Capsid Assembly is Regulated by Amino Acid Residues Asparagine 47 and 48 in The VP2 Protein of Porcine Parvovirus

Background: Porcine parvovirus (PPV) is a major cause of reproductive failure in swine, and has caused huge losses throughout the world. Viral protein 2 (VP2) of PPV is a major structural protein that can self-assemble into virus-like particles (VLP) with hemagglutination (HA) activity. In order to identify the essential residues involved in the mechanism of capsid assembly and to further understand the function of HA, we analyzed a series of deletion mutants and site-directed mutations within the N-terminal of VP2 in the Escherichia coli (E. coli) system. Results: Our results showed that deletion of rst 47 amino acids from the N-terminal of VP2 protein did not affect capsid assembly, and further truncation to residue 48 Asparagine (Asn, N) caused detrimental effects. Site-directed mutagenesis experiments demonstrated that residue 47Asn reduced the assembly eciency of PPV VLP, while residue 48Asn destroyed the stability, hemagglutination, and self-assembly characteristics of the PPV VP2 protein. These ndings indicated that the residues 47Asn and 48Asn are important amino acid sites to capsid assembly and HA activity. Results from Native PAGE inferred that macromolecular polymers were critical intermediates of the VP2 protein during the capsid assembly process. Site-directed mutation at 48Asn did not affect the association of monomers to form into oligomers, but destroyed the ability of oligomers to assemble into macromolecular particles, inuencing both capsid assembly and HA activity. Conclusions: These results demonstrated that PPV capsid assembly is a complex process that is regulated by amino acids 47Asn and 48Asn, which are located at the N-terminal of VP2 and closely related to the association of macromolecular particles. Our ndings provide valuable information on the mechanisms of PPV capsid assembly and the possibility of chimeric VLP vaccine development by replacing as much as 47 amino acids at the N-terminal of VP2 protein.

surrounding the icosahedral three-fold axes of parvovirus are related to host-receptor recognition, cell tropism, hemagglutination (HA) activity, and antigenicity [11,12]. Amino acid residues 93, 103, or 323, located in structural regions near the three-fold "spike" of the capsid, are responsible for the host range differences between canine parvoviruses and feline panleukopenia viruses [13]. Amino acids that determine tissue tropism were found to cluster on the viral surface, such as amino acid residues 317 or 321 of VP2 in minute virus of mice (MVM), and residues 381, 386, and 436 in PPV [8,14]. Apart from its important roles in determining tropism and host range, the capsid also contains major antigenic sites, hemagglutination sites, lack infectious virus and are inherently safer comparable to those of the native virion [15,16]. These bene cial characteristics make PPV VLP as the best vaccine candidate, which expressed in many different systems [17][18][19]. Recent years, VLP attracts broad scienti c interest due to their properties as versatile scaffold in nanotechnology, such as potential therapeutic agents, targeted drug delivery, novel vaccines and so on [20][21][22][23]. PPV VLP, excepting originated advantages from their particle integrity, its simple structure makes it as an ideal platform in biomedicine to study the mechanisms of capsid assembly and virus infection, as well as the development of new vaccines [24,25].
Virus self-assembly and virion morphogenesis are regulated by different types of protein subunits, such as monomers, dimers, trimers, pentamers, or hexamers. Both the assembly and stability are protected from the formation of multiple non-covalent interactions between protein subunits [26,27]. Trimers, as the intermediates, may exert morphogenetic control on the icosahedral parvovirus capsids during the assembly of MVM [28,29]. In polyomavirus, pentamers of VP1, the major capsid protein, associated with VP2 and VP3 complexes within the cytoplasm were shown to be responsible for cooperative nuclear translocation [30]. Key amino acid residues or most non-covalent interactions between protein subunits, such as hydrophobic, hydrogen bonds or salt bridges, may be critical for the assembly and stability of virions [31,32]. As intermediates and intermolecular interactions play an important role in the function of virus capsid, it is very meaningful to further study of this work. Knowledge of capsid assembly, stability, and dynamics of PPV virions is essential to understand the virus life cycle and promote the development of antiviral vaccines and antiviral agents based on inhibiting the assembly and uncoating process.
In order to identify the functional domains of the VP2 protein involved in capsid assembly and HA activity, we constructed ten deletion mutants and three site-directed mutants within the N-terminal domain of VP2 in the Escherichia coli (E. coli) system. Our results demonstrated that truncation of the Nterminal of VP2 at amino acid residue 47 did not affected VLP assembly and HA activity. Further truncation to residue 48 Asparagine (Asn, N) caused detrimental effects. Site-directed mutagenesis demonstrated that residues 47Asn-48Asn were the key amino acid sites controlling VLP assembly and HA activity. Our results also indicated that mutation of 48Asn did not affect the formation of monomers into oligomers, but prevented the assembly of oligomers into macromolecular particles, which eventually destroyed VLP formation.

Results
Expression of mutant VP2 proteins in E. coli cells In order to screen the residues involved in HA and assembly features of PPV capsids, we designed and expressed a series of deletion mutants and site-directed mutants of the VP2 protein in E. coli (Fig. 1). First, three N-terminal deletion mutants, with 30, 50, and 100 amino acid residues of the VP2 protein removed (namely NT-∆30aa, NT-∆50aa, NT-∆100aa, respectively), were constructed. Only the NT-∆30aa protein was successfully expressed in soluble form in E. coli and exhibited HA activity, reaching 2 16 in vitro. However, the NT-∆50aa and NT-∆100aa mutant proteins were insoluble, and none of them could agglutinate guinea pig erythrocytes (Table 2 and Fig. 2). Subsequently, the NT-∆35aa, NT-∆40aa, NT-∆45aa, NT-∆46aa, NT-∆47aa, NT-∆48aa, and NT-∆49aa mutants were expressed as soluble forms in the supernatant of E. coli cells (Fig. 2). Among these soluble mutant VP2 proteins, NT-∆30aa, NT-∆35aa, NT-∆40aa, NT-∆45aa, NT-∆46aa, and NT-∆47aa exhibited HA activity, with the HA titers reaching 2 16 , 2 16 , 2 16 , 2 15 , 2 16 , and 2 10 , respectively. Both NT-∆48aa and NT-∆49aa did not have HA activity ( Table 2), indicating that the rst 47 amino acid residues were responsible for the HA activity of PPV capsids. Interestingly, both 47 and 48 amino acid residues were asparagine. Next, we then designed and constructed another three mutants, N47K, N48K, and N47KN48K, using site-directed mutagenesis, where we replaced 47Asn and/or 48Asn amino acid with lysine (Lys, K) in the VP2 protein ( Fig. 1). As shown in Fig. 3a, three mutants were expressed as soluble forms. The results from the HA test showed that only the N47K mutant exhibited HA activity, with the titers of 2 14 , and the N48K and N47KN48K mutants showed no HA activity (Fig. 3b). These results further indicated that the 47Asn and 48Asn residues are crucial amino acids that play an important role in the structure and function of PPV capsids.
Puri cation of mutant VP2 proteins Five mutant VP2 proteins-NT-∆30aa, NT-∆47aa, NT-∆48aa, N47K, and N48K-and the native VP2 protein were chosen for further analyses. Fractions containing native VP2 or mutant VP2 proteins were eluted with 50 mM Tris containing 150 mM NaCl and 150 mM imidazole through Ni-NTA a nity chromatography. The eluates were then monitored continuously by UV-absorption at 280 nm, and the active fractions were recovered through size-exclusion high-performance liquid chromatography (SE-HPLC). Our SDS-PAGE analysis results showed that the purity of these proteins was over 90% (Figs. 4a-4f). As shown in the chromatogram, mutant VP2 and native VP2 protein had only one main peak, whereas the N47K protein had two main peaks (Fig. 4). As compared to the native VP2 protein, the main peak of the VP2 deletion mutants was observed at about 8.5 minutes, except for NT-∆48aa, which was observed at 7.3 minutes ( Fig. 4g and Table 3). As for the two site-directed mutants, the main peak of N48K, which appeared at around 7.5 minutes, was earlier than that of N47K and native VP2 ( Fig. 3h and Table 3). Moreover, the main peak of NT-∆48aa was in accord with N48K, which was almost at 7.3 minutes-1 minute earlier than that of the native VP2 protein ( Fig. 4i and Table 3). Detailed information regarding the elution times of the VP2 mutant proteins is summarized in Table 3.
Identi cation of key amino acid residues in assembly of PPV capsids Self-assembly properties of mutant VP2 proteins in vitro were con rmed by transmission electron microscopy (TEM) and dynamic light scattering assay (DLS). Results from TEM showed that regular VLP were observed in the native VP2 protein and NT-∆30aa, NT-∆47aa, and N47K, but not in NT-∆48aa and N48K (Fig. 5). NT-∆48aa and N48K particles showed some amount of aggregation or multimeric structure with completely irregular particles, which differed morphologically from regular VLP (Figs. 5d and 5f). Consistently, results from DLS demonstrated that the native VP2 protein, NT-∆30aa, NT-∆47aa, and N47K self-assembled into particles in solution, with an average diameter of 25 nm, but NT-∆48aa and N48K existed as non-uniform particle sizes, as indicated in Fig. 6 and Table 3. Additionally, we further investigated whether the N-terminal deletion and site-directed mutant proteins retained the ability to react with the conformational monoclonal antibody 10A9, which only recognizes assembled capsids. As shown in Fig. 7, NT-∆30aa, NT-∆47aa, N47K (peak 1), and full-length VP2 reacted with the conformational monoclonal antibody 10A9 with titers above 1:640,000, but N47K (peak 2), NT-∆48aa, and N48K did not. These results indicated that 47Asn and 48Asn are key amino acid residues of the VP2 protein, which have a signi cance effect on capsid assembly.

Analysis of mutant VP2 proteins by sucrose gradient sedimentation
Sucrose gradient sedimentation was conducted to analyze the stability and aggregation states of mutant VP2 proteins. As shown in Fig. 8a, the native VP2 protein was detected mainly in fraction 14, with one peak in the 45% sucrose gradient. The NT-∆30aa mutant behaved similarly to the native VP2 protein, and particles were observed primarily in the 45% sucrose gradient. This suggested that, similar to the fulllength VP2 protein, the NT-∆30aa protein was assembled into VLP with high stability (Fig. 8b). As compared with the full-length VP2 protein, numerous unassembled VP2 proteins in the NT-∆47aa, NT-∆48aa, N47K, and N48K mutants accumulated and migrated as different protein subunits within the 5-45% sucrose gradient (Figs. 8c-8f). These results implied that the 47Asn and 48Asn residues affected the stability of the VLP structure, resulting in a large amount of intermediate aggregation, which reduced assembly e ciency and VLP formation.
Native PAGE was employed to estimate and compare unassembled subunits or intermediates in native VP2 and mutant N48K protein during the VLP formation process. As shown in Fig Fig. 9a). When amino acid residue 48 was mutated from an Asn to Lys, numerous unassembled proteins were detected in the N48K mutant, as indicated in Fig. 9b. Three new unassembled subunits in the N48K mutant, f at 720 kDa, h 480 kDa (upper), and i at 242 kDa, were observed, excepting two unassembled subunits g at about 720 kDa (below) and j at about 146 KDa, located in the same position with c and e intermediates in VP2 protein (Lane 1 in Fig. 9). Importantly, these ve unassembled subunits of N48K accumulated and migrated as major proteins in different sucrose layers (Fig. 9b). Notably, macromolecular polymers a and b were not detected in the N48K mutant (lane 1 and 14 in Fig. 9). The results from liquid chromatography-tandem mass spectrometry (LC-MSMS) analysis showed that these intermediates (a-j) were all VP2 or mutant VP2 proteins, which represented different unassembled VP2 subunits during the VLP formation process (Fig. S1). Based on the results described above, we preliminary inferred that macromolecular polymers a (at 1236 kDa) and b (at 1048 kDa) may be critical intermediates of the VP2 protein, and site-directed mutation at the 48Asn amino acid impeded aggregation of the VP2 macromolecular particles, which is an important factor that rendered the VP2 protein defective in VLP formation during the capsid assembly process.
Discussion PPV VP2 has both antigenicity and immunogenicity and can self-assemble into VLP in vitro. These VLP do not contain genetic material as found in the native virion, which makes these VLP suitable for studying viral packing and assembly, vaccine design, and biological nanomaterials. In our previous study, we investigated the HA and self-assembly characteristics of full-length VP2 protein capsids in vitro, and VP2 VLP showed a prominent advantage in immunogenicity and immunoprotection [19,33]. However, the mechanisms of self-assembly of PPV capsids is still unclear. In this study, we constructed a series of mutants with N-terminal deletions and site-directed mutations in the full-length VP2 protein in order to identify regions that regulated VLP formation and HA activity. Our results demonstrated that PPV capsid assembly is a complex process that is regulated by amino acids 47Asn and 48Asn, which are located at the N-terminal of VP2 and closely related to the association of macromolecular particles.
Structural proteins, as well as their critical residues, play an irreplaceable role in protein self-association and capsid assembly. Deletion of ve C-terminal residues or merely glutamic acid 257 of VP3 in the infectious bursal disease virus was found to promote capsid assembly [34]. Deletion of 27 residues or less from the N-terminal of the VP2 protein in porcine circovirus type 2 resulted in a similar icosahedral shape and immunogenicity as compared with the full-length VLP [35]. Additionally, deleting up to 125 amino acid residues from the N-terminal region of the capsid protein and truncating the C-terminal of the capsid region to 601 amino acid residues in hepatitis E virus was conducive for the self-assembly of viral particles [36]. In order to locate the key domains that may impact HA activity and assembly features of PPV capsids, ten mutant VP2 proteins were expressed in an E. coli system. As shown in Fig. 2, mutant proteins were expressed in the supernatant of E. coli cells, except NT-∆100aa and NT-∆50aa proteins, which aggregated as inclusion bodies. However, among these soluble mutant VP2 proteins, six mutants, NT-∆47aa, NT-∆46aa, NT-∆45aa, NT-∆40aa, NT-∆35aa, and NT-∆30aa, had HA activity; mutants NT-∆49aa and NT-∆48aa did not have HA activity (Table 2). To investigate the self-assembly and biological functions of the N-terminal deletion mutants, three mutants, NT-∆30aa, NT-∆47aa, and NT-∆48aa, were chosen for further study. Results from TEM and DLS analysis con rmed that typical VLP formed in the NT-∆30aa and NT-∆47aa mutants, with a particle size about 25 nm, but not in NT-∆48aa (Figs. 5-6 and Table 3). Results from the indirect enzyme-linked immunosorbent assay (ELISA) showed that NT-∆30aa, NT-∆47aa, and VP2 were recognized by the conformational monoclonal antibody 10A9, with titers above 1:640,000. However, NT-∆48aa was not recognized by the antibody 10A9, which only recognizes assembled VP2 capsids, implying NT-∆48aa did not properly assemble. The results described above suggested that when the 48Asn amino acid was removed from the VP2 protein, the protein lost the ability to hemagglutinate guinea pig erythrocytes and was unable to self-assemble into VLP, indicating that 47Asn and 48Asn are the critical residues involved in the function of HA activity and capsid assembly during the VLP formation process.
At physiological conditions, homogeneous VLP assembled, while at acidic or basic pHs, with low ionic strength, the major assemblies were small intermediates [37]. It has been reported that residues involved in hydrogen bonds, hydrophobic interactions, and salt bridges or heterologous peptide insertions may have an effect on the stability and assembly of viral proteins [25,31,38].Site-directed mutations in capsid protein has proved to modify the properties or function of virus capsids [39][40][41]. In this study, deleting the rst 47 amino acid residues within the N-terminal region of the VP2 capsid protein had no signi cant effect on HA and self-assembly characteristics of PPV capsids. However, deletion of amino acid 48 resulted in loss of HA activity and VLP formation. Considering that amino acids 47 and 48 are both asparagine, three mutants (N47K, N48K, and N47KN48K) were designed and constructed by replacing 47Asn and/or 48Asn amino acid with lysine in the VP2 protein. As shown in Fig. 3a, the mutants were expressed as soluble proteins in E. coli. Results from the HA assay indicated that only the N47K mutant exhibited HA activity, with 2 14 titers (Fig. 3b). The ability of puri ed N47K and N48K to self-assemble into VLP in vitro was assessed by TEM, DLS, and ELISA assays. Results from TEM and DLS revealed that VLP were observed in the N47K mutant, but not in the N48K mutant (Figs. 5-6). As shown in Fig. 7, the N48K mutant was not recognized by the conformational monoclonal antibody 10A9, which only reacts with assembled capsids. Site-directed mutations or deletion of amino acids within the N-terminal of the VP2 protein changed the morphology and function of PPV VLP. Sucrose gradient sedimentation was employed to analyze the effects of mutations on the formation and stability of PPV capsids during the VLP formation process. Notably, results from sucrose gradient sedimentation showed that the mutant VP2 proteins were distributed within the 5-45% sucrose, implying that there were unassembled protein subunits that accumulated, as seen in the NT-∆47aa, NT-∆48aa, N47K and N48K mutants, but not the NT-∆30aa mutant and native VP2 protein (Fig. 8). This suggested that 47Asn-48Asn residues are critical amino acids involved in the self-assembly process and stability of PPV VLP. Residue 47Asn reduced the assembly e ciency of PPV VLP, while residue 48Asn destroyed the stability, hemagglutination, and selfassembly characteristics of the PPV VP2 protein. We preferred to deem that because the stability structure of PPV VLP was destroyed, and hence made VP2 capsids lost its HA activity. However, further research needs to be performed in order to better understand the underlying mechanism.
Virus capsids are assembled from stable forms or assembly intermediates. These structures may be soluble monomers or small oligomers (e.g., dimers, trimers, pentamers, hexamers), and can be detected in the dynamic balance of assembled-unassembled capsids, depending on the virus and conditions [26]. The trimer was con rmed as the transient intermediate existing during the assembly of MVM VLP, and capsid formation was dependent upon stronger intertrimer interactions that were equally spaced in an equatorial belt surrounding each trimer [28,29]. Present studies have shown that 48Asn residue in the VP2 protein directly impacted the structure or function of viral particles, and deletion or mutation at this site rendered the VP2 protein defective in assembling capsids. As shown in Fig. 8f, results from sucrose gradient sedimentation showed that a number of unassembled subunits existed in the N48K mutant, which might represent different oligomers or polymers formed by the VP2 protein. Native PAGE was employed to evaluate the native conditions of these unassembled subunits in the N48K mutant during the VLP formation process. Comparing the N48K mutant with the native VP2 protein, intermediates a (at about 1236 kDa) and b (at about 1048 kDa) were detected as major protein subunits in VP2 capsids, while intermediates f (at 720 kDa), g (720 kDa below), h (at 480 kDa upper), i (at 242 kDa), and j (at 146 kDa) aggregated in the N48K mutant (Fig. 9). These intermediates in the N48K mutant appeared as small oligomers, such as dimers (j at 146 kDa), trimers (i at 242 kDa), pentamers (h at 480 kDa upper), as well as a complex (f at 720 kDa and g at 720 kDa below), and accumulated in large amounts, which eventually had a signi cant negative affect on the formation of VLP. Importantly, the intermediates a and b were not observed in the N48K mutant, which were two important intermediates of VP2 protein during the VLP assembly process. We preliminarily deduced that the macromolecular particles, intermediate a located at about 1236 kDa and intermediate b at about 1048 kDa, may be formed by pentamers or trimers. Although the mechanisms of PPV capsid assembly remains unclear, our data suggested that sitedirected mutagenesis at 48Asn did not affect the association of monomers to form into oligomers, but destroyed the ability of oligomers to assemble into macromolecular particles. This resulted in adverse effects on VLP formation and HA activity.

Conclusions
In summary, we investigated the key amino acid sites within the N-terminal region of the VP2 protein, which may control and affect VLP formation and HA activity of PPV capsids. Deletion of the rst 47 amino acids within the N-terminal had no effect on VLP assembly, whereas deletion of the rst 48 amino acids resulted in loss of VLP formation. Site-directed mutagenesis experiments indicated that residue 47Asn reduced the assembly e ciency of PPV VLP, while residue 48Asn had a detrimental effect on the stability and function of PPV capsids. Further research revealed that mutation at 48Asn impeded the ability of oligomers to assemble into macromolecular particles, which eventually had a negative effect on VLP formation. These ndings provided new insights into understanding the mechanisms of PPV capsid assembly and valuable information regarding designing chimeric VLP vaccines by replacing the rst 47 amino acid residues at the N-terminal of the VP2 protein.

Construction of capsid VP2 mutants
Ten mutants with amino acids deleted from the N-terminal region of the VP2 capsid protein and three site-directed mutations in VP2 were constructed using the E. coli expression system (Fig. 1). The VP2 gene was truncated at the N-terminal by inserting PCR-ampli ed fragments anked by Nde I and Xho I restriction enzyme sites into the pET28a vector (Novagen, Malaysia) described previously [33], as indicated in Table 1. Recombinant vectors were con rmed by using enzyme digestion and DNA sequencing (Sangon Biotech, Shanghai, China). Three site-directed VP2 mutants were obtained in vitro by using the Fast Site-Directed Mutagenesis Kit (Tiangen, Beijing, China). The pET28a plasmid, containing the full-length VP2 gene (1737 bp), was ampli ed by three pairs of mutation primers (Table 1). Brie y, ampli cation reactions were carried out in a total volume of 50 µL, containing 2 µL of template DNA, 2 µL of each forward and reverse primer (10 µM), 10 µL of 5× Fast Alteration Buffer, 1 µL of Fast Alteration DNA Polymerase (1.0 U/μL), and 33 µL of RNAse-free water. Polymerase chain reaction (PCR) procedure was followed as described below. Initial denaturation was at 2 minutes at 95°C followed by 18 cycles, each cycle consisting of 20 seconds at 94°C, 10 seconds at 55°C and 2.5 minute at 68°C. The nal extension was 5 minutes at 72°C. After ampli cation, the PCR product was digested with Dpn I restriction enzyme at 37°C for 1 hour, and then 5 µL of the digestion product was transformed into FDM competent cells to screen for mutants. Desired mutants were identi ed by DNA sequencing.
Expression and puri cation of mutated VP2 protein VP2 and mutant VP2 protein in the supernatant of E. coli cells were puri ed by a two-step method, using Ni-NTA a nity chromatography (Merck, Darmstadt, Germany) followed by SE-HPLC. Clari ed cell lysates were pumped onto a Ni-NTA column balanced with 10 bed volumes of buffer A. After washing with 10 bed volumes of buffer B (150 mM NaCl, 50 mM Tris, 25 mM imidazole, pH 8.0), VP2 or mutant VP2 proteins were eluted with buffer C (150 mM NaCl, 50 mM Tris, 150 mM imidazole, pH 8.0). Fractions were further separated by SE-HPLC, using a Superdex™ 200 10/300 GL (GE Healthcare, Pittsburgh, PA, USA) column as described below. A Superdex™ 200 10/300 GL column was equilibrated with 10 bed volumes of buffer A at a rate of 1 mL/min. Subsequently, samples were applied to the Superdex™ 200 10/300 GL column. After binding, fractions were further eluted with buffer A. Elution fractions were collected and determined by 12% (v/v) SDS-PAGE. Fractions containing VP2 protein or mutant VP2 proteins were ltered with a 0.22 µm lter (Merck Millipore, Darmstadt, Germany) and quanti ed using a Micro BCA™ protein assay kit (Solarbio, Beijing, China). Importantly, all puri cation procedures were carried out at 4°C, and puri ed fractions were stored at −80°C to retain biological activity.

SDS-PAGE and western blotting analysis
SDS-PAGE and western blotting were conducted to detect protein expression as described below. First, 5× loading buffer was added to each cell lysate or fraction. Then mixtures were heated at 100°C for 10 minutes, resolved by 12% SDS-PAGE, and nally stained with Coomassie blue (Beyotime, Shanghai, China). For WB, samples were transferred from SDS-PAGE to PVDF membranes (Bio-Rad, Richmond, California, USA). VP2 or mutant VP2 proteins were detected using a horseradish peroxidase (HRP)conjugated anti-his monoclonal antibody (Jackson ImmunoResearch, Philadelphia, PA, USA) at a dilution of 1:5,000. Membranes were observed with enhanced chemiluminescent reagent (NCM Biotech, Suzhou, China) or 3-amino-9-ethylcarbazole solution (ZSGB-BIO, Beijing, China), and examined by light microscopy (Leica DMI3000B) (Feica Microsystems, Wetzlar, Germany).

HA analysis
Supernatants containing VP2 or mutant VP2 were diluted with PBS for standard HA assays. To perform a HA test, 50 μL of clari ed cell lysates containing VP2 or mutant VP2 were 2-fold diluted with PBS in a 96-V-shaped-well microtiter plate. Then 50 μL of guinea pig erythrocytes (0.8%) were added to each well. The plate was incubated at 37°C for 1 hour. HA titers were de ned as the highest dilution that completely agglutinated guinea pig erythrocytes.

Indirect enzyme-linked immunosorbent assay
An indirect enzyme-linked immunosorbent assay was carried out using the conformational antibody 10A9, which only recognizes assembled capsids, as prepared in our previous study. First, 96-well plates were coated with puri ed mutant VP2 proteins, wild-type VP2 protein, or pET 28a empty vector, and then blocked with 5% skim milk in PBST at 4°C at least 8 hours. The conformational antibody 10A9 was used as the primary antibody and added into 96-well plates at 37°C for 30 minutes. After washing with PBST for ve times, an HRP-conjugated goat anti-mouse IgG antibody (Jackson ImmunoResearch, Philadelphia, PA, USA) was used as the secondary antibody and incubated in the plate at 37°C for 45 minutes. Finally, antibody binding was analyzed by adding TMB solution (Solarbio, Beijing, China), and 2 M H 2 SO 4 was added to stop the reaction. The optical density at 450 nm was recorded using a microplate reader (Omega Bio-Tek, Inc., Norcross, GA, Germany). All the experiments were performed in triplicate.

Transmission electron microscopy analysis
Transmission electron microscopy analysis was performed using a negative-staining method. Puri ed proteins were stained with 2% phosphotungstic acid (Solarbio, Beijing, China) for 1 minute and air-dried for 5 minutes at room temperature after removing excessive sample with lter paper. Grids were observed and imaged using a JEM-1400 (JEOL, Tokyo, Japan) at an acceleration voltage of 100 kV, with a magni cation of 120,000×.
Dynamic light scattering assay A dynamic light scattering assay was measured at room temperature using the Nano Particle Analyzer (Malvern, Beijing, China). Samples were diluted to an identical concentration of 0.1 mg/mL before measurement, and all determinations were conducted at a scattering angle of 90° and equilibrated to 25°C. At least 20 acquisitions were collected for each sample. This assay was performed in triplicate for data analysis.
Sucrose gradient sedimentation analysis VP2 and mutant VP2 proteins were puri ed in sucrose gradients. A four-step gradient of 5%, 15%, 30%, and 45% (wt/vol) sucrose was prepared in Tris-HCL buffer. All gradients were fractionated by bottom puncture, prepared in 12-mL tubes by loading 2 mL/step, and incubated at 4°C at least 8 hours to allow diffusion into a continuous gradient. Samples were rstly precipitated with 40% saturated ammonium sulfate (Sinopharm, Beijing, China), dialyzed with Tris-HCL buffer, loaded onto the gradients, and then ultracentrifuged at 200,000 ´ g in a P40ST rotor at 4°C for 36 hours (Hitachi, Ibaraki, Japan). After ultracentrifugation, samples in each fraction were electrophoresed via SDS-PAGE and identi ed by WB as described above. Native morphology of these samples was further examined using the NativePAGE™

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
All data generated or analyzed during this study are included in this published article and its supplementary information les.