key: cord-0074127-tfvqmk92
authors: Xia, Ming; Hoq, Md Rejaul; Huang, Pengwei; Jiang, Wen; Jiang, Xi; Tan, Ming
title: Bioengineered pseudovirus nanoparticles displaying the HA1 antigens of influenza viruses for enhanced immunogenicity
date: 2022-01-28
journal: Nano Res
DOI: 10.1007/s12274-021-4011-x
sha: 051f7509141fd85f595665eea2e4c1e179f57070
doc_id: 74127
cord_uid: tfvqmk92

Even with implementation of current influenza vaccines, influenza still claims up to 500,000 lives worldwide annually, indicating a need for a better vaccine strategy. We have developed a technology to generate unique S(60)-HA1 pseudovirus nanoparticles (PVNPs) that display the receptor-binding HA1 domains of influenza viruses. Each self-assembled S(60)-HA1 PVNP consists of a T = 1 icosahedral S(60) nanoparticle that resembles the inner shell of norovirus capsid and 60 surface-displayed HA1 antigens that are excellent vaccine targets. Soluble S(60)-HA1 PVNPs presenting HA1 antigens of H7N9 influenza virus subtypes have been produced efficiently in large amount. Their three-dimensional (3D) structures have been solved by cryogenic electron microscopy. The PVNP-displayed HA1 antigens react with HA-specific antibody, and retain authentic sialic acid binding specificity and hemagglutinate human erythrocytes. The PVNPs are highly immunogenic, eliciting high titers of HA1-specific antibodies in mice and the mouse sera strongly inhibited hemagglutinations of homologous and heterologous influenza virus HA proteins. Therefore, the S(60)-HA1 PVNPs may provide useful reagents to study influenza viruses and offer a potential new vaccine tactic to fight the deadly influenza disease. [Image: see text]

Influenza viruses (IVs), members of the family Orthomyxoviridae, are single-stranded, negative-sense RNA viruses causing deadly, epidemic influenza (flu) disease in both humans and animals. WHO and other organizations estimated that seasonal flu leads to 250,000-500,000 deaths per annum globally despite of applications of the current flu vaccines [1] [2] [3] . Thus, flu continues to be a significant global health burden, and new flu control and prevention strategies with improved efficacy are urgently needed. Structurally, each virus is covered by a lipid envelope, two surface proteins, known as hemagglutinin (HA) and neuraminidase (NA), respectively, protrude outward from the envelope, forming the viral spikes. They are numbered distinctly to name various influenza A virus (IAV) subtypes through different HA-NA combinations. The HA and NA play key roles in IV replication cycles, and thus are excellent targets of vaccines and antiviral drugs.

Vaccines remain the most effective approach to control and prevent flu. However, US Centers for Disease Control and Prevention (CDC) records indicate that the effectiveness of conventional flu vaccines based on egg-grown, inactivated IVs was low, ranging from 10% to 60% [4] , explaining the continual flu prevalence with significant mortality and morbidity. Most of the vaccines are produced by growing the corresponding IV strains in fertilized chicken eggs, which often leads to adaptive mutations and thus altered antigenicity of the vaccine IVs after many cycles of propagations in eggs. Such mutations have been shown to be an important factor contributing to the low effectiveness of egg-based flu vaccines [5] [6] [7] [8] .

For example, the egg-grown contemporary H3N2 vaccine strains used since 2016 were found antigenically mismatched circulating H3N2 strains [9, 10] , explaining the low efficacy of H3N2 vaccine during the winter seasons of 2016/2017 (33%) and 2017/2018 (22%) [4, 11] . Accordingly, a study to compare human H3N2 antibody responses elicited by egg-based (Fluzone®), cellbased (Flucelvax®), and recombinant HA-based (Flublok®) flu vaccines during the 2017/2018 winter season showed that the recombinant HA-based flu vaccines exhibited significantly higher neutralizing antibody titers than those induced by the egg-and cellbased flu vaccines [8] . Therefore, the recombinant HA-based flu vaccine with original HA sequences represents an approach for higher protective efficacy than the egg/cell-grown vaccines with adaptative mutations.

The immunodominant HA1 domains constitute the heads of IV HA trimers, interacting with host sialic acid receptors to initiate IV infection and thus are an ideal vaccine target. The commercial use of the Flublok® vaccine has demonstrated the feasibility and usefulness of the recombinant HA-based flu vaccine approach. However, the relatively smaller size and low valence of a 2 Materials and methods

DNA sequences encoding the HA1 antigen of an A/H7N9 strain (GenBank AC#: AUN86892.1, A/Jiangsu/TM261/2017/H7N9, amino acid C60 to N285, 226 aa), were codon-optimized to Escherichia Coli using the codon adaptation tool at http://www.jcat.de and synthesized by GenScript (Piscataway, NJ). The DNA fragment was cloned into the previously made pET-24b (Novagen)-based plasmid that was generated for production of the S R69A -VP8* fusion protein [20] by replacing the VP8*-encoding sequences using the HA1-encoding DNA fragment. R69A refers to the mutation in the NoV S domain to remove the exposed protease site [20] . A linker (HHHH) was added between the S and HA1 domains and a His tag was fused to the C-terminus of HA1 ( Fig. 1(a) ). A further plasmid without the His-tag encoding sequences was also made for tag-free S-HA1 protein production. As a control for comparisons, a further plasmid for expression of glutathione S-transferase (GST)-HA1 fusion protein were made by cloning the HA1-encoding DNA fragment into the pGEX-4T-1 vector (GE Healthcare Life Sciences) [22] .

Recombinant proteins were expressed using the E. coli (BL21, Arctic) system induced by 0.4 mM isopropyl-β-Dthiogalactopyranoside (IPTG) at 13 °C overnight as described previously [20, 23, 24] . Soluble His-tagged proteins were purified with His-tag binding cobalt resins (Thermo Fisher Scientific, Waltham, MA, USA) according to the instructions of the manufacturers, eluted with 150 mM imidazole (Sigma-Aldrich, St. Louis MO, USA) in 50 mM phosphate buffered saline (PBS, pH 7.2). Soluble GST-tagged proteins were purified using the GSTbinding resin (Glutathione Sepharose 4B GST-tagged protein purification resin, GE Healthcare Life Sciences) as described previously [24] .

The tag-free S-HA1 protein was expressed in E. coli as described above. For purification, clarified bacterial lysis was incubated with ammonium sulfate [(NH 4 ) 2 SO 4 ] at 1.2 M end concentration for 30 min to selectively precipitate the S-HA1 proteins as reported Production and characterization of His-tagged S-HA1 PVNPs. (a) Schematic construct of the His-tagged S-HA1 proteins. S, modified norovirus S domain; hinge, the hinge region of norovirus VP1; HA1, the HA1 antigen of H7 subtype; His, His tag. (b) SDS-PAGE of the purified S-HA1 protein (arrow) at 5 2 kDa. Lanes 1/2/3, purified proteins of three preparations; M, protein standards with indicated molecular sizes. (c) An elution curve of gel-filtration chromatography of the purified S-HA1 protein through a size exclusion column. The major PVNP elution peak and a minimal peak of S-HA1 protein monomers are indicated. The two elution peaks were calibrated using the S 60 -VP8* nanoparticles (~ 3.4 mDa) [20] and GST dimers (52 kDa) with their elution positions shown by star symbols on a red dashed line. Y-axis shows UV (A 280 ) absorbances (mAU), while X-axis indicates elusion volume (ml). (d) A representative micrograph of transmission electronic microscopy (TEM) showing the S-HA1 PVNPs.

previously [25] . Briefly, after centrifugation with an Avanti J26XP centrifuge (Beckman Coulter) using a JA-17 rotor at 10,000 rpm for 20 min, the protein pellet was collected, washed twice using 1.2 M (NH 4 ) 2 SO 4 solution (in PBS), and then solved in 20 mM tris buffer (pH 7.5). The protein was analyzed and further purified by gel-filtration, ion exchange chromatography, and cesium chloride (CsCl) density gradient centrifugation (see below).

Purified proteins were analyzed by SDS-PAGE using 10% separating gels and were quantitated by SDS-PAGE using serially diluted bovine serum albumin (BSA, Bio-Rad) with known concentrations as standards on same gels [26] .

This was performed as described previously [23, 24, 27] using an AKTA Fast Performance Liquid Chromatography system (FPLC, AKTA Pure 25L, GE Healthcare Life Sciences) using a size exclusion column (Superdex 200, 10/300 GL, 25 mL bed volume, GE Healthcare Life Sciences) to analyze the sizes of the S-HA1 proteins as evidence of PVNP formation. The column was calibrated using gel filtration calibration kits (GE Healthcare Life Sciences), while the elution peaks of the PVNPs and S-HA1 monomers were determined using the previously made S 60 -VP8* nanoparticles (~ 3.4 mDa) [20] and GST-dimer (52 kDa). Relative protein concentrations in the effluent were reported as A 280 absorbance.

This was carried out using the same AKTA FPLC system (see above) using a HiPrep Q HP 16/10 column (20 mL bed volume, GE Healthcare Life Sciences) to analyze and purify the tag-free S-HA1 protein. A previous established procedure [25] was modified and applied here. After the column was equilibrated using 7 column volumes (CVs) of 20 mM Tris buffer (pH 8.0, referred as buffer A), protein samples were loaded. The column was washed using 7 CVs of buffer A to remove the unbound proteins, followed by elution of the bound proteins using 7 CVs of buffer B (1 M NaCl in buffer A) via a linear gradient elution manner (0%-100% B). The column was then stripped by 7 CVs of buffer B (100% B), followed by an equilibration using 7 CVs of buffer A. Relative protein concentrations in the effluent were reported as A 280 absorbance, while the elution positions of the proteins were shown as percentages of buffer B.

This was performed as described previously [20] . Briefly, 0.5 mL of (NH 4 ) 2 SO 4 precipitated S-HA1 protein (see above) was mixed with 10 mL of CsCl solution with a density of 1.3630 g/mL and packed in a 12-mL centrifuge tube. After centrifugation at 288,000× g for 45 h in the Optima L-90K ultracentrifuge (Beckman Coulter), the gradient was fractionated by bottom puncture into 21 fractions (~ 0.5 mL each). CsCl densities of fractions were determined according to the refractive index.

TEM was used to inspect the morphology of the PVNPs. After staining with 1% ammonium molybdate, purified S-HA1 proteins were observed under an EM10 C2 microscope (Zeiss, Germany) at 80 kV at magnifications between 15,000× and 30,000× as described elsewhere [20] .

Specific pathogen free (SPF) BALB/c mice at ~ 6 weeks of age were purchased from the Jackson Laboratory (Bar harbor, ME, USA) and maintained under SPF conditions at the Division of Veterinary Services of Cincinnati Children's Hospital Medical Center (CCHMC). Mice were randomly divided into four groups with 6-8 mice each (n = 6-8) and each group was administered with one of following immunogens: (1) S 60 -HA1 PVNPs (S 60 -HA1); (2) GST-HA1 fusion protein (GST-HA1); (3) S 60 nanoparticle without HA1 antigens as platform control (S 60 ); and (4) PBS as diluent control (PBS). Immunogens were administered with Alum adjuvant (Imject Alum, Thermo Fisher Scientific, USA) at 25 μL/dose (20 μg/mouse/dose). Endotoxin contaminations of the recombinant protein immunogens were removed using Endotoxin Removal Spin Columns (Thermo Scientific TM Pierce, USA). After this treatment, all immunogens for mouse immunization contained less than 1.5 endotoxin units. Mice were immunized intramuscularly (i.m.) in the thigh muscle three times with injection volume of 50 μL at 2-week intervals. Bloods were collected 2 weeks after the final immunization via heart puncture and sera were processed from bloods via an established protocol [26] .

EIAs were used to determine HA1 specific antibody titers as described elsewhere [28, 29] . Briefly, commercial recombinant H7 HA protein (Sino Biological) was coated on 96-well plates at 1 μg/mL. After blocked with 5% (w/v) skim milk, plates were incubated with mouse sera at serial dilutions. The bound antibodies were measured by goat-anti-mouse IgG-horse radish peroxidase (HRP) conjugate (1:5,000, MP Biomedicals). IgG titers were defined as maximum dilutions of sera with positive signals. A positive signal was determined by a mean of negative values + 3 × standard deviation. EIAs were also used to measure S-HA1 PVNPs in the fractions of CsCl density gradient. In this case, the fractions were diluted in 800 folds in PBS and coated on microtiter plates. The PVNPs were detected using our in-house made antibody against NoV-like particle (VLP) [30] .

CryoEM was carried out to reconstruct the three-dimensional (3D) structures of the S-HA1 PVNPs as described in our previous studies [26, 31, 32] . Briefly, 3 µL of purified PVNPs were dropped onto graphene oxide coated lacey grids (Tedpella #01896), then blotted for 4 s before flashed frozen into ethane using Cp3 cryoplunder (Gatan, Inc., Pleasanton CA, USA) at 90% humidity. Lowelectron-dose images (~ 20 e/Å2) were recorded on Talos F200C cryo-TEM, recorded by Ceta 16M camera at a nominal magnification (57,000×) with a calibrated pixel size of 2.59 Å. The collected micrographs were imported into cryoSPARC/v3 for further processing. A total of >14,000 particles were selected with Gautomatch v0.53 (http://www.mrc-lmb.cam.ac.uk/kzhang/). Multiple homogeneous class averages were separated using twodimensional (2D) classification in CryoSPARC 3.1 [33] . The Histagged particles were further divided to two classes based on their size difference, among which 6,200 separated particles for the larger, and 5,284 particles for the smaller PVNPs were subjected to ab-initio 3D reconstruction for initial model generation and final homogeneous refinement for further analysis. Similarly, 1,064 separated tag-free PVNPs particles were selected for initial model generation, followed by homogeneous refinement. CryoEM map fitting was conducted using UCSF Chimera software (http://www.rbvi.ucsf.edu/chimera; version 1.15) using in the crystal structures of the inner shell of the 60-valent feline calicivirus (FCV) VLP (PDB code: 4PB6) [34] and the HA1 heads of H7 HA trimer (PDB code: 4LN3) [35, 36] . The structure of the S 60 nanoparticle was modeled using the crystal structure of the 60valent FCV inner shell (PDB code: 4PB6) using the same UCSF Chimera software.

This was used to measure interactions between the S-HA1 PVNPs and sialoglycans. Sialoglycans of (1) Neu5Acα2-8Neu5Acα2-8Neu5Acα-sp-PAA-biotin, (2) Neu5Acβ2,6(Galβ1,3)GalNAcα-PAA-biotin, (3) α-Neu5Ac2,3βGal1,4Glc(3'-sialyllactose)-PAAbiotin, (4) sialyl Le a -PAA-biotin, and (5) sialylLe x -PAA-biotin were from GlycoTech (Rockville, MD), while glycans of Neu5Acα2,3Galβ1,3(Fuca1,4)GlcNAcβ-SpNH-PAA-biotin and Neu5Acα2,3Galβ1,4(Fucα1,3)GlcNAcβ-SpNH-PAA-biotin were from the Consortium for Functional Glycomics (http://www.functionalglycomics.org/). All glycans were conjugated with polyacrylic acid (PAA) and labeled with biotin. Both PVNPs and the S 60 nanoparticle control at 10 μg/mL were coated on microtiter plates (Thermo Scientific) at 4 °C overnight. After blocking with 5% (w/v) nonfat milk, the plates were incubated with the glycans at 2 μg/mL for 60 min. The bound glycans were detected by streptavidin-HRP conjugates (Jackson ImmunoResearch Laboratories) at 1:5,000 dilution.

This was performed to test the antigenic reactivity of the S-HA1 PVNPs with HA specific antibody as described elsewhere [22] . Briefly, native and boiled S-HA1 PVNPs were serially diluted and coated on microtiter plates (Thermo Scientific TM , USA) at 4 °C overnight. After blocking, commercial H7 HA specific antibody (Sino Biological) at 1:1,000 or appropriate dilutions was added to the coated wells. The bound antibody was detected by corresponding HRP-conjugated secondary antibody (Thermo Scientific TM , USA) at 1:2,000 dilution.

The assays were used measure hemagglutination titers of recombinant HAs and S-HA1 PVNPs using human RBCs that were provided by the Translational Core Laboratory at CCHMC, as described elsewhere [22] . Briefly, 500 µL of 0.5% RBCs in 10 mM PBS were mixed with serially diluted HAs or PVNPs in 96well V-bottom plates and incubated at 4 °C for 60-100 min, using the S 60 nanoparticles and PBS as negative controls. Hemagglutination titers were defined as the highest dilutions of the HAs or PVNPs that caused hemagglutination. For HI assays, HAs (Sino biological) or PVNPs at 4× hemagglutination titers were incubated with serially diluted serum antibody for 60 min before they were mixed with RBCs for hemagglutination assays. The HI titers were defined as the maximum serum dilutions that prevented hemagglutination.

All animal studies were completed in compliance with the recommendations in the Guide for the Care and Use of Laboratory Animals (23a) of the National Institute of Health (NIH). The protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of the Cincinnati Children's Hospital Research Foundation (Animal Welfare Assurance No. A3108-01).

Statistical differences between two data groups were calculated by GraphPad Prism 9.0 (GraphPad Software, Inc.) using unpaired Student's t-test. Differences were considered non-significant when P-values > 0.05 (marked as ns), significant when P-values < 0.05 (marked as *), highly significant when P-values < 0.01 (marked as **), and extremely significant when P-values < 0.001 (marked as ***), respectively.

S-HA1 fusion protein containing the HA1 antigens of an H7N9 IV ( Fig. 1(a) ) was produced via the E. coli system. The Cterminally His-tagged, soluble S-HA1 protein was purified at yields of ~ 20 mg/L bacterial culture using His-tag binding cobalt resin. SDS-PAGE showed the purified protein at an expected size of ~ 52 kDa ( Fig. 1(b) ). Gel-filtration chromatography of the protein revealed a single major elution peak with high molecular size like the previously made S 60 -VP8* nanoparticles [20] (Fig.  1(c) ), while the elution peak corresponding to the S-HA1 protein monomers was minimal, indicating that vast majority of the proteins self-assembled into PVNPs or large complexes. TEM inspection of the major gel-filtration elution peak demonstrated the self-formation of the S-HA1 H7 PVNPs ( Fig. 1(d) ). Major PVNP population had sizes between 16 and 20 nm with some variations.

To assess potential impact of the His-tag on PVNP formation and explore alternative production and purification approaches, tagfree S-HA1 protein ( Fig. 2(a) ) was also produced and purified by selective precipitation using (NH 4 ) 2 SO 4 [25] at concentrations of 0.7 and 1.2 M. This resulted in high protein yields at 30-40 mg/L bacterial culture ( Fig. 2(b) ). Gel-filtration of the (NH 4 ) 2 SO 4purified, resolved protein revealed a similar single major elution peak of PVNPs (Fig. 2(c) ), confirming that the fusion protein selfassembled into PVNPs. The protein was also analyzed by an anion exchange chromatography (Fig. 2(e) ), in which vast majority of the target protein was eluted in the major peak corresponding to 31.7% buffer B, equivalent to an ionic strength of 317 mM NaCl and 20 mM Tris. SDS-PAGE of the major elution peaks of the gelfiltration and ion exchange revealed the target proteins at high purity ( Fig. 2(d) ), supporting that the two chromatography approaches may be used to further purify the (NH 4 ) 2 SO 4precipitated, tag-free S-HA1 proteins.

The (NH 4 ) 2 SO 4 -purified protein was analyzed further by CsCl density gradient centrifugation, resulting in a visible protein band at a density of ~ 1.311 mg/cm 3 (Fig. 2(f) ). The gradient was fractionated into 21 portions, followed by detection of the PVNPs in the factions by EIA assays using the S domain-specific antibody, confirming the protein band as the S-HA1 protein ( Fig. 2(g) ). Finally, the protein band was examined by TEM, revealing typical PVNPs (Fig. 2(h) ) with better homogeneity than the His-tagged ones (comparing Fig. 2(h) with Fig. 1(d) ). These data indicated that, while the His-tags did not substantially affect the PVNP formation, tag-free PVNPs showed better production yields and improved PVNP homogeneity.

The 3D structures of the S-HA1 PVNPs were reconstructed using cryoEM approach. By focusing on the major populations, we noted and reconstructed two types of the His-tagged PVNPs that have distinct diameters of ~ 21 and ~ 16 nm, respectively. Both PVNPs were reconstructed at 12-Å resolution (Figs. 3 and 4) and they are in T = 1 icosahedral symmetry, consisting of 60 S-HA1 proteins, thus are named S 60 -HA1 PVNPs. Fitting of the crystal structure of the S 60 -nanoparticle (PDB ID: 4PB6) [20, 34] into the cryoEM density map of the 21-nm PVNP inner shell region indicated that their shapes match well, but the PVNP inner shell appears slightly smaller, particularly in the protruding regions at the five-fold axes (Fig. 3(c) ). This confirmed the identity of the S 60nanoparticle inner shell of the PVNP (Fig. 3, comparing 3 (b) with 3(h)) and revealed that the surface displayed HA1 antigens force the wall of the S 60 -nanoparticle to go inward structurally, resulting in a compact interior shell. Similar scenario also occurred to the 16nm, His-tagged PVNP, in which both the wall of the S 60nanoparticle shell and HA1 antigens go further inward, leading to a more compact PVNP with smaller diameter (Figs. 4(a)-4(c) ).

The above fitting also showed that the 60 HA1 antigens are located along the sunken grooves among the protrusions at the five-fold axes of the S 60 nanoparticle (Figs. 3(c) , 3(g), 4(e)). Noteworthy, the 60 HA1 antigens are arranged into 20 trimers on each PVNP, corresponding to the triangle patterns of the exposed C-termini of the NoV S domains on the S 60 nanoparticle (Figs. 3(g), 3(h), 4(e), 4(f)). These trimer-like HA1s may help to retain their authentic structures and receptor-binding function (see below). When the crystal structures of H7N9 HA1 trimers (PDB ID: 4LN3) [36] were fitted into the cryoEM density maps of the trimer-like HA1 regions, they fitted fine despite some discrepancies on the outside edges (Figs. 3(d) and 4(d) ), confirming their structural identities and features. The PVNP structural model in PDB format resulted from the fitting indicated a smaller central lumen than that of the S 60 -nanoparticle [20] (Figs. 3(e) and 3(f) ). Finally, the two S 60 -HA1 PVNPs in two color schemes were generated to show the inner shell and the surface trimeric HA1 antigens, respectively (Figs. 3(g) and 4(e)).

Only the 21-nm PVNP was recognized in the major populations of the tag-free PVNPs. Its 3D structure was solved through cryoEM method at 15-Å resolution, revealing the same T = 1 icosahedral symmetry (Fig. 5) like the His-tagged PVNPs. Similar organization consisting of the S 60 nanoparticle inner shell and 60 exterior, trimer-like HA1 antigens like the His-tagged PVNPs were witnessed by fitting of the S 60 -nanoparticle and HA1 trimer crystal structures into corresponding regions of the cryoEM density maps, and by structural comparisons between the tag-free and the His-tagged PVNPs (Figs. 3 and 5) . However, differences between the tag-free PVNPs and the tagged ones were also noted. For example, the five-fold axis protrusions of the 21-nm/tag free PVNP go less outward compared with those of the 21-nm/Histagged PVNP (comparing Fig. 3(b) with Fig. 5(b) ). As a result, the grooves among the five-fold axis protrusions of the 21-nm/Histagged PVNP appear to be deeper than those of the 21-nm/tagfree PVNPs (Figs. 3(g) and 5(e)). We also noted that the roots of the HA1 antigens of the 21-nm/tag-free PVNP appear to submerge into the S 60 -nanoparticle wall unlike those of the 21-nm/His-tagged one (comparing Fig. 3 (b) with Fig. 5(b) ).

EIA assays showed that the commercial antibody (Sino Biological) specific to IAV H7 HA recognized the His-tagged S 60 -HA1 PVNPs, as shown by the high and dose-dependent EIA signals ( Fig. 6(a) ), indicating that the PVNP-displayed HA1s retained their natural conformations. These reactivity signals were mostly gone, when the PVNPs were boiled to destroy conformational epitopes, leaving mainly linear epitopes. As negative controls, the S 60 nanoparticles without HA1 antigens did not react with the antibody (Fig. 6(a) ). The tag-free PVNPs also revealed similar antigenic reactivity to the H7 HA-specific antibody (data not shown). Due to the similarity of the two PVNP types, downstream experiments were performed using the His-tagged PVNPs by taking advantage of their easy purification with higher purity.

The sialic acid binding ability and specificity of the S 60 -HA1 PVNPs were determined by EIA-based glycan binding assays. The results showed that the PVNPs bound only 2,3-linked, but not 2,6- or 2,8-linked sialoglycans ( Fig. 6(b) ), consistent with the known receptor binding nature of their parental H7N9 avian IVs [35, 37, 38] . Accordingly, the S 60 -HA1 PVNPs agglutinated human RBCs at a high hemagglutination titer of 122 ng/mL (Fig. 6(c) ). As negative controls, the S 60 nanoparticles did not bind any tested sialoglycans and did not cause hemagglutination. These data collectively demonstrated that the PVNP-displayed HA1 antigens retain the authentic structure, conformation, receptor binding function, and hemagglutination nature, supporting that the S 60 -HA1 PVNPs may serve as a reagent for IV study.

Numerous literatures show that antigens with pathogen associated molecular patterns (PAMPs) in polyvalent nature induce stronger immune responses compared with the mono-or low-valent ones [39−42] . To prove that this is true for the PVNPs, mice were immunized with the S 60 -HA1 PVNPs using dimeric GST-HA1 proteins for comparison. After three immunizations, mouse sera were collected to determine HA1-specific IgG titers using recombinant H7 HA1 protein (Sino Biological) as capture antigen. The outcomes showed that the PVNP-elicited HA1-specific IgG titer (80,000) was 8-fold higher than that (10,000) induced by the GST-HA1 protein (P = 0.0018, Fig. 7(a) ), supporting the above notion. Statistic differences between data groups are shown as ** for highly significant with P-values < 0.01, and *** for extremely significant with P-values < 0.001.

We then measured the inhibition effects of the mouse sera against hemagglutination of the S 60 -HA1 PVNPs, revealing high HI titers (44, 373) of the PVNP-immunized sera that was about 9.5 folds greater than that (4,693) of the mouse sera after immunization with the GST-HA1 protein (P = 0.0008, Fig. 7(b) ), consistent with their IgG titers. We also determined HI titers of the sera against hemagglutination of various commercial recombinant HA proteins (Sino Biological). The results showed that the S 60 -HA1 PVNP-immunized mouse sera showed very high HI titer (14, 933) against hemagglutination of the homologous H7 (A/ Zhejiang/1/2013/ H7N9) HA protein (Fig. 7(c) ). Noteworthy, the sera also showed high HI titer (667) to heterologous subtype H3 (A/California/7/2004/H3N2) HA protein that clusters with the H7 subtype in the IAV group 2, but relatively low HI titer (150) to heterologous subtype H1 (A/NewYork/18/2009/H1N1) HA proteins that belongs to the IAV group 1 (Fig. 7(c) ). As negative controls, the S 60 nanoparticle did not elicit HA1-specific antibody ( Fig. 7(a) ) and the resulted mouse sera did not inhibit the hemagglutinations by the S 60 -HA1 H7 PVNPs (Fig. 7(c) , control) or other tested recombinant HA proteins (data not shown). The above data support that the S 60 -HA1 PVNPs may serve as a flu vaccine candidate.

In this study we designed and generated the S 60 -HA1 PVNPs by taking advantage of the self-formation of our recently invented S 60 nanoparticles [20] to display the key neutralizing HA1 antigens of IVs for enhanced immunogenicity. The self-assembled S 60 -HA1 PVNPs can be produced through the prokaryotic E. coli system, as well as the eukaryotic mammalian cell system (to be published data). The bacterially expressed PVNPs have been well characterized structurally and functionally in this study. Each PVNP consists of an S 60 nanoparticle core that resembles the inner shell of NoV capsid and 60 surface-displayed HA1 antigens extending from the inner shell, making the PVNPs resembling virus-like particles. Remarkably, the exposed HA1 antigens are arranged into trimers, like those on the HAs of authentic flu viruses. Functional data showed that the PVNPs were recognized by HA-specific antibody, bound specific sialic acid receptors, and agglutinated human RBCs. These outcomes indicated that the PVNP displayed HA1 antigens retain the original structures, conformational epitopes, receptor binding functions, and PAMPs. Thus, the S 60 -HA1 PVNPs may be used as reagents to study IVs. As a result of polyvalent presentations of the HA1 antigens by the PVNPs, and the well-preserved PAMPs of both HA1 antigens and NoV S 60 -nanoparticles, the S 60 -HA1 PVNPs elicited strong immune responses. These were shown by the high HA1-specific IgG titers in the PVNP immunized mouse sera, as well as their high HI titers against hemagglutination of the S 60 -HA1 PVNPs and various recombinant HA proteins. The mouse sera after immunization with the S 60 -HA1 PVNPs exhibited very high HI titers, reaching to 14,933, against hemagglutination of the homologous H7 HA proteins. This strongly suggested that the S 60 -HA1 PVNPs may serve as a potent vaccine candidate against H7 IAVs, such as the highly pathogenic H7N9 avian IAVs that caused five epidemic waves from 2013 to 2017 in China, claiming 511 human lives [43] . The high HI titer resulted apparently from the specific binding of the serum antibody to the HA1 antigens of the H7 HA proteins and thus inhibited the hemagglutination ability of the HA protein. Same principle should also apply in vivo to confer protection. Due to the lack of an animal biosafety level 3 (BSL3) facility here, we are reaching out to identify collaborators with an appropriate animal model to study the protective efficacy of our S 60 -HA1 H7 PVNP vaccine candidate. If it succeeds, our S 60 -HA1 H7 PVNPs will serve as a bird flu vaccine to be used in poultry farms and/or high-risk human populations, such as those who have contact or work with poultry.

Noteworthy, the S 60 -HA1 PVNP-immunized mouse sera also inhibited hemagglutination of heterologous H3 (A/California/7/2004/H3N2) HA protein with a high HI titer of 667, which was significantly higher than that (150) against H1 (A/New York/18/2009/H1N1) HA proteins (P < 0.0001). These data implied certain cross protection of future S 60 -HA1 H7 PVNP vaccine against the two heterologous IAV subtypes. The higher HI titers of the H7 PVNP-immunized sera against the H3 HA protein than that of the H1 HA protein may be explained by the facts that H7 and H3 subtypes are genetically closer related, clustering in the same IAV group 2, while H1 belongs to IAV group 1, being genetically further away from the H7 subtype.

IV HA1 antigens are known for their variable sequences, leading to antigenic drifts and epochal evolution of flu viruses. However, a conserved antigenic supersite has been reported at or near the receptor binding site (RBS) that is targeted by broadly neutralizing antibodies [44] [45] [46] . Similarly, by analyzing the HA1 sequences of major H1 IVs circulated over the past century, a naturally protective epitope of limited variability near the RBS has been identified and proposed as a target of a universal flu vaccine [47] . Displaying the HA1 domain alone by the polyvalent S 60 nanoparticle may elicit substantially high proportion of antibody specific to the antigenic supersite and/or the conserved protective epitope. This may explain the observed high cross subtype HI titers. In fact, a previous study showed that bacterially expressed, oligomerized HA1s elicited cross-clade protective immunity in ferrets [48] . Thus, while a PVNP cocktail with multiple HA1 antigens of predominant IVs may offer broad protection (see above), an alternative approach would be to generate a pool of PVNPs that display a collection of native and/or artificially designed protective epitopes of limited variability [47] for broad protection against designated IVs. Finally, to the question that the PVNP containing only HA1s without other IV antigens might not elicit complete protective immunity as the primary vaccination, we wish to mention that our PVNPs will be a good choice of a boosting vaccine, because vast majority of human adults have been infected and/or immunized with one or more IVs and/or flu vaccines and thus have already memories to the conserved IV epitopes. In this case, the immunity against the variable HA1 antigens of currently circulated IVs may be the key to confer protection against the new IV strains.

We noted that the PVNPs formed by the S-HA1 proteins were heterologous in sizes. This was not unexpected, because in vitro expression of NoV VP1s often leads to VLP formation at heterologous sizes, including those in T = 1 (~ 31 nm), 3 (~ 46 nm), and 4 (~ 50 nm) icosahedral symmetry, consisting of 60, 180, and 240 NoV VP1s, respectively [49, 50] . By focusing on the major populations of the His-tagged H7 PVNPs, we have reconstructed the 3D structures of two PVNPs with distinct diameters (21 nm vs. 16 nm) by cryoEM method, revealing the same T = 1 icosahedral symmetry, being composed of 60 copies of the S-HA1 fusion proteins. This size difference appears to be resulted from their different ways of homotypic interactions among the S-HA1 fusion proteins during PVNP assembly. We speculated that the even larger PVNPs observed in this study may represent those in T = 3 icosahedral symmetry. Further study is necessary to clarify this scenario.

Despite size variations, the PVNPs should share the basic organization comprising of the icosahedral inner shell formed by the NoV S domains and the surface displayed HA1 antigens. As explained above, the polyvalent PVNPs and their well-preserved PAMPs enhance the immunogenicity of the HA1 antigens, leading to the high immune responses towards the HA1 antigens. Thus, the size variations should not affect the potential of the PVNPs as an excellent IV immunogen. Such size heterogeneity of PVNPs may be reduced by using a mammalian cell expression system that is known to provide better folding environments for viral structural proteins. More importantly, the HA1 antigens are glycosylated, the mammalian cell system offers a better chance to reconstruct their original glycosylation and other post translational modifications, making the resulted PVNPs a better flu vaccine candidate. On the other hand, the above explained protective epitope of limited variability near the RBS has been shown to be peptide epitope without glycosylation [47] . This justifies the observed high HI effects of the PVNP immunized sera in this study, supporting our ultimate goal to develop the PVNPs into a useful flu vaccine.

The readily available PVNPs displaying the HA1 antigens of H7 subtype in this study and those that will be generated in future studies based on our established procedures will offer convenient and useful reagents to study flu viruses. These PVNPs can be produced either via the prokaryotic E. coli system quickly at a low cost or through a eukaryotic system (our unpublished data) potentially for better structural and functional integrity of the HA1 antigens. During this study, we observed that the PVNPs remained structurally intact after storage at -80 and -20 °C for at least a year, as well as 4 °C for at least six months (MT, unpublished data). In addition, the PVNPs were shown to retain their structural integrity after lyophilization and rehydration treatments (MT, unpublished data), offering a further useful approach for PVNP preservation. The PVNPs with polyvalent HA1 antigens will be able to better mimic the HA features of flu viruses, providing stronger antigenic reactivity, greater receptor binding avidity, and higher immune responses, than those of commercially available recombinant monomeric or trimeric HA proteins. Thus, our PVNPs will offer new and better tools to study flu viruses.

A technology has been developed to generate the self-assembled S 60 -HA1 PVNPs displaying the key neutralizing HA1 antigens of IVs as a promising tool for IV study and as a potential flu vaccine candidate. These have been supported by our extensive characterizations in the structures, functions, and immune responses of the PVNPs. Our study offers a new strategy to study and fight the deadly flu illness and provide a multifunctional platform for novel vaccine development against other infectious diseases.

Global burden of influenza as a cause of cardiopulmonary morbidity and mortality

Estimates of global seasonal influenzaassociated respiratory mortality: A modelling study

Seasonal influenza vaccine effectiveness

A structural explanation for the low effectiveness of the seasonal influenza H3N2 vaccine

Low 2012-13 influenza vaccine effectiveness associated with mutation in the egg-adapted H3N2 vaccine strain not antigenic drift in circulating viruses

Chasing seasonal influenza-the need for a universal influenza vaccine

Comparison of human H3N2 antibody responses elicited by egg-based, cell-based, and recombinant proteinbased influenza vaccines during the 2017-2018 season

Contemporary H3N2 influenza viruses have a glycosylation site that alters binding of antibodies elicited by egg-adapted vaccine strains

Antibodies against egg-and cell-grown influenza A(H3N2) viruses in adults hospitalized during the 2017-2018 influenza season

Effects of influenza vaccination in the united states during the 2017-2018 influenza season

Self-assembling influenza nanoparticle vaccines elicit broadly neutralizing H1N1 antibodies

Hemagglutinin-stem nanoparticles generate heterosubtypic influenza protection

Mosaic nanoparticle display of diverse influenza virus hemagglutinins elicits broad B cell responses

Elicitation of broadly protective immunity to influenza by multivalent hemagglutinin nanoparticle vaccines

Accurate design of megadalton-scale two-component icosahedral protein complexes

The production of hemagglutininbased virus-like particles in plants: A rapid

Novel hemagglutinin nanoparticle influenza vaccine with Matrix-M TM adjuvant induces hemagglutination inhibition, neutralizing, and protective responses in ferrets against homologous and drifted A(H3N2) subtypes

Influenza hemagglutinin nanoparticle vaccine elicits broadly neutralizing antibodies against structurally distinct domains of H3N2 HA

Bioengineered norovirus S 60 nanoparticles as a multifunctional vaccine platform

Immune response and protective efficacy of the S particle presented rotavirus VP8* vaccine in mice

Saliva as a source of reagent to study human susceptibility to avian influenza H7N9 virus infection

The P domain of norovirus capsid protein forms a subviral particle that binds to histo-blood group antigen receptors

The P domain of norovirus capsid protein forms dimer and binds to histo-blood group antigen receptors

A nanoparticle-based trivalent vaccine targeting the glycan binding VP8* domains of rotaviruses

Norovirus P particle, a novel platform for vaccine development and antibody production

Polyvalent complexes for vaccine development

Development and evaluation of two subunit vaccine candidates containing antigens of hepatitis E virus, rotavirus, and astrovirus

A trivalent vaccine candidate against hepatitis E virus, norovirus, and astrovirus

Norovirus and histo-blood group antigens: Demonstration of a wide spectrum of strain specificities and classification of two major binding groups among multiple binding patterns

Terminal modifications of norovirus P domain resulted in a new type of subviral particles, the small P particles

Noroviral P particle: Structure, function and applications in virus-host interaction

Algorithms for rapid unsupervised cryo-EM structure determination

Structure determination of feline calicivirus virus-like particles in the context of a pseudooctahedral arrangement

Structures and receptor binding of hemagglutinins from human-infecting H7N9 influenza viruses

Structural analysis of the hemagglutinin from the recent 2013 H7N9 influenza virus

Preferential recognition of avian-like receptors in human influenza A H7N9 viruses

Glycan receptor binding of the influenza A virus H7N9 hemagglutinin

Norovirus capsid protein-derived nanoparticles and polymers as versatile platforms for antigen presentation and vaccine development

Recent advancements in combination subunit vaccine development

Nanoparticles of norovirus

Subviral particle as vaccine and vaccine platform

Epidemiology, evolution, and pathogenesis of H7N9 influenza viruses in five epidemic waves since 2013 in China

Cross-neutralization of influenza A viruses mediated by a single antibody loop

Broadly neutralizing human antibody that recognizes the receptor-binding pocket of influenza virus hemagglutinin

Receptor mimicry by antibody F045-092 facilitates universal binding to the H3 subtype of influenza virus

A naturally protective epitope of limited variability as an influenza vaccine target

Bacterial HA1 vaccine against pandemic H5N1 influenza virus: Evidence of oligomerization, hemagglutination, and cross-protective immunity in ferrets

High-resolution cryo-EM structures of outbreak strain human norovirus shells reveal size variations

Structural heterogeneity of a human norovirus vaccine candidate

The research described in this study was supported by the National Institute of Allergy and Infectious Diseases (NIAID, No.