First-generation N-terminal domain supersite public antibodies retain activity against Omicron-derived lineages and protect mice against Omicron BA.5 challenge

ABSTRACT

Monoclonal antibodies to SARS-CoV-2 can offer prophylactic and therapeutic protection against severe disease, with particular utility for immunosuppressed and vulnerable populations. With the constant emergence of new variants, understanding the neutralizing potency of monoclonal antibodies to dynamic spike protein epitopes is crucial. We show that a set of VH1-24-derived N-terminal domain (NTD)-directed antibodies, isolated from a convalescent donor early in the pandemic, displayed remarkable neutralization resilience against many Omicron SARS-CoV-2 variants, including BA.2, BA.5, and BQ.1.1. Neutralization potency to these Omicron variants is associated with slower off-rates to the spike protein. Structural characterization of the most potent NTD antibody, WRAIR-2008, revealed a conserved mode of interaction shared with other antibodies of the same multi-donor class. WRAIR-2008 protected mice from weight loss following BA.5 challenge and reduced infectious viral titers in the lungs. Our study highlights the retention of neutralization activity and protection of first-generation VH1-24-derived NTD-directed antibodies to specific Omicron variants and provides valuable insights into the shifting landscape of SARS-CoV-2 variants that are vulnerable to select monoclonal antibodies.

IMPORTANCE

As SARS-CoV-2 circulating variants evolve, it is important to understand the vulnerabilities of these viruses to neutralizing antibodies. Within this manuscript, we describe first-generation antibodies isolated following infection with WA-1 that retain viral neutralization to subsequent Omicron variants by targeting a site of viral vulnerability called the NTD. This work highlights the shifting landscape of SARS-CoV-2 variants and provides mechanistic insights into how antibodies from prior infections may play a role in preventing subsequent SARS-CoV-2 variant infections.

INTRODUCTION

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causative agent of coronavirus disease of 2019 (COVID-19), brought an unprecedented global pandemic with an estimated 776 million cases worldwide, over 7 million deaths (https://data.who.int/dashboards/covid19), and 15–20 million Americans affected by long COVID (1). The rapid deployment of vaccines and other countermeasures curbed the burden on the population by reducing morbidity levels and the number of severe cases of COVID-19. Alongside vaccines, monoclonal antibodies (mAbs) were developed as an important part of the clinical tool kit, with particular benefits for people who did not respond to vaccination or were at high risk of developing severe disease (2). MAbs have also been shown to benefit patients with persistent SARS-CoV-2 infection, a condition that may favor the development of highly divergent viral variants (3). As of December 2024, the vast majority of EUAs for SARS-CoV-2 mAbs have been revoked due to lack of effectiveness against circulating variants (2).

The N-terminal domain (NTD) of SARS-CoV-2 spike in the prefusion conformation is readily accessible at the spike trimer periphery and contacts the centrally-located receptor-binding domains (RBD) (4, 5). The NTD is an important target of antibody responses as it harbors several neutralizing epitopes, such as the antigenic NTD supersite, recognized by potently neutralizing mAbs (6–8). In addition, at least three other NTD neutralizing epitopes have been defined on the basis of binding competition (9). As antibodies targeting such epitopes do not efficiently compete with virus binding to the human angiotensin-converting enzyme 2 (ACE2) receptor, NTD-directed antibodies typically display more modest neutralization potency in vitro compared to mAbs that target the RBD. However, small animal studies have drawn attention to the elevated protective potency of NTD mAbs in vivo, in part due to their unique ability to engage spike-expressing cells and recruit downstream Fc effector functions (8, 10, 11), the latter contributing to protection even in the absence of neutralization (12–14).

Early studies identified several potent NTD neutralizing mAbs against the original and D614G SARS-CoV-2 strains (6–8, 10, 15–19), including the identification and characterization of WRAIR-2008 (10). WRAIR-2008 is one of many VH1-24 public class-antibodies described at the beginning of the pandemic that target the NTD supersite (6–10). However, subsequent dominant variants such as Alpha (B.1.1.7), Beta (B.1.351), Delta (B.1.617.2), and Omicron BA.1 harbored mutations in the NTD antigenic supersite that caused reduction or loss of neutralization activity from most of these first-generation mAbs (20–22), a likely consequence of immune pressure generated by infection and vaccination. Many escape mutations were located within the most common contact sites in the NTD loops N3 (Alpha, Delta, BA.1) and N5 (Beta). Whether these first-generation antibodies were able to retain neutralization activity to other Omicron variants remains unknown.

Here, we describe a set of first-generation VH1-24 public antibodies isolated early in the pandemic that bind the spike NTD of multiple variants and retain neutralization activity against Omicron variants BA.2, BA.5, and BQ.1.1. Cryo-electron microscopy (cryo-EM) analysis of WRAIR-2008 in complex with spike revealed conserved interactions among this class of public VH1-24 antibodies, largely mediated by contacts between WRAIR-2008 heavy chain complementary determining regions and the N3 and N5 loops in the NTD. WRAIR-2008 protected mice prophylactically from weight loss following challenge with Omicron BA.5, with the most potent neutralizer WRAIR-2008 reducing infectious virus in the lungs more effectively than its weakly neutralizing counterpart WRAIR-2025. When administered 1 day after challenge, WRAIR-2008 was able to significantly reduce infectious viral load in the lungs, demonstrating both prophylactic and therapeutic protection. Altogether, these data further characterize the NTD site of vulnerability on the spike glycoprotein that continues to evolve as new variants emerge and may inform the development of pan-sarbecovirus countermeasures that target this domain.

RESULTS

First-generation NTD-directed mAbs retain neutralization and Fc effector function activity against SARS-CoV-2 Omicron variants

We previously reported the identification of monoclonal antibodies from a convalescent donor infected with the original SARS-CoV-2 strain (WA-1) early in the pandemic (10). Using single B cell sorting approaches with a stabilized spike trimer and a spike-ferritin nanoparticle (23), we characterized 117 mAbs for their ability to bind and neutralize pre-Omicron variants (10). Of those, 14 mAbs demonstrated potent and/or broad neutralization and some provided both prophylactic and therapeutic in vivo protection in the K18/hACE2 transgenic mouse model against WA-1 challenge. Seven of the mAbs targeted the NTD antigenic supersite, and seven others recognized epitopes on the RBD belonging to RBD antibody classes I/II, IV, and V (24). With the emergence of Omicron variants harboring extensive mutations in both NTD and RBD, we sought to evaluate whether these antibodies retained activity against antigenic variants and gain an understanding of the mutations responsible for viral escape. Using our previously qualified pseudotyped neutralization assay (10), we tested our set of first-generation NTD- and RBD-directed mAbs against an updated SARS-CoV-2 pseudovirus panel that included the prominent Omicron lineages to understand how changes in spike sequence and structure impacted neutralization capacity.

Neutralization activity against all pseudoviruses was observed with the class V RBD mAbs that target a cryptic epitope conserved across sarbecoviruses (25) albeit with slightly reduced potency against Omicron-derived variants (Fig. 1A). The most potent class I/II RBD mAbs neutralized Delta and Epsilon variants but failed to neutralize any variants that appeared after BA.1 emergence, consistent with the pattern of escape mutations observed in Omicron lineages against this class of ACE2-competing antibodies. Conversely, NTD-directed antibodies followed a different pattern, where mutations in Epsilon and Omicron BA.1 abolished the neutralizing activity of all supersite-targeting NTD antibodies. However, Omicron strains such as BA.2, BA.5, and BQ.1.1 remained sensitive (Fig. 1A and B). In particular, WRAIR-2008, WRAIR-2037 (a clonal variant of WRAIR-2008), and WRAIR-2039 exhibited potent neutralization of these three variants with 50% virus neutralizing concentrations (IC50) ranging from 6 to 29 ng/mL, similar to what was observed for WA-1 (Fig. 1A and B). Lack of neutralization was observed against later Omicron variants XBB.1.5 and JN.1 strains, with modest neutralization of Alpha, Beta, and Delta variants observed for a few NTD mAbs, which typically evaded neutralization by first-generation NTD-directed antibodies (20, 26). We previously observed that all NTD-directed antibodies displayed neutralizing curves against WA-1 with a maximum plateau around 70%, indicating incomplete neutralization (10). In contrast, WRAIR-2008 was able to neutralize Omicron variants, such as BA.2, BA.5, and BQ.1.1 more effectively, with maximum neutralization nearing 90%, indicating a higher activity against Omicron compared to the ancestral WA-1 strain (Fig. 1C).

Fig 1.

Neutralization potency and breadth of a panel of NTD- and RBD-directed antibodies against SARS-CoV-2 variants. (A) Neutralization activity of NTD and RBD mAbs with therapeutic potential against a panel of pseudotyped viruses (pSV) representing pre-Omicron and Omicron SARS-CoV-2 variants. The heat map indicates IC50 values (ng/mL) ranging from very potent (dark blue), to intermediate (yellow), to weakly neutralizing (red) and non-neutralizing (white), calculated using a five-parameter regression analysis in two independent experiments. (B) Neutralization curves of the NTD supersite mAbs as measured in the pSV assay against the Omicron BA.2, BA.5, and BQ.1.1 strains. Plotted are the mean ± s.e.m. from at least two independent experiments. (C) Left, neutralization curves of WRAIR-2008 as measured in the pSV assay against the WA-1 and Omicron BA.5 strains. Plotted are the mean ± s.e.m. from five independent experiments. Right, comparison of the % maximum neutralization observed at 25 µg/mL, against WA-1 and BA.5 (Mann-Whitney test, P = 0.0079). Horizontal bar indicates geometric mean.

Since engaging Fc effector functions have been found to be important for in vivo protection using NTD supersite antibodies (8, 10), we tested whether these mAbs maintained robust Fc effector function against BA.5, as they did against WA-1 (10). WRAIR-2008, in addition to WRAIR-2037 and WRAIR-2039, all exhibited robust binding to BA.5 spike-expressing cells, leading to strong antibody-dependent complement deposition (ADCD) and antibody-dependent cellular cytotoxicity (ADCC) (Fig. S1). Phagocytic activities such as antibody-dependent cellular phagocytosis (ADCP) and antibody-dependent neutrophil phagocytosis (ADNP) were also detected at high levels. WRAIR-2025 displayed reduced opsonization activity to BA.5 spike as well as diminished activity across all Fc effector functions tested, likely due to lower affinity to the BA.5 spike trimer (see below). This effect was less pronounced at higher antibody concentrations, suggesting that increased occupancy and/or antibody density on the target cells could compensate for the reduced affinity. Combined, these data demonstrate that despite neutralization escape by Alpha, Beta, and Delta, and Omicron BA.1 variants, first-generation NTD supersite mAbs retained neutralization activity and the ability to engage Fc effector functions against Omicron strains BA.2, BA.5, and BQ.1.1.

Binding and affinity of VH1-24 NTD supersite antibodies vary against SARS-CoV-2 variants

The NTD-directed neutralizing antibodies that retained activity against Omicron variants were derived from the same public heavy chain VH1-24 gene, previously described in several studies and isolated from multiple donors (6, 7, 9, 12, 15, 16, 18, 19, 27, 28) (Fig. S2A). However, differences in D and J gene usage, characterized by a variety of HCDR3 residues and lengths from 11 to 19 residues, as well as pairing with highly divergent light chains (10) likely account for the diversity of neutralization observed across mAbs and SARS-CoV-2 variants. Similarly, while the WRAIR NTD-directed neutralizing antibodies target the same “supersite” epitope, based on prior mutagenesis and epitope binning analysis (10), they demonstrated differential neutralization potency (Fig. 1A). WRAIR-2025 and WRAIR-2035 exhibited more modest neutralization of BA.2, BA.5, and BQ.1.1, compared to WRAIR-2008, with IC50 ranging from 286 to 10,800 ng/mL. To explore those differences further, we performed binding studies with the BA.5 stabilized spike trimer and our panel of NTD-directed IgG1 antibodies. While WRAIR-2004 was unable to bind, explaining the lack of neutralization activity against BA.5, all the other mAbs displayed robust binding of similar magnitude compared to WA-1 (Fig. 2A). Affinity measurements (association and dissociation rates) were evaluated for WRAIR-2008 and WRAIR-2025 using the WA-1 and BA-5 spike trimers. The binding association rates to the BA.5 trimer did not differ between WRAIR-2008 and WRAIR-2025 IgG; however, the dissociation rate of WRAIR-2025 to BA.5 was much faster compared to WRAIR-2008, causing a loss in binding from picomolar to sub-nanomolar affinity and likely explaining the reduction in neutralization activity observed (Fig. 2B). In comparison, WRAIR-2008 and WRAIR-2025 bound the WA-1 trimers with comparable association rates and no discernible dissociation, resulting in picomolar affinity (Fig. 2B). A similar analysis performed with the equivalent Fab fragments highlighted an overall decrease in affinity by several orders of magnitude, compared to IgG. WRAIR-2008 Fab still showed improved affinity to the BA.5 trimer over WRAIR-2025 (Fig. S3A). This data further support earlier observations that, in contrast to RBD class I/II mAbs, NTD supersite-targeting antibodies rely on IgG dimeric binding to neutralize SARS-CoV-2 since monomeric Fab fragments fail to do so (7, 8, 10).

Fig 2.

Binding and affinity of NTD supersite antibodies against SARS-CoV-2 stabilized spike variants. (A) Left, binding of NTD supersite mAbs (200 nM) to the BA.5 spike trimer measured by biolayer interferometry (BLI). Right, ratio of binding responses between BA.5 and WA-1 spike trimers. (B) Top, binding kinetics of WRAIR-2008 and WRAIR-2025 IgG against the BA.5 spike trimer measured by BLI (by fitting [in gray] to a 1:1 binding model). The binding equilibrium constant (KD) values are indicated in each graph. Bottom, summary of binding affinity constants and fit parameters to both WA-1 and BA.5 trimers. (C) Top, binding responses (nm) using BLI of the NTD supersite mAbs at 200 nM against SARS-CoV-2 VoC spike trimers. MAbs are shown, from left to right, WRAIR-2004,-2008, -2025,-2035, -2037,-2039, and -2196, with WRAIR-2008 indicated by a red bar for all VoCs. Bottom, summary of binding affinity constants of WRAIR-2008 against the Alpha (B.1.1.7), Beta (B.1.351) and Delta (B.1.617.2) spike trimers measured by BLI as described in panel B. Error bars indicate standard deviation from mean calculated from three independent experiments.

To expand on these results, we tested the WRAIR NTD mAbs for binding against spike trimers representing pre-Omicron and Omicron variants (Fig. 2C). Despite the lack of neutralization of Alpha, Beta, and Delta variants by WRAIR-2008, robust binding to these spike trimers was observed, comparable to WA-1 and BA.5 (Fig. 2C). However, upon closer inspection, WRAIR-2008 IgG displayed faster dissociation constants to Alpha, Beta, and Delta trimers resulting in decreased binding affinity (Fig. 2C). K_D and IC₅₀ did not correlate perfectly across variants suggesting that other factors may influence neutralization activity. Together, these data suggest that WRAIR NTD IgG high affinity and stable binding to spike associates with neutralization activity to SARS-CoV-2 variants.

Cryo-EM of WRAIR-2008 in complex with spike

Previously, epitope mapping of our panel of NTD-targeted neutralizing antibodies by shotgun alanine mutagenesis (10) suggested that residues Y145, K147, and W152, all within the NTD N3 loop, are critical interface residues for WRAIR-2008 and other IGHV1-24*01-derived NTD neutralizing mAbs. Additional residues in the NTD N5 loop, such as R246, Y248, and P251, also reduced antibody binding upon mutagenesis to alanine (10). To verify these molecular interactions and provide a structural rationale for the neutralization breadth observed for WRAIR-2008, we performed cryo-EM analysis of WA-1-stabilized spike (HexaPro) (29) in complex with WRAIR-2008 Fab (Fig. S4 and S5). After non-uniform refinement, we obtained a reconstruction with a global resolution of 3.23 Š(Fig. 3A; Fig. S5). Notably, in the 2D class averages, we only observed spike in the closed (3 RBD-down) conformation with 3 Fabs bound per spike (Fig. S5). To improve the resolution and further clarify the molecular contacts across the NTD-WRAIR-2008 interface, we executed symmetry expansion, masked the NTD-Fab region of the complex prior to particle subtraction, and performed local refinement of this region, which resulted in a 3.89 Šreconstruction (Fig. 3A, inset, Fig. S5B). Contacts between the NTD and WRAIR-2008 heavy chain (HC) include hydrogen bonds between the side-chain amino groups of NTD residues K147, K150, and R246 and the main-chain carboxyl atoms of WRAIR-2008 heavy chain T30 and E53, respectively, and salt-bridges between, primarily, WRAIR-2008 HC E71 and E31 with NTD residues K147 and R246 (Fig. 3B). Charge complementarity at the WRAIR-2008-NTD interface is supplemented by weaker hydrogen bonds (>2.5 Šapart) and extensive van der Waals (VDW) interactions between Y145, W152, and Y248 with surrounding WRAIR-2008 HC residues T100a, R100c, and E31. WRAIR-2008 engagement with the NTD is almost exclusively driven by the HC, with a total buried surface area of 813 Ų, while the light chain-NTD interface only buries a total of 72 Ų (Fig. 3C). Amino acids Y145, K147, and W152, which were shown to be critical contact residues by shotgun mutagenesis, are 84.5%, 95.1%, and 67.8% buried at the interface, while R246 and Y248 are 70.8% and 86.1% buried, respectively.

Fig 3.

Cryo-EM structural analysis of WRAIR-2008 in complex with the SARS-CoV-2 spike. (A) Top view (left) and side view (center) of the density map of WRAIR-2008 bound to SARS-CoV-2 HexaPro spike (S) with all RBDs in the down position. The S trimer is colored light gray, with the RBD dark gray and the Fv regions of the WRAIR-2008 heavy and light chains colored dark blue and light blue, respectively. (Inset at right) Focused density map of the WRAIR-2008 Fv-NTD complex. (B) Refined atomic model of the WRAIR-2008 Fv interface with the NTD N3 (left and right) and N5 (center) loops. Hydrogen bonds or salt bridges are indicated with dashed lines. W152 is buried at the interface of N3 with the WRAIR-2008 heavy chain. (C) Buried surface area (BSA) for the WRAIR-2008 heavy and light chain CDR loops and the NTD N1 (residues 22–26), N3 (residues 141–156), and N5 (residues 246–260) loops. (D) Surface representation of the refined atomic model of the S protein, with individual protomers of the trimer colored light gray, dark gray, or cyan. (Inset at left) The NTD interface with heavy and light chain contact residues of contributing CDRs shown as sticks and labeled as per antibody coloring scheme. Loops are colored brown (N1), tan (N3), and orange (N5). (E) Orientations of the CDR H3 loops of other NTD supersite-targeted mAbs, in the context of the NTD N3 and N5 (colored as shown in [D]). WRAIR-2008, dark blue; 4A8 (PDB 7C2L), dark yellow; TXG-0078 (PDB 8SWH), tan; DH1050.1 (PDB 7LCN), purple; COVOX-159 (PDB 7NDD), green; 4–8 (PDB 7LQV), lavender. The “YYY” motif present in the CDRH3s of WRAIR-2008, 4A8, and TXG-0078 are shown as sticks.

All HCDR loops contribute to the WRAIR-2008-NTD interface although the interaction is dominated by the extended HCDR3 (Fig. 3C and D). Comparison of the WRAIR-2008 HC sequence with those of other VH1-24-derived mAbs that have been structurally characterized reveals high sequence similarity across HCDR1 and HCDR2, with considerable variation observed in both length and composition of the HCDR3s (Fig. 3E; Fig. S2 and S6). WRAIR-2008 shares a stretch of 4–5 aromatic residues (F100e-Y100i), a “YYY” motif, with NTD-targeted mAbs 4A8 (15), 1–87 (6), and TXG-0078 (27), a unique feature of the IGHJ6 gene that results in the sidechains extending perpendicular to the main chain in opposite directions. This serves to both stabilize the latter half of the HCDR3 against the light chain and situate the HCDR3 between the NTD N3 (residues 141–156) and N5 (residues 246–260) loops (Fig. 3E; Fig. S2B). VH1-24-derived NTD supersite antibodies engage the NTD spike through a conserved mode of recognition involving multiple residues in the NTD N3 and N5 loops, with residues Y145 and/or K147 (N3) and R246 and/or Y248 (N5) playing key roles in the interaction (Fig. 3; Fig. S6).

Structural explanation for binding and neutralization breadth

We endeavored to provide a structural explanation for the variation in neutralization capacity and potency of WRAIR-2008 across viral variants (Fig. 1A) by comparing the WA-1 NTD with variant NTDs and variant NTDs with the WRAIR-2008-bound WA-1 NTD. These efforts were somewhat obstructed by the limited number of NTD structures with resolved N3 and N5 loops. As the structural rearrangements of the N3 and N5 loops of the NTD are quite extensive upon mAb binding (Fig. S7), we initially compared structures of unbound WA-1 NTD with variant NTDs, followed by comparison of WRAIR-2008-bound WA-1 NTD with variant NTDs (Fig. S7). WRAIR-2008 is unable to neutralize Alpha, Beta, Delta, Epsilon, and Omicron BA.1, XBB.1.5, and JN.1 variants (Fig. 1A) primarily due to the shared deletion at Y144, as all of these variants, with the exception of Epsilon, feature this deletion. Additional deletions and point mutations likely further destabilize these loops and alter the binding surface, thus preventing or weakening interaction with WRAIR-2008. Although a structure of the Epsilon variant NTD with a fully-resolved N3 loop is not available, we speculate that the W152C substitution sufficiently disrupts the VDW interactions at the interface to alter binding and subsequent neutralization by WRAIR-2008.

To extricate the molecular details of WRAIR NTD-targeted mAbs with similar neutralization breadth, we crystallized Fab fragments of WRAIR-2008 and WRAIR-2039 (Fig. S8A and B; Table S3) and used the WRAIR-2008 crystal structure as the template to create a homology model of WRAIR-2037, a clonal relative of WRAIR-2008 (Fig. S8C). All three mAbs demonstrate similar neutralization capacity and breadth (Fig. 1A) and are derived from the same germline gene but possess HCDR3s of differing lengths (Fig. S2C). WRAIR-2008 and WRAIR-2039 neutralize SARS-CoV-2 WA-1, as well as Omicron strains BA.2, BA.5, and BQ.1.1 with similar potency. The neutralization capacity of WRAIR-2037 only extends further to variants Beta and Delta, with moderate potency. This reduced neutralization capacity could be explained by differences in charge distribution between WRAIR-2008 and WRAIR-2037. While there are 10 residue variations between these two mAbs, only 3 are found within the HCDR3, wherein R100c, F100e, and Y100f in WRAIR-2008 are substituted for T100c, Y100e, and S100f in WRAIR-2037 (Fig. S2). The Beta variant has a single deletion in the N3 loop (Y144) that may shift H146 into proximity for salt-bridge formation with Y100e of WRAIR-2037 (Fig. S7C). Although Beta features an additional deletion in the N5 loop (242-244, LAL), the antibody T100c, Y100e, and S100f substitutions must be sufficient to overcome the 242–244 deletion and subsequent shift in residues in this region, and allow for binding to occur. In the context of the Alpha variant, the N5 deletion is not found, which WRAIR-2037 cannot neutralize. Similarly, the N3 deletion in the Delta variant (156–157), along with the G142D and R158G mutations, may also shift the stretch of basic residues within the N3 loop (146-HKNNK-150) into interaction proximity with the carboxyl and hydroxyl sidechains of the WRAIR-2037 HCDR3. In support of these observations, WRAIR-2037 Fab was able to bind to Beta and Delta variants with measurable affinity (Fig. S3B).

The crystal structure of WRAIR-2039 Fab was used to create a homology model of WRAIR-2025 (Fig. S8D), which also features a shortened HCDR3 compared to WRAIR-2008 and WRAIR-2037 (Fig. S2A). WRAIR-2025 weakly neutralizes BA.5 (10,800 ng/mL), which is reflected in the loss of binding affinity to the BA.5 spike, as compared to WRAIR-2008, as well as BA.2 and BQ.1.1 (Fig. 1A and 2B; Fig. S3A). As these VoCs all contain the G142D mutation as the only alterations in the N3 and N5 loops, residue variation outside of the NTD that results in differences in spike dynamics must account for reduction in neutralization potency by WRAIR-2025.

Protection against BA.5 challenge in K18/hACE2 mice

Having established that VH1-24-derived public NTD antibodies retained neutralization activity against Omicron strains BA.2, BA.5, and BQ1.1, we next examined whether they would protect from SARS-CoV-2 Omicron viral challenge in vivo using the stringent transgenic mouse K18-hACE2 model. We elected to use the BA.5 strain due to its enhanced pathogenicity compared to other Omicron strains. Omicron BA.5 was shown to exhibit higher infection of lung cells compared to its BA.1 and BA.2 counterparts (30, 31).

Monoclonal antibodies WRAIR-2008 and WRAIR-2025 were selected for their potent and modest neutralization activity against BA.5, respectively (Fig. 1A), to test for in vivo protection in two separate experiments. The first experiment assessed prophylactic protection where groups of 15 mice were administered WRAIR-2008 at 1 and 10 mg/kg and WRAIR-2025 at 10 mg/kg 24 h prior to intranasal BA.5 challenge. Five animals were humanely euthanized at day 2 post-challenge to measure infectious virus in the lungs, while the other 10 animals were followed daily over 14 days for weight and signs of clinical illness. Significant weight loss was observed in animals that received the isotype IgG control, peaking at day 6 post-challenge (Fig. 4A). All animals that received the NTD-directed antibodies were protected from weight loss, inclusive of the group treated with mAb WRAIR-2025, despite its modest neutralization activity against BA.5 (IC50 > 10 µg/mL) (Fig. 1; Table S1). However, significant differences in infectious virus titers could be observed in lung tissue across treated groups. The higher dose of WRAIR-2008 afforded the best protection with an almost 1.5 log reduction in infectious viral load, followed by the lower dose of WRAIR-2008 (~1 log reduction) and WRAIR-2025 (< 1 log reduction).

Fig 4.

WRAIR-2008 prevents BA.5 infection in vivo and reduces lung viral load when administered prophylactically and therapeutically. (A) Prophylactic treatment in the K18-hACE2 SARS-CoV-2 mouse model. Antibodies were infused intravenously at the indicated dose 24 h into groups of mice (n = 15) prior to intranasal challenge with ~1 × 10⁵ PFU of SARS-CoV-2 BA.5. SARS-CoV-2 viral loads in lung tissue were measured 2 days after challenge in a subset of animals (n = 5 per group) by plaque assay. Points or bars indicate the mean group value with error bars indicating standard deviation. The remaining mice (n = 10 per group) were assessed daily for weight and clinical symptoms. (B) Therapeutic treatment in the K18-hACE2 SARS-CoV-2 mouse model. Antibodies were infused intravenously at the indicated dose 24 h after challenge, performed as indicated above. SARS-CoV-2 viral loads in lung tissue were measured 2 days after mAb administration in a subset of animals (n = 5 per group) by plaque assay. Points or bars indicate the mean group value with error bars indicating standard deviation. The remaining mice (n = 10 per group) were assessed daily for weight and clinical symptoms. (A, B) For weight loss (day 4 through 8) and viral load in lungs, asterisks indicate significance compared to the MZ4 antibody isotype control group, as assessed by one-way ANOVA with Dunnett’s multiple-comparisons test. For all analysis, ****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05; NS, P > 0.05.

The second experiment tested the therapeutic efficacy of WRAIR-2008 at 10 and 1.5 mg/kg, where mice were first challenged with viral strain BA.5 prior to receiving the antibodies one day later. In this setting, weight loss was observed in both isotype control and treated groups, peaking at day 6 post-challenge (Fig. 4B). However, mice that received WRAIR-2008 at either 10 mg/kg or 1.5 mg/kg exhibited a significant dose-dependent reduction in infectious viral load in the lungs, compared to the isotype control, indicating that this class of NTD-directed antibody can also provide benefits when administered therapeutically, in addition to prophylactically.

DISCUSSION

Neutralizing antibodies targeting the NTD of SARS-CoV-2 have been isolated from multiple donors and represent an important fraction, both quantitatively and qualitatively, of the humoral responses to infection and vaccination (19, 32). MAbs developed for treatment and protection against SARS-CoV-2 have been focused on those targeting the spike RBD and typically lose neutralization capacity with novel emerging variants. The need to further understand, develop, or elicit mAbs that target domains outside the RBD may be crucial for improved vaccines or combination mAb therapies with higher genetic barrier to viral escape.

Within this study, we characterized a set of 14 first-generation monoclonal antibodies, isolated from a SARS-CoV-2 infected participant in 2020, for their ability to neutralize across dominant SARS-CoV-2 variants. Seven of the mAbs targeted the NTD supersite, belonging to the VH1-24 public clonotype, and potently neutralized Omicron variants BA.2, BA.5, and BQ.1.1. One notable exception was WRAIR-2025, which displayed a severely reduced in vitro BA.5 neutralization titer. This diminished neutralization titer was explained by a faster dissociation rate against the BA.5 variant trimer resulting in lower affinity.

The differences in affinity between very similar mAbs WRAIR-2025 and WRAIR-2008 provided the opportunity to assess whether binding antibodies could still offer protection in vivo in the absence of a strong in vitro neutralization profile. This allowed us to explore the in vivo protective effects of this class of NTD-directed mAbs by uncoupling neutralization and in vivo Fc effector functions. Using a BA.5 challenge and weight loss model in K18-hACE2 transgenic mice, we observed that WRAIR-2008, which potently neutralized BA.5 in vitro, conferred full prophylactic protection and reduced infectious virus in the lungs, as we previously observed for WA-1 (10). In the same setting, WRAIR-2025 also fully protected from weight loss despite modest neutralization (IC50 > 10 µg/mL). NTD-directed non-neutralizing antibodies have previously been shown to provide some degree of protection in vivo, albeit not to the extent of what we observed here for WRAIR-2025 with complete protection from weight loss (14, 33, 34).

The ability of NTD-targeting mAbs to generate in vivo protection at low doses or in the absence of potent neutralization is deserving of further study, but we can hypothesize several reasons to explain the functionality of NTD-targeting antibodies using Fc effector functions. Avidity is one concept that may have allowed WRAIR-2025 to compensate for its lower affinity for spike, where multivalent target binding and clustering on the surface of viruses and/or infected cells facilitated the requisite binding levels to enable effector function activation, resulting in virus and/or infected cell clearance by the immune system. Such a role for Fc effector functions has been documented in FcgR knocked-out mice, where Fc effector functions, in particular through alveolar macrophages and FcgRIII, were essential for WA-1-based vaccine-mediated protection against the antigen-shifted BA.5 variant (35). Alternatively or in combination with avidity effects, the antibody angle of approach to the spike may be an important parameter. WRAIR-2008 approaches the spike from the top, in a membrane-distal orientation, ensuring that the mAb Fc domain is pointing away from the viral surface and positioned for recruitment of Fc effector cells. In contrast, non-neutralizing NTD antibodies, such as DH1052 (12, 14) and CV3-13 (33), only afforded modest protection in prophylactic studies. They target a site on the outer edge of NTD with an angle of approach parallel (CV3-13) or pointing toward the viral or cellular membrane (DH1052), ensuring that their Fc domains are not as readily accessible as that of NTD supersite antibodies.

Structural analysis using cryo-EM revealed the spike in the closed (3 RBD-down) conformation, with three Fabs bound per spike. This conformation does not appear to be a result of symmetry expansion, as the same data processed without symmetry expansion did not display any 2D class averages with an open S trimer and resulted in a nearly identical 3D reconstruction. Structural studies showed that WRAIR-2008 engagement with NTD was almost exclusively driven by the heavy chain (HC) and dominated by the extended HCDR3. Comparison of the WRAIR-2008 HC sequence with those of other VH1-24-derived mAbs that have been structurally characterized reveals high sequence similarity across HCDR1 and HCDR2, with considerable variation observed in both length and composition of the HCDR3s. This variation and the affinity of the HCDR3 with the NTD N3 and N5 loops may contribute to the variation in neutralization capacity within this public clonotype.

Our analysis of binding affinity with IgG and Fab highlighted differences in affinity that tracked with neutralization potency (e.g., for the BA.5 variant). While not fully understood, the neutralization mechanism of this class of antibodies relies on binding affinity, as IgGs formed a highly stable complex with spike. In contrast, Fab fragments engaged spike with much lower affinity, consistent with their inability to neutralize. Other benefits of the larger IgG molecule may include steric hindrance of putative co-receptor(s) as well as partial inhibition of ACE2 interaction (7, 8, 10), which may all contribute to viral inhibition. Nonetheless, a direct correlation between K_D and IC₅₀ could not be established as mutations/deletions within or outside the NTD may impact neutralization in a variant-specific way—which has been difficult to model from our structural data—possibly by altering spike dynamics and/or shifting the antibody angle of approach. We note that the exceptional binding breadth of WRAIR-2008 across SARS-CoV-2 spike variants extended well beyond those it could neutralize, suggesting that this class of antibody may offer additional benefits through non-neutralizing Fc effector functions.

Significant portions of the NTD are surface-exposed in the prefusion spike trimer structure, representing an accessible target for viral inhibition. As such, the NTD contains multiple mutational hotspots, with antigenic supersite N1, N3, and N5 loops, as well as additional N1 and N2 loop residues under constant pressure as the virus adapts to its host and evolves to escape the immune response. This leads to the situation where NTD-targeting mAbs also display decreased activity, but examples such as TXG-0078 indicate that the VH1-24 family of antibodies (27) can have robust breadth across both alpha- and beta-coronavirus lineages. In addition, these public antibodies are primed for recall through vaccination or infection and represent a segment of long-lived immunity that offers an important capability against novel variants. Pre-existing immunity to the NTD could have greater capacity than previously acknowledged. While NTD-targeting mAbs can utilize a variety of heavy chain variable genes, the VH1-24 public clonotype provides a common pathway for the generation of protective antibodies. As new variants continue to emerge, further assessment of the protective capacity of these type of high affinity mAbs can provide insights relevant for both vaccine and therapeutic development.

MATERIALS AND METHODS

Production of recombinant proteins and antibodies

The coding sequence for the SARS-CoV-2 WA-1 (GenBank, MN908947) stabilized spike trimer (S-2P) was a generous gift from J. McLellan. The S-2P sequence was subcloned into the pCMVR vector with C-terminal AviTag and poly-histidine tags. Four additional stabilizing mutations were added using the QuikChange multisite-directed mutagenesis kit (Agilent) to create the HexaPro variant (29) which was expressed in Expi293F cells (Thermo Fisher) and purified using Ni-NTA agarose (Qiagen) affinity chromatography and size exclusion chromatography on an Enrich 650 column (Bio-Rad). The biotinylated SARS-CoV-2 S-2P stabilized trimer for variants D614G, Alpha, Beta, Gamma, Delta, Epsilon, BA.1, BA.5, BQ1.1, and XBB.1.5 was produced as previously described (36). Monoclonal antibodies described here have been reported previously (10) and were expressed by co-transfecting plasmids encoding paired heavy and light chains into Expi293F cells. Antibodies were purified 4–5 days after transfection using AmMag Protein A magnetic beads and the AmMag SA purification system (GenScript), according to the manufacturer’s recommendations, with low pH elution and buffer exchanged into PBS. Alternatively, for larger scale preparations, antibodies were purified from cleared cell supernatants using Protein A agarose (Pierce) affinity chromatography. The purity and stability of mAbs was assessed by SDS–PAGE and Coomassie staining in both reducing and non-reducing conditions.

SARS-CoV-2 pseudovirus neutralization assay

A luciferase-based SARS-CoV-2 pseudovirus neutralization assay was used, as described previously (10). Briefly, SARS-CoV-2 pseudovirions (pSV) were produced by co-transfection of HEK293T/17 cells with a plasmid encoding SARS-CoV-2 S and an HIV-1 NL4-3 luciferase reporter plasmid (pNL4-3.Luc.R-E-, NIH AIDS Reagent Program). For all variants, the S expression plasmid sequence was 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. The following sequences for SARS-CoV-2 S glycoprotein variants were used: WA-1 (GenBank, YP_009724390), Alpha (GenBank, UFQ05186), Beta (GenBank, WNH06868), Gamma (GenBank, WNH06844), Delta (GenBank, UAL04647), Epsilon (GenBank, QPJ72086), BA.1 (GenBank, UUL70262), BA.2 (GenBank, WGP43633), BA.5 (GenBank, UPN16705), BQ.1.1 (GenBank, UWM38596), XBB.1.5 (GenBank, WDD59628), and JN.1 (GenBank, WVR23492). A D614G variant was also made from the Wuhan Hu-1/WA-1 construct using the Q5 site-directed mutagenesis kit (NEB). Infectivity and neutralization titers were determined using ACE2-expressing HEK293 target cells (Integral Molecular) in a semi-automated assay format using robotic liquid handling (Biomek NXp Beckman Coulter). Samples were diluted 1:40 in growth medium and serially diluted, and then 25 µL/well was added to a white 96-well plate. Purified mAbs started at a concentration of 1 mg/mL. An equal volume of diluted SARS-CoV-2 pSV was added to each well, and plates were incubated for 1 h at 37°C. Target cells were added to each well (40,000 cells/well), and plates were incubated for an additional 48 h. RLUs were measured with the EnVision Multimode Plate Reader (Perkin Elmer) using the Bright-Glo Luciferase Assay System (Promega). Neutralization dose–response curves were visualized in Prism 10, using a 4-parameter nonlinear regression analysis. Fifty percent inhibitory concentration (IC50) were calculated using 5-parameter logistic regression analysis from at least two independent experiments performed in triplicates (R package nplr). Assay equivalency was verified by participation in the SARS-CoV-2 Neutralizing Assay Concordance Survey (SNACS) run by the Virology Quality Assurance Program and External Quality Assurance Program Oversite Laboratory (EQAPOL) at the Duke Human Vaccine Institute, sponsored through programs supported by the National Institute of Allergy and Infectious Diseases, Division of AIDS.

Binding kinetics by biolayer interferometry

Biotinylated SARS-CoV-2 S-2P proteins (36), representing major variants, were immobilized on streptavidin sensors at ~1.5 nm. After establishing baseline against kinetics buffer (0.01% BSA, 0.002% Tween 20 in phosphate-buffered saline, pH 7.4), binding responses were obtained after a 450 s association phase with antibodies at 200 nM. A loaded sensor in kinetics buffer was used for reference subtraction. Binding association and dissociation constants to each respective protein were determined, from at least four concentrations of antibody (IgG or Fab). Binding curves were obtained after a 100 or 200 s association and up to 1,200 s dissociation step and fitted to a 1:1 binding model using the global fit function in the data analysis software 10.0 (FortéBio). Acceptable parameters of fit were R² > 0.95 and c² < 3. All assays were performed at 30°C with agitation set at 1,000 r.p.m. on an Octet RED96 instrument (FortéBio).

Measurements of antibody Fc effector functions using recombinant proteins

Biotinylated SARS-CoV-2 BA.5 S-2P stabilized trimer was incubated with yellow-green streptavidin-fluorescent beads (Molecular Probes) for 2 h at 37°C. Ten microliters of a 100-fold dilution of beads–protein mixture was incubated for 2 h at 37°C with 100 µL of monoclonal antibodies (8 serial 5-fold dilution starting at 25 µg/mL). Beads were then washed before adding effector cells for ADCP or ADNP.

ADCP

ADCP was measured as previously described (37). THP-1 cells (20,000 cells per well; Millipore) were used as effector cells. After 19 h incubation at 37°C, the cells were fixed with 2% formaldehyde solution, and fluorescence was evaluated on a LSRII flow cytometer (BD Bioscience). The phagocytic score was calculated by multiplying the percentage of bead-positive cells by the geometric mean fluorescence intensity (MFI) of the bead-positive cells and dividing by 10⁴.

ADNP

Fresh peripheral blood leukocytes from human (50,000 cells/well) were used as effector cells after red blood cell lysis with ACK lysing buffer (ThermoFisher Scientific). After 1 h incubation at 37°C, the cells were washed, surface stained, and fixed with 4% formaldehyde solution, and fluorescence was evaluated on a LSRII (BD Bioscience). Antibodies used for flow cytometry were anti-human CD3 AF700 (clone UCHT1) and anti-human CD14 APC-Cy7 (clone MϕP9; both BD Biosciences) and anti-human CD66b Pacific Blue (clone G10F5, Biolegend). The phagocytic score was calculated by multiplying the percentage of bead-positive neutrophils (SSC high, CD3− CD14− CD66+) by the geometric MFI of the bead-positive cells and dividing by 10⁴.

Measurements of antibody Fc effector functions using cell surface-expressed spikes

SARS-CoV-2 BA.5 S-expressing FreeStyle 293F and CEM-NK^R (NIH AIDS Reagent Program (Cat# 458; RRID:CVCL_X622) cells were generated by transfection with linearized plasmid encoding a codon-optimized full-length SARS-CoV-2 BA.5 S protein (GenBank, UPN16705). Stable transfectants were single-cell sorted and selected to obtain a high-level spike surface expressing clone.

Opsonization

293F-BA.5-S cells were incubated with 100 µL of monoclonal antibodies (8 serial 5-fold dilution starting at 25 µg/mL) for 30 min at 37°C. Cells were washed twice and stained with anti-human IgG PE (Southern Biotech). Cells were then fixed with 4% formaldehyde solution, and fluorescence was evaluated on a LSRII (BD Bioscience).

Antibody-dependent complement activation (ADCD)

An ADCD assay was adapted from reference 38. Briefly, 293F-S-S2A cells were incubated with monoclonal antibodies as described above and washed twice and resuspended in R10 media. Cells were washed with PBS and resuspended in 200 µL of guinea pig complement (Cedarlane), which was prepared at a 1:50 dilution in Gelatin Veronal Buffer with Ca²⁺ and Mg²⁺ (Boston BioProducts). After incubation at 37°C for 20 min, cells were washed in PBS 15 mM EDTA (ThermoFisher Scientific) and stained with an anti-guinea pig complement C3-FITC (polyclonal, ThermoFisher Scientific). Cells were fixed with 4% formaldehyde solution, and fluorescence was evaluated on a LSRII (BD Bioscience).

CD16 reporter assay (ADCC)

CEM-NK^R-BA.5-spike cells (100,000 cells per well) were incubated with monoclonal antibodies as described above. Cells were washed and 200,000 Jurkat-Lucia NFAT-CD16 cells (Invivogen) were added to each well in 100 µL of IMDM 10% FBS. The cells were then centrifuge for 1 min at low speed and co-cultured for 24 h at 37°C. Fifty microliters of Quanti-Luc was added to 20 µL of co-culture supernatant, and luminescence was measured immediately on a luminometer (2104 Multilabel reader, PerkinElmer).

Fab production

WRAIR IgGs in PBS buffer (pH 7.4) were combined with LysC protease (New England Biolabs) at a 1:2,000 mass ratio. The reaction was incubated in a water bath at 37°C and allowed to proceed for 2 h. Proteolysis was assessed by SDS-PAGE. To isolate Fabs, the reaction mixture was passed twice through protein A beads (Cytiva). The flow-through was collected and assessed by SDS-PAGE. Fab that was not promptly used was concentrated to ~10 mg/mL, snap frozen in LN₂, and stored at −80°C.

X-ray crystallography and structure analysis

WRAIR-2008 Fab (14 mg/mL) and WRAIR-2039 Fab (10 mg/mL) were screened for crystallization conditions using an Art Robbins Gryphon robot, 0.4 µL drops, and a set of ~1,000 conditions. Screens were observed daily using a Jan Scientific UVEX-P system. Crystals used for data collection were grown in 0.17 M ammonium sulfate, 25.5% (wt/vol) PEG 4000, 15% (vol/vol) glycerol (JCSG+ suite from Qiagen) or 0.12 M alcohols, 0.1 M buffer system 1 pH 6.5, 30% (vol/vol) precipitant mix 1 (Morpheus screen from Molecular Dimensions) for WRAIR-2008 and WRAIR-2039 Fabs, respectively. Crystals were cryoprotected with addition of glycerol to 25% (vol/vol).

Diffraction data were collected at the AMX beamline at the National Synchrotron Light Source-II (Brookhaven National Lab) for WRAIR-2008 or beamline 24-ID-E (NE-CAT) at the Advanced Photon Source (Argonne National Lab) for WRAIR-2039. Diffraction data were indexed, integrated, and scaled utilizing either autoPROC for WRAIR-2008 or RAPD, NE-CAT’s automated data processing pipeline, for WRAIR-2039. Phenix xtriage was used to analyze all scaled diffraction data output from autoPROC or RAPD for measurement value significance, completeness, asymmetric unit volume, and possible twinning or pseudotranslational pathologies. Mild anisotropy was identified in the WRAIR-2008 data and corrected using the UCLA Diffraction Anisotropy Server (https://srv.mbi.ucla.edu/Anisoscale/). Both Fab structures were determined by molecular replacement using MoRDa (39) in CCP4i2 (40) or PHASER (41) in Phenix. Fabs CM25 (PDB 7M8J) and FC08 (PDB 7DX4) were used as the search models for the heavy and light chains of WRAIR-2039. WRAIR-2039 was used as the search model for WRAIR-2008. Structures were iteratively built in Coot (42) and refined using Phenix Refine (43) or REFMAC5 (44). Non-crystallographic symmetry was applied during refinement of WRAIR-2008. Model quality was assessed with MolProbity (45). All data collection and refinement statistics are reported in Table S3.

Cryo-electron microscopy data collection

Sample was prepared for cryo-EM by combining WRAIR-2008 Fab with WA-1 HexaPro in a 3:1 molar ratio, followed by a 30 min incubation at 37°C. The complex was isolated by gel filtration chromatography using a Superdex 200 Increase 10/300 GL column (Cytiva) in PBS. Fractions containing the complex were concentrated to ~0.6 mg/mL, and glycerol was added to a final concentration of 2% (vol/vol). A volume of 2.5 µL of the HexaPro-WRAIR-2008 Fab complex was deposited onto a Quantifoil R 0.6/1 Cu 400 mesh grid. Excess fluid was removed by blotting with Whatman #1 paper for 10 s with a 2.1 mm offset and then vitrified in liquid ethane using a Leica EM GP Automatic Plunge Freezer.

Data collection was performed on a JEOL CryoARM 200 using SerialEM 4.1.0beta with a nominal magnification of 60 kX and a pixel size of 0.873 Å/pixel on an Ametek Gatan K3 detector in counting mode. Ninety frame movies with a total dose of ~50 e⁻/Å (2) were motion corrected and CTF was estimated using patch CTF in CryoSPARC Live v4.4.0. Particle picking, extraction, 2D classifications, ab initio reconstruction, homogeneous refinement, and non-uniform 3D refinement were all completed in CryoSPARC v4. Particle picking was performed using a circular Gaussian blob with a diameter of 150–300 Å with the closed spike (EMDB 11203; low-pass filtered) as the template. Iterative rounds of 2D classification were used to remove classes without clear secondary structural features or intact spike trimer. Particles from classes that contained visible Fab Fv density were used for ab initio reconstruction with a single class output; the resulting reconstruction was refined with the homogenous and non-uniform 3D refinement options in CryoSPARC. C3 symmetry was imposed throughout.

Local non-uniform refinement was employed to improve resolution at the Fab-NTD interface. Symmetry expansion was performed on the particle set for non-uniform refinement of the spike-Fab complex, followed by particle subtraction and local refinement in CryoSPARC.

Cryo-EM model building and refinement

SARS-CoV-2 HexaPro in the 3-RBD down conformation was built into the EM density map using PDB 6GZE (46) as the initial template. The WRAIR-2008 Fab variable region was modeled using the WRAIR-2039 Fab crystal structure as the template. Initial model fitting was performed using Isolde (47) in ChimeraX (48). Automated and manual model building were iteratively performed using real space refinement in Phenix and Coot. Sharpened maps were created by providing half maps to the Resolve Cryo-EM tool in Phenix (49). Geometry validation and structure quality assessment were performed using MolProbity, CaBLAM (50), and EMRinger (51), as implemented in Phenix. A summary of the cryo-EM data collection, reconstruction, and refinement statistics is shown in Table S2.

Structure analysis and figure preparation

Structural biology applications used in this project were compiled and configured by SBGrid (52). Interactive surfaces were analyzed with PISA (https://www.ebi.ac.uk/pdbe/pisa/). Structure figures were prepared using PyMOL v.2.5.2 (Schrodinger, LLC) and ChimeraX.

In vivo protection studies in K18-hACE2 transgenic mice

All research in this study involving animals was conducted in compliance with the Animal Welfare Act, and other federal statutes and regulations relating to animals and experiments involving animals and adhered to the principles stated in the Guide for the Care and Use of Laboratory Animals, NRC Publication, 1996 edition.

The research protocol was approved by the Institutional Animal Care and Use Committee of the Trudeau Institute and US Army Medical Research. K18-hACE2 transgenic mice were obtained from Jackson Laboratories. Mice were housed in the animal facility of the Trudeau Institute and cared for in accordance with local, state, federal and institutional policies in an NIH AAALAC International-accredited facility. For the prophylactic protection studies, on day −1, groups of 15 male and female K18-hACE2 mice (8–10 weeks of age) were injected intravenously with the purified antibodies at the indicated dose. On study day 0, all mice were inoculated with 1 × 10⁵ PFUs of SARS-CoV-2 Omicron BA.5 via intranasal instillation, a challenge dose determined to cause weight loss. In the therapeutic study, mice (8–10 weeks of age) were inoculated with SARS-CoV-2 BA.5 24 h before being injected intravenously with the indicated antibody cocktail. All mice were monitored for clinical symptoms and body weight twice daily, every 12 h, from study day 0 to study day 14. Mice were euthanized if they displayed any signs of pain or distress as indicated by the failure to move after stimulated or inappetence, or if mice had greater than 25% wt loss compared to their study day 0 body weight. From each group, a subset (five) of mice were euthanized 2 days after antibody administration, and lung tissue was harvested for the measurement of infectious virus titers in the lower respiratory tract using a plaque reduction neutralization test (PRNT).

Statistical analysis

Neutralization is the geometric mean of the IC50 values calculated using 5-parameter logistic regression from at least two-independent experiments performed in triplicates (R package nplr). A two-tailed Mann–Whitney t-tests was used to verify the existence of significant differences between % neutralization values obtained with WRAIR-2008 (25 µg/mL) against SARS-CoV-2 WA-1 and BA.5. In the animal study, one-way ANOVA with Dunnett’s multiple comparisons tests were used to assess significance in weight changes and viral loads across groups compared to the isotype control antibody-treated animals. Fold change in binding to the BA.5 S trimer was calculated relative to the binding obtained with the WA-1/2020 S trimer. Non-neutralizing mAbs were assigned the IC50 of 25 µg/mL, the mAb starting concentration in the assay. All tests, except for the five parameter logistic regression performed in R (version 4.3.1) and R studio (2023.06.1+524), were performed in Prism (version 10, GraphPad Software). Data were graphed using Prism software (version 10, GraphPad Software).

DATA AVAILABILITY

The associated data for the cryoEM structure reported in this paper are available via Protein Data Bank (PDB: 9ECZ, 9MI3) and the Electron Microscopy Data Bank (EMD-47928, EMD-48284). Data for crystallographic complexes are available from the PDB under accession codes 9ECX and 9ECY. Source data are provided within this paper. All other data are available from the corresponding authors upon request.

ETHICS APPROVAL

Research was conducted in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals and experiments involving animals and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals, NRC Publication, 1996 edition.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/mbio.01036-25.

Supplemental material. mbio.01036-25-s0001.pdf.

Figures S1-S7 and Tables S1-S3.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.