Protective Non-neutralizing anti–N-terminal Domain mAb Maintains Fc-mediated Function against SARS-COV-2 Variants up to BA.2.86-JN.1 with Superfluous In Vivo Protection against JN.1 Due to Attenuated Virulence

Severe acute respiratory syndrome CoV-2 has caused millions of deaths worldwide since it first emerged in 2019 (1). mAbs were successfully used as therapeutics early in the pandemic (2). These Abs disrupted the ACE2 receptor-spike receptor-binding domain (RBD) interaction, thus neutralizing the virus and protecting the host (3). However, with the emergence of variants, particularly Omicron, the Abs mostly lost the ability to neutralize the virus (4, 5). None of the clinically approved Abs are functionally active against the emerging Omicron subvariants XBB and BQ1.1 (6). This loss of function in these neutralizing Abs (nAbs) was primarily caused by the extensive mutations in the RBD in the Omicron subvariants at sites that most nAbs target. The emergence of these variants has led researchers to focus on generating pan-neutralizing Abs (7, 8). It remains to be seen whether this strategy will be successful against future variants of SARS-CoV-2.

Abs with potent Fc-mediated function have not been given the same level of attention as neutralizing Abs. Recently, we have shown that two potent opsonizing Abs, Ab81 IgG3 and Ab94 IgG1, targeting the RBD and N-terminal domain (NTD), respectively, strongly protect against the Wuhan strain in transgenic K-18 mice (9, 10). Notably, these Abs are non-neutralizing (nnAbs), mediating their protective effect through Fc-mediated effector functions. So far, only four studies have been published showing that four nnAbs can be protective where three bind to the NTD domain (called Ab94 IgG1, DH1052 IgG1, and CV3-13 GASDALIE mutant) and one against the RBD (Ab81 IgG3) (9–12). Other emerging findings with convalescent sera or nAbs with modified Fc effector functions show that Fc effector functions are crucial for anti–SARS-CoV-2 viral control (13–16). The protection granted by vaccines against severe disease by mutated variants like Omicron can be partially explained by intact Ab Fc effector functions directed against non-RBD sites (16). Thus, a viable strategy complementing the pan-universal nAb approach could be to generate opsonic Abs against conserved epitopes, avoiding the RBD and NTD sites targeted by most nAbs. However, no work has investigated whether these protective nnAbs, such as CV3-13, DH1052, Ab81, and Ab94, maintain binding, function, and in vivo protection against mutated variants such as the currently circulating JN.1.

Almost all mutations occurring in the RBD and NTD cluster around neutralizing epitopes, leaving many other epitopes unaltered. This localized mutagenesis led to our hypothesis that opsonizing Abs, which can be protective, could be used for targeting these escape variants. We observed that our anti-RBD mAb Ab81 lost binding eventually to BA.2.86. At the same time, the anti-NTD mAb Ab94 was still functional to all variants tested from Omicron to BA.2.86/JN.1. We performed in vivo experiments with the JN.1 virus in K18-hACE2 mice and discovered surprising results on the JN.1 virulence in this animal model with significantly attenuated phenotype, even compared with parent BA.2.86 in other published work (17). Still, clinical score assessment revealed that Ab94 showed some protective effect against this variant.

Expi293F suspension cells were purchased from Life Technologies (ThermoFisher) and routinely cultured in 125-ml Erlenmeyer flasks (Nalgene) in 30 ml of Expi293 medium (Life Technologies) in an Eppendorf s41i shaker incubator at 37°C with 8% CO₂ at 120 rpm. The cells were routinely passed and split to a density of 0.5 × 10⁶ cells/ml every 3 to 4 d. The day before transfection, the cells were seeded at a density of 2 × 10⁶ cells/ml. The next day, the cells were seeded at 7.5 × 10⁷ cells in 25.5 ml of Expi293 medium. Transfection with heavy and L chain plasmids was performed using the Expifectamine293 kit (Life Technologies) according to the manufacturer’s instructions. The plasmids for the heavy chains were generated previously (9, 10). Briefly, 20 μg of heavy and L chain plasmid, respectively, were mixed with 2.8 ml of OptiMEM (Life Technologies) and 100 μl of Expifectamine and incubated at room temperature for 15 min. Afterward, the transfection mix was added to the Expi293F cells. The following day, 1.5 ml of enhancer 1 and 0.15 ml of enhancer 2 (both from the Expifectamine293 kit) were added, and the cells were cultured for another 72 h.

The cells were removed from the cell culture medium by centrifugation (400g, 5 min), and the supernatant was transferred to a new tube. To capture the IgGs from the medium, protein G Sepharose 4 Fast Flow (Cytiva) was added to the medium and incubated end-over-end at room temperature for 2 h. The beads were collected by running the medium bead mix through a gravity flow chromatography column (Bio-Rad) and washed twice with 50 ml of PBS. The Abs were eluted using 5 ml of HCl-glycine (0.1 M, pH 2.7). Tris (1 M, pH 8, 1 ml) was used to neutralize the pH. The buffer was exchanged to PBS using Amicon Ultra-15 centrifugal filters (Σ) with a molecular cut-off of 30,000 Da. The concentrations of the purified Abs were spectrophotometrically measured with the IgG setting of the Nanodrop (Denovix).

Wuhan Spike protein was generated by transfecting Expi293F cells with 40 µg of plasmid containing the gene for the Spike protein (CS/PP Spike encoding a secretable version of the protein was used to allow purification from cell culture supernatants), donated previously to us by Dr. Florian Krammer’s laboratory (18). Omicron (158-40589-V08H26-100), BA.2 (158-40589-V08H28-100), BA.4 (158-40589-V08H32-100), BA.5 (158-40589-V08H33-100), XBB (158-40589-V08H40-100), BQ1.1 (158-40589-V08H41-100), XBB1.5 (BPS-101677-2), and BA.2.86 (158-40589-V08H58-100) was acquired from Sino Biological. Then, 25 µg of Spike protein was biotinylated according to the instructions of EZ-Link Micro Sulfo-NHS-LCBiotinylation Kit (ThermoFisher). Fluorescent (APC) streptavidin microsphere beads (63 µl from stock) (1 µm, Bangs Laboratories) were conjugated with the biotinylated Spike protein (25 µg) according to the manufacturer’s instructions, as done previously (9, 10).

The binding assays were performed in a 96-well plate precoated with 200 µl of 2% BSA (in PBS) at 37°C for 30 min 250,000 Spike-coated beads were used in all wells, with an Ab concentration of 1 µg/ml. The beads were opsonized at a volume of 100 µl in 1× PBS at 37°C for 30 min on a shaking heat block. The wells were washed twice with PBS. To assess Ab binding to Spike beads, 50 µl of (1:500 diluted) a Fab-specific fluorescent secondary Ab (109–546-097, Jackson ImmunoResearch) was used to create a fluorescent signal. The secondary Ab was left to incubate with the Spike–bead Ab complex at 37°C for 30 min on a shaking heat block. 100 µl of PBS was added to the wells before analysis in the flow cytometer. The beads were analyzed using a Beckman Coulter Cytoflex flow cytometer, which acquired 10,000 beads per sample. The data were processed using FlowJo. The gate for Spike beads was set based on forward and side scatter (Supplemental Fig. 1A). The gate for Spike beads positive for Abs was set based on reactivity to a non-reactive IgG control mAb (Supplemental Fig. 1B).

To study binding kinetics to Spike trimer or NTD, we immobilized a high-capacity amine sensor chip (Bruker) with anti-human IgG (Fc) Ab (Cytiva BR-1008-39) at 25 µg/ml in 10 mM sodium acetate buffer pH 5 at flow rate 10 µl/min and contacting 300 s. This was done in a MASS-16 biosensor instrument (Bruker). Running buffer PBS + 0.05% Tween 20. Ags were acquired from Nordic biosite (Omicron 158-40589-V08H26-100, BA.2 158-40589-V08H28-100, BA.4 158-40589-V08H32-100, BA.5 158-40589-V08H33-100, XBB 158-40589-V08H40-100, XBB.1.5 158-40592-v08h146-100, and BQ.1.1 158–40589-V08H41-100). The Abs were diluted in PBS and injected over the surface for 90 s at 10 µL/min. The running buffer was PBS with 0.01% Tween 20. All variants of the Spike trimer were analyzed at 20 to 1.25 nM concentrations. The XBB.1.5 RBD was also analyzed at concentrations ranging between 20 and 1.25 nM. The Ags were injected at these concentrations and were allowed to interact with the sensor for 2 min, with a flow rate of 30 µl/min, followed by a dissociation for 6 min. After each cycle, the surface was regenerated with 3 M MgCl. All experiments were performed once. The data were analyzed using Sierra Analyzer software version 3.4.3 (Bruker) program to determine apparent dissociation constants (K_D) and constant rates (k_A).

For ELISA avidity measurements, 1 µg/ml of BA.2.86 spike protein trimer (158-40589-V08H58-100, Sino Biological) in PBS was immobilized onto ELISA high-bind plates (Sarstedt) overnight at 4°C. Similarly, for the XBB1.5 S1 (158-40591-V08H47-100, Sino Biological) and Wuhan S1 (154–31814, Sino Biological) affinity measurements, 1 µg/ml was coated overnight. The wells were washed with PBST (1× PBS with 0.05% Tween 20) three times and then blocked with 2% BSA/PBS for 1 h at room temperature. After three washes, the primary Abs (anti-spike mAb) were added (100 µl of mAbs with a concentration of 10 µg/ml) and incubated for 45 min at 37°C. The wells were washed three times with PBST after that. A rabbit anti-human heavy and L chain–HRP secondary Ab (Rockland Bionordika, 609-103-123) at a dilution of 1:5000 in PBS was added and left to incubate for 45 min at 37°C. The wells were washed three more times with PBST. Finally, 100 μl of developing reagent (20 ml of sodium citrate, pH 4.5 + 1 ml 2,2′-azino-bis [3-ethylbenzothiazoline-6-sulfonic acid] diammonium salt, 0.2 g in 10 ml of water, Σ + 0.4 ml of 0.6% H₂O₂) was added. OD₄₅₀ was recorded after 30 min. For XBB1.5 and Wuhan S1-subunit affinity analysis, a one-site nonlinear regression model using GraphPad Prism was used to determine K_D, where B_Max and K_D were left unconstrained. The data were plotted with GraphPad Prism.

THP-1 cells (Sigma-Aldrich) were cultured as described previously (19). In all experiments, 1 × 10⁵ THP-1 cells were used. In the experiments, the ratio of cells to beads ranged from 6.25 to 25 for the curves (multiplicity of prey [MOP], 6.25 to 25) and MOP 25 for the XBB experiments. For all experiments, the Spike beads were opsonized with 2.5 µg/ml of Abs in a volume of 100 µl of sodium media (pH adjusted to 7.3 with NaOH; 5.6 mM glucose, 10.8 mM KCl, 127 mM NaCl, 2.4 mM KH₂PO₄, 1.6 mM MgSO₄, 10 mM HEPES, 1.8 mM CaCl₂). Opsonization was performed for 30 min at 37°C on a shaking heat block (300 rpm) in a volume of 100 µl. During this incubation period, THP-1 cells were counted (using a Bürker chamber), and the medium was exchanged from RPMI to sodium medium. THP-1 cells (50 µl, 2 × 10⁶/ml) were added to each well, and the cells were allowed to phagocytose beads for 30 min at 37°C on a shaking heat block (300 rpm). The 96-well plate was then incubated on ice for 15 min and analyzed directly in a Beckman Coulter Cytoflex flow cytometer. A gate for THP-1 cells was set up based on their forward and side scatter. The gate for association was set with a negative control of cells only. The analysis stopped after 5,000 events were captured in the THP-1-gate. The data were analyzed in the program FlowJo by setting similar gates as described above (Supplemental Fig. 2A). The FlowJo-processed data were further analyzed in GraphPad Prism, in which the bead signal (APC-A) of the THP-1 gate association gate, percentage of associating cells, and phagocytosis score were plotted to compare the different Abs. The phagocytosis score was calculated based on the percentage of bead⁺ cells (associating cells) multiplied by the median fluorescence intensity (APC-A) of that population as described previously (20, 21).

Neutrophil phagocytosis assays were done as described previously (10) on a 96-well plate, with similar gating strategies employed (Supplemental Fig. 2C). In short, neutrophils were isolated from whole blood from volunteering healthy donors using polymorphprep (Abbot). Ethical approval for this study protocol was acquired from the Swedish Regional Ethical Committee (Etikprövningsnämnden Lund, 2015/01801). The spike beads were opsonized at MOP 30 with 2.5 μg/ml of monoclonal Abs and 1% Ab-depleted serum (Peelfreez). After 30 min of opsonization at 37°C on a shaking heat block, 1 × 10⁵ neutrophil cells in a volume of 50 µL were added to the opsonized beads and left to phagocytose for 30 min at 37°C on a shaking heat block. Neutrophils were counted by using an XN-350 hematology analyzer (Sysmex). After the phagocytosis step, the 96-well plate was put on ice for 20 min, and the cells were stained with an anti-CD18 marker at 2 μg/ml 20 µl volume (BV421 mouse anti-human CD18, BD Biosciences). After staining, the plate was analyzed on a Beckman Coulter Cytoflex flow cytometer, and the data were acquired by gating for size and granularity followed by CD18-positive signal. Association and phagocytosis scores were analyzed similarly as with the THP-1 cells.

Fourteen 8-wk-old female K18 hACE2 (B6.Cg-Tg(K18-ACE2)2Prlmn/J) mice were split into two groups of seven mice each for Ab94 IgG1 and vehicle. Abs were administered in one single dose, i.p., 8 h before virus-inoculation under anesthesia. The animals were inoculated intranasally with 10⁵ PFU of SARS-CoV-2 (JN.1 strain, produced at the Institute of Virology in Universitätszmedizin Berlin as done previously [22]) in 25 µl under 3.5% isoflurane anesthesia. The mice’s body weights and health status were recorded daily, and the animals were euthanized if they lost more than 20% of their body weight or showed severe deterioration in health status. The infection proceeded for 8 days before the animals were all euthanized and the experiment terminated. Blood, tissue, and bronchoalveolar lavage (BAL) were harvested and stored accordingly. All the animal experiments were performed under the approval of the regional animal experimental ethics committee in Stockholm (approval no. 2020-2021). BAL fluid analysis for viral titers was analyzed as described previously using quantitative PCR (qPCR) (9, 10).

Statistical analysis was performed in GraphPad Prism. For comparisons between clones when more than two treatments were analyzed, a one-way ANOVA with multiple comparisons test was used, with correction for multiple comparisons with Dunne or Tukey tests, depending on the comparisons being made. A two-tailed Mann–Whitney U test was used to compare the two treatments.

Analysis of spike mutations was performed using the lineage comparison tool at GISAID, which was recently described (23), by aligning the strains of interest with a reference strain B.1.

In previous work, we established a set of mAbs from B cells of previously SARS-CoV-2–infected subjects (Fig. 1A). We found that six mAbs (Ab11, Ab36, Ab66, Ab77, Ab81, and Ab94) were non-neutralizing but opsonic and capable of triggering phagocytosis (9). Two mAbs (Ab57 and Ab59) were both neutralizing and opsonic. The non-neutralizing Ab94 and Ab81 have both been shown to be protective against stringent infection (10⁵ PFU) of Wuhan wild-type (WT) strain in a hACE2-K18 mouse model (9, 10). Ab81 IgG3 and Ab94 IgG3 were assessed in vivo due to their potent Ab-dependent cellular phagocytosis (ADCP) and Ab-dependent neutrophil phagocytosis (ADNP) function, whereas the IgG1 version of Ab94 was used as a proof of concept for nnAb function due to its average opsonic ability. In the case of Ab94, protection has been demonstrated both in therapeutic and prophylactic models (Fig. 1A). Ab81 was originally an IgG1 in the original donor, whereas Ab94’s subclass is unknown. However, whether Ab81 and Ab94, which bind to the RBD and NTD, respectively, would retain binding and protective function against the heavily mutated spike variants was unclear. It is well known that RBD mutations occur in neutralizing epitopes and sites relevant for ACE2 receptor binding (4–6). For NTD, there exist five common sites that anti-NTD nAbs target: residues 14–26, 67–79, 141–156, 177–186, and 246–260 (N1-N5 loops) (24, 25). Furthermore, residues 14–20, 140–158, and 245–264 have been identified as Ag supersites for anti-NTD nAbs (24, 25). Epitopes outside the neutralizing residues are considered non-neutralizing and potentially opsonic.

We looked at all mutations in the relevant variants from Omicron up to EG.5.1 and BA.2.86. Interestingly, although the spike protein has undergone extensive mutations in both the RBD and NTD, most mutations occur in the neutralizing sites. In contrast, non-neutralizing sites generally have few mutations (Fig. 1B, 1C; Supplemental Fig. 3A, 3B). The Omicron variant added mutations in the N2 loops (A67V, del69-70) but also in more distal sites in the NTD not associated with neutralizing mAbs (T95I, del211, L212I, and INS214EPE). The Omicron BA.2, 4, and 5 subvariants added more mutations in the N1 (T19I, L24S, and del25/27) and N3 loop (G142D). Only one mutation was added outside the neutralizing loops (V213G) (Supplemental Fig. 3C). It is evident that most of the Omicron mutations occurred in the neutralizing epitopes in the N1–N3 loops and some mutations on residues 211–214, which are not associated with neutralizing Abs. The SARS-Cov-2 variants further mutated in BQ1.1 and XBB, with a majority of the NTD mutations in the N2–N4 neutralizing epitopes (V83A, del144, H146R, and Q183E) and one mutation between the N4–N5 loop (V213E) (Supplemental Fig. 3D). We looked at the more heavily mutated variants XBB.1.5, 1.16, 1.19, 2.3x, CH1.1, and EG.5.1. It was even more clear that there is evolutionary pressure to mutate in the neutralizing epitopes with mutations primarily in the N3 (K147E, W152R, and F157L), N4 (E180V) and the N5 loop (G252V, D253G, and G257S) (Supplemental Fig. 3E). EG.5.1 added a mutation between the N1–N2 loop (Q52H). Finally, the newly emerged BA.2.86 has continued to add mutations in or close to the neutralizing epitopes: in the N1 (R21T), N3 (F157S and R158G), and N5 loops (H245N and A264D) (Supplemental Fig. 3F), although the other mutations from BA.2.86 and JN.1 correspond to non-neutralizing sites such as S50L, V127F, 211del, L212I, and L216F. Taken together, most of the mutations in the SARS-CoV-2 variants concerning the NTD end up in the neutralizing epitopes N1–N5 and some, with unknown significance, outside these loops, primarily in residues 211–216 (Fig. 1B). Thus, Abs targeting residues outside of these loops can possibly retain their binding and function against these mutated variants. However, even if these mAbs bind to nonmutated epitopes, they might lose binding and function due to allosteric changes from other mutations. We decided to see whether our protective Ab94 and Ab81 retain binding to their non-neutralizing but protective epitopes in the mutated SARS-CoV-2 variants.

Although we do not have complete information on the epitope sites for our anti-RBD nnAbs (Ab11, Ab36, Ab66, Ab77, and Ab81), we wanted to use these as well to test the hypothesis of using nnAbs to target, theoretically, less mutation-prone epitopes. In addition, we wanted to see whether the nAbs Ab59 and Ab57 maintain their binding to the mutated variants. We used spike-coated microsphere beads to assess the reactivity of our mAbs against the Wuhan spike protein by incubating the beads (coated with Wuhan, Omicron, BA.2, BA.4, and BA.5 spike protein) with our anti-spike Abs. We assessed reactivity by adding fluorescent anti-Fab secondary Abs (Supplemental Fig. 1A, 1B). We analyzed IgG binding to the spike-coated beads with flow cytometry as done previously (9, 10). All clones were analyzed as IgG3 except for Ab11, which was of the IgG1 subclass because it previously had shown reduced affinity in the IgG3 subclass. Of the eight clones, which are reactive to the Wuhan spike protein (Fig. 1A, 1D), six showed reactivity to the Omicron variant (BA.1) (Fig. 1E). Interestingly, only the nAb clones (Ab57 and Ab59) were those that lost binding to the Omicron variant. Contrary to those, the nnAbs Ab11, Ab36, Ab77, Ab81, and Ab94 retained strong reactivity, whereas Ab66 showed weaker binding. The same result was seen for the Omicron subvariants BA.2, BA.4, and BA.5 (Fig. 1E–H), reinforcing that opsonic Abs can potentially bind to more conserved epitopes. In contrast, neutralizing Abs are at significant risk of losing binding and function.

Because the bead-based assay was used as a screening method to assess maintained binding, we wanted to pursue a more in-depth analysis of the reactivity of our non-neutralizing mAbs against these Omicron variants. To investigate the binding kinetics of our clones toward the spike variants with greater resolution, we performed surface plasmon resonance assays (SPR) with the multivalent spike trimer as done previously (10). For these assays, we chose to focus on clones 66, 77, 81, and 94 in the IgG3 subclass because these clones were shown to have potent in vitro Fc effector functions previously in this subclass (10). The results showed that the apparent dissociation constant (K_D) and association rate constant (k_A) are equivalent for all tested clones when comparing the reactivity toward the Wuhan spike protein and the variants Omicron, BA.2, BA.4, and BA.5 (Fig. 2A–D, 2E). Ab66 was an exception that showed reduced binding to BA.5 (3.75 nM) compared with WT (0.65 nM). Interestingly, Ab94 showed stronger binding to the mutated variants compared with WT (for instance, 2.5 nM to Omicron but 8.5 nM to WT), ranging from 3-fold better to more than 100-fold (BA.2) (Fig. 2D, 2E). Ab81 showed subnanomolar binding toward all variants tested. We further analyzed the avidity of clones 81 and 94 against the newer variants BQ.1.1 and XBB. We focused on these two clones because they were both shown to be protective against a lethal Wuhan infection in a K18-hACE2 mouse model (9, 10). These SPR experiments showed that both clones retain a high affinity toward BQ1.1 and XBB (Fig. 2F, 2G). Ab81 shows reduced binding toward BQ1.1 (1 nM) and subnanomolar to XBB (0.35 nM). Ab94 binds with similar avidity to XBB as to the Wuhan spike (k_A and K_D) but, interestingly, binds 5-fold better to BQ1.1 following a pattern similar to that seen before with Omicron and its variants (Fig. 2F, 2G). Finally, we analyzed the binding ability of Ab81 against the RBD of XBB 1.5 due to its mutation in the RBD site F486P. This experiment showed a similar result that Ab81 has a subnanomolar affinity toward even this mutated variant of XBB (Fig. 2F, 2G) comparable to the affinity to Wuhan RBD measured previously (0.2 nM) (10). We did not analyze Ab94 for this variant due to a lack of access to the whole spike trimer or NTD compatible with SPR. In summary, the results show that the opsonic nnAbs retain strong binding to variants that emerged from Omicron, whereas the nAb clones did not. Finally, the previously protective clones Ab81 and Ab94 seemingly show high-affinity binding against these heavily mutated spike trimer variants. The highly protective Ab94 binds even better to the mutated variants, except for XBB, which is comparable to the WT spike. These data reinforce the hypothesis that protective nnAbs, such as Ab81 and Ab94, can maintain strong reactivity toward mutated SARS-CoV-2 variants, whereas nAbs, such as Ab57 and Ab59, lose their function.

Having established our clones’ binding properties, we decided to investigate whether this translates to intact in vitro effector functions in the form of Fc-mediated phagocytosis. We opted to test their function on BQ.1.1 (the circulating variant in Sweden during the fall of 2022) spike trimer-coated microsphere beads. We have previously shown that our mAbs are opsonic in a flow cytometry–based phagocytosis assay of spike beads with the THP-1 cell line (9, 10) and that the results are relevant for biological protection in vivo. First, we tested the individual mAbs in both IgG1 and IgG3 subclasses for clones 81 and 94. We looked at the percentage of cells that were bead-positive and the median signal for beads in this population, which is a metric for bead quantity (Supplemental Fig. 2B). Because the phagocyte–bead interaction is dependent on these metrics, we calculated a phagocytosis score based on the percentage of cells that are bead-positive multiplied by the median fluorescence signal (bead signal) of that population, as described in other works (20, 21).

Interestingly, both subclasses of IgG1 and IgG3 for the mAbs show potent Fc-mediated function compared with the negative control (Xolair, human IgG1 anti-IgE). This was observed in the percentage of cells that are interacting with the BQ1.1 spike-coated beads (bead⁺ cells), with all single IgGs showing comparable performance in this metric compared with the negative control (Fig. 3A). Furthermore, we analyzed the number of beads being phagocytosed in cells associated with beads by quantifying the bead signal. In this analysis, IgG3 Ab81 stood out as the most potent opsonin, followed by Ab94 IgG3 (Fig. 3B). Ab81 and Ab94 IgG1 outperformed the negative control as well. Finally, we analyzed the phagocytosis score to acquire a comprehensive overview of the bead–phagocyte interaction. This analysis shows that Ab81 IgG3 enhanced bead uptake by the THP-1 cells more than 4-fold compared with the negative control, and Ab94 IgG3 showed similar activity with a 3.5-fold enhancement (Fig. 3C). No difference was seen between the IgG subclass controls (Fig. 3D). The IgG1 versions of Ab81 and Ab94 were 2.7-fold more potent than the negative control, trailing their IgG3 variants. Taken together, the results show that the opsonic clones 81 and 94 retain Fc-mediated function against the BQ1.1 variant and that the IgG3 versions are, as is the case with Wuhan, more potent opsonins.

In a previous study, we showed that combining a potent anti-NTD Ab (Ab94) with that of a potent anti-RBD (Ab81) generated a superior opsonic performance (10). We named this mixture DuoMab, and the IgG3 version of these mAbs showed the largest Fc-mediated effector functions against the Wuhan strain (10). Having established potent function in both subclasses (IgG1 and IgG3) of Ab81 and Ab94 against BQ1.1, we set out to test whether their potency could be enhanced by combining them in the same two-antibody mixture as done against the Wuhan strain (DuoMab) but against BQ1.1. We performed similar experiments by incubating opsonized spike-coated beads with THP-1 cells. Only the DuoMab IgG3 outperformed the single mAbs at equivalent concentrations. DuoMab IgG1 and the individual mAbs Ab81 IgG1 and Ab94 IgG1 had very similar associating cells (Fig. 3E). There was an increase in association with DuoMab IgG3 (Fig. 3E). When looking at phagocytosis scores, there was a 1.3-fold difference for the DuoMab IgG1 versus single IgG1 (Fig. 3F). Similarly, as with the percentage of bead⁺ cells, DuoMab IgG3 had higher phagocytosis score compared with single IgG3 mAbs (Fig. 3F). Compared to the negative control, DuoMab IgG3 was the most potent opsonin, with a 5-fold higher phagocytosis score (Fig. 3F).

Furthermore, because the ratio of prey (spike-coated beads in this case) is a known factor influencing phagocytosis outcome (19), we decided to vary the ratio of prey to phagocytes, the MOP, as described previously (19). We plotted the MOP against the percentage of associating cells and used nonlinear regression analysis to generate the MOP₅₀ value (equivalent to EC₅₀) to compare the cocktails versus single mAbs. A lower MOP₅₀ value signifies an Ab that mediates more efficient phagocyte–prey interaction and is a more potent mediator of Fc-dependent phagocytosis. We observed that although Ab81 and Ab94 IgG3 are both more efficient than the control (Ab81 IgG3 MOP₅₀ = 9, Ab94 IgG3 MOP = 8, control MOP₅₀ = 48; Fig. 3G), they are both most effective when used in conjunction as DuoMab IgG3 (MOP₅₀ = 6; Fig. 3G). Similarly, as for the IgG3 variants, the IgG1 single mAbs outperformed the negative control (Ab81 IgG1 MOP₅₀ = 14, Ab94 IgG1 MOP₅₀ = 12, Control MOP_{50 =} 48; Fig. 3H) but were less efficient than their IgG3 counterparts. In this article, the DuoMab IgG1 mixture was more effective than both individual IgG1s (MOP₅₀ = 9). However, although DuoMab IgG1 was more potent than its IgG1’s single mAbs, it was not more effective than Ab94 IgG3 and was inferior to both DuoMab IgG3 and Ab81 IgG3, respectively (Fig. 3G). Further analysis indicates that DuoMab IgG3 exhibits a cooperative function with a Hill-slope constant (n_H) above 1 (Fig. 3J). We did not observe such cooperative functions for the individual mAbs nor DuoMab IgG1.

To extend our findings with the potent functions of DuoMab IgG3 compared with IgG1, we performed similar experiments with XBB and XBB.1.5 spike trimer-coated beads. Before those experiments, we assessed whether the G252A mutation in XBB.1.5 influenced Ab94’s binding to this variant. We did not observe a difference by ELISA for the mutated XBB.1.5 S1 subunit compared with the Wuhan, in which Ab94, exhibits like Ab81, retains strong binding to the XBB.1.5 variant (Supplemental Fig. 4A). Concerning opsonic function, our phagocytosis experiments with XBB and XBB.1.5-coated beads results reinforced our findings with the Wuhan and BQ1.1 strain, with IgG3 DuoMab being highly opsonic compared with IgG1 DuoMab in terms of overall phagocytosis performance (Fig. 3K). For XBB.1.5, unlike XBB, the DuoMab IgG1 performed worse compared with IgG3 measured as a phagocytosis score; however, this can possibly be attributed to biological variability of using a different THP-1 cell batch and beads because these experiments were done on different occasions. Our results show that both Ab81 and Ab94 exhibit potent Fc-mediated function against BQ1.1 and that engineering these mAbs in the IgG3 subclass and combining them in a two-antibody mixture generates even more potent functions against XBB, BQ.1.1, and XBB1.5.

During August 2023, a new subvariant of BA.2, BA.2.86, emerged and replaced most of the circulating variants with additional mutations primarily in the RBD and NTD region of the spike protein (Fig. 4A). This variant further developed into the current (as of 2024 February) dominating variant JN.1 with an additional mutation, L455S, in the RBD according to sequence data deposited in the GISAID database (23). We assessed the reactivity of our protective anti-RBD and anti-NTD mAbs Ab81 and Ab94, respectively, against the BA.2.86 spike protein, the parent to JN.1, by ELISA. We performed all our assays on the BA.2.86 spike protein because we could not acquire the JN.1 spike protein then. Interestingly, these results showed that Ab81 lost binding completely against the BA.2.86 variant due to the extensive mutations in the RBD domain (Fig. 4A). Compared to BA.2, XBB, XBB.1.5, and BQ1.1 (Ab81 binds to all mentioned variants), BA.2.86 has acquired I332V, K356T, R403K, V445H, N450D, L452W, N481K, Del483, and E484K in the RBD domain (23), whereas residues 332, 356, 403, 450, 452, 481, 483, and 484 are not mutated in BA.2, XBB, XBB.1.5, and BQ1.1. Thus, any of these novel mutations were most likely responsible for Ab81’s loss of binding rather than V445H mutation because XBB.1.5 has a mutation V445P. Ab94, In contrast, maintained intact binding to BA.2.86 spike protein (Fig. 4A) despite adding 10 additional mutations not present in XBB.1.5, XBB, BQ1.1, and parent BA.2 (R21T, S50L, V127F, F157S, R158G, N211I, del212, L216F, H245N, and A265D) (23). We performed both in vitro (THP-1) and ex vivo phagocytosis experiments (neutrophils) to assess Fc-mediated opsonic function against BA.2.86-coated microsphere beads. For the neutrophil experiments, we added 1% Ab-depleted serum to include potential complement-mediated phagocytosis. These experiments showed that Ab94 retains its Fc function against this heavily mutated variant. Large differences were seen with the IgG3 mAb, being almost 2-fold superior to the original IgG1 version for both ADCP and ADNP, in line with our findings with Wuhan, BQ1.1, XBB, and XBB.1.5 (Fig. 4B).

Having established that Ab94 exhibits intact Fc-mediated function against BA.2.86-coated beads, we wanted to see whether this translates to intact in vivo protective effects. In previous work, Ab94 IgG3 did not perform as well in vivo in the K18-hACE2 mouse model against Wuhan compared with Ab94 IgG1. The discrepancy between our experimental results and in vivo work with Ab94 concerning both subclasses are hypothesized to be due to differences in phagocytes and Fc receptors between the human in vitro and ex vivo experiments and murine in vivo model. We thus proceeded with in vivo experiments with only Ab94 IgG1 because we did not expect Ab94 IgG3 to perform better against JN.1 than against Wuhan. We then assessed whether our non-neutralizing anti-NTD mAb had any protective function against JN.1 challenge in a prophylactic in vivo model of K18-hACE2 mice (Fig. 4C, 4D). We treated seven mice with 200 µg of Ab94 IgG1 or just vehicle 8 h before inoculating the mice intranasally with 10⁵ PFU of JN.1 SARS-CoV-2. The high dose of the virus, which is 10–100 times higher than given in most experiments in which neutralizing mAbs have been evaluated against Wuhan (7, 8, 14), was given because we did not know whether this variant would, like Omicron (26), be attenuated and not as virulent as Wuhan in this animal model. In addition, we wanted the experiment to be comparable to the previous work in which 100,000 PFU of Wuhan virus was given in the exact same experimental setup. We assessed protective function by looking at survival data over 9 days, reduction in mean body weight loss, and titers of virus in BAL fluid as done previously. We also employed a clinical scoring system in which body posture, piloerection, abnormal respiration (shortness of breath, wheeze, cough), and weakness/fatigue were assessed on a scale where 0 indicated unaffected, 1 indicated mild symptoms, 2 indicated moderate symptoms, and 3 indicated severe symptoms, similarly as in the other published work (27). Interestingly, we observed that at what was previously a lethal dose of Wuhan, JN.1 at the same inoculation dose did not cause any weight loss or mortality in all 14 mice except for one animal in the Ab94 IgG1-treated group, which experienced very sudden rapid weight loss (at day 5) and was euthanized; this mouse was interpreted as an outlier (Fig. 4E). When looking at clinical scores, we noted that mice in the Ab94-treated group (except the outlier, which developed significant weight loss after day 5 and was euthanized at day 7) did not show many clinical symptoms (Fig. 4F). The mice in the control group experienced higher clinical scores than those in the Ab94 IgG1 group at days 2, 7, and 9. At the termination of the experiment (day 9), the mice in the control group were in worse clinical condition, with an average score of 1.14 (3 mice being unaffected, 2 having a score of 1, one having a score of 2, and one having a score of 4) than those in the Ab94 IgG1 group, which only had one mouse with clinical symptoms (score of 2) (Fig. 4G). Thus, Ab94-treated mice had more than 3-fold better mean clinical score. Furthermore, we looked at viral titers in terminal BAL fluid (day 9) by qPCR, which would give an answer regarding viral load in the lower airways. These results showed that the negative control group had a 24.6 CT value (cycle threshold), and the Ab94 IgG1 group titer was 24.5, respectively. These values are in a range similar to that of Ab94-treated mice exhibited against Wuhan in our previous work (median CT value 24.3, marked out in the graph in gray shades with mean ± SEM). At the same time, the vehicle group against Wuhan had a much lower CT value (17.8, marked as mean ± SEM in gray shades) (Fig. 4H). This suggests that the JN.1 exhibits a low viral load after 9 d postexposure in the lower airways and that the mouse immune system, which is naive against SARS-CoV-2, can eliminate the virus to low levels without severe morbidity and mortality. Thus, any clear benefit from non-neutralizing mAb treatment is difficult to observe in this model against this particular viral strain, despite positive results in clinical scoring. Finally, the outcome of the animal experiment reveals important novel insights into JN.1 virulence, which is severely attenuated as compared with other SARS-CoV-2 strains, even recently mutated ones such as parent BA.2.86, which still inflicts weight loss in hamster models known for acquiring mild SARS-CoV-2 disease (17).

mAb therapy against SARS-CoV-2 was successful in the initial phase of the pandemic (2). These Ab therapeutics neutralized Abs that targeted specific RBD sites important for interaction with the ACE2 receptor. Unfortunately, mutated variants such as Omicron and its sublineages showed extensive mutations in these neutralizing epitopes, leading all previously clinically approved mAbs to lose binding and function (4–6). Although many neutralizing mAbs generated initially against the Wuhan strain lost binding and function, researchers have successfully developed mAbs that neutralize previously relevant mutated variants (7, 8). However, when testing these nAbs against emerging variants, several studies have shown a reduced neutralizing effect (IC₅₀) compared with the Wuhan strain in the newer variants (7, 8, 28, 29). It appears likely that the trend will continue and that these nAbs will also eventually lose neutralizing capacity or are at a large risk of doing that.

Neutralizing Abs constitute a fraction of the adaptive Ab response against the SARS-CoV-2 virus. Most Abs that bind to the spike Ag are not neutralizing (30). However, this does not necessarily imply that they do not serve any protective immune function. In previous work, we showed that all our Ab clones that bind to the spike protein promoted efficient phagocytosis through Fc effector functions (9). The importance of Fc-mediated effector functions of anti-spike mAbs has not been the scientific community’s focus as an important immune mechanism against severe disease by SARS-CoV-2. For other viral pathogens such as adenovirus (31), West Nile virus (32), HIV-1 (33), respiratory syncytial virus (34, 35), human CMV (36), and influenza virus (37–39), Ab Fc-mediated effector functions are important for viral clearance. Thus, it is not surprising that more evidence is emerging showing that non-neutralizing Abs mediate protective Fc effector function against SARS-CoV-2. The evidence ranges from work with polyclonal Abs from convalescent patients (13, 15), modification of amino acids affecting Fc effector function of neutralizing Abs (14), to non-neutralizing Abs (9–12).

In this article, we show that two protective nnAbs, Ab81 and Ab94, retain high-affinity binding against variants up to XBB.1.5 and BQ1.1 and retain Fc-mediated opsonic function against BQ1.1, XBB, and XBB.1.5. The mutations in the BA.2.86 variant RBD led to loss of binding for Ab81, while Ab94 remained unaffected. Ab94 was derived from Swedish patients in March 2020 (9), early during the pandemic. No variants of concern had yet surfaced. This means that these protective non-neutralizing Abs did not arise as a host response to any spike variants and that it is likely that the human immune system has been able to generate such protective nnAbs in response to SARS-CoV-2 infections all through the pandemic. This ability is likely common also in response to commonly used spike-based vaccines. The ability to generate protective Abs against conserved non-neutralizing epitopes contributes to the ability of vaccination using early vaccine variants and previous, early infections to remain relatively effective in preventing serious disease, also following infection with emerging viral variants (40).

In our previous work (10), Ab94 and Ab81 were both protective against an infection of authentic SARS-CoV-2 in K18-hACE2 mice in a prophylactic setting. This is even more striking when considering that survival benefits were shown in a model suboptimal for human IgGs effector functions because human IgG (especially IgG3) promotes weaker phagocytosis by mouse phagocytes (41, 42). Additionally, human IgG3 has a much shorter half-life in mouse circulation than human IgG1 (43). Therefore, the protective effects shown by these two mAbs could possibly be even better in a humanized murine model than the K18-hACE2 model due to possibly suboptimal activation of murine Fc receptors by human IgG subclasses as compared with murine IgG subclasses. Furthermore, Ab94 IgG1 was also shown to be protective in a therapeutic model, in which it was comparable to clone 59, a potent nAb with IC₅₀ of 19 ng/ml (9). This comparison suggests that the protective benefit of nnAbs is comparable to that of nAbs. Notably, the mice were inoculated with a lethal dose of SARS-CoV-2 in both infection models (10⁵ PFU) (9, 10). This dose is higher than that used in most studies that have shown the protective benefits of nAbs. Most researchers have tested the nAb protective effect against doses ranging from 10³ PFU to 10⁴ (7, 8, 14, 44–46), which are sublethal in comparison (47). Thus, not only did Ab94 retain binding and function against the new variants 4 years after its generation, but its Fc-mediated function has also been shown to be protective against death and weight loss in infection models in which mice were challenged with a higher, lethal viral dose of Wuhan strain compared with those commonly used to assess nAb protectiveness (7, 8, 44–47).

In this work, we report, to the best of our knowledge, the first instance of assessment of JN.1 virulence in the K18-hACE2 mice model. Despite using a very high dose of the virus (100,000 PFU), we observed low virulence of this variant with no body weight loss and survival being 100% for all groups except for one mouse (interpreted as a clear outlier). This somewhat aligns with BA.2.86 and EG5.1 virulence in golden hamsters. However, the hamsters experienced weight loss, which is important to highlight because those animals are more resilient against weight loss and death than K18-hACE2 mice (48, 49). Thus, JN.1 exhibits much less virulence than the parent BA.2.86 when comparing across models, in which based on the results in the hamster model with BA.2.86 (17), JN.1 would have been expected to cause weight loss if its virulence would be comparable to BA.2.86. Interestingly, we observed that viral titers in the lower airway for all groups were comparable to what was previously protective (Ab94-treated) Wuhan viral levels in mice. We interpret these results in two ways. One, studying Fc-mediated effector functions against JN.1 in this model reveals that a previously strongly protective mAb against Wuhan, inoculation dose-for-dose, adds encouraging trends in clinical scoring but no added benefit in preventing weight loss (because it does not occur), survival (because no mice in control did not die), or reduction of viral titers at the end of the experiment (because of low levels of virus). For future work, larger doses of the virus and nasal swabs taken at early time points (days 2, 4, and 6 postinfection) are likely better to study the protective effects of non-neutralizing mAbs. The second interpretation is that it seems as though JN.1, through its L455S mutation, has traded lower affinity to the hACE2 receptor compared with BA.2.86 while being more infective and immune evasive than BA.2.86 and other contemporary variants such as HV.1 and JD1.1 (50, 51). Taken together, these changes have seemingly led to a much less virulent infectious pathogen but with high fitness to become dominant in 2023 and early 2024.

Our results further raise questions on the importance of different immune mechanisms in protection against JN.1 and its subvariants and how much selection pressure is added to the non-neutralizing epitopes targeted by previously protective mAbs such as Ab94. More work is needed to fully understand these intriguing host–pathogen interactions, building on the results established here. Our work reports a detailed timeline on how two protective nnAbs against spike protein (of which only four have been reported in published work) maintain and lose their binding and functional activity against the SARS-CoV-2 variant, with which the BA.2.86 variant made a large negative impact on the anti-RBD clone. Based on our work, it is suggested that the development of non-neutralizing Abs against spike would benefit by avoiding the RBD domain and focusing on the NTD domain instead, such as Ab94, which has been shown here to be robust against mutations (10 novel mutations from BA.2.86 compared with its then contemporary variants of concerns) by targeting a more conserved epitope with potentially still protective effects against JN.1 variant in the form of clinical score assessment.

Finally, our work here raises questions on the protective benefit of nnAbs that are equal to Ab94 IgG1 in opsonic ability, ADCP, such as Ab11 (for instance), because they maintain binding to Omicron and its subvariants. Our work highlights their promise because only two of our nnAbs have been tested in vivo, and both were protective against stringent Wuhan infection (100,000 PFU) and, as shown in this article exemplified by Ab94, maintain function against mutated variants after 4 years of viral evolution. More work is needed to understand the extent of nnAb-based protection. Adding further potential, nnAbs can be enhanced in various ways, such as by IgG3 hinge-engineering in the Fc domain to potently enhance opsonic ability, which we have done for three anti-spike mAbs, Ab11, Ab36, and Ab77, recently (52). Future work should investigate whether using Ab-engineering strategies on nnAbs to enhance immune function against these heavily mutated variants like JN.1 is a feasible approach to enhance non-neutralizing protective function.

A.I. and P.N. have filed a patent pending for the Abs described in this article. The other authors have no financial conflicts of interest.

We thank the people in the Department of Clinical Microbiology of Lund University Hospital for RNA extraction and qPCR and employees at Scantox, especially Michelle Gustafsson, for in vivo work.