Multimerization of secretory IgA enhances antiviral activity by aggregating influenza A virus particles

Abstract

Secretory immunoglobulin A (sIgA), the first line of defense against mucosal epithelial infections, is a multimer. Although multimeric sIgA shows higher antiviral activity than monomeric IgA, the mechanism underlying this enhancement and the functional differences among IgA subclasses remain unclear. Here, we generated recombinant monomeric and multimeric IgA1 and IgA2m2 antibodies from clone F045-092, which targets the head domain of influenza A virus (IAV) hemagglutinin (HA) and inhibits viral entry, and compared their functional properties. Monomeric IgA1 inhibited viral entry but not viral release. Multimerization of IgA1 enhanced viral entry inhibition and conferred the additional function of inhibiting viral release. In contrast, monomeric IgA2m2 inhibited both viral entry and release and its multimerization enhanced these inhibitory effects. Importantly, treatment of IAV particles with multimeric sIgA or monomeric IgA2m2 induced viral aggregation via IgA-mediated tethering of neighboring virus particles, as visualized via cryo-electron tomography. Analysis of F(ab′)₂ and Fab fragments from F045-092 IgA2m2 demonstrated that bivalency is essential for antibody-mediated viral aggregation (AVA) by monomeric IgA2m2, further supporting a direct tethering mechanism among virus particles. In conclusion, multimerization of the HA head-targeting IgA tends to enhance the inhibition of viral entry and release through AVA, contributing to its enhanced antiviral activity. Additionally, different IgA subclasses with identical variable regions can exhibit distinct properties in the monomeric form, and subclass switching may confer additional functions. Our findings contribute to the development of virus-aggregating antibodies as a new therapeutic approach that efficiently inhibits viral entry and release at the mucosal epithelium.

Multimeric secretory IgA efficiently neutralizes influenza A virus by cross-linking viral particles, and distinct inhibitory effects on viral release are observed among different IgA subclasses.

Introduction

Influenza A virus (IAV), a member of the Orthomyxoviridae family, causes influenza, an acute respiratory illness. The enveloped viral particle contains two membrane-spanning glycoproteins, hemagglutinin (HA) and neuraminidase (NA), which form spike structures on its surface. HA, which consists of the head and stalk domains, is responsible for viral entry by binding to sialic acid receptors via the head domain and inducing membrane fusion between the virus particle envelope and endosomal membrane via the stalk domain¹. NA possesses neuraminidase activity in its head domain and plays a critical role in viral release from infected cells by cleaving the sialic acid linkage between budding viral particles and the host cell surface, preventing viral aggregation at the cell surface². HA serves as the primary antigen of the IAV, and anti-HA antibodies are predominantly produced following vaccination and natural infection with IAV.

IgA plays a crucial role in preventing IAV infection in the human respiratory mucosal epithelium^3–5. Human IgA is classified into two subclasses: IgA1 and IgA2. These subclasses have major differences in the amino acid sequences of the heavy chain constant region, particularly in the lengths of the hinge regions⁶ (Supplementary Fig. 1). IgA1 has a longer and more flexible hinge, which allows its Fab arms to adopt a wide range of orientations, including both T-shaped and Y-shaped configurations. In contrast, IgA2 has a shorter hinge enriched in proline residues, which restricts Fab arm flexibility and favors a more compact T-shaped conformation⁷. These structural differences may lead to variations in their properties and functions, such as the efficiency of epitope recognition and the ability to bridge distinct epitopes. IgA2 is further divided into three allotypes based on genetic polymorphisms: IgA2m1, IgA2m2, and IgA2mn⁸. The distribution of IgA subtypes varies among tissues; IgA1 comprises nearly 90% of the IgA in the respiratory tract, whereas IgA2 accounts for 60% in the colon^9–11. While IgA exists as a monomer in the serum, secretory IgA (sIgA) secreted at the mucosal surface is predominantly multimeric, such as dimers, trimers, and tetramers^12,13. sIgA multimers consist of two or more IgA monomers linked by a 15 kDa polypeptide J chain (JC) through disulfide bridges, along with a secretory component (SC), an approximately 80 kDa proteolytic fragment derived from the polymeric immunoglobulin receptor expressed on the basolateral surface of the mucosal epithelium^6,14–16. Recent structural analysis revealed that the JC serves as a structural scaffold for IgA multimerization by interacting with the Fc tailpieces, and the SC stabilizes the multimeric assembly through covalent linkage^15,16. Dimeric, trimeric, and tetrameric sIgAs against IAV are produced in the human upper respiratory mucosa via intranasal vaccination, where multimerized sIgA exhibits higher neutralizing potency against IAV than monomeric IgA¹². Analyses of monoclonal recombinant multimeric sIgA1 and sIgA2m2 have demonstrated a greater neutralizing potency against IAV than their monomeric IgA counterparts^17,18. Enhanced antiviral activity of multimeric sIgA has been observed against the human immunodeficiency virus (HIV) and SARS-CoV-2^19–25, suggesting common characteristics of sIgA. However, these studies did not examine the differences in IgA subclasses, and it remains unclear how multimeric sIgA enhances antiviral activity and whether variations in IgA subclasses affect antiviral properties.

To understand the mechanisms underlying the enhanced anti-IAV activity of multimeric sIgA and to elucidate the differences in anti-IAV activity between IgA subclasses, in this study, we generated recombinant monomeric IgA1 and IgA2, as well as multimeric sIgA1 and sIgA2, all with identical variable regions in the heavy and light chains (i.e., the identical antigen recognition site), which were derived from an anti-HA-neutralizing antibody, clone F045-092, which targets the receptor-binding site (RBS) of the HA head domain^26,27 (Supplementary Fig. 2). We then compared the anti-IAV activity of the respective forms of recombinant IgA, focusing on the specific steps of viral replication that were inhibited. These findings provide critical insights into how the molecular form of HA head-targeting IgA impacts its antiviral function and may inform the design of IgA-based strategies for combating IAV infection.

Results

Preparation of recombinant IgA1 and IgA2m2 from clone F045-092

A previous study demonstrated that the multimerization of clone F045-092 IgA enhanced its ability to inhibit viral replication compared to the monomeric form¹⁷. To further elucidate the impact of IgA multimerization on anti-IAV activity, we produced recombinant monomeric IgA1 and IgA2m2 and multimeric sIgA1 and sIgA2m2 from clone F045-092¹⁷. We analyzed the IgA2m2 allotype as a representative of the IgA2 subclass because of its superior multimerization efficiency and well-characterized structural features associated with the JC and secretory components^16,28. To produce recombinant F045-092 IgA1, Expi293F cells were co-transfected with the expression plasmids for the α1 heavy (A1) chain of clone F045-092, the light chain (Lambda) of clone F045-092, JC, and SC. Recombinant F045-092 IgA2m2 was produced by expressing the α2 heavy (A2m2) chain of F045-092 instead of the α1 heavy chain. The supernatants from the transfected cells were harvested, and then sIgA1 and sIgA2m2 of F045-092 were purified from the culture supernatants via affinity chromatography, followed by separation using size exclusion chromatography (SEC). At an absorbance of 280 nm, three peaks were observed for F045-092 IgA1 (Fig. 1A, top panel), whereas two peaks were observed for F045-092 IgA2m2 (Fig. 1B, top panel). Blue native polyacrylamide gel electrophoresis (BN-PAGE) of the SEC fractions indicated that fractions 7 and 8 of the F045-092 IgA1 sample primarily contained tetrameric sIgA1, fractions 11 and 12 predominantly contained dimeric sIgA1 with a minor amount of tetrameric sIgA1, and fraction 16 contained monomeric IgA1 (Fig. 1A, middle panel). Consistently, negative-staining electron microscopy (EM) and high-speed atomic force microscopy (HS-AFM) revealed the presence of large molecules, approximately 30 nm in size, in fraction 7. In contrast, smaller molecules, approximately 15 nm in size, were observed in fraction 16 (Fig. 1A, bottom panels). In the F045-092 IgA2m2 sample, fractions 5–8 predominantly contained tetrameric sIgA2m2, while fractions 12–13 contained monomeric IgA2m2 (Fig. 1B, middle panel). Although we were unable to identify the additional higher-molecular weight band, the double-band at approximately 720 kDa observed in fractions 5–8 may represent tetrameric and pentameric IgA2m2 as suggested by structural analyses of the IgA2m2 Fc–JC–SC complex¹⁶ Negative-staining EM and HS-AFM confirmed the presence of larger molecules in fraction 6 and smaller molecules in fraction 12 (Fig. 1B, bottom panels), similar to the IgA1 samples. In addition, sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) under reducing condition showed that both the JC and SC were detected in the multimeric sIgA fractions but not in the monomeric IgA fraction (Fig. 1C). Thus, we successfully produced recombinant F045-092 IgA1 and IgA2m2 in both multimeric and monomeric forms, although dimeric IgA1 devoid of tetrameric IgA1 could not be prepared.

Fig. 1. Generation of different forms of recombinant IgA derived from clone F045-092.

A Top: size exclusion chromatography (SEC) profiles of recombinant IgA1 from the clone F045-092. The collected sample fractions were indicated by the gray and white columns. Middle: Blue native polyachrylamide gel electrophoresis (BN-PAGE) analysis of the recombinant IgA1 fractions from SEC. The fraction numbers correspond to those in the SEC profile. M indicates size marker. Bottom: Negative-staining electron microscopy (EM) and high-speed atomic force microscopy (HS-AFM) images of recombinant IgA1. The fraction numbers correspond to those in the SEC profile. The arrowheads indicate each Fab. Scale bars: 20 nm. B Top: SEC profiles of recombinant IgA2m2 from the clone F045-092. Collected sample fractions were indicated by gray and white columns. Middle: BN-PAGE of the recombinant IgA2m2 fractions obtained from SEC. The fraction numbers correspond to those in the SEC profile. M indicates size marker. Bottom: Negative-staining EM and HS-AFM images of recombinant IgA2m2. The fraction numbers correspond to those in the SEC profile. The arrowheads indicate each Fab. Scale bars: 20 nm. C Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) of IgG and SEC-fractionated F045-092 IgAs. Fraction numbers of IgA correspond to those in (A), (B). M indicates size marker.

To assess the anti-IAV activity of various forms of recombinant F045-092 IgA, we performed a neutralization assay for the IAV (A/Victoria/361/2011 [H3N2] strain) and determined the minimum neutralization concentrations (MNCs). Both multimeric sIgA1 and sIgA2m2 exhibited higher neutralizing activity than their respective monomeric IgAs and IgG (Supplementary Fig. 3A), consistent with a previous study¹⁷. Based on the assumed molecular weights of 160 kDa, 400 kDa, and 700 kDa for monomeric, dimeric, and tetrameric IgA, respectively, the mean MNC values were 2460 pM for monomeric IgA1110 pM for dimeric/tetrameric sIgA1, 63 pM for tetrameric sIgA11096 pM for monomeric sIgA2m2, and 70 pM for tetrameric sIgA2m2, corresponding to approximately 16-fold to 39-fold lower effective concentrations in molar units (or approximately 4-fold to 9-fold lower in weight-based units) compared with monomeric IgA. These results indicate that multimerization of F045-092 IgA enhances its anti-IAV activity during influenza virus replication, independent of the IgA subclass.

Inhibitory activity of recombinant F045-092 sIgA1 and sIgA2m2 on IAV replication

To determine the replication step at which multimeric sIgA (dimeric/tetrameric sIgA1, tetrameric sIgA1, and tetrameric sIgA2m2) exerts enhanced inhibitory activity, we first assessed its impact on viral entry using immunofluorescence assays. The respective IgAs were serially diluted four-fold, mixed with 100 plaque-forming units (PFU) of IAV, and used to infect Madin–Darby canine kidney (MDCK) cells. At 10 h post-infection (hpi), the number of infected cells was counted by immunostaining with an anti-IAV nucleoprotein (NP) antibody (Fig. 2A, Supplementary Fig. 3B, and Table 1). The IC₅₀ values of tetrameric and dimeric/tetrameric sIgA1 were approximately 35-fold and 28-fold lower, respectively, than that of the monomeric form in molar units (corresponding to 8-fold and 11-fold lower in weight-based units), and the IC₅₀ value of tetrameric sIgA2m2 was approximately 149-fold lower in molar units (34-fold lower in wight-based units) than that of the monomeric form. Notably, when compared at equivalent weight concentrations (i.e., corresponding to the same total number of Fab arms), tetrameric sIgA displayed substantially stronger inhibitory activity than monomeric IgA, indicating functional advantages of multimerization beyond simple Fab valency. These results indicate that multimerization enhanced the inhibitory activity of IgA against viral entry, independent of the IgA subclass.

Fig. 2. Antiviral activities of different forms of F045-092 recombinant IgA and IgG.

A Virus entry inhibition assay. Herein, 100 plaque-forming units (PFU) of influenza A virus (IAV) were incubated with serially diluted recombinant IgA, followed by infection of Madin–Darby canine kidney (MDCK) cells. At 10 hours post-infection (hpi), virus-infected MDCK cells were fixed, and virus entry was assessed by counting IAV NP-positive cells using immunofluorescence assays. Relative infectivity was normalized to the control without antibodies, which was set at 100%. Data were indicated as the mean ± SD of six technical replicates. Clone36 IgG, an anti–SARS-CoV-2 spike (S) protein antibody isolated in our laboratory, was used as a negative control. The data for the negative control were obtained from three technical replicates. B Virus release inhibition assay. MDCK cells were infected with IAV at a multiplicity of infection (MOI) of 2, and treated with respective forms of IgA from 3 hpi. Both cells and supernatants were collected at 12 hpi. The samples were subjected to western blotting using an anti-M1 antibody. Anti-α-tubulin antibody was used as an internal control. The results shown are representative data from three independent experiments. C Ultrathin section electron microscopy images of virus-infected cells treated with IgA. MDCK cells were infected with IAV at an MOI of 2, treated with respective forms of IgA from 3 hpi, and the cell surface was examined at 12 hpi using transmission electron microscopy. Scale bars: 1 μm. D Scanning electron microscopy (SEM) images of virus-infected cells treated with IgA. MDCK cells were infected with IAV at an MOI of 2, treated with different forms of IgA from 3 hpi, and the cell surface was examined at 12 hpi using SEM. Scale bars: 1 μm. E Neuraminidase inhibition assay. Relative neuraminidase activity at the indicated concentration of each IgA form or IgG was calculated from the optical density at 450 nm (OD₄₅₀), which was normalized using the OD₄₅₀ in the absence of IgA. Zanamivir was used as a positive control. Clone36 IgG (an anti-SARS-CoV-2 S protein antibody isolated in our laboratory) was used as a negative control. Data were indicated as the mean ± SD of four or five technical replicates.

Table. 1.

Inhibitory activity of each form of F045-092 IgA on influenza A virus entry

Formation and subclass	IC50 (ng/ml)	IC50 (pM)
IgG	1324 ± 508	8827 ± 3387
Monomeric IgA1	168 ± 53	1051 ± 332
Dimeric/tetrameric sIgA1	15 ± 4*	37 ± 10**
Tetrameric sIgA1	22 ± 7*	31 ± 10**
Monomeric IgA2m2	515 ± 153	3216 ± 954
Tetrameric sIgA2m2	15 ± 5**	21 ± 7**

The number of infected cells was normalized to the number of cells in the absence of antibodies. The half-maximal inhibitory concentration (IC₅₀) of the antibody with standard error of the mean was calculated using nonlinear regression analysis. Molar concentrations were calculated using the following estimated molecular weights: IgG, 150 kDa; monomeric IgA, 160 kDa; dimeric sIgA, 400 kDa; and tetrameric sIgA, 700 kDa. Data were obtained from three (for control) or six replicates. The Mann–Whitney U test was used to determine the P values of each subclass of monomers with dimer or tetramers. All analyses were performed using GraphPad Prism 9 software. *p < 0.05, **p < 0.01.

Since several monoclonal antibodies in multimeric sIgA form have been reported to inhibit viral release from infected cells^29,30, we assessed whether multimeric F045-092 sIgA also inhibited viral release. At 3 hpi, each form of IgA was added to IAV-infected cell cultures, and the supernatants and cell lysates were harvested at 12 hpi. The samples were then subjected to western blotting (WB) using an anti-IAV M1 antibody to compare the amounts of viral particles released into the supernatants of each sample. The intensities of the M1 protein in the cell lysates were comparable among all virus-infected samples (Fig. 2B), suggesting that the addition of any form of IgA at 3 hpi had little effect on viral entry and protein expression. In contrast, in the supernatants, treatment with dimeric/tetrameric and tetrameric sIgA1 decreased the intensity of M1 in a concentration-dependent manner, whereas treatment with monomeric IgA1 had little effect, even at the highest concentration (Fig. 2B, upper left panel), suggesting that multimeric sIgA1 efficiently inhibited viral release. Tetrameric sIgA2m2 also efficiently inhibited viral release, and it appeared to do so at a lower concentration than tetrameric sIgA1 (Fig. 2B, upper right panel), indicating that viral release inhibition by multimeric sIgA is likely independent of IgA subclass. Intriguingly, unlike monomeric IgA1, monomeric IgA2m2 inhibited viral release to a similar extent as multimeric sIgA1 (Fig. 2B).

To further confirm whether tetrameric sIgA and monomeric IgA2m2 inhibit viral release, we examined IAV-infected cells treated with the respective forms of IgA after viral entry using ultrathin section transmission electron microscopy (TEM). Similar to untreated virus-infected cells, sparsely distributed budding virus particles were observed on the surface of virus-infected cells treated with monomeric IgA1 (Fig. 2C, second panel from the left). In contrast, a large number of budding viral particles accumulated on the surface of cells treated with tetrameric sIgA and monomeric IgA2m2 (Fig. 2C, right 3 panels). Consistently, scanning electron microscopy (SEM) revealed a large number of viral particles on the cell surface after treatment with tetrameric sIgA and monomeric IgA2m2 (Fig. 2D, right 3 panels). Collectively, these results demonstrate that, although IgA2m2 appears to have superior potency, multimeric sIgA inhibits viral release from infected cells independent of IgA subclass. Notably, monomeric IgA2m2 also possesses this ability, although its virus entry-inhibitory activity is relatively low (Fig. 2A).

We then sought how multimeric sIgA and monomeric IgA2m2 inhibit viral release. Since it is well known that virus release is hindered by NA inhibitors such as zanamivir³¹, we hypothesized that NA activity might be inhibited by steric hindrance from the binding of multimeric sIgA and monomeric IgA2m2 to the HA head region on virus particles and cell surfaces. Accordingly, an enzyme-linked lectin assay (ELLA) was performed to assess the impact of IgA treatment on NA activity³². Mixtures of IAV particles and serially diluted IgA were incubated on fetuin-coated plates for 16–18 h, followed by treatment with horseradish peroxidase (HRP)-conjugated peanut agglutinin. NA activity was then assessed by measuring the optical density at 450 nm (OD₄₅₀) after the substrate addition. We found that all tested forms of IgA1 and IgG rarely exhibited neuraminidase inhibition (NI) activity (Fig. 2E and Supplementary Fig. 3C), indicating that multimeric sIgA1 inhibits virus release independently of NA activity. In contrast, both forms of IgA2m2 exhibited NI activity in a concentration-dependent manner, with the tetrameric sIgA2m2 exhibiting the highest activity (Fig. 2E and Table 2). Collectively, these results suggest that, in the case of IgA1, factors other than NI activity likely play a primary role in viral release inhibition. NI activity may also contribute to the inhibition of viral release in the case of IgA2m2.

Table. 2.

Neuraminidase inhibition activity of each form of F045-092 IgA

Formation and subslass	EC50 (ng/ml)	EC50 (pM)
IgG	>10,000	>40,000
Monomeric IgA1	>10,000	>40,000
Dimeric/tetrameric sIgA1	>10,000	>40,000
Tetrameric sIgA1	>10,000	>40,000
Monomeric IgA2m2	1129 ± 107	7062 ± 670
Tetrameric sIgA2m2	47 ± 7*	67 ± 10**

The Optical density at 450 nm values in the absence of antibodies were set as 100%. Data were obtained from five replicates, and the half-maximal effective concentration (EC₅₀) values with standard error of the mean were calculated using nonlinear regression analysis. Molar concentrations were calculated using the following estimated molecular weights: IgG, 150 kDa; monomeric IgA, 160 kDa; dimeric sIgA, 400 kDa; and tetrameric sIgA, 700 kDa. The P values between the IgA2m2 monomer and IgA2m2 tetramer were determined using the Mann–Whitney U test. All analyses were performed using GraphPad Prism 9 software. *p < 0.05, **p < 0.01.

Multimeric sIgA and monomeric IgA2m2 aggregate purified IAV particles

Based on these results, we hypothesized that multimeric sIgA aggregates virus particles, leading to the inhibition of virus release from the cell surface. Accordingly, to examine the size of potential viral aggregates induced by sIgA, purified IAV particles were incubated with each form of IgA for 1 h and subjected to sucrose density gradient ultracentrifugation. After fractionation, the samples were analyzed via WB using anti-M1 and anti-IgA antibodies to detect the viral particles. Similar to the IgA-untreated sample, M1 was predominantly detected in fractions 4–6, with a peak in fraction 5, when the IAV particles were treated with monomeric IgA1 (Fig. 3A, second panel from the top). In contrast, M1 was primarily detected in fractions 7 and 8 after treatment with multimeric sIgA and monomeric IgA2m2 (Fig. 3A, lower four panels), suggesting the formation of aggregates comprising IAV particles. WB with an anti-IgA antibody revealed an IgA distribution pattern similar to that of M1 (Supplementary Fig. 4), suggesting that IgA was associated with the IAV particles within aggregates. To further assess whether the IAV particles and IgA were indeed aggregated, we analyzed their morphology in the M1 peak fraction from each sample using negative-staining EM. As expected, large virus aggregates were observed in samples incubated with multimeric sIgA (dimeric/tetrameric sIgA1, tetrameric sIgA1, and tetrameric sIgA2m2) and monomeric IgA2m2, but not in samples incubated with monomeric IgA1 (Fig. 3B). These results indicate that multimeric sIgA of F045-092 exhibits antibody-mediated virus aggregation (AVA) activity independent of IgA subclass, whereas monomeric IgA2m2 of F045-092 exhibits subclass-spesific AVA activity.

Fig. 3. Virus aggregation induced by recombinant F045-092 IgA.

A Sucrose density gradient ultracentrifugation of influenza A virus (IAV) particles treated with respective forms of IgA. Briefly, IAV particles were incubated with each form of IgA and subjected to 20–60% sucrose-gradient ultracentrifugation. After fractionation, the samples were analyzed via western blotting using an anti-M1 antibody. B Negative-staining electron microscopy images of IAV particles treated with different forms of IgA. Following fractionation, each peak fraction was examined via transmission electron microscopy. The fraction numbers correspond to those shown in Fig. 3A. Scale bars: 1 μm. C Dynamic light scattering analysis of IAV-IgA complex size distribution. Purified IAV particles were incubated with each form of IgA or phosphate-buffered saline, and the size distribution was measured using the DLS-8000 series. The X axis represents the distribution of particle diameters, while the Y-axis represents the intensity distribution. The results shown are representative data from three independent experiments.

To measure the sizes of the viral aggregates in solution, purified IAV particles were incubated with the respective forms of IgA and subjected to dynamic light scattering (DLS), in which Brownian motion was measured to estimate the size distribution of particles in solution (Fig. 3C). Untreated IAV particles showed a single peak at approximately 100 nm, consistent with the typical size of IAV particles^33,34. IAV particles treated with monomeric IgA1 showed a peak similar to that of the untreated IAV particles (Fig. 3C, left panel). In contrast, IAV particles treated with multimeric sIgA (dimeric/tetrameric sIgA1, tetrameric sIgA1, and tetrameric sIgA2m2) and monomeric IgA2m2 showed additional peaks at approximately 1000 nm (dimeric/tetrameric sIgA1, tetrameric sIgA1, monomeric IgA2m2, and tetrameric sIgA2m2) or much larger (tetrameric sIgA1, monomeric IgA2m2, and tetrameric sIgA2m2), in addition to a peak at approximately 100 nm (Fig. 3C, both panels). Taken together, these results demonstrate that multimeric sIgA and monomeric IgA2m2 from F045-092 possess AVA activity, which is likely sufficient to retain viral particles on the cell surface and inhibit viral release. In addition, our findings revealed that antiviral activities vary between IgA subclasses despite sharing an identical variable region, highlighting the distinct functional characteristics of each IgA subclass.

Multimeric sIgA and monomeric IgA2m2 tether distinct virus particles

A possible mechanism by which multimeric sIgA (dimeric/tetrameric sIgA1, tetrameric sIgA1, and tetrameric sIgA2m2) and monomeric IgA2m2 of F045-092 exhibit AVA involves the tethering of distinct viral particles by binding to HA proteins on their surfaces. Therefore, we performed cryo-electron tomography (cryo-ET) of the viral aggregates to determine whether IgA was located between virus particles (Fig. 4). Low-magnification images showed that IAV particles tended to aggregate after treatment with multimeric sIgA (dimeric/tetrameric sIgA1, tetrameric sIgA1, and tetrameric sIgA2m2) and monomeric IgA2m2 (Fig. 4, upper panels). Denoised 3D reconstructed slice images clearly showed an array of HA and NA spikes of uniform height with no additional density outside the spikes of viral particles untreated with IgA (Fig. 4, PBS, bottom panel). For viral particles treated with monomeric IgA1, additional densities, which likely corresponded to the partial densities of IgA sliced in the tomogram, were observed outside the spikes (Fig. 4, bottom panels, blue arrowheads). Similarly, additional densities, likely corresponding to IgA, were observed between viral particles within aggregates treated with multimeric sIgA (dimeric/tetrameric sIgA1, tetrameric sIgA1, and tetrameric sIgA2m2) and monomeric IgA2m2 (Fig. 4, bottom panels, red arrowheads), suggesting that multimeric sIgA and monomeric IgA2m2 of F045-092 physically tether neighboring viral particles to induce their aggregation (Supplementary Fig. 5A, B).

Fig. 4. Ultrastructural analysis of the IAV-F045-092 IgA complex using cryo-electron tomography (cryo-ET).

The purified influenza A virus (IAV) particles were incubated with IgA, rapidly frozen, and subjected to cryo-ET. Upper panels show low-magnification images; middle panels show magnified views of the area boxed with the dotted line in the upper panels; bottom panels show further magnified, computationally sliced images of the areas boxed with the dotted line in the middle panels. Blue arrowheads indicate additional structures outside the spikes on the IAV particles, and red arrowheads indicate additional structures located between the IAV particles. White scale bars in low-magnification images, 1 μm; in high-magnification images, 100 nm. Black scale bars in the magnified slice images, 20 nm.

Importance of IgA2m2 bivalency for virus aggregation

One unique functional difference among IgA subclasses is that F045-092 IgA2m2 retains the ability to mediate AVA, even in its monomeric form. Given that the HA on IAV particles can bind to sialic acid in the Fc region of IgA³⁵, we hypothesized that the Fc region of IgA2m2 may contribute to virus aggregation. We also assumed that the bivalency of IgA might be essential for tethering neighboring viral particles. Therefore, we digested monomeric IgA2m2 with pepsin and papain to produce its F(ab′)₂ and Fab, respectively^36,37, followed by purification using SEC and characterization using SDS-PAGE and WB (Supplementary Fig. 6). Fractions containing proteins of approximately 100 kDa under non-reducing condition and 37 kDa and 25 kDa under reducing condition, corresponding to IgA2m2 F(ab′)₂ (Supplementary Fig. 6B, fractions 8), and those containing proteins of 50 kDa under non-reducing conditions and 25 kDa under reducing conditions, corresponding to IgA2m2 Fab fragments (Supplementary Fig. 6D, fraction 9), were used in subsequent experiments.

Following incubation with purified IAV particles, negative-staining EM demonstrated that, in addition to the tetrameric sIgA2m2 and monomeric IgA2m2 from F045-092, the IgA2m2 F(ab′)₂ fragment induced virus aggregation, whereas the IgA2m2 Fab fragment did not (Fig. 5A). Negative-staining immuno-EM using an anti-human IgA α chain antibody conjugated to 4 nm gold beads showed that both IgA2m2 F(ab′)₂ and IgA2m2 Fab bound to IAV particles (Fig. 5B), indicating that the IgA2m2 Fab fragment exhibits minimal AVA activity despite its ability to bind to virus particles. These results demonstrate that the bivalency of IgA2m2, but not its Fc region, is essential for mediating AVA.

Fig. 5. Characterization of different forms of F045-092 IgA2m2.

A Negative-staining electron microscopy (EM) images of purified influenza A virus (IAV) particles incubated with different forms of F045-092 IgA2m2 (tetrameric IgA2m2, monomeric IgA2m2, IgA2m2 F(ab′)₂, and IgA2m2 Fab). Scale bars: 1 μm. B Negative-staining immuno-EM images of IAV-IgA2m2 complexes. IAV particles were incubated with different forms of IgA2m2 (tetrameric IgA2m2, monomeric IgA2m2, IgA2m2 F(ab′)₂, and IgA2m2 Fab), incubated with an anti-α-chain antibody conjugated to 4 nm gold beads, and stained with uranyl acetate. The left panels show low-magnification images; the middle panels show the magnified views of the white dotted box; and the right panels show further magnified views of the white dotted box in the middle panels. Red circles indicate gold particles. Scale bars in low-magnification images: 1 μm; in high-magnification images: 100 nm. C Kinetics of IgA association and dissociation rates as measured using bio-layer interferometry. IAV Hemagglutinin was immobilized onto the sensor, and each form of serially diluted different forms of IgA2m2 (tetrameric IgA2m2, monomeric IgA2m2, IgA2m2 F(ab′)₂, and IgA2m2 Fab) was applied. Association and dissociation rates were measured and used to calculate the binding avidity. Raw data were plotted using purple lines, and fitting curves estimated by Prsim9 were added as black lines.

Given that different forms of IgA2m2 may have different avidity potentials for HA, which could affect their AVA potency (e.g., Fab may fail to aggregate virus particles because of its potentially low affinity for HA), we analyzed their apparent binding kinetics using bio-layer interferometry (BLI). Briefly, the His-tagged HA ectodomain of IAV was immobilized on a Ni-NTA sensor chip, and serially diluted IgA (tetrameric sIgA2m2, monomeric IgA2m2, IgA2m2 F(ab′)₂, and IgA2m2 Fab) were applied to measure the association and dissociation rates (Fig. 5C and Table 3). Consistent with previous reports¹⁷, the apparent affinity of tetrameric sIgA2m2 was the highest among the IgA2m2 forms, consistent with the high avidity potential provided by the presence of eight Fab arms. In addition, apparent affinity of monomeric IgA2m2 was substantially higher than those of IgA2m2 F(ab′)₂ and IgA2m2 Fab (Table 3). These results support the idea that increasing valency enhances the avidity potential for HA and that the Fc region (or intact IgA structure) contributes to exerting higher apparent affinity. However, the difference in apparent affinity between IgA2m2 F(ab′)₂ and IgA2m2 Fab was modest, with IgA2m2 F(ab′)₂ showing a 2.7-fold lower apparent affinity (higher Kd value) than IgA2m2 Fab. These results suggest that bivalency is essential for the AVA activity of monomeric IgA2m2 and that a higher avidity potential for the HA head region further enhances its potency. Finally, we examined the apparent binding kinetics of IgA1 in monomeric and multimeric forms (Table 3). The apparent affinity of monomeric IgA1 was comparable to that of IgG, and its multimerization increased the avidity potential for HA, similar to IgA2m2. The result is consistent with greater inhibition of virus budding (Fig. 2B–D) and AVA activity (Fig. 3) observed for multimeric IgA1. Importantly, although monomeric IgA1 was unable to induce virus aggregation, its apparent affinity was approximately 23-fold higher than that of IgA2m2 F(ab′)₂, confirming that apparent affinity is not a critical determinant of AVA activity.

Table. 3.

Binding kinetics of fragmented IgA2m2 against influenza A virus

Formation and subslass	Kon (1/Ms)	Koff (1/s)	Apparent Kd (nM)
IgG	2.0E5	9.5E-2	481#
Monomeric IgA1	2.3E5	3.3E-2	140#
Dimeric/tetrameric sIgA1	1.2E5	2.1E-4	2#
Tetrameric sIgA1	2.4E5	7.7E-4	3#
IgA2m2 Fab	6.1E3	7.3E-3	1205
IgA2m2 F(ab′)2	4.4E3	1.4E-2	3286#
Monomeric IgA2m2	4.9E4	1.8E-3	38#
Tetrameric IgA2m2	1.5E5	2.5E-3	16#

^#The binding kinetics were measured by bio-layer interferometry using an Octet N1 system. Calculated kinetic parameters were obtained by global fitting using a simple 1:1 binding model. For multimeric sIgA, monomeric IgA, and IgG, this model does not account for the effects of multivalency or avidity; therefore, the kinetic parameters for tetrameric and dimeric/tetrameric forms should be interpreted as apparent values rather than true rate constants. In contrast, Fab fragments have no avidity potential.

Discussion

It has been reported that multimeric sIgA exhibits superior antiviral activity against various viruses such as IAV, SARS-CoV-2, and HIV^{17–25,38,39}. However, the mechanisms underlying the enhanced antiviral activity and the differences in antiviral activities between IgA subclasses remain largely unclear. In this study, we compared the antiviral activities and functional properties of recombinant monomeric and multimeric IgA1 and IgA2m2 generated from clone F045-092, which target the RBS on the head domain of IAV HA. We demonstrated that, in addition to the enhanced inhibition of viral entry, both multimeric sIgA1 and sIgA2m2 inhibited viral release by tethering and aggregating virus particles. In addition, we revealed that monomeric IgA2m2, but not monomeric IgA1, inhibited viral release by inducing viral aggregation, which requires bivalency to tether neighboring viral particles. These results indicate that AVA activity contributes to the enhanced antiviral activity of multimeric sIgA and that different IgA subclasses with identical variable regions potentially exhibit distinct antiviral properties in the monomeric form. Our findings suggest that virus-aggregating antibodies can serve as new therapeutics to block both viral entry and release.

It is well known that neutralizing antibodies targeting the head domain of HA inhibit the entry step of the viral replication cycle by sterically blocking the interaction between HA and the sialic acid receptor⁴⁰. Our recombinant monomeric IgA1 and IgA2m2, derived from the F045-092 neutralizing antibody targeting the head domain of HA, inhibited viral entry; multimerization of IgA enhanced this inhibitory activity (Fig. 2A). This is likely due to the larger size and higher binding avidity potential of multimeric sIgA for HA than monomeric IgA (Table 3), which would bind more stably to HA on IAV particles through multiple arms and efficiently block the interaction between HA and the sialic acid receptor. Similarly, the IgM of F045-092 also exhibits higher neutralization potency and a broader breadth of neutralization compared to IgG⁴¹, which is consistent with our observations that multimerization and antibody architecture critically influence the functional properties. In addition, multimeric sIgA would inhibit viral entry into cells through clathrin-mediated endocytosis, the main route of IAV entry^42,43, by aggregating virus particles prior to cell entry. In our study, multimeric sIgA incubated with IAV formed large virus aggregates exceeding 1000 nm in diameter, independent of IgA subclass (Fig. 3B, C), which surpassed the transport size limitation of clathrin-mediated endosomes (approximately 200 nm)⁴⁴. Thus, multimeric sIgA would enhance antiviral potency by increasing binding avidity potential to viral particles and promoting viral aggregation, leading to the inhibition of viral entry via clathrin-mediated endocytosis (Fig. 2A).

Some multimeric sIgAs that target HA have been reported to inhibit IAV release^29,30; however, the underlying mechanisms remain to be elucidated. In our study, both multimeric sIgA1 and sIgA2m2 of F045-092 inhibited virus release and caused the accumulation of viral particles on the cell surface (Fig. 2C, D). Given that antibodies against HA can sterically hinder the access of NA to sialic acid molecules⁴⁵, thereby exerting NI activity, multimeric sIgA is thought to have a greater potential to exhibit higher NI activity than monomeric IgA owing to its larger size. However, the NI activity did not differ notably between monomeric IgA1 and tetrameric sIgA1 (Fig. 2E), suggesting that NI activity was not the primary factor driving viral aggregation in the case of IgA1. Instead, given that multimeric sIgA1 showed a higher binding avidity potential for HA (Fig. 5C and Table 3), tethering HA proteins of different neighboring virus particles via the multivalent Fab arms of multimeric sIgA1 is likely a primary factor in virus aggregation, as shown in our cryo-ET analysis (Fig. 4). In contrast, because both monomeric and tetrameric IgA2m2 exhibited NI activity (Fig. 2E and Supplementary Fig. 3C), virus aggregation in the case of IgA2m2 can be achieved by the combined effects of NI activity and direct tethering of neighboring virus particles.

Interestingly, we found functional differences between monomeric IgA1 and IgA2m2 of F045-092 despite sharing the same variable regions and targeting an identical epitope on the HA head domain: only monomeric IgA2m2 retained AVA activity (Fig. 3B, C). Given that the Fab fragment, but not F(ab′)₂, of IgA2m2, lost AVA activity, the bivalency of IgA2m2 is essential for viral aggregation, likely because each Fab arm binds to a distinct virus particle, tethering two different virus particles. However, the reason for the functional differences between monomeric IgA1 and IgA2m2 remains unknown. One possible explanation for how structural differences between IgA subclasses affect their functions is that each IgA subclass is a product of a separate gene and contains numerous amino acid sequence differences, including glycosylation sites in the heavy chain constant regions⁶ (Supplementary Fig. 1). Supporting this idea, structural modeling based on X-ray and neutron scattering data suggests that the two Fab arms of monomeric IgA1 and IgA2 adopt different angles^7,46. The extended and flexible hinge of IgA1 may facilitate interactions between HA molecules located within the same viral envelope, but may be less suited for bridging HA molecules across distinct viral particles. In contrast, the restricted and compact hinge of IgA2, which favors a more compact T-shaped conformation of Fab arms, may be particularly suited for crosslinking HA molecules between different viral particles or between viral particles and the cell surface, thereby promoting virus aggregation (Supplementary Fig. 5B), as predicted by a geometric model based on the structures of Fab fragments bound to HA⁴⁷. Unexpectedly, we found that F045-092 IgG, which has a hinge length comparable to that of IgA2m2, also aggregated virus particles even in its monomeric and F(ab′)₂ forms (Supplementary Fig. 6F, G, and 7), suggesting that the conformation and flexibility of the Fab arms as well as bivalency are important for exerting AVA activity in monomeric form. Structural differences among IgG, IgA1, and IgA2m2, including hinge architecture, disulfide bonding, and glycosylation may further influence the efficiency and geometry of multivalent interactions with HA on different virus particles, potentially contributing to AVA activity across these antibody classes and subclasses. However, whether monomeric IgA2m2 targeting HA generally tends to exhibit AVA activity remains uncertain. Since this ability is thought to depend on the epitope site on HA and the angle of the two Fab arms of IgA on the epitope⁴⁷, monomeric IgA2m2 antibodies targeting the HA head region, such as F045-092, may tend to exhibit AVA activity. Supporting this notion, several IgA and IgG antibodies targeting the HA head region have been reported to inhibit viral release from infected cells^29,30,40,48. However, future studies using a panel of IgA2m2 and IgG antibodies recognizing distinct epitopes within the HA head region will be required to generalize this observation. It would also be of interest to examine whether IgA antibodies targeting the HA stalk region also exhibit AVA activity. Thus, further analysis using various monomeric IgA1 and IgA2m2 derived from different anti-HA clones recognizing distinct epitopes will be necessary to clarify the mechanism underlying AVA mediated by monomeric IgA.

Monomeric IgA1 and IgA2m2 of F045-092 exhibited comparable neutralization activities (Supplementary Fig. 3A), indicating that they possess comparable abilities to inhibit viral replication. However, they showed marked differences in their ability to inhibit viral entry, with monomeric IgA1 exhibiting greater activity (Fig. 2A, Supplementary Fig. 3B, and Table 1), and viral release, with monomeric IgA2m2 exhibiting greater activity (Fig. 2B). These data demonstrate that the neutralizing activity of these antibodies reflects the combined effects of inhibiting both viral entry and release and that both activities contribute substantially to the suppression of viral replication. It is possible that the slightly reduced neutralizing activity of monomeric IgA1 compared with monomeric IgA2m2 and IgG might be due to the absence of viral release inhibition. However, whether an antibody with strong entry-inhibitory activity exhibits a reduced ability to inhibit virus release (or vice-versa) upon IgA subclass switching remains to be elucidated in future studies for the development of virus-aggregating antibodies that efficiently inhibit both virus entry and release.

IgA1 is the predominant IgA subclass in the upper respiratory tract, and our findings that dimeric and tetrameric sIgA1 efficiently inhibit viral entry and release suggest a key role for multimeric sIgA1 in mucosal defense against IAV infection. By acting at the site of initial infection, multimeric sIgA1 can effectively limit viral replication and spread. In contrast, IgG, which is abundant in the circulation, is expected to encounter IAV primarily in the lower respiratory tract following extravasation under inflammatory conditions, where its ability to inhibit both viral entry and release may contribute to protection against severe pneumonia. The biological relevance of AVA mediated by monomeric IgA2m2 remains unclear, as monomeric IgA is not the dominant form at mucosal surfaces in vivo. Nevertheless, this observation suggests that IgA subclass–specific structural features can influence antiviral mechanisms independently of multimerization, with potential implications for IgA-based antiviral strategies.

In conclusion, we demonstrated that multimerization of F045-092 IgA targeting the HA head domain enhances anti-IAV activity by aggregating virus particles and that monomeric IgA subclasses with identical variable regions potentially exhibit distinct functions, such as AVA. Thus, multimeric sIgA with AVA activity likely inhibits both viral entry and release at the mucosa, playing a key role in protecting against IAV infection. In addition, since NA inhibitors such as oseltamivir⁴⁹ and zanamivir³¹ show strong protective efficacy against influenza, viral release can be a promising target for the development of new antivirals. Similar to monomeric IgA2m2 derived from F045-092, monomeric virus-aggregating antibodies could serve as novel therapeutics against IAV infection.

Materials and methods

Cells and virus

MDCK cells (kindly provided by Yoshihiro Kawaoka, The University of Tokyo) were maintained in minimal essential medium (MEM) (Gibco, Gaithersburg, MD, USA), supplemented with 5% newborn calf serum (Cytiva, Marlborough, MA, USA) at 37 °C under 5% CO₂. Expi 293 F cells (Thermo Fisher Scientific, Waltham, MA, USA) were grown in Erlenmeyer flasks (Corning, NY, USA) in Expi293^TM Expression Medium (Gibco) at 37 °C under 8% CO₂. The IAV strain (A/Victoria/361/2011 [H3N2]) was propagated in MDCK cells, and the viral titers were determined by plaque assay using MDCK cells.

Expression and purification of recombinant IgAs and IgG

Plasmids expressing F045-092 secretory IgA subunits (pC-hSC-His, pC-hJC, pA1-bVAbF045, pA2m2-bNAbF045, and pL-bNAbF045) were prepared as previously described¹⁷. Similarly, recombinant IgA1 and IgA2m2 were prepared as previously described¹⁷, with minor modifications. Briefly, a total of 120 μg of expression plasmids (40 μg of α1 or α2 heavy chain, 40 μg of λ light chain, 20 μg of JC, and 20 μg of SC), along with 192 μl of polyethyleneimine (PEI) (Polysciences, Warrington, PA, USA), were incubated in 10 ml of Opti-MEM (Sigma-Aldrich, Burlington, MA, USA) at room temperature. After 25 min of incubation, the mixture was added to 100 ml of Expi 293F^TM expression medium (3.0 × 10⁶ cells/ml, 95% viability) in an Erlenmeyer flask, and the cells were cultured for 5–7 days. Then, the supernatants and cells were harvested, and cell debris was removed by centrifugation at 1750 × g for 10 min at 4 °C. The supernatants were then filtered through a 0.45 μm filter (Merck Millipore, Burlington, MA, USA), incubated with Capture Select^TM IgA affinity matrix (Thermo Fisher Scientific) overnight at 4 °C, and then IgA samples were eluted according to the manufacturer’s protocol. Finally, the eluted F045-092 recombinant IgA was neutralized using 1 M Tris-HCl (pH 9.0). Purified F045-092 IgA was then concentrated using a 30 kDa cut-off Amicon Ultra (Merck Millipore), and the buffer was exchanged with phosphate-buffered saline (PBS) (137 mM NaCl, 8.1 mM Na₂HPO₄, 2.68 mM KCl, 1.47 mM KH₂PO₄). The F045-092 IgA solution was then condensed to less than 500 μl and subjected to SEC on a Superose 6 Increase 10/300 GL column (Cytiva) using an AKTA Pure™ 25 system (Cytiva). Next, 500 μl of each sample was fractionated, and the fractionations were analyzed via BN-PAGE at 4 °C. BN-PAGE was performed using NativePAGE 4–16% Bis-Tris gels (Thermo Fisher Scientific) with NativeMark (Thermo Fisher Scientific) as a molecular weight marker. After electrophoresis, the gels were bleached according to the manufacturer’s protocol. Gel images were acquired using ChemiDoc (Bio-Rad, Hercules, CA, USA).

For generating the F045-092 IgG, each plasmids expressing IgG1 heavy chain and λ light chain were used as previously reported¹⁷. Briefly, a total of 100 μg of expression plasmids (50 μg of IgG1 heavy chain, 50 μg of λ light chain), along with 160 μl of PEI, were incubated in 10 ml of Opti-MEM at room temperature. After 25 min of incubation, the mixture was added to 100 ml of Expi 293F^TM expression medium (3.0 × 10⁶ cells/ml, 95% viability) in an Erlenmeyer flask, and the cells were cultured for 5–7 days. Then, the supernatants and cells were harvested, and cell debris was removed by centrifugation at 1750 × g for 10 min at 4 °C. The supernatants were then filtered through a 0.45 μm filter, and subjected to Hitrap protein G column (Cytiva) using AKTA START system(Cytiva). The eluted F045-092 recombinant IgG was neutralized using 1 M Tris-HCl (pH 9.0). Purified F045-092 IgG was then concentrated using a 30 kDa cut-off Amicon Ultra (Merck Millipore), and the buffer was exchanged with phosphate-buffered saline (PBS) (137 mM NaCl, 8.1 mM Na₂HPO₄, 2.68 mM KCl, 1.47 mM KH₂PO₄). All of the IgAs and IgG samples were preserved at −80 °C until use.

Negative-staining electron microscopy

First, in-house carbon-coated 600-mesh Cu grids (Gilder, Grantham, England) were glow-discharged. Next, 5 µl of the sample was applied to the grid, stained with 2% uranyl acetate or 2% phosphotungstic acid (TAAB, Aldermaston, England), and observed using an HT-7700 TEM (Hitachi High-Tech Corporation, Tokyo, Japan). For immunoelectron microscopy, purified IAV particles were fixed with 2% paraformaldehyde and applied to formvar-coated 400 mesh Ni grids (VECO, Eerbeek, Netherlands). The grids were then incubated with each IgA sample (100 μg/ml), followed by incubation with goat anti-human IgA alpha chain conjugated to 4 nm gold beads (1:10 dilution, Jackson ImmunoResearch, West Grove, PA, USA). After washing, the samples were fixed in 1% glutaraldehyde, negatively stained with 2% uranyl acetate, and observed using an HT-7700 TEM (Hitachi High-Tech Corporation).

High-speed atomic force microscopy (HS-AFM)

For HS-AFM observation, each IgA sample was diluted to 20 μg/ml (tetrameric sIgA) or 100 μg/ml (monomeric IgA) with 10 mM phosphate buffer (pH 7.4) and mounted onto freshly cleaved mica without surface modification. After 10 min of incubation, the mica surface was washed with distilled water and visualized using an HS-AFM system (Nano Explorer; Research Institute of Biomolecule Metrology Co., Ltd., Ibaraki, Japan). Images were collected at a two images/s using cantilevers with a 0.1 N m⁻¹ spring constant and a resonance frequency in water of 0.6 MHz (BL-AC10DS; Olympus, Tokyo, Japan). To acquire high-resolution images, electron beam-deposited tips were fabricated using phenol or ferrocene powder, as described by Uchihashi et al.⁵⁰. Images 300 × 300 pixels in size were obtained. Further analysis using Gaussian filtering was performed using Kodec 4.4.7.39⁵¹.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE)

The purified protein was mixed in 2× Tris-glycine SDS sample buffer (Thermo Fisher Scientific), heated at 95 °C for 5 min, and subjected to SDS-PAGE using a 4-15% gradient gel (Bio-Rad). Electrophoresis was performed at 25 °C for 30 min using a constant voltage (200 V) in Tris-glycine buffer (25 mM Tris, 192 mM Glycine, and 0,1% SDS). The gels were stained by CBB satine one (Nacalai Tesque, Kyoto, Japan) for 1 hour. Then, gels were destained by ultrapure water and acquired images by ChemiDoc (Bio-Rad).

Purification of virus particles and their complexes with IgA

MDCK cells grown in TC175 flasks (Corning) were infected with IAV at a multiplicity of infection (MOI) of 0.00001 and propagated in 1 × MEM containing 0.3% BSA (Thermo Fischer Scientific) (MEM/BSA) containing 1 μg/ml TPCK (tosyl sulfonyl phenylalanyl chloromethyl ketone)-trypsin. Two days post-infection, the supernatants were harvested and centrifuged to remove the cell debris. Then, the virion-containing supernatant was subjected to ultracentrifugation at 10,000 × g for 2 h at 4 °C through a 30% (w/v) sucrose cushion using an Optima XE-90 Ultracentrifuge (Beckman Coulter, Brea, CA, USA) with an SW32Ti rotor (Beckman Coulter) to obtain purified IAV particles. The pellets were resuspended in PBS. To purify the IAV-IgA complex, purified IAV (2 × 10¹⁰ PFU/ml) was mixed with 200 μg/ml of each IgA sample for 1 h. Next, sucrose density gradient ultracentrifugation at 10,000 × g for 2 h at 4 °C under 20–60% sucrose concentration was performed using an Optima XE-90 Ultracentrifuge (Beckman Coulter) with an SW41Ti rotor (Beckman Coulter). The fractions were collected every 1 ml from the top.

Neutralization assays

Two-fold serially diluted IgA samples (ranging from 5 μg/ml to 2.4 ng/ml) were incubated with 100 PFU of IAV for 1 h at 37 °C and then added to MDCK cells grown in 96-well plates. Negative controls (wells with no antibody in the diluent) were prepared for each plate. Four days post-infection, the cytopathic effects (CPE) in each well were assessed under a light microscope. Wells without visible CPE were evaluated as fully neutralized, whereas CPE-positive wells were scored as non-neutralized. The lowest concentration that did not show CPE was defined as the minimum neutralization concentration (MNC). The MNC data were plotted using GraphPad Prism 9 software (GraphPad Software, San Diego, CA, USA).

Virus entry inhibition assay

The IgA samples and IgG were serially diluted four-fold and mixed with IAV (2000 PFU/ml) for 1 h. Then, the antibody-virus mixtures were added to MDCK cells grown in a 96-well plate (2.5 × 10⁴ cells/100 μl). At 8 to 10 hpi, cells were fixed with 2% paraformaldehyde, permeabilized with 0.1% Triton X-100 (MP Biomedicals, Irvine, CA, USA) in PBS, and blocked with 20% Blocking One (Nacalai Tesque). The cells were then incubated with a mouse anti-NP polyclonal antibody (1:2000 dilution, ab20343, Abcam, Cambridge, England), followed by incubation with an Alexa 488-conjugated anti-mouse IgG secondary antibody (1:1000 dilution, A11001, Thermo Fisher Scientific). After nuclear staining with Hoechst 33342 (1:2000 dilution, Thermo Fisher Scientific), the infected cells were visualized under a fluorescence microscope (BZ-800) (KEYENCE, Osaka, Japan) using a 4× objective. Nine fields per well were captured using a 3 × 3 tiling pattern, and the images were merged and analyzed using the KEYENCE software (BZ-H4A, Keyence, Osaka, Japan). A uniform threshold was applied across all the wells to quantify the number of infected cells. Data were plotted, and IC₅₀ was calculated using GraphPad Prism 9 software (GraphPad Software). The Mann–Whitney U test was used to calculate the p-values compared to each subclass of monomeric IgA. All analyses were performed using GraphPad Prism 9 (GraphPad Software).

Virus release inhibition assay

MDCK cells grown in 12-well plates were infected with IAV at an MOI of 2 (1 × 10⁷ PFU/ml). At 1 hpi, unbound IAV was removed by washing with 0.3% bovine serum albumin in MEM (MEM/BSA). At 3 hpi, the medium was replaced with MEM/BSAcontaining various concentrations of IgA (0.1–1 µg/ml). At 12 hpi, both cells and supernatants were collected, lysed in 2× Tris-glycine SDS sample buffer (Thermo Fisher Scientific), and subjected to western blotting.

Scanning electron microscopy (SEM)

IAV-infected MDCK cells were treated with IgA at 3 hpi and fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer on ice at 12 hpi. The samples were post-fixed with 1% osmium tetroxide on ice, dehydrated using a graded ethanol series, and substituted with t-butanol. The samples were then freeze-dried using a VFD-30 device (Vacuum Device, Inc., Ibaraki, Japan). After coating with osmium tetroxide using HPC-1SW (Vacuum Device Inc.), the samples were observed using a Regulus8220 microscope (Hitachi High-Tech Corporation).

Neuraminidase inhibition (NI) assay

The NI assay was performed as described previously^18,32, with minor modifications. Briefly, high-binding 96-well flat-bottom plates (Corning) were coated with 100 μl of fetuin (F3385, Sigma-Aldrich) at a concentration of 25 μg/ml for 24 h. Serial dilutions of IAV in PBST/BSA (PBS containing 0.05% Tween-20 and 1% BSA) were added to the fetuin-coated plates, and incubated for 16–18 h. After washing with PBST (PBS containing 0.05% Tween-20) six times, 2 μg/ml of HRP-conjugated peanut agglutinin (L7759, Sigma-Aldrich) was added and incubated for 2 h at room temperature in the dark. Subsequently, 100 μl of 1-Step Ultra TMB-ELISA Substrate solution (Thermo Fisher Scientific) was added to each well and incubated for 10 min at room temperature in the dark. The color reaction was stopped by adding 1 N H₂SO₄ (Nacalai Tesque). The absorbance was measured OD₄₅₀ using a GloMax microplate reader (Promega, Madison, WI, USA), with a reference wavelength of 600 nm. The NA activity corresponding to 90% of the maximal signal was defined and used as the baseline as well as for the NI activity measurements.

To assess the NI activity, 1 × 10⁸ PFU of IAV was mixed with two-fold serial dilutions of antibodies and incubated for 1 h. The IAV-IgA mixture was then added to the fetuin-coated plates. After 16 h of incubation at room temperature, the plates were washed with PBST and incubated with 2 μg/ml of HRP-conjugated peanut agglutinin (L7759, Sigma-Aldrich) for 2 h in the dark. Afterward, 100 μl of 1-Step Ultra TMB-ELISA Substrate solution (Thermo Fisher Scientific) was added to each well and incubated for 10 min in the dark at room temperature. The color reaction was stopped by adding 1 N H₂SO₄ (Nacalai Tesque). The OD₄₅₀ was measured (reference wavelength of 600 nm) using a GloMax plate reader (Promega). Neuraminidase activity was normalized to the negative control (virus-only wells), set to 100%. The half-maximal effective concentrations (EC₅₀) were calculated using GraphPad Prism 9 software (GraphPad Software).

Western blotting

Virus-infected cells, the supernatants, and purified viruses were dissolved in 2× Tris-glycine SDS sample buffer (Thermo Fisher Scientific), heated at 95 °C for 5 min without a reducing agent, and subjected to SDS-PAGE using a 4–15% gradient gel (Bio-Rad). Electrophoresis was performed at 25 °C for 30 min using a constant voltage (200 V) in Tris-glycine buffer (25 mM Tris, 192 mM Glycine, and 0,1% SDS). The proteins were transferred onto Immobilon-P transfer membranes (Merck Millipore) and blocked using Blocking One solution (Nacalai Tesque). The membranes were then incubated overnight at 4 °C with rabbit polyclonal anti-M1 (1:10000 dilution, GTX126921, GeneTex, Irvine, CA, USA), rabbit anti-α-tubulin (1:10000 dilution, PM054, MBL, Tokyo, Japan), goat anti-lambda chain (1:10000 dilution, A80-116P, Bethyl Laboratories, Montgomery, TX, USA), or mouse anti-IgA2 Fc region antibodies (1:10000 dilution, 9140-01, Southern Biotech, Birmingham, AL, USA). Afterward, the membranes were washed with PBST and incubated with HRP-conjugated anti-rabbit (1:10000 dilution, NA934, Cytiva), HRP-conjugated anti-mouse (1:10000 dilution, NA931, Cytiva), HRP-conjugated anti-Goat IgG (1:10000 dilution, ab6741, Abcam), or HRP-conjugated anti-human IgA+IgG+IgM(H + L) secondary antibodies (1:10000 dilution, 609-103-130, Rockland Immunochemicals, Limerick, PA, USA). Protein signals were detected using Chemi-Lumi One Super reagent (Nacalai Tesque).

Ultrathin section electron microscopy

MDCK cells infected with IAV and treated with IgA were fixed with 2.5% glutaraldehyde in 0.1 M cacodylate buffer on ice for 1 h and post-fixed with 1% osmium tetroxide on ice for 1 h. The fixed samples were stained en bloc with 1% uranyl acetate in water, dehydrated using an ethanol gradient, substituted with propylene oxide, and embedded in epoxy resin. Ultrathin sections ( ~ 50 nm) were cut using an ultramicrotome (EM UC7; Leica, Wetzlar, Germany), stained with 2% uranyl acetate and Reynold’s lead citrate, and observed using an HT7700 TEM (Hitachi High-Tech Corporation, Japan) at 80 kV.

Dynamic light scattering (DLS)

Purified IAV (2 × 10¹⁰ PFU/ml) was inactivated using 0.02% β-propiolactone and mixed with 200 μg/ml of each IgA sample for 1 h. DLS measurements were performed in cuvettes at 25 °C using a DLS-8000 series instrument (Otsuka Electronics Co., Ltd., Osaka, Japan). Particle size distribution was analyzed using the CONTIN method with software DLS-8000 series⁵².

Cryo-electron tomography (Cryo-ET)

Glow-discharged holey carbon grids (Quantifoil Cu 200 mesh R1.2/1.3, Jena, Germany) were treated with 10 nm fiducial gold beads (Electron Microscopy Sciences, Hatfield, PA, USA), and a mixture of IAV (2 × 10¹⁰ PFU/ml) and IgA (200 μg/ml) was applied to both sides of the grid. The grids were blotted for 7.5 s at 8 °C under 100% humidity and rapidly vitrified in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific). Images were acquired using a Glacios TEM operated at 200 kV and equipped with a Falcon IV camera (Thermo Fisher Scientific) installed at the Institute for Life and Medical Sciences, Kyoto University. Tilt series were recorded at 92000× magnification and a pixel size of 1.5 Å/pixel, with a −5 μm defocus. All tilt series were acquired using a dose-symmetric scheme over a range of -60° to +60° with 3° increments, yielding a total electron dose of 123 e⁻/Ų. The acquired tilt series data were preprocessed using MotionCor2 and the CTFFIND-4.1 plugin in RELION-5^53–55. The corrected tomograms were further aligned by tracking the gold fiducial and reconstructed by weighted back projection with a SIRT-like filter using IMOD⁵⁶. The tomograms were 8 × binned and denoised using IsoNet to visualize the enlarged tomogram slice of the images⁵⁷.

Expression of recombinant HA ectodomain

The expression of the recombinant HA ectodomain was performed as previously described¹⁷, with minor modifications. Briefly, Expi293F cells (3 × 10⁶ cells/ml, 50 ml) were transfected with 50 μg of a plasmid expressing HA ectodomain fused with a thrombin cleavage site and a 6 × His tag using polyethyleneimine and cultured for 5–7 days. After removing cell debris via centrifugation, the supernatants were loaded onto a Ni-NTA affinity column (Qiagen, Hilden, Germany). After washing with 10 CV of washing buffer (Tris-HCl [pH 8], 150 mM NaCl), the recombinant HA protein was eluted using 5 CV of elution buffer (Tris-HCl [pH 8], 150 mM NaCl, 500 mM imidazole). The elute was concentrated using Amicon Ultra with a 30 kDa molecular weight cut-off (Merck Millipore), and the protein concentration was measured using a BCA assay kit (Invitrogen).

Preparation of antibody fragments

To generate the F(ab′)₂ fragment, 5.0 mg/ml of purified monomeric IgA2m2 or IgG was digested with 0.1 mg/ml of pepsin in sodium acetate buffer (pH 4.5) at 37 °C for 16 h. The reaction was terminated by adjusting the pH to 8.0 with 1 M sodium hydroxide. The resulting F(ab′)₂ fragments were separated by SEC using a Superdex 200 Increase 10/300 GL column (Cytiva) on an AKTA Pure^TM 25 system (Cytiva). The peak fractions were subjected to SDS-PAGE, followed by CBB Blue staining (Nacalai Tesque) to evaluate the molecular weight.

To generate the Fab fragment, 5.0 mg/ml of purified monomeric IgA2m2 was incubated with 60 μg/ml of papain in a reaction buffer containing 20 mM EDTA and 20 mM L-cysteine (pH 7.0) at 37 °C for 6 h. The reaction was quenched by adding iodoacetamide to a final concentration of 30 mM. The Fab fragments were further purified via SEC using a Superdex 200 Increase 10/300 GL column (Cytiva) on an AKTA Pure^TM 25 system (Cytiva). Peak fractions were examined, and those lacking a detectable Fc region while retaining the lambda chain signal were defined as Fab-containing fractions and used for subsequent analyses.

Bio-layer interferometry (BLI) assay

Binding kinetics were measured using an Octet N1 system (SARTORIUS, Göttingen, Germany). Briefly, an Ni-NTA biosensor (SARTORIUS) was hydrated and equilibrated in PBST for 10 min before use. After baseline acquisition in PBST, 400 nM recombinant HA in PBST was immobilized onto the Ni-NTA sensors for 2 min according to the manufacturer’s instructions. Unbound HA was removed by washing with PBST. Subsequently, different concentrations of IgA (95–1000 nM) were applied to the HA-immobilized sensors for binding analysis. Buffer- and antibody-only controls were included to account for non-specific binding. Association and dissociation signals were recorded, and the kinetic curves were corrected by subtracting the background responses. Kinetic parameters, such as the association rate constant (k_on), dissociation rate constant (k_off), and equilibrium dissociation constant (K_D), were calculated via global fitting using Octet N1 system software. For figure visualization, Prism 9 was used for plotting the fitted curve and raw data.

Statistics and reproducibility

Statistical analyses for the virus entry inhibition assay and the neuraminidase inhibition (NI) assay were performed using the Mann–Whitney U test with GraphPad Prism 9 (GraphPad Software). A p-value of <0.05 was considered statistically significant. The corresponding p-values are indicated as *p < 0.05 and **p < 0.01. All experiments were repeated three to six times.

Author contributions

K.Y., Y.M., and T.N. designed the study. K.Y., Y.M., Y.F-F., A.H., C.N., T.I., Y.S., and M.N. performed the experiments; K.Y., Y.M., S. S., S. K., T.S., and T.N. wrote the manuscript with input from all co-authors. All the authors have read and approved the final manuscript.

Peer review

Peer review information

Communications Biology thanks Thomas Langer and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Theam Soon Lim and Laura Rodríguez Pérez. A peer review file is available.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09783-9.