Introduction

Neutralizing antibodies (NAbs) play a pivotal role in defending the host against pathogenic assaults, particularly from viral invaders. They are progressively being employed as passive antiviral agents in both prophylactic and therapeutic strategies, as well as in guiding vaccine design1,2,3. However, the continuous viral evolution and immune evasion substantially challenge the potential application of NAbs in the development of therapeutics and vaccines4. Broadly neutralizing antibodies (bNAbs), characterized by their capacity to recognize conserved epitopes across diverse viral strains, variants, or species, highlight the importance of understanding the mechanisms underlying broad neutralization5,6,7. This understanding will not only hasten the development of new antibody therapeutics but also inform the rational design of next-generation vaccines.

It’s known that the most common mechanisms of neutralization include obstructing virus-receptor recognition, disrupting the membrane fusion process, and preventing the release of progeny virions8. Recent studies have identified an extraordinary neutralizing mechanism where NAbs inactivate viruses by inducing irreversible structural disruption of various viral components, ranging from spike proteins of enveloped viruses (e.g., the SARS-CoV-2 spike9,10,11,12,13 and HIV envelope protein14,15,16,17), to the integral capsid of non-enveloped viruses (e.g., Hepatitis E virus18, Coxsackievirus19,20 and Norovirus21,22). The inherent meta-stable nature of these viral elements, manifesting in allowing cascading conformational rearrangements crucial for viral infection, may be an essential prerequisite of their disruption by NAbs. Structural information indicates that most NAbs target relatively cryptic epitopes, ordinarily shielded by glycan or other viral components, signifying these regions as potential viral vulnerabilities22,23,24. Due to their relative inaccessibility to the immune system, these epitopes tend to remain highly conserved, bestowing NAbs with a broadly neutralizing capacity. The presented conserved epitopes always represent the Achilles’ heel in the dynamic landscape of rapidly evolving viruses25,26,27. Such a perspective offers a novel strategy for pathogen elimination that is broadly effective against virus evasion. While a binding-induced steric hindrance model has been proposed to explain how NAbs disrupt viral structures12,18, the precise molecular underpinnings of this process remain largely elusive, partly due to methodological constraints in inspecting the ultrafast-scale reaction dynamics of virus destabilization.

We previously identified a rotavirus (RV) bNAb, 7H13, which targets the viral VP4 ‘spike’ protein28. RV entry involves a process akin to the action of enveloped virus fusion proteins, wherein VP4 undergoes large-scale conformational rearrangements, leading to membrane perforation29. In this study, we demonstrated the broad neutralizing capacity of the VP4-specific bNAb 7H13 against diverse P-genotypes of RV strains. By combining cryo-electron microscopy (cryo-EM) and cryo-electron tomography (cryo-ET) analysis, we showed that 7H13 induces irreversible damage to viral VP4, thereby blocking the adsorption process of the virus. Employing a low-temperature, time-resolved cryo-EM approach, we captured a series of intermediate states of viral immune-complexes and elucidated the high-resolution structure of the VP4:7H13 complex. Our findings indicate that 7H13 asymmetrically binds to a conserved epitope on the capsid-proximal side of VP4. The 7H13 targeting the right-hand side of VP4 induces steric clashes with a single VP4 residue, F418, triggering localized structural alterations and leading to the destabilization of the VP4 protrusion. Structure-guided mutations designed to reduce steric collisions mediated by I54 in the 7H13 heavy chain confirmed the essential role of this residue in activating the ‘molecular switch’ of F418 and initiating VP4 disruption. Our study provides a detailed understanding of the intricate mechanism behind antibody-mediated RV spike disruption, offering a paradigm for exploring such distinct neutralizing mechanism where antibodies destabilize viral components.

Results

In vitro and In vivo characterization of rotavirus bNAb 7H13

The infectious RV virion, a triple-layer particle (TLP), has the outermost layer formed by two structural proteins, VP4 and VP7, which define the P and G genotypes of the virus, respectively30. The VP4 protein is activated by cleavage into the N-terminal VP8* and C-terminal VP5*, which retain associated and are involved in mediating cell attachment and membrane penetration, respectively31,32. We previously reported a panel of mouse-derived NAbs targeting RV VP4, with several demonstrating the ability to cross-neutralize P[8], P[6], and P[4] genotype viruses28. Among these NAbs, 7H13 stood out for its significant cross-neutralizing potential, characterized by a low IC50 across all three P genotypes. In this study, we further evaluated 7H13’s broad neutralizing effect through extensive in vitro and in vivo analyzes. The reported human bNAb 41#, which recognizes an epitope at VP5*–VP8* junction and exhibits the ability to neutralize six P genotypes, was selected as a control33. To circumvent potential biases from differing constant region frameworks across species, we engineered a chimeric antibody, C7H13, by integrating the variable regions of 7H13 into a human antibody framework.

A panel of 14 RV strains, comprising 11 diverse P genotypes and 7 host species, was utilized in our in vitro neutralization evaluation. 7H13, C7H13, and 41# preferentially neutralized human RVs with sub-micromolar efficacy (IC50: 0.10–1.55 μg/mL), whereas unrelated monoclonal antibodies (mAbs) specific to SARS-CoV-2 with either human antibody constant domain (Fc) or murine Fc were ineffective against these RV strains (Fig. 1a and Figure. S1). Compared to 41#, 7H13 and C7H13 exhibited comparable neutralizing breadth and displayed significantly higher efficacy against the P[7] genotype of porcine RVs (Fig. 1a). Immunofluorescence assay confirmed the binding capability of 7H13/C7H13 to these RV strains, showing overall binding efficacy greater than that of 41# (Figure. S2), indicating higher binding affinity.

Fig. 1: Characterization of NAb 7H13 in vivo and in vitro.
Fig. 1: Characterization of NAb 7H13 in vivo and in vitro.The alternative text for this image may have been generated using AI.
Full size image

a Neutralization profile of 7H13 (murine Fc) and C7H13 (human Fc) against a panel of RV strains. 41# (human Fc) serves as positive control, while 5C6 (human Fc) and A2H4 (murine Fc) are negative controls. Refer to Figure S1 for detailed datasets. Cartoon illustrations were created in BioRender. Yang, H. (2024) https://BioRender.com/c98a036. b Schematic of the experimental design to assess the prophylactic effect of NAbs against RV infection and fecal shedding in a murine model. Created in BioRender. Yang, H. (2024) https://BioRender.com/s21c678. c Protective efficacy of 7H13 and m41# against fecal virus shedding in adult mice (n = 5 per group) following RV EDIM challenge. Data are log10 transformed and presented as mean ± SEM. d Cross-inhibition between C7H13 and m41# as evaluated by blocking ELISA. Blocking mAbs (100 μg/mL) were pre-incubated with microplates before the addition of blocked mAbs, and the binding of blocked mAbs was detected by HRP-conjugated secondary antibodies. Data are shown as mean ± SEM of three technical replicates. e Fold change in neutralizing activity of NAbs over time. Fold change represents the variation in IC50 across time points (1, 2, 4 and 6 h), calculated as the ratio of IC50 at 1 h relative to those at subsequent time points. Refer to Figure S3 for detailed datasets. f Schematic of the experimental setup for pre- and post-attachment neutralization assays. Created in BioRender. Yang, H. (2024) https://BioRender.com/j09b061. g Neutralizing activities of NAbs against SA11 and Wa strains in pre- and post-attachment assays. Data are shown as mean ± SEM of three technical replicates. Source data are provided as a Source Data file.

We next evaluated the in vivo protective efficacy of 7H13 and an engineered mouse Fc version of 41# (m41#, a modification of 41# incorporating the 7H13 framework) in mitigating RV-induced viral shedding. An adult mouse model was used to ensure proper antibody transcytosis. Mice were challenged with a mouse RV strain EDIM (2.5×107 copies/mouse) one day after intravenous administration of either 7H13 or m41# (5 mg/kg) into the tail vein. Daily fecal samples were collected over seven days to monitor viral shedding (Fig. 1b). In contrast to control mice, which showed continuous viral shedding for five days post-infection, those treated with 7H13 or m41# exhibited no detectable viral excretion (Fig. 1c), demonstrating 7H13’s strong in vivo efficacy in preventing RV infection and shedding.

The blocking experiment conducted with 7H13 and 41# revealed that 7H13 recognizes an epitope distinct from that of 41# (Fig. 1d). Notably, 7H13 uniquely displayed time-dependent neutralizing activity, increasing in potency from 1 to 6 h of pre-incubation with RV, a property not observed with 41# or m41# (Fig. 1e and Figure S3). This suggests that 7H13 may employ an atypical virus inactivation mechanism distinguishing to that of 41#. 7H13’s neutralization action was further elucidated by comparative scenarios including the pre- and the post-attachment assays (Fig. 1f). In the pre-attachment assay, the virus was pre-incubated with antibody at 37 °C for 1 h before cell infection. In post-attachment assay, the virus was first absorbed onto cells at 4 °C for 1 h, followed by incubation with antibody at 4 °C for an additional 2 h to ensure virus-antibody interaction. Cell cultures were then performed at 37 °C for 14 h for detection. The neutralizing potency of 7H13 in the pre-attachment assay was ~6-fold higher than in the post-attachment assay (Fig. 1g), implying that 7H13 effectively impedes the initial binding of RV to host cells. Conversely, 41# displayed no significant disparity in neutralizing activity between two assays, confirming its ineffectiveness at blocking RV attachment, aligning with prior research33 (Fig. 1g). Subsequent assays using fluorescently labeled VP4 protein confirmed that 7H13 could inhibit VP4’s attachment at elevated concentrations (Figure S4), yet it interestingly does not interact with VP8* (Figure S5), the key mediator of RV attachment. Collectively, these findings suggest that 7H13 may impede viral attachment in an unconventional and highly effective mechanism that does not rely on directly targeting the receptor-binding sites.

Cryo-EM reveals 7H13’s unusual binding mode and binding induced VP4 destabilization

To elucidate the mechanism underlying 7H13-mediated broad neutralization, we prepared the purified simian RV SA11 (Figures S6a–d) and performed cryo-EM analysis on its complex with the antigen-binding fragment (Fab) of 7H13. The Fab demonstrated neutralizing activity comparable to the full-length mAb (Figures S6e–g). We first characterized the cryo-EM structure of unbound SA11 at an overall resolution of 3.39 Å (Figures S7a and Table S1), revealing well-defined density for the capsid shell with typical TLP features, yet displaying relatively weaker density for the VP4 decorated on the outermost layer (Fig. 2a). Focused 3D classification specifically on VP4 revealed its adoption of the atypical asymmetric ‘upright’ conformation (Fig. 2b, c). Partial occupancy of VP4 in TLPs, consistent with previous studies34, was observed (Figure S8a). Localized reconstruction enhanced the map quality to a resolution of 3.19 Å, highlighting a progressively blurred VP4 density towards the apical VP8* lectin domain, possibly due to structural dynamics between the prone VP5* and the upright VP5* projection (Figure S9). To this end, we implemented further local refinements using a mask that encompassed only the three VP5* β-barrel domains, resulting in a 3.84 Å resolution density map (Figure S9 and Table S2). This facilitated confident modeling of the intact VP4 spike except for the VP8* lectin domain.

Fig. 2: The unusual binding mode of 7H13 and binding-induced VP4 destabilization.
Fig. 2: The unusual binding mode of 7H13 and binding-induced VP4 destabilization.The alternative text for this image may have been generated using AI.
Full size image

a Radially colored surface representation of reconstructed RV strain SA11, showing external and sectional views of the intact virion. Symmetry axes are annotated with geometric markers. b Ribbon diagram of VP4 model, color-coded by domain: VP8* α in orange, VP8* lectin and its linker to VP8* α in purple, VP5* β-barrel (VP5*A, *B and *C) in shades of pink, VP5* C-terminal foot in yellow. c Focused 3D classification of SA11, with VP4 densities color-coded by domain. VP6 and VP7 are depicted in deep and light gray, respectively, and clipped to visualize the VP4 foot. d SA11:7H13 complex incubated at 37 °C for 5 minutes. e, f Focused 3D classifications of SA11-7H13 (37 °C-5min). e shows two 7H13 bound to opposite sides of the VP4, differentiated by green and blue. f shows disrupted VP4, characterized by noise-dominated densities in pink. g SA11:7H13 complex incubated at 37 °C for 1 h. h Focused 3D classification of SA11-7H13 (37 °C-1h), highlighting extensive VP4 disruption. i Schematic representation of disrupted VP4. A semi-transparent surface representation of VP4 density segmented from SA11:7H13 (37 °C-1h) in h is overlaid with atomic models. Disrupted VP4 segments and their connectors to the spike base are illustrated with patches and dotted lines, respectively, colored to match their domains. j, k Linear diagrams of VP4 domain organization before (j) and after (k) 7H13 binding, with amino acid positions numbered. Disrupted domains are shaded in gray in (k). l SDS-PAGE analysis of SA11 following incubation with 7H13-Fab at two time points. M: molecular mass protein marker; C: SA11 control.

We next incubated the virus with an excess of 7H13 Fabs at 37 °C for 5 minutes and 1 h, respectively, to characterize the structures of the SA11:7H13 immune-complexes. Intriguingly, the cryo-EM structure from the 5-minute incubation (SA11:7H13-5min), resolved at 3.64 Å (Figure S7b and Table S1), exhibited substantially blurred electron density for VP4 compared to the native virion structure (Fig. 2d). The localized 3D classification revealed that only a small fraction of VP4 had unambiguous density, with two 7H13 bound to either side of the dimeric VP5* projection (Fig. 2e), revealing that 7H13 binds to an epitope on the capsid-proximal side of VP4 and far from the VP8* domain. However, the majority of VP4 presented blurred density, except for its integral foot region, indicating significant structural disruption or potential loss of some VP4 components (Fig. 2f). Extending the incubation to 1 h (SA11:7H13-1h) resulted in a structure at a resolution of 3.8 Å, where VP4 density became nearly indiscernible (Fig. 2g and Table S1). Localized 3D classification on SA11:7H13-1h dataset could no longer detect identifiable density for VP4 or bound antibodies, with nearly all existing VP4 losing the portions that protrude externally from the capsid (Fig. 2h and Figure S8c). As a control, we resolved the structure of SA11:41# immune-complex which was prepared by incubating SA11 with 41# Fab at 37 °C for 1 h (SA11:41#-1h). The resulting density map showed no evidence of VP4 disruption upon the 41# binding, consistent with previous reports (Figure S10). These findings indicate that 7H13 binding potentially disrupts the VP4 spike over time. Notably, the overall quantity of VP4 in TLPs remained relatively constant across the SA11 and SA11:7H13-1h datasets, as evidenced by the consistent proportion of sub-particles featuring VP4 foot density in localized 3D classifications (Figures S8a and S8c). These observations suggest that 7H13 binding leads to disruption of VP4’s partial structure, including the three VP5* β-barrel domains and the apical VP8* lection domains (Fig. 2i–k), potentially unveiling a novel mechanism for RV neutralization.

To delve deeper into this unusual neutralization mode, we first sought to examine whether 7H13 binding induces the shedding of VP4 or its component. SA11 was incubated with excess 7H13 Fab at 37 °C for 30 minutes and 6 h, and viral components were analyzed via SDS-PAGE. No variations in viral composition were observed among the untreated virus and those incubated with different antibodies or for varying durations (Fig. 2l), indicating that VP4 was not cleaved but rather underwent a conformational change upon 7H13 binding. Further high-performance size exclusion chromatography (HPSEC) analysis comparing untreated SA11 with SA11:7H13 suggested that the bound 7H13 could detach from the disrupted VP4 (Figure S11), potentially due to the conformational changes at its binding epitope. This observation could explain 7H13’s previously demonstrated time-dependent neutralizing activity. Collectively, these results suggest that 7H13 may exert irreversible damage on VP4, dissociating thereafter to engage anew with other intact VP4 molecules, significantly amplifying its neutralizing capacity.

Cryo-ET reveals the disordered VP4 on virion upon 7H13 binding

To extend our understanding of the conformational changes in VP4 induced by 7H13, cryo-electron tomography (cryo-ET) was employed to determine the in situ structures of RV when interacting with 7H13 (Table S3). Tomographic reconstruction of the untreated SA11 virion was firstly performed, showcasing its characteristic wheel-like structure, adorned with elongated viral VP4 that manifested in a ‘Y’-shaped configuration, aligning with established atomic models (Fig. 3a, b). Subsequent missing-wedge correction and denoising processes yielded 3D densities that clearly depicted SA11, confirming an unsaturated decoration of VP4 on the virion surface (Fig. 3c and Figure S12a–c), corroborating our localized classification results. The atomic structure of SA11 TLP fitted well within the tomographic density (Fig. 3d, e and Figure S12b), albeit with minor discrepancies in certain spikes (Figure S12d), attesting to the inherent dynamic nature of VP4.

Fig. 3: Cryo-ET reveals the disordered VP4 on virion upon 7H13 binding.
Fig. 3: Cryo-ET reveals the disordered VP4 on virion upon 7H13 binding.The alternative text for this image may have been generated using AI.
Full size image

a Representative tomographic slice of the native SA11 virion. b Magnified view of VP4 from a, showing intact VP4 morphology. c Segmented SA11 volume with radial coloring to highlight the core-to-periphery gradient. Blue and pink arrows indicate areas detailed in (d and e) respectively. d, e Close-up views of VP4 from the native SA11 virion. The virus is shown as semi-transparent surfaces overlaid with atomic models of RRV strain (PDB: 4V7Q). f Representative tomographic slice of the SA11:7H13 complex after 1-h incubation at 37 °C, showing altered VP4 morphology. g Magnified view of VP4 from (f) showing disrupted VP4 in the presence of 7H13. h Segmented SA11:7H13 volume with radial coloring, highlighting disordered regions of VP4. Blue and pink arrows indicate areas detailed in i and j, respectively. i, j Close-up views of disordered VP4 densities from the SA11:7H13 complex. The complex is shown as semi-transparent surfaces overlaid with composite SA11:7H13 model, generated by superimposing VP4:7H13 structure onto the virus model (PDB: 4V7Q). Scale bars: 50 nm in (a and f); 10 nm in (b and g).

For cryo-ET analysis of the SA11:7H13 complex, preparations were conducted at 37 °C for 30 minutes before vitrification. The reconstructed tomograms depicted the virus’s triple-layered architecture intact but showcased varying VP4 morphologies, evidencing conformational rearrangement rather than complete disintegration of VP4 (Fig. 3f). Most VP4 spikes diverged from their original ‘Y’-shape configuration, consistent with the above cryo-EM findings (Fig. 3g). The denoised 3D density map of SA11:7H13 illustrated irregular spike patterns, in stark contrast to the atomic model of VP4:7H13, suggesting the destabilized and disorganized state of VP4 following 7H13 binding (Fig. 3h–j and Figure S12e). Notably, the identification of spikes with enlarged yet irregular density volume suggests the persistent binding of some 7H13 to VP4 at this stage, despite the structural changes had occurred (Figure S12f). Furthermore, electron densities that partially matched the VP4:7H13 underscores that VP4 disruption might not occur instantaneously but rather progress over time, indicative of a gradual change in the thermodynamic stability of VP4 molecules triggered by 7H13 (Figure S12h). Taken together, our cryo-ET analysis provides further compelling evidence that 7H13 binding distinctly compromises the structural integrity of viral VP4.

Biased binding mode of 7H13 results in temperature-dependent VP4 destabilization

Given that the structural destabilization of VP4 may be a time-dependent process, we hypothesized that such dynamic perturbations could be mitigated at low temperatures. Accordingly, we lowered the incubation temperature to 4 °C and conducted a time-resolved cryo-EM analysis, preparing SA11:7H13 complexes at reaction times of 20 seconds, 5 minutes, 1 h, and 12 h, respectively. The resulting structures, with resolutions of 3.59 Å (20 sec), 3.47 Å (5 min), 3.53 Å (1 h), and 3.35 Å (12 h) (Figures S7d–g and Table S1), exhibited stark differences from those obtained at 37 °C, revealing clear densities for intact VP4 with 7H13 bound (Fig. 4a–e). Notably, the 20-sec incubation showed exceptionally weak densities for bound 7H13 (Fig. 4b), indicative of an incomplete virus-antibody reaction. These findings demonstrated that at 4 °C, 7H13 can bind to SA11 without inducing VP4 destabilization.

Fig. 4: Cryo-EM analysis reveals temperature and reacting time are dependent for the destabilization of VP4 by 7H13.
Fig. 4: Cryo-EM analysis reveals temperature and reacting time are dependent for the destabilization of VP4 by 7H13.The alternative text for this image may have been generated using AI.
Full size image

a–f Comparative structural analyzes of the native SA11 virion (a) and SA11:7H13 complexes following incubations under varying conditions: at 4 °C for 20 seconds (b), 5 minutes (c), 1 h (d), 12 h (e), and 4 °C for 1 h followed by 37 °C for another 1 h (f). VP7, VP4, left-7H13, and right-7H13 are colored light gray, pink, green, and blue, respectively. g, h 3D class averages of the SA11:7H13 complex for 20-second (g) and 5-minute (h) incubations, with VP4 represented as semi-transparent surfaces overlaid with ribbon diagrams, color-coded by domain: head in purple, body and stack in pink, and foot in yellow. i Schematic illustration of the hypothesized sequential VP4 disruption process. The color scheme corresponds to that used in h, with VP5* β-barrel domains shown in shades of pink.

Further localized classification of these datasets unveiled a biased binding mode of 7H13. Specifically, under the 20-sec incubation condition, classification results showed pronounced density for 7H13 on the left side of VP4 (hereafter referred to as left-7H13), while density on the right side (referred to as right-7H13) was absent (Fig. 4g and Figure S8d). Increasing the reaction time to 5 minutes enabled the visualization of well-defined density for the VP4 with both sides bound with 7H13 (Fig. 4h). This observation was consistent at 1-h and 12-h reaction times, with only a minor fraction of VP4 experiencing structural disruption after 12 h of incubation (Figures S8f and S8g). However, sequential incubation, initially at a low temperature (4 °C for 1 h) followed by physiological temperature (37 °C for 1 h), led to the disruption of nearly all viral VP4 (Fig. 4f and Figure S8h), highlighting temperature as a crucial factor in 7H13-induced VP4 destabilization. These findings illustrate the inherent structural instability of the VP4:7H13 complex, which is exacerbated by temperature elevation, and culminating in VP4 disintegration. Collectively, our time-resolved cryo-EM strategy, capturing a series of critical structural snapshots, provides initial insights into the reaction pathway by which 7H13 disrupts VP4: the antibody sequentially binds to the left and right sides of VP4, triggering change in VP4’s thermodynamic stability that eventually leads to its destabilization and disintegration (Fig. 4i).

7H13-binding results in a single residue switch occurred in the interface of three VP5*

We next sought to elucidate the high-resolution structure of the VP4:7H13 complex and its interaction details. Integrating sub-particle datasets from cryo-EM experiments conducted at 4 °C for 5 minutes and 1 h—both unambiguously demonstrating VP4 binding with two 7H13—enabled us to reconstruct the VP4:7H13 complex at a resolution of 3.55 Å (Figures S13 and Table S2). This reconstruction allowed us to build an atomic model comprising three VP5* β-barrel domains and the variable domains of two bound 7H13. The two 7H13 bind to the dimeric VP5* projection (VP5*A and VP5*B) from opposing sides, in close spatial proximity to the capsid surface (Fig. 5a). Notably, the binding pattern of the two 7H13 deviates from strict two-fold symmetry, with the epitope recognized by the right-7H13 involving the third prone VP5* (VP5*C) compared to that of the left-7H13 (Fig. 5b-e).

Fig. 5: High-resolution structure reveals an asymmetric mode of 7H13 binding triggered localized conformational alteration of VP4.
Fig. 5: High-resolution structure reveals an asymmetric mode of 7H13 binding triggered localized conformational alteration of VP4.The alternative text for this image may have been generated using AI.
Full size image

a Composite model of SA11:7H13 complex, with VP6 and VP7 surfaces clipped for clarity. VP8* α is shown in orange, VP8* lectin and its linker to VP8* α in purple, VP5* β-barrel in shades of pink, VP5* foot in gold, VP6 and VP7 in deep and light gray, respectively. Left- and right-7H13 are depicted in green and blue, respectively. b, c Atomic model of VP4:7H13, with VP4 depicted as a surface and 7H13 as ribbons. c represents a 180° rotated view relative to b. d, e Interactions of left- (d) and right-7H13 (e) with VP4. Epitopes are represented as semi-transparent surfaces with outlined boundaries, color-coded according to the corresponding VP5* β-barrel domains. f-h Detailed interactions between VP4 and 7H13. f Interactions between left-7H13 and the VP5* projection, with key residues shown in stick representation. Hydrogen bonds and salt bridges are indicated by green and orange dashed lines. g Close-up view of W52 and I54 from left-7H13 heavy chain, depicted in stick and semi-transparent surface representations, inserting into the pocket of VP5*A, visualized as hydrophobic (yellow) and charged (turquoise) surfaces. h Interactions between right-7H13 FR3 and the VP5*C hydrophobic loop. i Sequence conservation of 7H13 epitopes. Black asterisks mark residues recognized by both left- and right-7H13, while red asterisks indicate those unique to right-7H13. j–m Structural rearrangements induced by 7H13 binding. j Superimposition of unbound VP4 (white ribbon) and VP4:7H13 complex (colored ribbon and surface). k Close-up view of the VP4:left-7H13 binding interface. l Rearrangements of the VP5*A stem-loop F418 induced by right-7H13 HCDR2, highlighted by a yellow arrow. m Rearrangements of VP5*C hydrophobic loop induced by right-7H13 FR3. In l and m, the residues in 7H13 that dominantly contribute to inter-VP4-right-7H13 collisions are shown in stick and semi-transparent surface. n, o Cross-sectional comparison of VP5*A stem-loop after (n) and before (o) right-7H13 binding. The stem-loop is rendered in yellow, with other domain elements colored as in a.

For the left-7H13, the core epitope region is focused on a bulge at the bottom of the VP5*A, predominantly composed of a three-stranded antiparallel β-sheet (Fig. 5d). Additionally, a distal stem-loop (aa.414-421) from the opposing VP5*B also interacts with the left-7H13 (Fig. 5d). The interaction between the VP4 and left-7H13 buries a surface area of ~828 Å2, primarily mediated through polar interactions, including six hydrogen bonds and two electrostatic interactions (Fig. 5f). Among these, a salt bridge is formed between 7H13 and D417 at the tip of the VP5*B stem-loop (Fig. 5f). The W52 and I54 of the left-7H13 heavy chain variable domain (VH) are embedded into two adjacent shallow grooves on VP5*A, further stabilizing the interaction (Fig. 5g). For the right-7H13, although it recognizing a discrepant epitope region, its binding mode and detailed interactions are nearly identical to those of the left-7H13 (Fig. 5e and Figure S14). Besides, the VH of the right-7H13, mainly its framework region 3 (FR3), spans across VP5*C, establishing additional hydrogen bonds and electrostatic interactions with its hydrophobic loop (aa.284-291), thereby increasing the interface area to ~1015 Å2 (Fig. 5e, h). An epitope-conservancy analysis performed using 6713 group A rotavirus VP4 sequences confirmed the high conservancy of the 7H13 epitopic regions, with 21 of 24 sites showing conservation levels between 90% to 99% (Fig. 5i). Furthermore, 6 of 8 critical residues involved in electrostatic interactions—Q266, D270, K290, D417, E463 and R467—showed a level of conservation greater than 98% (Fig. 5i). The conserved nature of the main 7H13 epitope on VP5* is postulated to result from its critical role in stabilizing the meta-stable VP4 structure.

The structural analysis above suggests that right-7H13, with its larger binding interface, might have a stronger affinity for VP4 compared to left-7H13. This assumption, however, contrasts with findings from our low-temperature time-resolved cryo-EM analysis, which demonstrates the preferential binding of left-7H13. To investigate further, we compared atomic models of VP4 with and without the binding of 7H13 (Fig. 5j). No discernible changes were observed at the left-7H13 binding interface (Fig. 5k), whereas significant conformational alterations were identified at the right-7H13 binding site, especially in the VP5*A stem-loop (Fig. 5l). Located at the juncture of three VP5* subunits, this stem-loop appears to stabilize the interface by anchoring the F418 into a hydrophobic pocket formed by the other two VP5* (Fig. 5l). Binding of right-7H13 induces a conspicuous collision between F418 of VP5*A and heavy chain complementarity-determining regions 2 (HCDR2) of 7H13, particularly residue I54, causing F418 to shift from its original position and be replaced by the I54 (Fig. 5l). This results in the outward deflection of the tip of VP5*A stem-loop (Fig. 5l). Moreover, the VP5*C hydrophobic loop exhibits a slight movement away from the interface (Fig. 5m), suggesting that the VP5*C has negligible effects on the capacity of right-7H13 to bind VP4. In summary, these results provide a mechanistic explanation for the differential binding capabilities of 7H13 to the two sides of VP4. The VP5*A stem-loop, particularly residue F418, poses an energy barrier that right-7H13 must overcome to mediate the conformational rearrangement of the stem-loop, thereby facilitating its binding.

We wonder whether the destabilization of VP4 upon 7H13 binding is attributed to the conformational alteration in VP5*A F418, given its role as a structural bridge linking the three VP5* subunits. Notably, VP4 site 418 is highly conserved across RV genotypes, predominantly occupied by phenylalanine and, to a lesser extent, by tyrosine—both sharing structural similarities—suggests its essential role in stabilizing the metastable upright VP4 trimer (Fig. 5i). In the presence of right-7H13, its HCDR2 displaces F418 and inserts I54 into the hydrophobic cleft formed between VP5*B and VP5*C (Fig. 5n). This interaction disrupts molecular packing at the interface of three VP5* and exposes hydrophobic surfaces, potentially causing localized structural instability (Fig. 5n, o). It seems that the structural rearrangement of a single residue—F418—drives the destabilization and disruption of VP4. We therefore designate F418 as a ‘molecular switch’, with 7H13 activating it mainly through the key residue I54 in its heavy chain.

Structural guided mutagenesis confirms the VP4 disruption initiated by single F418 switch

To fully elucidate the mechanism underlying 7H13-induced VP4 destabilization, we first reduced the antibody concentration to one-tenth for preparing the SA11:7H13 complex, aiming for unsaturated 7H13 binding. Given the binding hindrance caused by the VP5*A F418 on the right side of VP4, we anticipated 7H13 would preferentially bind to the left-hand side under this condition. This hypothesis was confirmed by the reconstruction of the immune-complex (SA11:7H13-1/10) incubated at 4 °C for 1 h, which revealed the exclusive binding of 7H13 to the left side of VP4 (Figure S7i and S8i). Upon increasing the incubation temperature to physiological levels for one hour, the reconstructed map at a resolution of 3.46 Å still depicted well-defined viral VP4 density (Fig. 6a and Figure S7j, Table S1), suggesting that the VP4 disruption was effectively inhibited. The subsequent localized 3D classification confirmed that the VP4 was bound only with the left-7H13 and remained intact (Fig. 6b and Figure S8j). Moreover, the high-resolution localized reconstruction of the VP4:7H13-1/10 (4 °C), with a resolution of 4.19 Å (Fig. 6c and Figures S15, Table S2), confirmed that the left-7H13 binding has no effect on the VP5*A F418 (Fig. 6d). These findings are consistent with our anticipation attributing VP4 disruption directly to right-7H13 triggering the single residue switch.

Fig. 6: Structural validation of the biased binding of 7H13 and binding-induced F418 switch.
Fig. 6: Structural validation of the biased binding of 7H13 and binding-induced F418 switch.The alternative text for this image may have been generated using AI.
Full size image

a, b Cryo-EM reconstruction (a) and focused classification (b) of SA11:7H13 complex prepared at 1/10 antibody concentration and incubated at 37 °C for 1 h, showing VP4 (pink) bound exclusively with left-7H13 (green). c Localized reconstruction of SA11:7H13−1/10 complex incubated at 4 °C for 1 h. d Close-up view of VP5*A stem-loop from (c) depicted as a semi-transparent yellow surface overlaid with atomic model. The right-7H13 model (white) is superimposed to highlight the absence of collisions between its I54 residue and VP5*A F418 (both in stick). e Structural representations of I54 residue variants (I54A, I54S, and I54G) illustrate mutation-induced variations in side-chain size. f Neutralization activities of 7H13 and its mutants. Data are shown as mean ± SEM of three technical replicates. g, h Cryo-EM reconstruction (g) and focused classification (h) of SA11:7H13-I54G complex incubated at 37 °C for 1 h, showing VP4 bound by mutated left-7H13 (green) and right-7H13 (blue). i, j Localized reconstructions of the SA11:7H13-I54G complex incubated at 4 °C for 1-h, targeted to left-7H13 (i) and right-7H13 (j). k, l Superimpositions of VP4:left-7H13-I54G onto VP4:7H13 (k) and VP4:right-7H13-I54G (l), showing the mutation-induced alterations in binding postures. m Close-up view of VP5*A stem-loop from (j) with right-7H13-I54G depicted as a semi-transparent blue surface overlaid with the corresponding model, demonstrating that the 7H13-I54G mutant fails to induce conformational changes in VP5*A F418. Source data are provided as a Source Data file.

Next, we aimed to weaken the 7H13’s ability to mediate conformational changes in the VP5*A stem-loop through structure-guided mutagenesis. To achieve this, VH I54 was substituted with serine (7H13-I54S), alanine (7H13-I54A) and glycine (7H13-I54G), respectively, anticipating a gradual reduction in the steric collision induced by right-7H13 (Fig. 6e). As expected, these mutations resulted in a stepwise decrease in neutralization capacity, notably with 7H13-I54G exhibiting a significant loss of neutralizing ability (Fig. 6f). This indicates that the ability of 7H13 to disrupt VP4 is indeed closely related to HCDR2 and, in particular, to the I54, which is the key residue involving in triggering the F418 switch.

We then employed the 7H13-I54G mutant for cryo-EM analysis, preparing the viral immune-complex at 37 °C for 1 h. The resulting 3.43 Å density map revealed the intact VP4 density on the viral surface, demonstrating that the I54G mutation indeed inhibits VP4 disruption as expected (Fig. 6g and Figure S7k, Table S1). Compared to native virions or those treated with a one-tenth concentration of 7H13, the VP4 density appeared slightly weakened, indicating that partial VP4 might still undergo conformational changes upon 7H13-I54G binding. This hypothesis was corroborated by subsequent localized 3D classification, identifying some sub-particles with disrupted VP4 (Figure S8k). Nonetheless, the presence of sub-particles exhibiting intact VP4 simultaneously bound with two 7H13-I54G Fabs indicates that the I54G mutation weakens the ability of right-7H13 to trigger VP4 destabilization at physiological temperature (Figure S8k). Astonishingly, even after one hour of reaction, the density for left-7H13-I54G was more pronounced than that for right-7H13-I54G, with some sub-particles showing VP4 bound exclusively by the left-7H13-I54G (Fig. 6h and Figure S8k). This suggests that 7H13-I54G still retains a lower binding affinity for the right side of VP4.

To investigate the binding and neutralizing deficiency caused by the I54G mutation, we prepared the SA11:7H13-I54G complex at 4 °C for 1 h to resolve its high-resolution structure. The overall structure, at 2.54 Å resolution, showcased identifiable VP4 and bound antibodies similar to those observed under the 37 °C condition (Figures S7l and Table S1). Subsequent localized 3D classification and reconstructions with different masks yielded two structures: VP4 bound with the left-7H13-I54G (VP4:left-7H13-I54G, 3.01 Å) and the right-7H13-I54G (VP4:right-7H13-I54G, 3.13 Å) (Fig. 6i, j and Figure S16, Table S2). Intriguingly, these structures showed noticeable movements in the heavy chains of both left and right 7H13-I54G away from the antibody-antigen binding interface compared to native 7H13 (Fig. 6k, l). The binding posture alteration of left-7H13 could be attributed to the loss of hydrophobic interactions provided by I54. For right-7H13-I54G, as expected, it was inadequate to influence the conformation of VP4 F418 as well as the VP5*A stem-loop (Fig. 6m). However, the persistent collisions between F418 and the backbone of HCDR2, coupled with G54’s failure to reposition F418, necessitated slight conformation changes in HCDR2 to adapt to the interface. This likely left right-7H13-I54G in an unstable binding state, unable to overcome the energy barrier posed by the unswitched F418. Consequently, it undergoes a greater elevation away from the binding interface in comparison to the left-7H13-I54G (Fig. 6k), leading to weaker interactions with VP4, potentially resulting in eventual detachment from the VP4. These structural insights suggest that the right-7H13-I54G is inadequate to induce the conformation change in F418, thereby largely weakening its ability to disrupt VP4.

Taken together, the combination of comprehensive time-resolved cryo-EM and localized reconstruction provide a structural landscape of how 7H13 induces viral VP4 spike disruption by triggering a single residue (VP4 F418) switch. The conserved region surrounding VP4 F418 therefore represents a key site of vulnerability that may benefit for the development of universal vaccines and therapeutics against RVs.

Discussion

Antibodies are central to humoral immunity, serving as the frontline defense against viral threats through various mechanisms. Recent studies have illuminated an underappreciated mechanism whereby antibodies neutralize by directly compromising virus integrity, a phenomenon observed in both non-enveloped and enveloped viruses. These viral components share a meta-stable nature, undergoing specific alterations during infection. Regarding RVs, the VP4 protein undergoes substantial conformational changes upon attachment, facilitating membrane penetration, followed by the dissociation of the outermost capsid protein VP7 for the release of transcription machinery35. Previous studies have shown significant structural disruption in RV particles induced by the binding of NP1b mAbs 10G6 and 7G6, evidenced by the loss of the outer capsid layer36. In this study, we demonstrate that antibodies can also disrupt the RV VP4 spike, reinforcing the compelling theory that the viral proteins prone to conformational changes during infection potentially present specific structural vulnerability that antibody can exploit to cause their irreversibly disruption.

Despite the identification of numerous antibodies capable of disrupting viral components, the underlying mechanisms remain enigmatic, hindered by challenges in capturing the undestroyed intermediate state of the immune-complex. Several hypotheses have been raised; for instance, some antibodies are thought to mimic receptor functions to alter viral protein conformations, as in enteroviruses where antibodies initiate capsid uncoating by targeting receptor-binding regions37,38,39,40. Most disruption-inducing antibodies have been suggested to provoke virus disassembly through the facilitation of physical collisions, as their epitopes are identified in regions that are spatially inaccessible. An illustrative case is the 8C11 antibody specific to the Hepatitis E virus (HEV), which binds near the viral protrusion spike and causes substantial collisions with the capsid, culminating in capsid dissociation18. In this study, we applied cryo-EM and cryo-ET to uncover an alternative antibody-mediated viral disruption mechanism, which involves a single amino acid alteration in the RV VP4 spike following antibody 7H13 binding, triggering a significant structural destabilization in VP4. By employing a time-resolved cryo-EM approach and structure-guided mutagenesis, we resolved a series of intermediate state RV:7H13 complexes, depicting an intricate neutralization mechanism whereby 7H13 asymmetrically and sequentially targets two sides of VP4, with right-7H13 changing the conformation of a critical residue, VP5*A F418, through steric collisions, thereby instigating spike disruption. 7H13 was proposed to induce local structural instability around the key hydrophobic amino acid at the juncture of three VP5* subunits, propagating throughout VP4 in a manner reminiscent of a ‘butterfly effect’. This finding not only unveils a powerful antiviral mechanism but also accentuates the intricate equilibrium between viral evolution and infectivity. While these results highlight F418 as a structural vulnerability critical to 7H13-induced destabilization, further validation through mutagenesis of F418 would be instrumental in confirming its role in both viral infectivity and neutralization.

We speculated that antibodies, presumed to exert destructive effects through spatial collisions, operate via a mechanism akin to 7H13. They specifically target the virus’s vulnerable regions, triggering either localized or comprehensive structural destabilization. This notion is bolstered by prior findings, which suggests that antibodies capable of disrupting viral components generally bind to cryptic epitopes—sites typically shielded from the immune system. These epitopes become intermittently exposed due to the dynamic ‘breathing’ of viral capsids, characterized by reversible structural changes such as the arrangement of glycoproteins in enveloped virus or the entire capsid in non-enveloped viruses24,41,42, thereby facilitating antibody binding. Engagement with these regions may effectively immobilize the protein in a precarious state, thereby inducing structural disruption. This mechanism mirrors observations with antibodies targeting SARS-CoV-2 and norovirus. However, our research uncovered that 7H13 targets a distinctively virus-vulnerable site—a single residue F418—situated at critical junctures between subunit interfaces, pivotal for maintaining the trimeric stability of viral spike. This revelation highlights a previously underappreciated vulnerability within viral structures: conserved strategic points at interfaces between distinct subunits or domains within proteins critical for infection. Realizing this presents new possibilities for universal antiviral drug development, with a focus on targeting these conserved inter-subunit interfaces and identifying the vulnerability. An additional advantage of this strategy is the potential for antibody-induced conformational changes at inter-subunit junctures to obscure the antibody’s binding site, facilitating antibody dissociation and thus enabling their redeployment against other viral particles. 7H13 exemplifies this effect, achieving a progressively enhanced neutralizing impact, which implies that even at modest levels, these antibodies can exert robust neutralization. Considering the necessity of developing broadly-neutralizing RV vaccines due to the diversity of genotypes and their evolution over time43, our findings pave the way for innovative vaccine strategies. This includes using de novo protein design to create immunogens that mimic 7H13’s mechanism, offering enhanced protection against diverse and emerging RV strains.

In summary, our studies uncover a unique broadly neutralizing mechanism against RV, where the 7H13 antibody effectively disrupts the viral spike VP4 by acting on a single residue switch situated at the conserved interface among multiple VP4 subunits. This finding not only advances our understanding of RV but may also extend to other viruses, as similar virus-disrupting antibodies have been identified in a range of pathogens, including SARS-CoV-211,44,45 and HIV46,47. The use of time-resolved cryo-EM strategy in determining the intermediate states of RV:7H13 immune-complex provides robust methodology for exploring such antibody-mediated virus destabilization process. The investigation into the bNAb 7H13 and its neutralization mechanism offers a glimpse into an innovative aspect of the intricate dance between viral pathogens and their corresponding antibodies, providing a foundation for developing antiviral treatments and vaccines with broad efficacy, bolstering our arsenal against a diversity of viral diseases.

Methods

Cell lines

MA-104 (ATCC, Cat. #CRL2378.1), 293 T (ATCC, Cat. #CRL11268) and HT-29 cells (ATCC, Cat. #HTB-38) were maintained as monolayer cultures at 37 °C in Dulbecco’s modified Eagles’ medium (DMEM) (Sigma Aldrich, Cat. #D6429) with 10% fetal bovine serum (Gibco, Cat. #10099141). Expi293F (Invitrogen, Cat. #R790-07) cells were maintained in FreeStyle 293 expression medium (Gibco, Cat. #12338026). ExpiCHO (Thermo Fisher Scientific, Cat. #A29127) cells were maintained in ExpiCHO expression medium (Thermo Fisher Scientific, Cat. #A2910002).

Rotavirus isolates

The murine RV EDIM was kindly presented by the Institute of Pathogen Biology at Chinese Academy of Medical Sciences (Beijing, China) and was propagated in suckling mice. The porcine RV strain PROV-2, PROV-13, GZZY was kindly provided by Professor Tian Kegong (Henan Agricultural University, China). The lamb RV strain LLR was kindly provided by Beijing Wantai Biological Pharmacy Enterprise Co. Ltd (Beijing, China). The other RV strains were obtained from ATCC (ATCC, Manassas, VA, USA) and preserved by our laboratory. Propagation and titer determination of RVs except EDIM were performed as described previously48. EDIM was propagated in sucking mice and the viral concentration was determined by qRT-PCR as described in our previous studies49. The production and purification of SA11 particles were described previously50.

Mice

Female Balb/C mice of 8 weeks old were sourced from Slac Laboratory Animal Co., Shanghai, China, and housed in a specific-pathogen-free (SPF) environment at Xiamen University. The animals were maintained in SPF facilities under a 12-h light/dark cycle with unrestricted access to food and water. The environment was controlled with relative humidity between 45% and 65% and temperature between 20 °C and 24 °C. Viral infection experiments were carried out in a BSL-2 facility following the established guidelines for laboratory animal care and use.

Recombinant VP4* antigens expression and purification

To construct the expression plasmid of SA11-VP4*, the sequence encoding the aa.26-476 of SA11 VP4 (GenBank Accession #NC_011510), with a 6×His tag fused to the C-terminus was synthesized in Generalbio (Anhui, China), and cloned into the pTO-T7 vector developed in our laboratory49. The pTO-T7-VP4*-Gam plasmid was constructed by inserting the gene sequence encoding Acid Tolerant Monomeric GFP (Gamillus)51 to the N-terminus of VP4* (aa.26-476). The proteins were expressed in E.coli Bl21 (DE3) and purified by Ni-NTA (TransGen Biotech, Cat. #DP101-02) affinity chromatography as previously described28.

Antibody and fab preparation

The NAb 7H13 was produced in mouse ascites fluids and was affinity-purified with a Protein A column. The irrelevant SARS-CoV-2-specific mAb 5C6 and A2H4 were prepared in our lab by single B cell sequencing and mouse hybridoma techniques, respectively. To obtain the Fab fragments, the mAbs were digested with papain (Sigma Aldrich, Cat. #P4762) at a weight ratio 800:1 in 20 mM phosphate buffer (pH 7.0) containing 30 mM Cys and 50 mM EDTA overnight at 37 °C, then 30 mM iodoacetamide were added to quench the reaction. The resulting Fab fragment was then purified using Protein A agarose column (GenScript, Cat. #L00210) to remove the Fc fragment and residual full-length mAbs.

Cloning and expression of antibody

To extract variable region-coding sequences of 7H13, hybridoma cells were lysed. mRNA was reverse transcribed into cDNA with the following primers:5’- CCCAAGCTTCCAGGGRCCARKGGATARACIGRTGG-3’ for reverse transcription of the heavy chain variable region gene and 5’- CCCAAGCTTACTGGATGGTGGGAAGATGGA-3’ for reverse transcription of the light chain variable region gene. The PCR products were sent to Shanghai Bioengineering Co., Ltd for sequencing. The sequences of the antibody obtained were analyzed using an online analysis tool (IMGT; http://www.imgt.org).

To express C7H13 and 7H13 I54-mutants, the variable genes of antibody heavy and light chains were inserted into a pTT5 vector (Youbio, Cat. #VT2202) containing the constant region of human IgG1 heavy chain or light chain, respectively, via gene synthesis (Generalbio, China). Recombinant C7H13 antibodies were expressed in ExpiCHO cells (Thermo Fisher Scientific, Cat. #A29127), while 7H13 I54-mutants were expressed in Expi293F cells. Both were produced through transient co-transfection of equal amounts of heavy and light chain expression plasmids and purified from cell supernatants using Protein A agarose. Similarly, to express m41#, the variable gene regions of 41# antibody were inserted into the pTT5 vector containing the constant region of mouse IgG1 heavy chain or light chain. Recombinant m41# antibodies were expressed in ExpiCHO cells and purified from the supernatants using the same procedure as described for C7H13.

Neutralization assay

Micro-neutralization (MN) assays were carried out as described previously48. Briefly, serially diluted RV-specific mAbs were incubated with indicated RV strains at 37 °C for 1 h. The mixtures were then added to cells in 96-well plates (Nunc) for 14 h incubation. The cells were fixed with 100 μL per well of 0.1% (W/V) glutaraldehyde for 1 h at room temperature. The glutaraldehyde was removed and 0.3% Triton X-100 was added to each well for a permeabilization period of 30 min. HRP-conjugated anti-RV VP6 mAb, 2A9-HRP, was used for detection. After incubation for 1 h, plates were washed three times with PBST (0.05% Tween 20 in PBS). Then, a tetramethylbenzidine (TMB) substrate solution was added to the plates and incubated for 10 min. After patting dry, plates were scanned and blue-colored spots were counted by ImmunoSpot® 5S UV Analyzers (Cellular Technology Limited, Shaker Heights, OH, USA).

For the pre-attachment neutralizing assays, virus and mAb mixtures were incubated at 37 °C for 1 h, then added to MA-104 cells and incubated for 14 h at 37 °C for detection. In the post-attachment neutralizing assays, virus was first adsorbed onto cells at 4 °C for 1 h, followed by incubation with antibodies at 4 °C for 2 h to ensure binding. The mixture was then incubated at 37 °C for 14 h for detection. For the pre-incubation neutralization assays, virus and mAb mixtures were incubated for 1 h, 4 h, 5 h and 6 h at 37 °C before addition to the MA-104 cells.

Antibody blocking fluorescent protein adsorption experiment

MA-104 cells were seeded at a concentration of 2×105 cells/mL in 24-well plates and incubated at 37 °C for 20 h. Fourteen different RV strains infected the cells and incubated at 37 °C for 14 h. Then the cells were fixed using 4% paraformaldehyde for 15 min at room temperature. After washing three times, 0.3% Triton X-100 was added to each well for a permeabilization period of 10 min. Blocking was done by incubation with 10% goat serum at 37 °C for 1 h. Subsequently, the mAbs (2 μg/mL) were added and incubated at 37 °C for 1 h. Then, the cells were incubated with goat anti-human (Thermo Fisher Scientific, Cat. #A-11013) or donkey anti-mouse IgG (H + L) Cross-adsorbed secondary antibody, Alexa Fluor 488 (Thermo Fisher Scientific, Cat. #A-21202), diluted in 1:5000, at 37 °C for 1 h. After washing, the cells were observed under an EVOS M7000 microscope.

For the identification of mAb recognition epitopes through immunofluorescence experiments, eukaryotic expression plasmids of SA11-VP4, VP5, and VP8 were transfected into 293 T cells to detect the binding activity of mAbs with the transfected proteins. Anti-His FITC was employed to verify the expression of the transfected genes.

In vivo prevention experiment

Female Balb/C mice of 8 weeks old received tail vein injections of mAb at a dose of 100 μg per mouse. Each experimental group consisted of five mice. Twelve hours post-injection, the mice were fasted for one hour before administration of 500 μL 10% bicarbonate solution via gavage to neutralize gastric acid. Following this pretreatment, each mouse was then challenged by 500 μL 5×107 copies/mL RV EDIM (G16P[16]) by gavage, and fecal samples were collected daily from day 1 to 7 post-gavage.

Detection of rotavirus shedding in stools

Fecal samples were prepared into 10% fecal suspension (w/v) using PBS, and virus RNA was subsequently extracted using the virus DNA/RNA kit (GenMag Biotechnology Co., Ltd., Beijing, China). Then the extracted RNA was detected using a homemade universal NSP5 system as described previously49. RV shedding was quantified according to the standard curves. The assay’s lower detection limit was 10,000 copies/mL, and for the samples with no amplification curves, the virus load was set as 1000 copies/mL for data analysis.

Indirect Blocking ELISA

Indirect blocking ELISA was conducted to characterize the epitope recognized by the NAbs. Briefly, the blocking NAbs and control antibodies were diluted to the concentration of 100 μg/mL, added to VP4-coated microplates, and incubated at 37 °C for 30 min. Following the wash to remove non-bound antibodies, the blocked NAbs were diluted to a proper concentration (OD450/630 = 0.5–1) and added to the microplate, which were incubated for another 30 minutes at 37 °C. The microplates were then incubated with HRP-conjugated goat anti-mouse (Abcam, Cat. #ab97265) or anti-human IgG (Abcam, Cat. #ab97225) for detecting the binding of the blocked mAbs. Post-TMB color development and termination, the absorbance at 450 nm, using 630 nm as reference, was measured with an automated ELISA reader (AutoBio, China). The Blocking rate (%) was calculated using the formula:

$$b=\left(1-\frac{{Ob}}{{Ou}}\right)\times 100$$

where b is the blocking rate, Ob is the OD value of the blocked well, and Ou is the OD value of unblocked well.

Rotavirus SA11 propagation and purification

The SA11 virus was cultured using a two-layer cell factory (FeiFanT, China). MA-104 cells were cultivated to a density of 90–95% before adding the trypsin-activated virus (MOI = 0.1). The cytopathic effect (CPE) on the cells was monitored daily, and the culture was harvested once the CPE reaches ~80%.

The purification of RV was performed as reported previously50. Cells were disrupted by three cycles of freeze-thawing and cell debris removed by centrifugation at 7000 × g for 15 min. Supernatants were filtered through a 0.22 μm filter (Millipore, USA) and concentrated to around 200 mL by tangential flow filtration (PALL, USA). The viruses were further concentrated through a 30% sucrose cushion at 103,560 × g for 3 h using SW28 rotor (Optima L-90K, Beckman Colter, USA). The pellets were resuspended in TNC buffer (50 mM Tris-HCl, 0.15 M NaCl, 10 mM CaCl2, pH 7.5) on ice overnight. CsCl was added and adjusted to a final density of refractive index 1.369 before the samples were load into the rotor. After centrifugation at 207,200 × g for 24 h using a SW41 Ti rotor (Optima L-90K, Beckman Colter, USA), virus triple-layered particle (TLP) and double-layered particle (DLP) were separately collected and resuspended in TNC buffer. Virus particles were further purified by ultracentrifugation at 207,200 × g for 3 h using a SW60 Ti rotor (Optima L-90K, Beckman Colter, USA) and the final virus particle were resuspended in TNC buffer.

HPSEC

The purified samples of RV, RV:Fab complexes, and Fab fragments were analyzed using a Waters E2695 High-Performance Liquid Chromatography (HPLC) system (Waters Associates, Massachusetts, USA) equipped with a TSK Gel G5000PWXL analytical column (TOSOH, Tokyo, Japan) to evaluate their homogeneity. Protein presence in the eluate was detected by monitoring absorbance at 280 nm. HPSEC data were then processed and analyzed using Adobe Illustrator CS5 software.

Negative-staining electron microscopy

The purified RV TLPs were diluted in TNC buffer and then absorbed onto 200 mesh carbon-coated copper grids for 1 min. The grids were washed twice with double-distilled water and negatively stained with 2% phosphotungstic acid (pH 6.4) for 30 sec. Specimens were evaluated and imaged with the Thermo Fisher Tecnai T12 electron microscopy at 25,000× magnification.

Cryo-EM sample preparation and data acquisition

Aliquots (3 μL) of the purified RV SA11, along with its mixtures incubated with various concentrations of 7H13 Fab at differing temperatures, were loaded onto low-discharge (60 s at 20 mA) holey carbon Quantifoil grids (R1/2, 200 mesh). Grid preparation was carried out using a Vitrobot Mark IV (Thermo Fisher Scientific) set to 100% humidity and maintained at 4 °C. Image acquisition was conducted on two transmission electron microscopes (TEMs): 300 kV Tecnai G2 F30 (Thermo Fisher Scientific) and Titan Krios G4 (Thermo Fisher Scientific), utilizing SerialEM52,53 and EPU software (Thermo Fisher Scientific), respectively. For the Tecnai G2 F30 TEM, images were captured with a Gatan K3 direct electron detector in 40-frame movie mode at a nominal magnification of 31,000x, resulting in a physical pixel size of 1.0 Å (0.5 Å in super-resolution mode). Additional image datasets were collected on the Titan Krios equipped with a Gatan K3 and a Gatan BioContinuum HD Imaging Filter operated in zero-loss mode with a slit width of 20 eV, at a nominal magnification of 81,000x, which corresponds to a physical pixel size of 1.098 Å (0.549 Å in super-resolution mode).

Image processing and 3D reconstruction

Frames of each movie were aligned and averaged using MotionCor254 and the contrast transfer function (CTF) parameters were ascertained with Gctf 55. Micrographs with astigmatism, excessive drift, or contamination were discarded before reconstruction. Particles were automatically picked using the ‘Template picker’ in CryoSPARC V356. Several rounds of reference-free 2D classification were performed, and selected particles were subjected to homogeneous refinement using cisTEM 1.0.057—except SA11:7H13-I54G dataset, which the 3D refinement was processed in CryoSPARC. Sub-particles focusing on the viral spike were extracted, using box sizes of 360 pixels, and for the SA11:7H13-I54G dataset, 320 pixels. These underwent several rounds of 3D classification and refinement via Relion 3.158 and CryoSPARC. To obtain the high-resolution structure of each target (VP4, VP4:7H13, VP4:7H13-1/10, and VP4:7H13-I54G), we began with 3D classification using either a spherical mask of 240 pixels in diameter (SA11 and SA11:7H13 datasets) or a target-specific mask (SA11:7H13-1/10 and SA11:7H13-I54G datasets) to eliminate sub-particles without VP4 density. A two-step local refinement was then conducted in CryoSPARC, refining particles first without a mask, and then with a mask encompassing only the target areas. The refined sub-particles were re-imported into Relion for multiple rounds of localized 3D classification with a high tau value, allowing for the discarding of any suboptimal particles. Subsequently, sub-particles from the class with well-defined density underwent further local refinement in CryoSPARC. For the SA11:7H13-I54G data, distinct masks targeting the left- and right-7H13 were applied in the final 3D classification and refinement stages. The resolution of all density maps was determined by the gold-standard Fourier shell correlation curve, with a cutoff of 0.14358. Local map resolution was estimated with ResMap59.

Atomic model building, refinement, and 3D visualization

The cryo-EM structure of RV strain RRV capsid (PDB: 4V7Q60) served as a homology model for generating the initial models for this study. The initial atomic model for the variable domains of 7H13 Fab fragment and its I54G variant was generated by homology modeling using Accelrys Discovery Studio (DS) 2017 R2 software (https://www.3dsbiovia.com). Templates were initially fitted into the corresponding final cryo-EM maps using ChimeraX 1.461, followed by manual correction and adjustment through real-space refinement in Coot 0.9.662. The resulting models were refined with phenix.real_space_refine in PHENIX 1.19.263. Model statistics, including bond lengths, bond angles, all-atom clashes, rotamer statistics, Ramachandran plot statistics, etc., were closely inspected with Coot during the entire process. Final atomic models were validated using Molprobity64,65. The buried surface areas and the interactions were analyzed using PISA server (https://www.ebi.ac.uk/pdbe/pisa/) and the DS. All figures were generated using ChimeraX.

Cryo-electron tomography

The cryo-electron tomography data were collected using the Titan Krios TEM (Thermo Fisher Scientific) operated at 300 kV and equipped with a cold field emission gun, a Biocontinuum HD Imaging Filter (Gatan), and a K3 direct detection camera (Gatan). The data acquisition was facilitated by the Tomography 5 software (Thermo Fisher Scientific). Prior to capturing the tilt-series, montages of the grid squares were acquired at a magnification of 6500× with a defocus of −20 μm to localize RV virions. Tilt series were acquired in super-resolution mode at a nominal magnification of 81,000×, with a calibrated pixel size of 1.098 Å. A dose-symmetric acquisition scheme was employed with a tilt range from −60° to 60° with a 3° increment, and a target defocus ranging from −3 to −6 μm. Micrographs were divided into 10 frames with an exposure time of 0.26 seconds, resulting in a constant electron dose of ~3 e−/Å2, and the total accumulated electron dose for 41 tilt series was ~ 123 e−/Å2.

Tomogram reconstruction and volume rendering

Motion correction and CTF estimation of movies were performed using Warp 1.0.966. The tilt series were aligned in Imod 4.1167, using gold beads as fiducial marker. The tomograms were reconstructed using weighted back-projection (WBP) and WBP with a Simultaneous Iterative Reconstruction Technique (SIRT)-like filter equivalent to 10 iterations in Imod. The WBP tomograms were binned 4× and filtered using tomo_deconv (https://github.com/dtegunov/tom_deconv) to enhance the contrast, then visualized in Amira. The SIRT-like filtered tomograms were binned 12× and subjected to IsoNet68 for denoising and missing-wedge correction, then segmented and visualized in ChimeraX.

Sequence logos

21,259 sequences of RV VP4 were obtained from NCBI server. Only those covering VP5 domain (6713 sequences) were retained and subjected to Muscle69 for multiple sequence alignment. The Sequence logos for the 7H13 epitopes were generated by the application WebLogo (https://weblogo.threeplusone.com/)70. Default parameters of all bioinformatic tools were used for the analysis.

Ethics statement

All virus-related experiments were conducted in a biosafety level 2 (BSL-2) laboratory. All in vivo studies were performed in accordance with Institutional Animal Care and Use Committee guidelines and were approved by the Ethics Committee of Xiamen University Laboratory Animal Center.

Statistics and reproducibility

The SDS-PAGE analysis in Fig. 2l was successfully reproduced in three independent experiments. We acquired 6 tomograms for purified SA11 and 20 tomograms for SA11:7H13 complex (Table S3) from multiple frozen grids imaged over several independent sessions. The representative tomographic slices shown in Fig. 3a, f were selected from one tomogram for each condition (SA11 and SA11:7H13 complex), consistent with the features observed across the datasets. Similarly, the representative cryo-EM micrographs in Figure S7 and Figure S10 were selected from their corresponding datasets (Table S1).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.