Codon-optimized cDNAs encoding full-length spike from ancestor SARS-CoV-2 (GenBank accession no. QHD43416.1) were synthesized, cloned into the mammalian expression vector VRC8400 (51), and confirmed by sequencing. Spike variants D614G, BA.2.12.1 (T19I, L24S, P25-, P26-, A27-, G142D, V213G, G339D, S371F, S373P, S375F, T376A, D405N, R408S, K417N, N440K, L452Q, S477N, T478K, E484A, Q493R, Q498R, N501Y, Y505H, D614G, H655Y, N679K, P681H, S704L, N764K, D796Y, Q954H, N969K), BA.4/5 (T19I, L24S, P25-, P26-, A27-, H69-, V70-, G142D, V213G, G339D, S371F, S373P, S375F, T376A, D405N, R408S, K417N, N440K, L452R, S477N, T478K, E484A, F486V, Q498R, N501Y, Y505H, D614G, H655Y, N679K, P681H, N764K, D796Y, Q954H, N969K), BQ.1.1 (T19I, L24-, P25-, P26-, A27S, H69-, V70-, G142D, V213G, G339D, R346T, S371F, S373P, S375F, T376A, D405N, R408S, K417N, N440K, K444T, L452R, N460K_S477N, T478K, E484A, F486V, Q498R, N501Y, Y505H, D614G, H655Y, N679K, P681H, N764K, D796Y, Q954H, N969K), XBB.1 (T19I, L24-, P25-, P26-, A27S, V83A, G142D, Y144-, H146Q, Q183E, V213E, G252V, G339H, R346T, L368I, S371F, S373P, S375F, T376A, D405N, R408S, K417N, N440K, V445P, G446S, N460K, S477N, T478K, E484A, F486S, F490S, Q498R, N501Y, Y505H, D614G, H655Y, N679K, P681H, N764K, D796Y, Q954H, N969K), and XBB.1.5 (T19I, L24-, P25-, P26-, A27S, V83A, G142D, Y144-, H146Q, Q183E, V213E, G252V, G339H, R346T, L368I, S371F, S373P, S375F, T376A, D405N, R408S, K417N, N440K, V445P, G446S, N460K, S477N, T478K, E484A, F486P, F490S, Q498R, N501Y, Y505H, D614G, H655Y, N679K, P681H, N764K, D796Y, Q954H, N969K) were generated using the wild-type spike sequence and made via gene synthesis and cloning (GenScript). The spike genes from SARS-CoV-2-related CoVs (RaTG13, GenBank accession no. QHR63300.2; Pangolin GX, GenBank accession no. QIA48632.1), SARS-CoV and related CoVs (SARS-CoV Urbani, GenBank accession no. AAP13441.1; Frankfurt1, GenBank accession no. BAE93401.1; Civet SARS CoV 007/2004 S, GenBank accession no. AAU04646.1; WIV1, GenBank accession no. KF367457; SHC014, GenBank accession no. KC881005) were synthesized (GenScript). These full-length spike plasmids were used for pseudovirus production.

Diverse sarbeco-pseudovirus neutralization assay.

Spike-containing lentiviral pseudovirions were produced by cotransfection of packaging plasmid pCMVdR8.2, transducing plasmid pHR′ CMV-Luc, a TMPRSS2 plasmid, and S plasmids from human and animal coronaviruses (SARS-CoV-2 variants, SARS-CoV, SARS-CoV-2, SARS-CoV-related coronaviruses) into 293T cells using Lipofectamine 3000 transfection reagent (L3000-001; ThermoFisher Scientific, Asheville, NC) (52). 293 flpin-TMPRSS2-ACE2 cells (made at the VRC/NIAID/NIH by Adrian Creanga) were used for the neutralization assay. Cells were plated into 96-well white/black Isoplates (PerkinElmer, Waltham, MA) at 10,000 cells per well the day before infection of pseudovirus. Serial dilutions of antibody were mixed with titrated pseudovirus, incubated for 45 min at 37°C, and added to the cells in triplicate. Following 2 h of incubation, the wells were replenished with 150 μL of fresh medium. Cells were lysed 72 h later, and luciferase activity was measured with Microbeta (Perking Elmer). Percent neutralization and neutralization IC₅₀s and IC₈₀s were calculated as [(virus only control) – (virus plus antibody)]/(virus only control) × 100. Dose-response curves were generated with a five-parameter nonlinear function using GraphPad Prism 8.0.2. IC₅₀ and IC₈₀ titers are reported as the antibody concentration required to achieve 50% or 80% neutralization (50% or 80% inhibitory concentration).

WA-1 pseudovirus production and neutralization.

SARS-CoV-2 pseudovirions (PSVs) were produced by cotransfection of HEK293T/17 cells with a pcDNA3.1 encoding SARS-CoV-2 S protein and an HIV-1 NL4-3 luciferase reporter plasmid (pNL4-3.Luc.R-E-; National Institutes of Health [NIH] AIDS Reagent Program). The S expression plasmid sequence was derived from the Wuhan Hu-1 strain (GenBank accession no. NC_045512), which is also identical to the IL-1/2020 and WA1/2020 strains. The S expression plasmid sequence was also codon optimized and modified to remove the last 18 amino acids of the cytoplasmic tail to improve S incorporation into the pseudovirions and thereby enhance infectivity.

Test antibodies were diluted in growth medium and serially diluted, and 25 μL per well was added to a white 96-well plate. An equal volume of diluted SARS-CoV-2 PSVs was added to each well, and the plates were incubated for 1 h at 37°C. HEK293 target cells expressing ACE2 (Integral Molecular) were added to each well (40,000 cells per well). Plates were incubated for 48 h at 37°C. Then, Bright-Glo luciferase assay system substrate (Promega, Madison, WI) was added, and relative light units (RLUs) were measured using an EnVision multimode plate reader (Perkin Elmer, Waltham, MA). Dose-response curves were generated with a five-parameter nonlinear function using GraphPad Prism 9.5.0.

X-ray crystallography and structure analysis.

WRAIR-2151 and WRAIR-2063 Fabs were mixed with WA-1 RBD at a 1:1:1 molar ratio, incubated for an hour on ice, and concentrated down to 8.0 mg mL⁻¹. The resulting ternary complex was screened for crystallization conditions using an Art Robbins Gryphon crystallization robot, 0.2-μL drops, and a set of 1,200 conditions. Crystal drops were observed daily using a Jan Scientific UVEX-PS imaging system with automated UV and brightfield drop imaging. Initial crystallization conditions were optimized manually by mixing protein and reservoir solutions at 1:1 (vol/vol) ratios. Crystals used for data collection grew under the following crystallization conditions: 100 mM Bis-Tris propane HCl (pH 8.5), 200 mM sodium malonate dibasic, and 20% (wt/vol) polyethylene glycol (PEG) 3350 (Wizard Classic 3 and 4; Rigaku/Molecular Dimensions).

Diffraction data were collected at Advanced Photon Source (APS), Argonne National Laboratory beamline 24-ID-E, and were measured using a Dectris Eiger 16M pixel detector to a final resolution of 2.77 Å. Diffraction data indexing, integration, and scaling were carried out using the X-ray Detector Software (XDS) GUI (53). Data collection statistics are reported in Table 1.

The ternary complex structure was solved by molecular replacement using PHASER, and iterative model building and refinement were performed in Coot and Phenix (54–58). Phenix Xtriage was used to analyze all the scaled diffraction data output from X-ray detector software (XDS). Data were analyzed for measurement value significance, completeness, asymmetric unit volume, and possible twinning and/or pseudotranslational pathologies. To determine the structure of the ternary complex, we used the previously reported crystal structures of WRAIR-2151 Fab and SARS-CoV-2 RBD as search models (PDB 7N4M) (30). The heavy chain variable domain (V_H) of tremelimumab (PDB 5GGU) (59) and the light chain variable domain (V_L) of antibody 1260 (PDB 6B0E) (60) were used as the search models for WRAIR-2063. Iterative manual model building was performed in Coot. All structures were refined using Phenix refine with positional, global isotropic B-factor refinement and defined Translation/Libration/Screw (TLS) vibrational motion groups. The Ramachandran plot as determined by MOLPROBITY showed > 98% of all residues in the favored regions and ~2% of all residues in the allowed regions (Table 1) (61). Interactive surfaces were analyzed using PISA (www.ebi.ac.uk/pdbe/pisa/) and are provided in Table S1 in the supplemental material. Structure figures were prepared using PyMOL (The PyMOL Molecular Graphics System, version 2.1; Schrodinger, LLC). Software used in this work was curated by SBGrid (62).

Data and software availability.

The data for coordinates and structure factors are available under Protein Data Bank ID 8EOO. No novel software was used in this study.