Nasal delivery of secretory IgA confers enhanced neutralizing activity against Omicron variants compared to its IgG counterpart

Abstract

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and its multiple variants continue to spread worldwide, causing respiratory symptoms primarily through mucosal infection. The mucosa serves as the primary barrier against viral entry, in which secretory immunoglobulin A (sIgA) plays a critical role in preventing infection. Here, we engineered and characterized a neutralizing monoclonal antibody, ZW2G10, in IgG, monomeric, dimeric, secretory IgA1, and IgA2 formats. All seven forms of the ZW2G10 antibody showed similar thermal stability. sIgA, especially sIgA1, displayed enhanced neutralizing activity against Omicron-lineage BA.2.75, BA.2.76 and BA.4/5 pseudoviruses compared to IgG. Nasal administration of sIgA1 conferred robust protection against the BA.2.76 pseudovirus in ACE2 transgenic mice, and its protective efficacy was superior to that of IgG. The crystal structure of Omicron receptor binding domain (RBD) and ZW2G10 antibody fragment (Fab) complex revealed that ZW2G10 had no clashes with ACE2. Thus, nasal administration of sIgA may serve as a promising tool for the prevention and treatment of Omicron infection.

Graphical abstract

Mucosal tissues serve as the primary immune barrier against SARS-CoV-2, where secretory IgA (sIgA) predominates. Yu and colleagues report that, by increasing avidity, recombinant secretory IgA enhances its neutralizing ability, which is expected to be an effective tool for controlling future SARS-CoV-2 as well as other mucosal susceptible pathogens.

Introduction

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic continues globally, causing profound public health impacts. SARS-CoV-2 primarily relies on its surface spike protein to attach to the angiotensin-converting enzyme 2 (ACE2) receptor, which facilitates entry into host cells.¹ The respiratory mucosa serves as the initial site of infection, highlighting the critical role of mucosal immunity in preventing viral spread.^2,3 Immunoglobulin A (IgA), the predominant antibody on the mucosal surface,⁴ has been identified as a central mediator in the early humoral immune responses against SARS-CoV-2,⁵ whereas IgA deficiency is associated with more severe clinical symptoms.⁶ Additionally, IgA antibodies have demonstrated neutralizing activity against influenza, human immunodeficiency virus, and other viruses.^7,8

The concentration of IgA in the bloodstream is second only to that of IgG.⁹ In serum, IgA primarily exists as monomers, while on the mucosal surface, the predominant form of IgA is secretory IgA (sIgA).^10,11 The structure of monomeric IgA (mIgA) is similar to that of IgG. The constant region of the heavy chain of IgG is the γ chain, whereas that of mIgA is the α chain. In humans, IgA exists as two closely related subclasses, IgA1 and IgA2, with the main difference being the hinge region. The hinge of IgA1 is 13 amino acids longer than that of IgA2. Therefore, IgA1 is beneficial in antigen recognition with distantly spaced antigens but at the cost of increased susceptibility to proteolysis. In addition, IgA2 has a high degree of glycosylation.¹⁰ Two mIgAs are joined together to form a dimeric IgA (dIgA) via the joining chain (J chain). Additionally, sIgA also contains a secretory component (SC).¹² During the transport of dIgA from the basolateral surface of mucosal epithelial cells to mucosal surfaces, it binds to the polymeric Ig receptor (pIgR). The pIgR is cleaved to release the SC, which combines with dIgA to generate sIgA.¹³ The SC helps maintain the stable structure of sIgA and prevents its degradation in the gut or other mucosal environments.¹⁴

sIgA is widely distributed in tears, saliva, milk, and mucosal secretions in the respiratory, gastrointestinal, and urogenital tracts.¹⁵ Most sIgA molecules are dimers, but there are also multimers, such as tetramers.¹⁶ It is an important player in local mucosal immunity by providing multiple antigen-binding sites, exhibiting higher affinity for antigens, and displaying enhanced neutralizing activity.^17,18 In addition, sIgA also has nonspecific immune functions,^19,20,21 including forming antigen-antibody aggregates with pathogens to prevent them from penetration of the mucosal layer and enhance immune protection by interacting with other innate immune defense factors in mucosal secretions, such as mucin and lactoferrin. Interestingly, during sIgA secretion, pIgR mediates the transmembrane transport of dIgA-antigen complexes, thereby clearing invading soluble antigens or viral particles.²² However, the therapeutic potential of IgA has not been fully developed, and its pharmacological properties require further elucidation. One of the reasons is that it is difficult to obtain sufficient recombinant IgA with desired activity to meet research needs. Compared to IgG production, IgA has both monomeric and polymeric forms, which undoubtedly increases the complexity of its purification.

In recent years, the significance of IgA in mucosal immunity has become increasingly recognized, and its therapeutic potential in infectious diseases, oncology, and immunological disorders has been gradually explored.¹⁵ However, no IgA therapeutic agents have been developed into clinical trials. Here, we generated mIgA, dIgA, and sIgA based on the variable region gene derived from ZW2G10, a neutralizing monoclonal antibody discovered in a vaccinated donor. We found that the IgA counterparts retained the binding capacity to the spike protein of Omicron variants compared to their IgG parent. Moreover, compared to IgG, sIgA demonstrated stronger neutralizing activity against the SARS-CoV-2 pseudovirus both in vitro and in vivo, potentially attributable to its higher avidity for the S_ECD protein. Analysis of the structural basis for the neutralizing efficacy was conducted using the crystal structure of the ZW2G10 antibody fragment (Fab) bound to the Omicron receptor binding domain (RBD), revealing that ZW2G10 does not structurally compete with ACE2 binding to the receptor binding motif (RBM).

Results

Production and characterization of IgA antibodies

We have previously discovered a human monoclonal antibody (mAb), ZW2G10, which broadly targets the spike protein of different SARS-CoV-2 variants, including Omicron subvariants, and that confers potent neutralizing efficacy against SARS-CoV-2 pseudoviruses. To further evaluate the therapeutic potential of ZW2G10, different forms, including IgG (referring to the IgG1 subtype unless otherwise specified), mIgA1, dIgA1, and sIgA1 as well as mIgA2, dIgA2, and sIgA2, were constructed, produced, and characterized (Figure 1A). All seven antibody forms exhibited two common bands corresponding to heavy and light chains, while dIgA showed the additional J chain band, and in addition, sIgA showed the SC band. All components were consistent with their theoretical molecular weight, as revealed by SDS-PAGE under reduced conditions (Figure 1B). IgG is composed of 55 kDa heavy chains and 25 kDa light chains, while the heavy chains of IgA1 and IgA2 are approximately 70 kDa. IgA1 has a longer hinge region but lower glycosylation levels, so the molecular weight on the gel appeared similar to that of IgA2. The detailed differences in the amino acid sequences of the constant region of IgA1 and IgA2 heavy chains are listed in Figure S1. The J chain of dIgA and sIgA is approximately 25 kDa, and the SC of sIgA is 80 kDa. Further identification of each component was performed using western blotting (Figure 1C). HPLC chromatograms revealed a single symmetrical peak for each form (Figure 1D). The thermal stability test results showed that the melting temperatures (Tm) of IgG and all IgA antibodies were close, between 65.77°C and 68.16°C, indicating that these Igs have similar thermal stability (Figure 1E).

Figure 1.

Purification and characterization of IgA variants

(A) Schematic of IgG and IgA structures. The tailpieces are located at the C termini of the IgA. The dimeric and secretory forms of IgA2 are not depicted. (B) SDS-PAGE analysis of IgG and IgA under reduced conditions. The lane labeled M contains a ladder indicator. (C) Western blot analysis of IgG and IgA under reduced conditions. The purified products were analyzed by western blotting with an anti-human IgG Fc antibody, anti-human IgA Fc antibody, anti-lambda LC antibody, anti-J chain antibody, and anti-SC antibody. (D) The HPLC profile of ZW2G10 isotypes. Differently colored curves represent the detection peaks of PBS, IgG, mIgA, dIgA, and sIgA at 280 nm. (E) Thermal denaturation of ZW2G10-IgA variants, monitored by intrinsic tryptophan fluorescence. The left curve graph shows raw data, the barycentric mean (BCM) of the fluorescence emission spectra is plotted, and the right curve graph shows smoothed first derivatives used to calculate melting temperatures. The bar chart displays the Tm value of each antibody. Error bars indicate the mean ± SD from three replicates.

During antibody production, we noted that the yield of sIgA was relatively lower than of other forms, which has also been reported in other research.²³ Thus, we optimized the expression conditions of sIgA, including transfected vector ratios and the production collection time. In transfected HEK293T cell culture supernatants, sIgA achieved peak yield when the molar ratio of the four vectors (heavy chain:light chain:J chain:SC) was 4:4:1:1 (Figure S2A). Using this optimal transfection strategy, we transfected these vectors into Expi293F cells and observed that 5 days after transfection was the best time for collection of protein production (Figure S2B).

Binding and neutralizing capacity of purified IgG and IgA antibodies

To compare the binding properties of the ZW2G10 IgG and IgA forms, we performed an ELISA using the ectodomain of the spike protein (S_ECD) from different variants of SARS-CoV-2. All seven forms of ZW2G10 bound to S_ECD from SARS-CoV-2 wild type (WT) and Omicron BA.1, BA.2.75, BA.2.76, and BA.4/5 with one-half-maximal effective concentration (EC₅₀) values ranging from 0.038 to 1.393 μg/mL (Figure 2A). ZW2G10-IgG showed strong binding to all tested S_ECD with EC₅₀ values ranging from 0.038 to 0.093 μg/mL, while the maximum absorbance value of BA.2.76 was lower than those of other variants. The conversion of IgG to IgA did not significantly increase the binding capacity of ZW2G10 to S_ECD except for mIgA1 binding to the WT, BA.1, and BA.2.76 as well as dIgA1 binding to BA.2.76, as shown in Figure 2A; ZW2G10 in these IgA forms showed higher EC₅₀ values than IgG. We present the EC₅₀ values in molar concentration in Figure S3. Interestingly, ZW2G10 in IgA1 form, whether in monomeric, dimeric, or secretory form, displayed enhanced binding capability to all tested variants compared to the IgG form, except that mIgA1 showed binding similar to, but slightly weaker than, that of IgG to BA.2.75 and BA.4/5.

Figure 2.

Binding and neutralizing capacity of purified IgG and IgA antibodies

(A) The binding capacity of antibodies toward S_ECD proteins. The EC₅₀ and fold change differences between IgG and IgA antibody forms are indicated. (B) The neutralizing capacity of antibodies against pseudotyped SARS-CoV-2 variants. The IC₅₀ and fold change differences between IgG and IgA antibody forms are indicated. (C) The neutralizing capacity of antibodies against authentic SARS-CoV-2 variants. (D) K_on and K_off of the IgG and IgA1 antibody against the S_ECD proteins as determined by BLI. The K_D value differences between IgG and IgA1 antibody forms are indicated. Error bars indicate the mean ± SD from three replicates.

The neutralizing activity of ZW2G10 IgG and IgA antibodies was evaluated using a pseudotyped virus neutralization assay in HEK293T cells stably expressing ACE2 (Figure 2B). ZW2G10 IgG could effectively neutralize SARS-CoV-2 WT pseudotyped virus with an IC₅₀ value of 9.851 ng/mL. The neutralizing activity of IgG was slightly reduced against BA.2.75, moderately reduced against BA.4/5 and markedly reduced against BA.2.76. mIgA1 exhibited neutralizing activity comparable to IgG against all four pseudotyped viruses, while mIgA2 showed enhanced neutralization against BA.2.76 and BA.4/5. dIgA1 showed neutralizing capacity similar to mIgA2 against the WT, BA.2.75, and BA.2.76, whereas dIgA1 exhibited greater potency than mIgA2 against BA.4/5. sIgA1 and sIgA2 demonstrated the most potent neutralizing activity against all tested pseudotyped viruses, with IC₅₀ values for the WT, BA.2.75, and BA.4/5 below 7.981 ng/mL. Notably, sIgA1 neutralized BA.2.76 with an IC₅₀ value of 61.97 ng/mL, representing a 161.3-fold improvement compared to IgG. ZW2G10 IgG and the most potent form, sIgA1, were further evaluated for neutralizing activity against WT, BA.1, and BA.4/5 authentic viruses (Figure 2C). IgG neutralized the WT, BA.1, and BA.4/5 with IC₅₀ values of 17.08, 45.73, and 1509 ng/mL, respectively. Compared to IgG, sIgA1 exhibited enhanced neutralizing activity against the WT, BA.1, and BA.4/5 with IC₅₀ values of 3.09, 18.47, and 152.9 ng/mL, respectively. These results demonstrate that the dimeric form of IgA, particularly the secretory form of IgA1, could confer enhanced neutralizing activity against various SARS-CoV-2 variants compared to its IgG counterpart. In addition, we generated ZW2G10 as an IgG3 subtype with long hinge regions and evaluated its binding and neutralizing activity (Figure S4). ZW2G10 IgG3 demonstrated binding capacity comparable to IgG1 and mIgA1 against S_ECD of all five tested variants. IgG3 exhibited neutralizing activity similar to mIgA1 against all tested pseudotyped viruses except BA.2.76. Although IgG3 slightly enhanced its neutralizing activity against BA.2.76, its neutralizing activity against all tested pseudoviruses was significantly lower than that of sIgA1. These results suggest that the long hinge region may enhance antibody neutralization, but less significantly than the enhancement conferred by the dimeric form.

The enhanced neutralization of dIgA and sIgA was thought to be related to their increased avidity for antigens.²⁴ We further investigated the binding kinetics of the antibodies to S_ECD using a biolayer interferometry assay (Figure 2D). ZW2G10 IgG showed similar association and dissociation rates for BA.2.75, BA.2.76, and BA.4/5, with the equilibrium dissociation constant (K_D) ranging from 0.512 to 3.520 nM. Compared to IgG, mIgA1 showed similar K_D values for BA.2.76 and BA.4/5 but increased avidity for BA.2.75 (K_D, 0.094 nM). dIgA1 showed significantly increased avidity for BA.2.75 and BA.4/5, with similar association rates but much lower dissociation rates. The same results were observed for sIgA1. The increase in avidity of dIgA1 and sIgA1 to BA.2.76 was not as profound as for BA.2.75 or BA.4/5, but it was worth noting that the maximal shift values of IgG and mIgA1 to BA.2.76 were very low, while those of dIgA1 and sIgA1 were much higher (Figure S5). The increased avidity could partly explain the enhanced neutralization of dIgA1 and sIgA1, but there might be other mechanisms involved. SC has been reported to be involved in the immune response against mucosal pathogens²⁵; we thus tested the neutralizing efficacy of SC against Omicron pseudoviruses. The results demonstrated that SC could confer dose-dependent neutralizing activity against Omicron lineage pseudoviruses (Figure S6) and, thus, might mediate enhanced neutralization by sIgA.

In vivo protective efficacy of ZW2G10 IgG and sIgA1

To verify the in vivo protective efficacy of the ZW2G10 IgG and IgA formats, we used a mouse model that stably expressed ACE2 and was infected with a lentivirus-based SARS-CoV-2 pseudovirus. Infection was monitored by an in vivo imaging system (IVIS) at the indicated time points to track the expression level of the luciferase reporter gene. The imaging results showed a significant bioluminescence signal in the nasal cavity of mice on day 4 after infection, with luminescence intensity peaking on day 6 and slightly decreasing by day 7 (Figures 3A and 3B). To test the inhibitory efficacy of our antibodies, ZW2G10 IgG was administered intranasally 1 h before challenge with a BA.2.76 pseudovirus. Compared with the control group, prophylactic intranasal administration of 1.6 mg/kg (p < 0.01) or 0.16 mg/kg (p < 0.05) ZW2G10 IgG significantly reduced pseudovirus infection in mice (Figure 3C). The protection efficacy of sIgA1 was also tested using this assay. ZW2G10 sIgA1 showed robust inhibition activity against the BA.2.76 pseudovirus at both 1.6 and 0.16 mg/kg when administered 1 h before challenge and tested on day 6 after challenge (Figures 3D and 3E). Furthermore, the neutralization percentage was calculated on day 6 after infection, with values of 82.1% for 1.6 mg/kg IgG, 54.6% for 0.16 mg/kg IgG, 96.9% for 1.6 mg/kg sIgA1, and 83.3% for 0.16 mg/kg sIgA1 (Figure 3F). The sIgA1 showed better protective activity, which was consistent with the results of the in vitro neutralization experiments.

Figure 3.

Evaluation of ZW2G10 antibody protection efficiency in the mouse model of intranasal pseudovirus infection

(A) Construction and validation of the bioluminescent imaging mouse model. ZW2G10-IgG was administered 1 h before the inoculation with pseudovirus to evaluate protection efficiency. Average radiance was measured 1, 4, 6, and 7 days post infection. (B) Changes in the average radiance of bioluminescence signals at different time points. (C) Quantification of the average radiance on the sixth day after infection. (D) Bioluminescence signals in the nasal cavity of mice infected with pseudoviruses. ZW2G10-sIgA1 was administered 1 h before the inoculation with pseudovirus to evaluate protection efficiency. Average radiance was measured 6 days post-infection. (E) Bioluminescence signals on the sixth day after infection. (F) Comparison of protection efficiency between ZW2G10-IgG and ZW2G10-sIgA1 on the sixth day after infection. Data represent mean ± SD. An unpaired t test was used for comparisons of two groups. The asterisks shown in the figures refer to the level of significance: ∗p  ≤  0.05; ∗∗p ≤ 0.01.

Crystal structure of the Omicron RBD and ZW2G10 Fab complex

To elucidate the mechanism for ZW2G10 retaining its neutralizing activity against Omicron variants, we determined the crystal structure of the ZW2G10 Fab bound to the Omicron RBD (BA.1) (Figure 4A). The final complex structure was determined to a resolution of 2.6 Å (Table S1). The binding motif of ZW2G10 is not overlapping with ACE2 and is accessible in both up and down conformations of the RBD (Figure 4B). ZW2G10 binds to the Omicron RBD at an interface covering an area of 1,348 Å², mainly via the heavy-chain (HC) complementarity-determining regions CDRH2 and CDRH3. The intertwined contacts between the HC and the Omicron RBD accounted for 91% of the binding surface area contributed by the ZW2G10 HC and light chain (LC). There appeared to be an extensive hydrogen bond network between Asn343, Arg346, Asn354, Val445, Asn450, Thr470, and Asn481 from the Omicron RBD and Arg16, Asn54, Ile58, Lys65, Asn74, Gln82, Tyr107, and Glu111 from the ZW2G10 HC. Notably, there was also a salt bridge between Arg346 of the RBD and Glu111 of the HC CDR3 (Figure 4A).

Figure 4.

Crystal structure of the ZW2G10 Fab in complex with the SASR-CoV-2 Omicron RBD

(A) Interactions between the ZW2G10 Fab and the Omicron RBD; the hydrogen bonds are represented as yellow dashes. Left: the HC of the ZW2G10 Fab made main contact with the RBD; the HC and LC are colored violet and light pink, respectively. Right: magnified views of the binding interface. Residues Arg16, Asn54, Ile58, Lys65, Asn74, GLN82, Tyr107, and Glu111 of HC CDR2 and CDR3 formed an extensive hydrogen bond network with Asn343, Arg346, Asn354, Val445, Asn450, Thr470, and Asn481 from the Omicron RBD. (B) Comparison of the binding mode of ZW2G10 with ACE2 and other neutralizing antibodies, including S309 and LY-CoV1404. The RBD is presented as a surface and cartoon, colored white. ACE2 (light pink), ZW2G10 (violet), LY-CoV1404 (slate), and S309 (aquamarine) are shown as cartoons. (C) Footprints of various antibodies on the Omicron RBD with ZW2G10 outlined in violet, LY-CoV1404 in slate, and S309 in aquamarine. The major contacting residues of ZW2G10 are indicated and shown as sticks. (D) Amino acid mutation comparison of RBDs from SARS-CoV-2 WT, BA.1, BA.2.75, BA.2.76, and BA.4/5 variants.

Superimposition of the crystal structure of the ZW2G10-SARS-CoV-2 Omicron RBD complex onto the cryoelectron microscopy structure of the ACE2-SARS-CoV-2 RBD complex (PDB: 7WPB) revealed no clashes between ACE2 and ZW2G10. This finding was consistent with the inability of ZW2G10 to block ACE2 engagement, and ZW2G10 might perform its neutralization activity by blocking membrane fusion.²⁶ We compared the binding epitopes of the neutralizing antibody ZW2G10 with those of S309 and LY-CoV1404 (bebtelovimab) on the Omicron RBD (Figure 4C). The S309-derived antibody sotrovimab has been used previously in clinical therapy, and the binding motif of S309 was outside of the RBM. However, S309 is quite distinct from ZW2G10 in that it fails to neutralize Omicron BA.2 in vitro; evidence showed that S371F seems to be responsible for the impotency of S309. ZW2G10 forged tight interactions with the RBD and occupied a larger buried surface area than that of the LY-CoV1404. The RBD-binding site of LY-CoV1404 included residues Thr345, Arg346, Val445, and Tyr449 of the RBD, which were also bound by ZW2G10. However, the bulk of the LY-CoV1404 footprint overlaps with the RBM, which mechanistically allows LY-CoV1404 to achieve neutralization by blocking ACE2 binding.

Neutralizing mechanism of ZW2G10

ZW2G10 IgG could neutralize the SARS-CoV-2 WT pseudovirus with an IC₅₀ value of 9.851 ng/mL and BA.2.75 and BA.4/5 with IC₅₀ values of 20.16 ng/mL and 240 ng/mL, respectively. The neutralization capacity of ZW2G10 IgG against BA.2.76 was significantly reduced. The extensive interaction network between the ZW2G10 and RBD interface indicated that mutations G339H and G446S slightly weakened the neutralization of ZW2G10 IgG. Since mutations G339D and L452R changed the surface charge distribution and located at the opposite ends of the binding interface, which reduced the binding avidity, the neutralizing IC₅₀ of ZW2G10 against the BA.4 variant increased 24.4-fold. The R346T mutation largely contributed to the escape of the BA.2.76 variant, as R346 was positioned in the center of the hydrogen interaction network and formed three hydrogen bonds and four salt bridges with CDRH1 and CDRH3 of ZW2G10, including a main-chain interaction with Asn54 of CDRH1 (Figure 4A).

Omicron spike preferentially adopts the one-RBD-up conformation, and recent studies have shown that the S371L, S373P, and S375F substitutions enhanced the stability of the one-RBD-up conformation of Omicron.²⁷ ZW2G10 mainly made direct contact with the up-RBD of the Omicron S trimer but also touched the down-RBD from a neighboring protomer (Figure 5B). While ZW2G10 did not directly interact with the S371L, S373P, and S375F substitutions of Omicron RBD, we presumed that ZW2G10 may dissociate the Omicron spike by other mechanisms and analyzed the interaction interface in detail. Residues Phe486 and Tyr489 of the adjacent down-RBD are closely surrounded by hydrophobic residues Tyr369, Leu371, Ala372, Pro373, Phe374, Phe375, and Phe377 and forms extensive hydrophobic contacts with the up-RBD of the same Omicron spike trimer (Figure 5A). Superimposition of the crystal structure of ZW2G10-SARS-CoV-2 Omicron RBD complex over the SARS-CoV-2 Omicron spike structure (PDB: 7WVN) revealed that hydrophobic residues move toward the interface of the up-RBD (Figure 5C). The Val367 main chain makes hydrogen bonds with the N-glycan of Asn343 and the main chain of Tyr369, resulting in a 3.1 Å shift. The side chain of Asn370 establishes interaction with the hydrophobic residue Ala372, finally obliquely disrupting the hydrophobic interactions between the up-RBD and the down-RBD of the Omicron spike trimer (Figures 5B and 5C). As a result, the binding of ZW2G10 may decrease the stability of the one-RBD-up conformation, thereby exerting a neutralizing effect on the viral infection.

Figure 5.

The interface between the up-RBD and down-RBD in the apo SARS-CoV-2 Omicron spike and ZW2G10-bound Omicron spike

(A and B) The up-RBD of the Omicron spike and ZW2G10-bound Omicron spike are shown as a cartoon and colored green (A) and cyan (B), respectively. (C) Superimposition of the ZW2G10-bound Omicron up-RBD over the apo SARS-CoV-2 Omicron up-RBD. The down-RBD is colored white and shown as a surface and cartoon, and the locations of residues Phe486 and Tyr489 of the down-RBD are labeled and presented as sticks, and residues 369–375 and Phe377 of the up-RBD are labeled and presented as a surface and sticks. Asn343-glycan, involved in the interaction with the hydrophobic residues, is also indicated and shown as sticks.

The conversion of ZW2G10 IgG to mIgA1 did not increase the neutralization activity against SARS-CoV-2 variants. However, dimerization of IgA1 greatly increased the neutralizing efficacy of ZW2G10, denoted as indicated by lower IC₅₀ values compared to those of parental IgG. Furthermore, a significant increase in neutralization activity of ZW2G10 to BA.2.76 was only observed for the sIgA1 form, which implied an improvement of the binding avidity, as the SC fragment is involved in maintaining the stability of dIgA. To further explain the neutralization mechanism of the sIgA1 form of ZW2G10, we superimposed the RBD and ZW2G10 complex onto the solved sIgA1 structure. Due to the long and flexible hinge region of IgA1, only the structure with one Fab was reported; the ZW2G10 sIgA1 model was constructed with two Fabs. The neutralizing epitopes of SARS-CoV-2 display loosely on the S protein, spaced by 25 nm, which restricts the simultaneous binding by two arms of the same IgG.²⁸ Compared to IgG, sIgA1 has a longer hinge region, and the antibody molecules are closer to the T type than the classical Y type, giving the Fab segment more flexibility, allowing both arms to bind simultaneously to the two receptor-binding regions of the inter-spike, overcoming steric hindrance to enhance the neutralization capacity (Figures 6A and 6B). Thus, except for the SC component, the increased neutralization activity can also be attributed to the increased binding efficacy mediated by the inter-spike binding of sIgA1.

Figure 6.

Mechanism of the increased neutralizing potency of the ZW2G10 sIgA1 form

Structure models show inter-spike linking by ZW2G10 IgG (A) and sIgA1 (B) antibodies. The distance between the inter-spike that binds to the two arms of a single IgG molecule was about 13.6 nm, as measured by the binding footprint of the two up-RBDs. In the sIgA1 form, the longer hinge region enabled a more flexible binding mode, and the distance between the two up-RBDs of inter-spike that binds to the two arms of one sIgA1 molecule could reach 26 nm.

Discussion

SARS-CoV-2 Omicron lineages primarily infect the upper respiratory tract, where the local mucosal immune response is the first barrier against viral invasion. While IgG is the most abundant Ig in circulation, sIgA plays the dominant role in a mucosal environment. The multivalent structure and unique SC endow sIgA with superior properties, such as more potent neutralizing activity and a longer half-life in a mucosal environment. Here, we investigated the function and structure of ZW2G10, a human mAb with neutralizing activity against Omicron variants, and characterized the binding and neutralizing potency of monomeric, dimeric, and secretory forms of ZW2G10 IgA1 and IgA2.

Despite the excellent performance of sIgA, the complexity of its composition makes it more challenging to produce.²³ By optimizing the ratio of plasmids encoding each component, the yield of ZW2G10 sIgA1 in HEK293T cells improved by more than 4.4-fold when using a 4:4:1:1 M ratio compared to 1:1:1:1 (HC:LC:J chain:SC). The optimal plasmid ratio 4:4:1:1 is consistent with the molecular composition of sIgA1, as one sIgA molecule comprises 4 HCs, 4 LCs, 1 J chain, and 1 SC. The optimized condition may be extended to other mammalian- or plant-based expression systems to improve sIgA yield.

The conversion of IgG to IgA retains, but does not universally improve, the binding capacity of ZW2G10 to S_ECD, which may be related to the variable region, the constant region, as well as the antigen types. A comprehensive understanding of the functional characteristics of IgA antibodies still requires further investigation. Additionally, we conducted a preliminary evaluation of the activity of the constructed IgG3 subtype. Similar to IgA1, the primary distinction of IgG3 compared to other IgG subtypes lies in its hinge region. Our data indicated that the hinge region of IgG3 serves as a critical determinant mediating alterations in neutralizing activity, consistent with previous reports.^29,30 However, this enhancement in activity was no more pronounced than that conferred by dimerization, a finding that provides valuable insights for antibody class-switching studies.

Notably, when the binding EC₅₀ values were expressed as molar concentrations, dIgA1 and sIgA1 exhibited 3.15- and 2.248-fold stronger binding to BA.2.76 compared to IgG, respectively. This result aligned with the enhanced avidity of dIgA1 and sIgA1 toward S_ECD, as measured by biolayer interferometry (BLI). In contrast, IgA2 failed to exhibit superior binding to IgG, even in its dimeric form. The longer hinge region might play a role in the enhanced binding of IgA1, which may confer increased flexibility and favor a better attachment angle with S_ECD or even allow two Fabs to contact two protomers simultaneously in the same S_ECD trimer. The dIgA1 might crosslink two different S_ECD trimers for the benefit of the increased flexibility in the hinge region as well as the dimeric format, which results in the significantly enhanced binding capacity. Dimeric and secretory ZW2G10 IgA formats conferred stronger avidity and neutralizing activity compared with mIgA or mIgG. The relationship between potency and avidity has been investigated previously,²⁴ with the study indicating that IgA1 could simultaneously engage neighboring S antigens, resulting in intermolecular avidity effects and correspondingly slower dissociation rates. Indeed, compared to IgG and mIgA1, dIgA1 and sIgA1 displayed similar association rates but significantly slower dissociation rates for BA.2.75 and BA.4/5 S_ECD. This trend did not extend to BA.2.76 S_ECD, where different ZW2G10 formats showed varied association rates but similar dissociation rates, since the antibody molecules binding to the sensor were limited.

IgA1 and IgA2 have different structures and distributions. The primary structural distinction between IgA1 and IgA2 lies in the hinge region, which is 13 amino acids longer in IgA1, resulting in a T-shaped structure for IgA1 and a rigid Y-shaped structure for IgA2, similar to IgG.³¹ This structural feature confers IgA1 with greater flexibility, potentially enabling it to simultaneously engage two spike proteins, a mechanism that may contribute to the superior potency of ZW2G10 sIgA1 against BA.2.76 compared to sIgA2. The distribution of sIgA subclasses varies across mucosal sites, with 80%–90% of sIgA in nasal secretions being sIgA1.³² Nasal delivery of ZW2G10 sIgA1 may confer prophylactic protection against the BA.2.76 pseudovirus, though its efficacy against authentic viruses and emerging variants of concern requires further investigation. Both prophylactic and therapeutic efficacy of nasally delivered sIgA1 have been demonstrated.²³ Furthermore, nasal delivery of aerosolized sIgA1 to the upper and lower respiratory tract via a mesh nebulizer could achieve high local doses with minimal systemic exposure.³³ This study highlights the potential of sIgA-based passive immunotherapy or prophylactics as effective tools for controlling future SARS-CoV-2 epidemics and other mucosa-susceptible pathogens.

Materials and methods

Construction of IgA-related expression vectors

We amplified the HC variable-region genes (Chinese Patent Announcement CN114031685B) of the antibody from the previously established expression vector using specific primers. The amplified variable-region gene and primer sequences are detailed in Table S2. Subsequently, the IgA1/IgA2/IgG3 constant region, J chain, and SC sequence were synthesized by General Biology (Anhui, China) and cloned individually into the pcDNA vector. Amino acid sequences are provided in Table S2. Hpa I (GTTˆACC) and EcoR V (GATˆATC) restriction sites were introduced flanking the 5′ end of the IgA constant region, separated by a random spacer sequence. Following double digestion, the vector was homology recombined with the variable-region sequence to yield the complete IgA HC expression vector.

Production of IgG and IgA antibodies

Previously, we amplified the HC and LC variable-region genes of ZW2G10 from single cells and cloned them into an IgG expression vector. Subsequently, HC isotype switching was performed, and the variable region was amplified using specifically designed primers, followed by ligation into an expression vector containing the IgA HC constant region. After confirming vector sequence integrity, transient transfection was performed in Expi293F cells. For IgG and mIgA, HC and LC vectors were co-transfected at a 1:1 ratio; for dIgA, HC, LC, and J-chain vectors at 1:1:1; and for sIgA, HC, LC, J chain, and SC vectors at a 4:4:1:1 M ratio. After 5 days, supernatants were collected, and IgG1/IgG3 and IgA were purified via affinity chromatography using protein A/G (Cytiva) and LambdaFabSelect (Cytiva) prepacked columns, respectively. The purified antibodies were buffer exchanged into PBS, and then IgG3, mIgA, dIgA, and sIgA were separated using a HiLoad 26/600 Superdex 200-pg size-exclusion column (Cytiva). Target fractions were pooled, concentrated, and analyzed by SDS-PAGE and HPLC. For HPLC analysis, antibodies were loaded onto a TSKgel G5000PW_XL (7.8 mm × 30 cm, TOSOH) equilibrated with PBS (pH 7.4) using the HPLC system (Waters). The flow rate was set to 0.5 mL/min, and eluted proteins were monitored at 280 nm.

Western blot analysis

Under reducing conditions, the antibodies were loaded onto a 4%–12% polyacrylamide gel and transferred to a nitrocellulose membrane. The membrane was blocked with 5% bovine serum albumin (BSA) in PBST for 1 h at room temperature, using IgG as the negative control. After washing, the membrane was incubated for 1 h with either an anti-IgJ antibody (Abcam) diluted 1:1,000 in PBST to detect the J chain or an anti-IgA secret component antibody (Abcam) diluted 1:1,000 in PBST to detect the SC. After washing, the membrane was incubated for 1 h with a goat anti-mouse IgG H&L antibody (horseradish peroxidase [HRP]) (Abcam) or goat anti-rabbit IgG H&L antibody (HRP) (Abcam) diluted 1:10,000 in PBST. After washing, the membrane was incubated with Western HRP Substrate (Millipore) at room temperature, and immunoreactivity was detected using a chemiluminescence imaging system.

In vitro stability testing

Thermal stability testing of ZW2G10-IgA variants was analyzed using the UNcle device (UNchained Labs). Light scattering was determined by the barycentric mean of intrinsic tryptophan fluorescence emission spectra collected from 25°C to 95°C at a rate of 0.5°C/min, with all samples at a concentration of 0.5 mg/mL in PBS. Data were analyzed using the UNcle software, with the Tm defined as the peak of the first derivative of the melting curve.

Optimization of sIgA expression

To determine the optimal vector ratio, four sIgA-expressing vectors were transfected into HEK293T cells at varying ratios, and supernatants were collected 48 h post transfection. To determine the optimal post-transfection collection time, vectors were transfected into Expi293F cells using the predetermined optimal molar ratio of 4:4:1:1, and supernatants were harvested at designated intervals. Supernatants were diluted with PBST and used as test samples. Polystyrene microplates were coated overnight with 2 μg/mL anti-IgA secret component antibody (Abcam) in carbonate buffer (pH 9.6) at 4°C. After washing with PBST, plates were blocked with 2% BSA in PBST for 1 h at 37°C. After washing with PBST, a series of diluted purified sIgA and the samples to be tested were added to 96-well plates, incubated at 37°C for 1 h, and then washed with PBST. Anti-lambda LC antibody (Abcam) was added at a 1:5,000 dilution and incubated at 37°C for 1 h. After washing, TMB single-component substrate solution (Solarbio) was added and incubated at 37°C for 3 min. The reaction was terminated by adding ELISA stop solution (Solarbio). The absorbance was measured at 450 nm (reference, 630 nm). A standard curve generated from purified sIgA concentration and absorbance values was used to quantify sIgA levels in cell culture supernatants.

ELISA

The isolated antibody was assessed by ELISA for its binding capacity to the S_ECD protein. Polystyrene microplates were coated overnight with 2 μg/mL S_ECD protein in carbonate buffer (pH 9.6) at 4°C. All S_ECD proteins were purchased from Acro Biosystem. After rinsing with PBST, plates were blocked with 2% BSA in PBST for 1 h at 37°C. Serially diluted antibodies were added to the 96-well plates and incubated at 37°C for 1 h. After washing, an anti-lambda LC antibody (Abcam) was introduced at a 1:5,000 dilution and incubated at 37°C for 1 h. Post rinse, TMB single-component substrate solution was added and incubated at 37°C for 3 min, followed by the addition of ELISA Stop Solution (Solarbio) to halt the reaction. The absorbance was measured at 450 nm (reference: 630 nm). Data were analyzed by four-parameter nonlinear regression using GraphPad Prism 8.0 software.

Pseudovirus neutralization assay

The ACE2-293T cells (generated and maintained in our laboratory) used in this experiment were HEK293T cells stably overexpressing the ACE2 receptor. All pseudoviruses were purchased from Vazyme. For the neutralization assay, 50 μL pseudovirus (diluted to 5 × 10³ TCID₅₀) was incubated with serially diluted antibodies or SC protein for 1 h at 37°C in a 96-well plate, followed by addition of 2 × 10⁴ ACE2-293T cells per well. Blank controls (wells without pseudovirus or antibodies) and virus controls (pseudovirus without antibodies) were included. After 48 h of culture, cells were lysed and analyzed using Bright-Lite detection reagent (Vazyme). Luciferase activity was quantified with a microplate luminescence detector (Tecan, Spark 10M). The formula for calculating the percentage of neutralization is 100% − (sample signals − blank control signals) / (virus control signals − blank control signals) × 100%. Statistical analysis and IC₅₀ determination were performed in GraphPad Prism 8.0 using three-parameter logistical regression.

Plaque reduction neutralization test

To evaluate the neutralizing activities of ZW2G10 IgG and sIgA1 against authentic SARS-CoV-2 WT, Omicron BA.1, BA.4/5 variants, a plaque reduction neutralization test was conducted as described in a previous study.³⁴ Briefly, Vero E6 cells were cultured overnight in 24-well plates at 1.5 × 10⁵ cells per well. Antibodies serially diluted 1:4 in DMEM containing 2.5% fetal bovine serum (FBS) were incubated with equal volumes of 120 plaque-forming units of authentic virus diluted in DMEM containing 2.5% FBS at 37°C for 1 h. The antibody-virus mixtures were then added to the wells. Cells infected with virus without antibody addition were set as controls. After 1 h incubation at 37°C under 5% CO₂, the cell supernatant was discarded, and DMEM containing 2.5% FBS and 1% methylcellulose was added. After five days of culture, the plates were fixed with 8% paraformaldehyde, stained with 0.5% crystal violet, and rinsed thoroughly with water. Plaques were then enumerated. All operations were performed in a biosafety level 3 facility. The inhibition percentage was calculated by the following formula: neutralization (%) = (1 − sample plaque / blank plaque) (%). The IC₅₀ values of the antibodies were calculated using GraphPad Prism software.

BLI assay

The S_ECD proteins of SARS-CoV-2 were biotinylated using EZ-Link Sulfo-NHS-LC-LC-Biotin (Thermo Fisher Scientific). The biotinylated S_ECD proteins were diluted in PBS containing 0.2% Tween 20 to a concentration of 20 μg/mL and then immobilized onto streptavidin biosensors (Sartorius) for 120 or 180 s at 400 rpm. After washing with buffer for 60 s, the biosensors were immersed in wells containing the antibodies diluted in the same buffer for 300 s at 1,000 rpm, followed by incubation with the buffer for 600 s. K_D values were generated from the ratio of off-rate values to on-rate values. Data were analyzed using Data Analysis Software 12.2 (Sartorius).

Infection and prevention protection of pseudovirus in vivo

Animal studies were approved by the Institutional Animal Care and Use Committee of the Laboratory Animal Center of the Academy of Military Medical Sciences (approval IACUC-DWZX-2023-001) in compliance with the Laboratory Animal Guidelines for Ethical Review of Animal Welfare (GB/T35892-2018). A total of 16 5-week-old female hACE2 transgenic mice (Cyagen) were divided into four groups. Four mice were anesthetized with isoflurane and then administered 40 μL ZW2G10-IgG dissolved in PBS via the intranasal route. One hour later, the mice were anesthetized with isoflurane and then administered 40 μL pseudovirus (relative luminescence units = 10⁸) via the intranasal route. The mice were placed in a supine position for intranasal administration of antibodies and pseudoviruses, using a fine pipette tip (50 μL total) to drip the liquid into the nose. Four mice administered PBS solution intranasally 1 h before infection served as a virus control group, while four mice were intranasally administered PBS only as blank control group. In vivo bioluminescence imaging was performed using IVIS Spectrum (PerkinElmer) on the first, fourth, sixth, and seventh days after the pseudovirus challenge. In the sIgA1 protection assay, each group contained three mice, and the mice administered ZW2G10-sIgA1 were subjected to bioluminescence imaging on the sixth day after the challenge.

For in vivo bioluminescence imaging, each mouse was treated with 40 μL of D-luciferin (PerkinElmer) via intranasal administration at a dose of 40 mg/kg. Seven minutes after D-luciferin administration, mice were imaged on the IVIS Spectrum for 5 min. Regions of interest (ROIs) for the mice were drawn, and radiant efficiency (p/s/cm²/sr), which represents the total photon counts within the ROI, was measured. All images were processed using Living Image software (PerkinElmer), and the same fluorescence threshold was applied for group comparison. The formula for calculating the percentage of neutralization was 100% − (sample signals − blank control signals) / (virus control signals − blank control signals) × 100%.

Gene cloning, expression, and protein purification

The DNA encoding the Fab fragment of ZW2G10 was cloned into the pcDNA3.1 vector (Novagen) and transfected into the Expi293F mammalian cells for protein expression. Subsequently, the protein was further purified by nickel column and gel filtration on a HiLoad 16/600 Superdex 75 pg column (Cytiva). Target peak fractions containing the protein were concentrated for crystallization. The Omicron BA.1 RBD was expressed using the Bac-to-Bac baculovirus expression system (Invitrogen). The coding sequence for the Omicron RBD was codon optimized for expression in cell lines derived from Spodoptera frugiperda and cloned into the pFastbac1 Plasmid (Invitrogen) under control of the polyhedrin promoter, which is designed to fuse an N-terminal gp67 signal peptide and a C-terminal His₆ to the target protein. The recombinant pFastbac1 plasmid was used to transform DH10Bac E. coli (Invitrogen). Transfection and virus amplification were conducted with ExpiSF9 cells (Thermo Fisher Scientific), and the recombinant proteins were expressed in High Five cells (Invitrogen). Soluble proteins were recovered from cell supernatants by metal affinity chromatography using a HisTrap HP 5 mL column (GE Healthcare) and then purified by gel filtration on a HiLoad 16/600 Superdex 75 pg column (Cytiva). For crystallization, the Omicron RBD and ZW2G10 Fab complex was prepared by mixing at a 1:1 M ratio at 4°C overnight and further purified by gel filtration chromatography using a HiLoad 16/600 Superdex 200 pg column (Cytiva) equilibrated with 20 mM Tris-HCl and 150 mM NaCl (pH 7.5).

Crystallization, data collection, and structure determination

Crystallization trials were set up with commercial crystallization kits using the sitting drop vapor diffusion method. Normally, 1 μL of protein at the corresponding concentrations was mixed with 1 μL reservoir solution. The resultant drop was then sealed and equilibrated against 100 μL reservoir solution at 4°C or 16°C. Diffraction-quality crystals of the Omicron RBD and ZW2G10 Fab complex were obtained in 0.2 M ammonium acetate, 0.1 M sodium acetate (pH 4.6), and 15% (w/v) polyethylene glycol 4000 at 16°C with a protein concentration of 5.8 mg/mL. The complex crystals were further optimized, and higher levels of diffraction were finally obtained under the condition of 0.2 M ammonium acetate, 0.1 M 2-mopholinoethenesulphonic acid (pH 6.5), and 15% (w/v) polyethylene glycol 3350. Crystals were flash cooled in liquid nitrogen after a brief soak in reservoir solution with the addition of 15% (v/v) glycerol. X-ray diffraction data were collected under cryogenic conditions (100 K) at Shanghai Synchrotron Radiation Facility (SSRF) BL18U1. The data were indexed, integrated, and scaled using HKL3000.³⁵ The structure of the Omicron RBD and ZW2G10 Fab complex was determined by the molecular replacement method using Phaser.³⁶ The atomic models were completed with COOT³⁷ and refined with Phenix,³⁸ and the stereochemical qualities of the final models were assessed with MolProbity. Data collection, processing, and refinement statistics are summarized in Table S1. All structural figures were generated using PyMOL software.

Data availability

Data will be made available upon request. The coordinates and structural characteristics of the Omicron RBD (B.1.1.529) and ZW2G10 Fab complex have been deposited in the Protein DataBank (https://www.rcsb.org/; PDB: 8ZYF).

Acknowledgments

We thank the staff from the BL18U1 beamline at the Shanghai Synchrotron Radiation Facility (SSRF) for assistance during data collection.

Author contributions

Conceptualization, G.Z., X.C., and C.Y.; methodology, G.Z., P.H., H.Y., E.L., and X.C.; validation, H.S., J.H., T.F., Y.D., Jie Li, and Y.W.; investigation, G.Z., P.H., and X.C.; resources, X.C.; writing – original draft, G.Z., P.H., and H.Y.; writing – review & editing, Jianmin Li, C.Y., and S.C.; visualization, G.Z., P.H., H.Y., E.L., and X.C.; supervision, Y.C., S.C., and Jianmin Li; project administration, Y.C., S.C., and Jianmin Li.

Declaration of interests

The authors declare no competing interests.