Hemagglutinin-displaying influenza nanovaccines: progress and promise

Abstract

Influenza remains a major global health concern, with the ongoing seasonal epidemics causing millions of cases and up to 650,000 deaths worldwide annually. Current influenza vaccines only provide strain-specific and short-lived protection, exposing vulnerability to antigenic drift and reassortments. To overcome these limitations, next-generation vaccine platforms are being developed, with nanoparticle-based approaches showing promise. Displaying hemagglutinin (HA) on multivalent scaffolds enhances B cell receptor engagement, germinal center formation, and affinity maturation, while supporting durable and broadly protective humoral immunity. This review highlights recent published advances found in PubMed, Web of Science, and Google Scholar since 2020 in HA-displaying nanoparticle influenza vaccines, emphasizing strategies to improve immunogenicity, broaden protection across influenza strains and subtypes, and redirect responses toward conserved epitopes of HA.

1. Introduction

Influenza has posed a persistent threat to global health for over a century, with several pandemics driving our scientific understanding of infectious diseases. The 1918 H1N1 pandemic, the first influenza pandemic of the 20^th century and one of the deadliest disease events in history, caused an estimated 50 million deaths worldwide [1,2] and infected roughly one-third of the global population [3]. Subsequent pandemics in 1957 (H2N2), 1968 (H3N2), and 2009 (H1N1) further highlighted influenza’s capacity to cause significant disease burden. Beyond just these pandemics, seasonal influenza remains a substantial public health concern, causing up to 41 million illnesses annually in the U.S. alone and up to 650,000 influenza-related deaths worldwide [4,5].

Influenza viruses are enveloped, negative-sense, single-stranded RNA viruses, whose genome encodes both structural and accessory proteins. Among these, hemagglutinin (HA) is the most abundant surface glycoprotein, mediating viral attachment and entry into host cells. Neuraminidase (NA) facilitates the release of replicated virions by cleaving sialic acids from host cell receptors. In influenza A viruses - the causative agents of previous pandemics - 19 HA subtypes and 11 NA subtypes have been identified. These subtypes each undergo continuous antigenic drift through point mutations and antigenic shift through genetic reassortment, causing both seasonal epidemics and pandemics.

Vaccination is a key strategy for preventing and mitigating influenza infections, and inactivated and live attenuated H1N1, H3N2, and influenza B vaccines have been available in the market [6–8] with recombinant protein vaccines also recently becoming available in the 21^st century [9]. Current seasonal vaccines, however, tend to elicit strain-specific [10,11] and short-lived [12] immune responses and thus require annual reformulation and administration. This reformulation is reliant on predictions of which influenza strains will be circulating, but manufacturing concerns such as viral yield from egg culture, differing glycosylation patterns from egg and cell culture vaccines, and vaccine production lead times can also impact vaccine strain choice and antigenic match to circulating strains. As a result, the effectiveness of seasonal vaccines varies widely, ranging from 10% to 60% according the Centers for Disease Control and Prevention (CDC) [13]. Furthermore, these seasonal vaccines are unlikely to prevent infections from reassortment strains (as in 2009) or against other influenza A subtypes such as highly pathogenic avian influenza (HPAI) H5N1 or H7N9. In particular, the ongoing spread of clade 2.3.4.4b HPAI H5N1 viruses in birds and cattle[14], multiple spillovers to humans[15,16], and the potential for rapid adaptation to mammalian hosts [17–20] raises concerns about a potential pandemic. It is particularly concerning because the global population remains largely naïve to avian influenza strains, amplifying concerns about their pandemic potential.

To address these challenges, next-generation vaccine platforms have been explored, with nanoparticle-based approaches emerging as particularly promising [21–23]. Displaying HA on the surface of nanoparticles provides a multivalent and repetitive array of HA antigens that promotes B cell receptor clustering, enhances germinal center reactions, and supports affinity maturation [24,25]. In particular, presenting antigens in a highly ordered and repetitive array reduces the threshold for B cell activation and promotes early plasmablast differentiation, thereby increasing the activated B cell population[26]. Moreover, stronger signaling generated by B cell receptor clustering amplifies antigen internalization and presentation, which in turn facilitates more effective engagement with follicular helper T cells within the germinal center[27]. Multivalency also increases the avidity of the antigens [28], further strengthening the antigen-B cell interactions and leading to robust and durable humoral responses. In this review, we focus on advances since 2020 found in PubMed, Web of Science, and Google Scholar in nanoparticle-based influenza vaccines that display HA to improve vaccine immunogenicity, induce broader protection against influenza, or redirect immune responses toward the conserved regions of HA.

2. Nanovaccines that display HA to enhance immunogenicity

The display of hemagglutinin (HA) on a nanoparticle surface can often be used to enhance the strength of the immune response against HA. Matrix-M, a saponin-based nanoparticle adjuvant[29], is one platform for influenza nanovaccines.[30–33]. H5-MNP is a nanoparticle vaccine consisting of hemagglutinin from A/AW/SC/2021 (H5N1) anchored to a Polysorbate 80 micelle and adjuvanted with Matrix-M that complexes with HA to form H5-MNP (Figure 1A)[34]. Boosted intramuscular vaccination of non-human primates with the H5-MNP nanoparticle resulted in detectable hemagglutination inhibition (HAI) titers against A(H5N1) A/AW/SC/2021. Rhesus macaques boosted intramuscularly with the H5-MNP also exhibited 92.2-fold higher pseudovirus neutralizing titers against a matched H5 pseudovirus as compared to titers prior to boost. Using a competitive binding assay with monoclonal antibodies targeting the hemagglutinin stem, vestigial esterase domain (VE), and receptor binding site (RBS), all of which are known targets of neutralizing antibodies, the authors showed that H5-MNP immunization induced antibodies in NHPs targeting all three of these sites (Figure 1B).

Figure 1. Assembly and immunogenicity of H5-MNP nanoparticles.

A) Assembly of H5-MNP nanoparticles as monitored by negative-stain transmission electron microscopy. B) Concentrations of stem-targeting, vestigial esterase (VE)-targeting, and receptor-binding site (RBS)-targeting antibodies with equivalent binding to A/AW/SC/2021 H5 as sera from immunized nonhuman primates. Adapted from Patel et al. with permission from Springer Nature [34].

Shinde et al.[35] conducted a phase III clinical trial that compared the safety and immunogenicity of a quadrivalent Matrix-M based vaccine (qNIV) with a traditional quadrivalent inactivated vaccine (IIV4) in 2654 adults aged 65 or older. The safety profiles were comparable between the two vaccinated groups, and serious adverse events were infrequent and balanced. qNIV demonstrated higher hemagglutination-inhibition titers to all four vaccination strains and elicited broader binding against heterologous H3N2 and B-Victoria strains. Importantly, qNIV also induced markedly greater polyfunctional CD4⁺ T-cell responses than IIV4, addressing one of the limitations of the traditional influenza vaccines in older adults.

Several rationally or computationally designed nanoparticles have also been used as vaccine platforms. I53_dn5 is a self-assembling icosahedral nanoparticle composed of two components: I53_dn5A and I53_dn5B [36]. When complexed, the I53_dn5B subunits contain a threefold symmetry axis that allows for the display of intact antigen trimers such as HA. In 2022, Kraft et al. reported on the impact that shielding the underlying nanoparticle with glycosylation, PEGylation, or PASylation had on the immune response to the antigen, such as HA [37]. The various scaffold masking strategies did not enhance anti-HA IgG titers, though nanoparticle-displayed HAs whether with masked or unmasked scaffolds did induce higher anti-HA IgG titers than soluble HA. Ren et al. similarly used I53_dn5 nanoparticles displaying H5 hemagglutinins, though little difference was seen in antibody binding between nanoparticle-complexed H5 and free I53_dn5B-conjugated H5[38].

The I3-01 nanoparticle[39] and its variants form another common engineered nanoparticle platform. The SpyCatcher-mi3 system is a self-assembling I3-01-derived nanoparticle scaffold for vaccine assembly that utilizes the spontaneous isopeptide bond formation between the SpyTag peptide and SpyCatcher protein[40]. In Rahikainen et al., H3 HA from the strains A/Aichi/2/1968 (H3N2) and A/Victoria/361/2011 (H3N2) were also successfully displayed using the SpyCatcher-mi3 system and administered as vaccines to mice. These nanoparticle vaccines demonstrated stronger immunogenicity as compared to free antigen as determined by titers obtained in both an ELISA and indirect ELISA.

Other recent vaccine platforms have also utilized SpyTag/SpyCatcher conjugation to display antigens on the nanoparticle surface. The Cap protein of porcine circovirus type 2 has been modified to incorporate the SpyCatcher protein to display HA, and the resulting vaccine was strongly immunogenic in mice [41]. H5-E2 is a protein nanoparticle vaccine modified to display surface SpyTags that conjugate to H5-SpyCatcher antigen and flagellin-SpyCatcher TLR5 agonist while also containing internal cavity cysteines to conjugate the cysteine-modified TLR9 agonist CpG[42]. Nanoparticles displaying flagellin promoted the secretion of cytokines associated with both adaptive and innate immunity. Additionally, conjugation of H5 to E2 nanoparticles with or without TLR agonists significantly increased H5-specific IgG antibody response compared to free H5. Significantly increased homosubtypic cross-binding was also observed in vaccine groups consisting of H5-E2 with single or dual adjuvants conjugated. Lethal H5N1 challenges demonstrated that mice immunized with antigen bound to E2, adjuvanted or not, were fully protected with minimal morbidity observed in animals receiving H5-E2 conjugated with at least one adjuvant.

The P22-HA_head is a virus-like particle (VLP) vaccine consisting of the HA head domain from A/Puerto Rico/8/1934 (H1N1) incorporating the SpyCatcher protein covalently bonded to the P22 bacteriophage coat protein engineered to incorporate a SpyTag[43]. To determine whether the VLP could elicit a protective immune response against H1N1 infection, mice were immunized intratracheally with P22-HA, HA alone, or an admixture of P22 and HA. Mice immunized with P22-HA experienced insignificant weight loss as compared to other groups. Additionally, the VLP group demonstrated superior anti-HA IgG production and HA neutralization titers. While full survival was observed in mice inoculated with the admixture of P22 and HA head, anti-HA IgG and HA neutralization of antisera from this group was not as effective as in the full VLP group.

The MS2-SA VLP is a similar bacteriophage coat protein platform that consists of an AviTag-modified and biotinylated coat protein from the bacteriophage MS2 coated with streptavidin (SA)[44]. Hemagglutinin from A/Puerto Rico/8/1934 (H1N1) was then attached to the streptavidin via site-specific biotinylation of an AviTag inserted into the C-terminus of the antigen to yield PR8-HA VLP (Figure 2A). PR8-HA VLP, when subcutaneously inoculated in ferrets, elicited significantly higher neutralization titers than control animals, with results significantly enhanced by AddaVax or Quil-A adjuvant (Figure 2B). Ferrets immunized with adjuvanted VLP and intranasally challenged with PR8 ceased shedding virus after 3 days. Serum-neutralizing antibody titers persisted for over 3 years in groups immunized with adjuvanted VLP, with evidence that suggests adjuvanted PR8-HA VLP immunization is effective as a boost immunization as well.

Figure 2. Assembly of PR8HA-VLP and Serum Neutralization from Immunized Ferrets.

A) Schematic representation of PR8HA-VLP. B) Duration of neutralizing antibody titers after single PR8HA-VLP immunization. Adapted with permission from Chiba et al. under the terms of a Creative Commons Attribution License 4.0 (CC-BY 4.0) (https://creativecommons.org/licenses/by/4.0/) [44].

Membrane-enveloped nanoparticles have also found several applications. Hemagglutinins appended with a His-Tag can be chelated to cobalt-bearing liposomes, resulting in a lipid nanoparticle that offers higher protein density and orientation control as compared to conventional liposome vaccines[45]. These HA-liposomes demonstrate improved immunogenicity and protective efficacy against viral challenge at high doses as compared to free antigen. Immunization with HA-liposomes also resulted in significantly higher concentration of the particles in germinal centers and follicular dendritic cell areas as compared to soluble HA. HA-liposomes also enhanced induction of HA-specific B cells and T follicular helper cell (T_FH) responses in mice post-immunization. Co-expression of the influenza HA and M1 proteins in insect cells leads to the formation of VLPs that can be used as vaccines[46,47].

In 2020, a plant-derived quadrivalent virus-like particle influenza vaccine (QVLP) was studied in phase III clinical trials[48]. This vaccine is based on enveloped VLPs expressing HA on the membrane surface derived from transient expression of hemagglutinin protein in Nicotiana benthamiana. Two studies aimed to determine the efficacy of the QVLP as compared to a quadrivalent inactivated vaccine that included two influenza A strains and two influenza B strains with the two studies separated by age: one study spanning ages 18-64 and another study consisting of patients 65 and older. Notably, the QVLP elicited strong CD4⁺ T cell responses to the target proteins in both age groups. Consistent results were also seen across different vaccine lots in an additional phase III trial in 2021 [49]. Additionally, plant-based vaccines demonstrate an ability to strongly elicit T helper cell type 1 (Th1) and CD8⁺ T cell responses[50].

A plant-based VLP vaccine displaying H1 from A/California/07/2009 (H1N1) was used as the platform in a study exploring the elimination of the interaction between HA and sialic acids [51]. In this study, wild type H1 was compared to a counterpart with the Y98F mutation, which prevents sialic acid binding of the HA. Mice immunized with a single dose of either vaccine yielded similar H1-specific IgG titers, but vaccination with H1_Y98F-VLP resulted in significantly higher hemagglutination inhibition (HI) and microneutralization titers 21 days post-vaccination. Evaluation of the CD4⁺ T cell response in the spleen and bone marrow of mice following vaccination with 3 μg of either vaccine resulted in comparable and significant elicitation of IL-2⁺TNFα⁺IFNγ^γ CD4⁺ T cells compared to placebo group. Vaccination with H1_WT- and H1_Y98F-VLP against a viral challenge resulted in complete protection from lethal infection, but H1_Y98F-VLP-vaccinated mice did show significantly reduced lung viral titers five days post-infection compared to the H1_WT-VLP vaccinated mice.

HA-SAV is a self-assembling nanoparticle comprised of HA from A/California/04/2009(H1N1) and the amphiphilic block copolymer mPEG-b-PLA-NHS[52]. Bone marrow dendritic cells treated with the HA-SAV secreted higher quantities of immunostimulatory cytokines such as IL-6 and TNF-α compared to a soluble HA treatment. Intramuscular immunization of mice with HA-SAV results in higher titers of anti-HA IgG and HA inhibition than control groups or free HA, indicating proper stimulation of the humoral immune system by the nanoparticle. Additionally, HA-SAV demonstrated protective efficacy, where mice immunized with the vaccine had higher survival rates and lower weight loss as compared to PBS or free HA-immunized groups. Intracellular cytokine staining for IFNγ also demonstrated enhanced CD4⁺ and CD8⁺ T cell activation in HA-SAV-immunized mice as compared to other experimental groups.

3. Nanovaccines that display HA to induce broader protection against influenza

Another promising application of nanoparticle-based HA vaccines is to create more broadly protective vaccines capable of protecting against multiple types, subtypes, or strains of influenza. This section will review different approaches to accomplish this goal including the combination of multiple HA-containing nanoparticles, the generation of mosaic vaccines that contain different HAs on the same particle, and the display of chimeric and computationally designed HAs on nanoparticle vaccines.

Ferritin-derived nanoparticles are one popular platform for nanoparticle HA vaccines. As shown by Kanekiyo et al. in 2013, Heliobacter pylori ferritin nanoparticles contain a threefold symmetry axis in their crystal structure near the N-terminus of the ferritin monomer, meaning that ferritin could be used both as a trimerization domain and as a nanoparticle platform. As such. hemagglutinin can be genetically fused to ferritin to result in the self-assembly of eight intact HA trimers on each 24-mer ferritin nanoparticle[53]. This technology has continued to be developed by multiple research groups around the world. Tang et al. developed an H1-ferritin nanoparticle vaccine that protected mice and piglets from lethal H1N1 challenge [54]. Wang et al. developed a ferritin-based vaccine displaying the hemagglutinin from A/Darwin/9/2021 (H3N2) that protected mice from a heterologous A/Guizhou/54/1989 (H3N2) challenge after two doses[55].

Additional nanoparticle platforms have also been used to create broadly protective HA-based vaccines. Badten et al. developed a nanoparticle vaccine platform based on the E2 protein from Geobacillus stearothermophilus that self-assembles into a 60-mer particle[56]. To externally display a full hemagglutinin protein on the surface of the E2 particle, the authors conjugated a maleimide-linked tris-nitrilotriacetic acid (tNTA) to an engineered free cysteine (E279C) on E2. Nickel (II) was then added to be chelated by the tNTA groups, and recombinant H1 hemagglutinins from A/California/7/2009 (H1N1) with polyhistidine tags could then be loaded onto the nanoparticles. Sera from mice immunized with these H1-E2 nanoparticle vaccines when adjuvanted with the TLR4 agonist monophosphoryl lipid A showed significantly higher binding against both homologous H1 and heterosubtypic H5 hemagglutinins when compared to sera from mice immunized with similarly adjuvanted soluble H1, indicating the advantages of displaying the HA on a nanoparticle platform.

Liposomes, virosomes, and similar lipid-enveloped nanoparticles provide an additional platform for displaying hemagglutinin on particle surfaces. Sia et al. used a liposome containing cobalt-porphyrin phospholipid (CoPoP) to display His-tagged hemagglutinin trimers on the liposome surface while adjuvanting them with phosphorylated hexaacyl disaccharide[57]. Serum transferred from mice immunized with liposome vaccines displaying the H3 HA from A/canine/Illinois/11613/2015 (H3N2) protected 90% of mice from a lethal challenge with mouse-adapted A/Hong Kong/1/1968 (H3N2), a heterologous virus[57], with similar results also seen after intranasal delivery[58]. In a ferret challenge model, liposome vaccines displaying the hemagglutinin from A/Texas/50/2012 (H3N2) showed improved protection against a 2011 H3N2 strain when compared to AddaVax-adjuvanted soluble HA and control liposome[57].

Another approach enabled by nanoparticle vaccine platforms is the creation of mosaic vaccines in which multiple different hemagglutinins are displayed on the same nanoparticle. This technique, which was pioneered by Kanekiyo et al. in 2019 to display hemagglutinin receptor-binding domains on ferritin[59], is designed to preferentially bind and activate B cells that can recognize multiple different co-displayed antigens. Several other studies have further explored and developed this concept.

Boyoglu-Barnum et al. displayed full HA trimers on the i53_dn5 engineered nanoparticle[60]. As in other HA studies with this nanoparticle, the trimeric i53_dn5B subunit was genetically fused to the HA ectodomain and mixed with the i53_dn5A subunit to result in self-assembled nanoparticles. Two nanoparticle vaccine formulations were tested: a cocktail of homotypic nanoparticles each displaying one HA from the strains used in the 2017-2018 seasonal vaccines and a mosaic vaccine incorporating all four of these HAs into each particle (Figure A-B). Both nanoparticle formulations induced improved heterosubtypic binding than the seasonal inactivated vaccines in mice, ferrets, and nonhuman primates. Notably, both cocktail and mosaic formulations showed improved protection against heterosubtypic H5N1 and H7N9 in ferrets, though no significant difference between the cocktail and mosaic formulations was observed (Figure 3C). A serum transfer study from NHPs to mice showed corresponding results with no significant significance between cocktail and mosaic vaccines (Figure 3D). Similar outcomes between cocktail and mosaic nanoparticle vaccines displaying the full HA trimer were also seen in other platforms based on the AP205 coat protein and on the nanoscaffold mi3[61]. Thus, while the mosaic display of HA receptor binding domains did show a marked improvement in antibody breadth compared to a mixture vaccine[59], the advantage of mosaic display of full HAs over mixtures in eliciting more broadly binding antibodies appears to be less clear.

Figure 3. Structure and protective efficacy of I53_dn5 hemagglutinin mosaic vaccines.

A) Schematic of I53_dn5 mosaic and cocktail vaccines. B) Reconstruction of i53_dn5 vaccine and displayed hemagglutinin antigen. C) Survival curves and lung viral RNA from immunized ferrets challenged with H5N1 or H7N9. D) Survival of mice after serum transfer from immunized NHPs and challenge with H5N1 or H7N9. Adapted from Boyoglu-Barnum et al. with permission from Springer Nature [60].

Enveloped virus-like particles (VLPs) with an M1 core are another common platform to create homotypic and mosaic hemagglutinin vaccines. Park et al. created homotypic influenza VLPs containing M1 and HA proteins from A/Puerto Rico/8/1934 (H1N1) that also incorporated lipid-anchored cytokines GM-CSF and IL-12[62]. These influenza VLPs with lipid-anchored cytokines were able to protect aged mice from a heterologous A/WSN/1933 (H1N1) challenge, albeit with a higher dose than was required for young mice. Mao et al. created mosaic influenza A VLPs containing H1 and H3 hemagglutinins together with an M1 from A/California/04/2009 (H1N1) that after three doses protected mice against two different influenza B challenges[63]. Kang et al. designed a mosaic M1 VLP displaying H5 hemagglutinins from two different H5N1 clades that was able to protect 100% of chickens from lethal challenges from viruses in each clade while protecting 90-100% of chickens at one-tenth of the dose required for complete protection[64].

Lamson et al. developed a “beads on a string” antigen by genetically fusing eight hemagglutinin head domains with glycine-serine-serine intermediate linkers in sets of four and subsequently displayed these on ferritin nanoparticles via SpyTag/SpyCatcher conjugation[65]. After three doses, these nanoparticle vaccines induced high serum antibody endpoint titers in mice against vaccine-matched full HAs, though binding titers were reduced against mismatched HAs. These beads-on-a-string nanoparticle vaccines induced strong neutralization of matched H1N1, H3N2, and H5N1 viruses while neutralization of a matched H7N9 was near background. Interestingly, a DNA-delivered vaccine encoding an H1 HA head domain displayed on an IMX313 self-assembling heptamerization domain — another approach targeting only the HA head — was able to protect mice from a heterologous A/California/07/2009 (H1N1) challenge[66].

The impact of immunization with mosaic nanoparticle vaccines on helper T cell responses was evaluated by Mallajosyula et al. [67]. First, in individuals shown to have biased antibody responses against H1, H3, or influenza B hemagglutinins, the authors showed that antibody binding to these HAs was correlated to activation by HA peptides of the corresponding subtypes of CD4⁺ T_FH cells isolated from those individuals, indicating the potential for MHC II biases towards specific subtypes being important in driving antibody biases. They then hypothesized that mosaic display of HAs on a nanoparticle would help overcome this antibody bias by allowing antigen-presenting B cells to present peptides from all three HA subtypes even if the B cell is specific for only one. The authors used a lumazine synthase (LS) nanoparticle as their nanoparticle platform and covalently attached HAs from H1, H3, and influenza B via a sortase-catalyzed reaction between C-terminal LPETG amino acid motifs on the HAs with a pentaglycine tag on the LS nanoparticle monomers. This hypothesis was then tested in C57BL/6 mice that were predicted to have highest CD4⁺ T cell stimulation by influenza B peptides followed by H1 and H3 subtypes of influenza A (Figure 4A). Mice were either mock immunized, immunized with a mixture of soluble H1, H3, and influenza B HA, or immunized with an LS mosaic nanoparticle vaccine displaying all three hemagglutinins. The resulting antibody titers indicated that while the expected IgG binding bias towards influenza B and H3 hemagglutinin was borne out in the mixture-immunized mice, the mosaic-immunized mice showed no bias in serum IgG binding to HA (Figure 4B). This lack of IgG bias after mosaic vaccination was then further confirmed in a human organoid model (Figure 4C) where both a commercial inactivated influenza vaccine and a cocktail of homotypic HA-decorated LS nanoparticles showed a subtype bias in their antibody binding profiles whereas the mosaic nanoparticles showed much less of a subtype bias (Figure 4D). The mosaic nanoparticle also resulted in significantly increased T_FH activation compared to the nanoparticle cocktail and inactivated virus formulations (Figure 4E). Finally, incubation of the organoids with the phagocytosis inhibitor Pitstop 2 indicated that serum antibody binding to these HAs was reliant on successful uptake of antigen.

Figure 4. Analysis of T cell help for mosaic HA nanoparticles.

A) Predicted number of high-affinity peptides to MHC II allele found in C57BL/6. B) Area under the curve from IgG titration against H1, H3, and influenza B hemagglutinins from vaccinated mouse sera. C) Schematic of human tonsil organoid. D) ELISA binding of organoid cultures to H1, H3, B/Victoria, and B/Yamagata hemagglutinin after incubation with vaccine candidates. E) Activation of CD4⁺ helper T cells in tonsil organoids after incubation with vaccine candidates. F) Total antibody AUC with and without antigen uptake inhibitor Pitstop 2. Reproduced from Mallajosyula et al. Science, 10.1126/science.adi2396 2024, AAAS [67].

Nanoparticle vaccines have also been used to display chimeric or computationally designed hemagglutinins. In contrast to the mosaic approach that displays different hemagglutinins on the same nanoparticle, a hemagglutinin could be designed to combine domains, epitopes, or segments from multiple different influenza strains into a single hemagglutinin molecule that could then be displayed as the antigen.

Huang et al. designed a “Digitally Immune Optimised and Selected” H5 hemagglutinin (DIOSvax-H5_inter) based on a phylogenetic analysis of selected H5 amino acid sequences and displayed it on mi3 nanoparticles via SpyTag/SpyCatcher conjugation[68]. This nanoparticle vaccine elicited TNF⁺IFNγ^γ CD4⁺ T cells that could be activated by peptides from both Clade 2.1.3.2 H5 and Clade 2.3.4.4b H5 and also induced neutralizing antibodies against H5N1 viruses from 12 different clades with 100% of mice having inhibitory dilutions at 50% infectivity of more than double the detection limit, indicating that this single antigen when presented on a nanoparticle was capable of eliciting broad T cell and B cell derived immunity against H5N1.

A similar concept was deployed against influenza B by Zhao et al. in an M1 VLP platform[69]. The authors designed two hemagglutinin antigens, one per influenza B lineage, based on 4122 sequences from the B/Victoria lineage and 2745 sequences from the B/Yamagata lineage with the aim of maximizing the coverage of each lineage by predicted T cell epitopes from the designed sequence. Mice were then immunized either with PBS control, a commercial quadrivalent inactivated vaccine, or M1 VLPs displaying either the B/Yamagata HA (VLP-BY), the B/Victoria HA (VLP-BV), or both in a mosaic VLP (VLP-BI). VLP-BV and VLP-BI both induced highly cross-reactive sera against a broad range of influenza B hemagglutinins, and all VLP-immunized mice induced higher IFNγ responses than the inactivated virus, indicating a stronger Th1 bias, though the mosaic VLP also induced greater a greater IL-4 response than the homotypic VLPs or the inactivated virus, indicating a more potent Th2 response for the bivalent VLP. Finally, both homotypic VLPs provided complete protection in mice against challenges with lineage-matched influenza B viruses, but the protection offered against mismatched viruses was not evaluated.

Madapong et al. designed consensus H1 and H3 hemagglutinin antigens (H1-con and H3-con, respectively), evaluated their immunogenicity in mice either as soluble protein adjuvanted with AddaVax or displayed on the surface of tobacco mosaic virus (TMV) coat protein by chemical conjugation to engineered lysine residues in the TMV coat protein adjuvanted with AddaVax, and compared them to a commercial trivalent inactivated virus vaccine[70]. Both soluble and TMV-bound H1-con vaccines elicited significantly higher serum IgG binding than PBS control against 1934, 2007, and 2009 H1 hemagglutinins after two doses with the nanoparticle-bound HA outperforming the soluble H1-con in some cases. A similar trend was seen for the H3-con against 2004, 2005, and 2009 H3 hemagglutinins with the TMV-H1-con outperforming H1-con in all cases after the boost dose. Additionally, the TMV-H1-con demonstrated 100% protection in mice against A/Puerto Rico/8/1934 (H1N1) and A/Fort Monmount/1/1947 (H1N1) challenges while the commercial vaccine only had 20% protection against A/Fort Monmount/1/1947 (H1N1). Surprisingly, the TMV-H3-con vaccine only showed 20% protection against challenges with A/Texas/1/1977 (H3N1) and A/Aichi/2/1968 (H3N2) compared to 100% and 80% protection for the soluble H3-con vaccine against the 1977 and 1968 viruses, respectively.

Finally, the Ross group have developed multi-level consensus antigens termed “Computationally Optimized Broadly Reactive Antigens” (COBRAs) against different influenza A subtypes and then displayed these antigens together with neuraminidase on human immunodeficiency virus (HIV) Gag VLPs. A variety of H1 HA COBRA antigens have been developed with P1, an antigen designed to cover 1933-1957 and 2009-2011 H1 HAs, and its derivatives providing 100% protection in mice against A/California/07/2009 (H1N1) challenge but 0-60% protection against A/Brisbane/59/2007 (H1N1); the best performing derivative against A/Brisbane/59/2007, termed V3, included Sa and Cb regions that had been copied from X6, a COBRA sequence targeting 1999-2012 hemagglutinins[71]. Two more recent COBRA antigens Y2 and Y4 designed using sequences from 2014-2016 (Y2) and 2016-2019 (Y4) provided superior protection in mice relative to P1 and X6 against 2009 and 2018 H1N1 challenges with Y2 and Y4 both conferring complete protection compared to 100% and 60% for P1 against 2009 and 2018 H1N1s, respectively, and 80% and 40% for X6[72]. A large panel of second generation COBRA H3 HAs were also developed and displayed on HIV Gag VLPs with TJ-2 (based on 2002-2005 H3 HAs) best neutralizing a 2007 H3N2 virus while TJ-5 (based on 2008-2012 H3 HAs) potently neutralized H3N2 viruses from 2012-2016[73]. These TJ-2 and TJ-5 COBRA VLPs also induced broad neutralization against H3N2 strains in pre-immune ferrets[74] Meanwhile, additional antigens against H7[75] and influenza B[76] have been recently developed as well.

4. Nanovaccines that target the conserved stalk domain of HA

Targeting the conserved stalk domain of HA has also been a popular strategy to create a broadly protective influenza vaccine. In contrast to the highly plastic HA head domain that undergoes continuous antigenic drift, the HA stalk domain is more structurally constrained and less tolerant to mutations within and across subtypes [77]. Further, the broadly neutralizing antibodies that target the stalk such as CR6261 and CR8020 have shown cross-binding and cross-neutralization across multiple HA subtypes [78,79]. This property makes the HA stalk an attractive target for the development of broadly protective influenza vaccines. Many stalk-targeting approaches such as stalk-only antigens [80–86], sequential immunization with chimeric HAs [87–90], antigenic suppression[91,92], and antigenic reorientation[93,94] are currently being developed. Here, we will discuss recent applications of nanoparticle platforms in designing vaccines targeting the HA stalk domain.

Displaying full-length HA on a nanoparticle can induce an immune response towards the HA stalk due to pre-existing immunity when an individual is immunologically naïve to the HA head but not to the HA stalk. Houser et al. [95] performed a human phase I clinical trial to evaluate the safety and humoral response elicited by a ferritin nanoparticle vaccine displaying an H2 hemagglutinin (H2HA-Ferritin). The study was performed with 50 healthy adults who each received a single 20 μg dose of H2HA-Ferritin, homologous 60 μg doses of H2HA-Ferritin, or a heterologous prime boost regimen of full-length H2 HA DNA prime and H2HA-Ferritin boost. Vaccination in all regimens were well tolerated with no adverse events. The H2HA-Ferritin was compared between H2-naive adults and H2-exposed adults in both homologous and heterologous regimens to determine the effect of pre-existing immunity on the humoral response. The H2-exposed adults would have had pre-existing immunity towards the H2 HA head, likely diminishing the magnitude of subsequent stalk-targeting responses. In contrast, H2-naïve adults would likely have very limited pre-existing immunity against the H2 head domain while having some pre-existing anti-stalk immunity from immunization or infection with other subtypes. Indeed, H2HA-Ferritin immunization in H2-naïve adults increased antibody binding titers against heterologous group 1 HA stalks compared to before immunization, induced neutralization against H2N2, H5N1, and H6N1 after a boost, and yielded ADCC-activating antibodies against H1 HA stalk. In contrast, H2-exposed adults exhibited higher baseline IgG binding, neutralization, and ADCC activity but with modest increase after vaccination. Nevertheless, cross neutralizing antibodies from both H2-naïve and H2-exposed groups were primarily stalk-directed as the addition of extra stalk-only protein to neutralization assays with an H5N1 virus significantly inhibited neutralization (Figure 5).

Figure 5. Inhibition of H5N1 neutralization by competing antigens.

Percent inhibition of A/Vietnam/1203/04 (H5N1) neutralization due to the introduction of competing antigens: negative control (DSCav-1 RSV F protein), full-length H2 hemagglutinin, or H2 stalk -only construct. Geometric mean inhibition with 95% confidence intervals. Reproduced from Houser et al. with permission from Springer Nature [95].

Several studies have used nanoparticles to improve the immunogenicity of HA stalk-only constructs. Thrane et al. [96] displayed the H1 HA stalk (mini-HA) from Impaggliazzo et al. [80] on the coat protein from the bacteriophage AP205 using the SpyTag and SpyCatcher sequences fused to mini-HA and AP205 respectively to form capsid-like particle (CLP) vaccines termed CLP-HA_stem. In immunized mice, CLP-HA_stem induced stronger antibody responses against HA stalk, drove IgG2a isotype switching, and displayed greater protection against homologous and heterologous challenges compared to soluble mini-HA. Zhu et al. [97] displayed mini-HA on lumazine synthase nanoparticles using SpyTag/SpyCatcher conjugation and observed enhanced T cell responses and neutralizing antibodies in mice compared to those in mice immunized with the soluble mini-HA. Kar et al. [98] displayed the previously stabilized H1 HA stalk (pH1HA10) from Mallajosyula et al. [82] on MsDps2, encapsulin, and ferritin nanoparticles and compared their immunogenicity against soluble pH1HA10. However, the pH1HA10 nanoparticles did not show a notable improvement over the soluble pH1HA10 in the context of this study.

Multiple studies have also analyzed the humoral responses elicited by HA-stalk nanoparticle vaccines. Darricarrère et al. [99] evaluated the broadly neutralizing antibodies (bnAbs) elicited by immunization with stabilized H1 and H3 HA stalk-only HA antigens displayed on ferritin nanoparticles (H1ssF, H3ssF) [85,86] in cynomolgus macaques, a non-human primate (NHP). Previous studies with these constructs had shown protective responses in mice and ferrets without broad neutralization, but the addition of AF03 adjuvant in NHPs led to potent neutralization of H1N1 and H5N1 by H1ssF antisera and of H3N2, H10N8, and H7N9 by H3ssF antisera. Repertoire analysis revealed that H1ssF-induced cross-reactive antibodies were primarily derived from the IGHV3 class of heavy chain variable gene segments while H3ssF-induced antibodies were predominantly enriched in IGHV4 gene segments. Interestingly, most human bnAbs against group 1 HAs are derived from IGHV1-69[100], a gene segment which is not present in cynomolgus macaques. Despite the genetic deviation from human bnAbs, structural analysis revealed that the epitopes of macaque stalk-targeting bnAbs extensively overlapped with the epitopes of human stalk-targeting bnAbs such as CR6261, MEDI8852, and CR8020. As such, antibodies across multiple antibody gene segment classes may be potential bnAb precursors,, and sequential vaccination with stalk-only antigens from different groups may be able to direct bnAb development to uncommon but broadly-protective antibody motifs[101].

Moin et al. [102] developed a stabilized H10 HA stalk ferritin nanoparticle vaccine (H10ssF) and tested its immunogenicity in mice, ferrets, and cynomolgus macaques. In mice, H10ssF provided strong group 2 heterosubtypic protection with survival rates higher than the previous H3ssF and H7ssF vaccines in lethal challenges with H3N2 and H7N9, respectively. Critically, co-immunization with H1ssF and H10ssF elicited broad cross-group neutralizing responses across all three animal models tested (Figure 6). A bnAb isolated from co-immunized cynomolgus macaques (789-203-3C12) showed cross-group neutralizing activity and protective efficacy in mice against multiple group 1 and group 2 influenza strains.

Figure 6. Cross-binding and neutralizing antibody responses in mice, ferrets, and cynomolgus macaques elicited by co-immunization with H1ssF and H10ssF.

A) Antibody binding titers to group 1 and group 2 HAs in mice. B and C) Neutralizing antibody responses to pseudotyped lentiviruses expressing multiple group 1 and group 2 HAs in mice (B) and ferrets (C). D) Neutralizing antibody responses in cynomolgus macaques. Reprinted from Immunity, 55, Moin et al., “Co-immunization with hemagglutinin stem immunogens elicits cross-group neutralizing antibodies and broad protection against influenza A viruses”, P2405-2418.e7, Copyright (2022) with permission from Elsevier. [102]

With the recent success of the stalk-only constructs in animal models, several similar vaccines have gone through human clinical trials. Widge et al. [103] evaluated the safety and humoral responses of H1ssF in a human phase I clinical trial. Fifty-two healthy adults were immunized with a single 20 μg dose or two 60 μg doses of H1ssF. H1ssF was well tolerated and caused no adverse events amongst the participants. Analogous to the previous animal studies with H1ssF [86], cross-reactive responses against heterologous group 1 HAs were observed. Further, H1ssF elicited both enhanced ADCC activity in antibodies directed against H1 HA stalk compared to the pre-immunization baseline and durable cross-reactive neutralizing antibody responses that persisted for more than a year after vaccination.

Andrews et al. [104] analyzed the humoral response after H1ssF vaccination from the two dose sera from the same phase I clinical trial. H1ssF vaccination resulted in a robust H1 HA stalk specific plasmablast response, and two HA stalk binding populations were observed in flow cytometry (Figure 7A) with improved anti-stalk response after H1ssF vaccination compared to the commercial quadrivalent inactivated vaccine but equivalent anti-stalk response across H1ssF doses and age groups (Figure 7B–C). The two plasmablast populations bound to two distinct HA stalk epitopes (Figure 7D–G): 1) a central stem epitope-binding population that expressed antibodies with heterosubtypic binding and neutralization across group 1 HAs and viruses, typically encoded by IGHV1-69 class antibodies; and 2) an anchor epitope-binding population that expressed antibodies with binding and neutralization restricted to H1 HAs and viruses respectively, typically encoded by IGHV3+NWP class antibodies (Figure 7H). The binding specificity of these two populations were confirmed by introducing glycans into the central and anchor epitopes at positions 45_HA2 and 27_HA2 respectively (Figure 7I–K) which inhibited the binding of known antibodies targeting these epitopes. (Figure 7L). These glycan variants similarly inhibited the binding of each population of plasmablasts to HA, confirming their specificity for the central and anchor epitopes (Figure 7M). Moreover, H1ssF vaccination induced a stronger memory B cell response towards HA stalk than the seasonal quadrivalent influenza vaccine but had comparable binding and neutralization profiles. Importantly, participants with pre-existing immunity against H2 had fewer anchor-targeted B cells and more central epitope-focused B cells than the H2 naive participants, which led to more cross-binding.

Figure 7. HA stalk epitopes targeted by H1ssF-elicited B cells.

A). Flow cytometry plot of plasmablasts from H1ssF-immunized humans labeled with H1 stalk-only and H1 full HA ectodomain. B and C) Number of cells with specificity for H1 stalk domain per million B cells when grouped by (B) vaccine or (C) patient age and vaccine dose. D) Percentage of B cells specific for H1 stalk that were in the lower population as determined by flow cytometry. E) Gene class paitrings for heavy and light chains for antibodies isolated from the upper stalk population, lower stalk population, or neither. Lines between upper half and lower half correspond to observer heavy chain and light chain pairings, with the most common gene classes assigned colors. F) Location of central and anchor epitopes on HA. G) Percentage of antibodies in upper, lower, and negative classes encoded by VH1-69 and VH3+NWP gene classes. H) Structure of an isolated mAb 204-1B06 complexed with H1. I, J, and K) Cartoon depicting interference of glycans at 27_HA2 and 45_HA2 with F_ab binding to the HA stalk at central and anchor epitopes. L) Binding area under the curve (AUC) of mAbs against wildtype H1 NC99 (H1-WT), H1 with a glycan at 27_HA2 (H1-N27), H1 wth a glycan at 45_HA2 (H1-N45), and H1 with glycans at both 45_HA2 and 27_HA2 (H1-N45/27). M) Flow cytometry plots of memory B cells labeled with wild-type H1 variants. Reproduced from Andrews et al., Science Translational Medicine, 10.1126/scitranslmed.ade497 2023, AAAS [104].

Casazza et al. [105] performed a human phase I trial evaluating the safety and humoral responses of H10ssF. 25 healthy adults were immunized with a single 20 μg dose or two 60 μg doses of H10ssF. In parallel to the H1ssF clinical trial, H10ssF was well tolerated without any adverse events amongst the participants. H10ssF vaccination resulted in strong sera antibody binding against H10 HA stalk and full length H10 HA, remaining significantly above baseline throughout the duration of the study (40 weeks). Neutralizing antibody titers against H10N8 rose after both doses, and ADCC activity was observed against full-length H10 HA. Strong sera binding against group 2 HAs was also observed, but the neutralizing titers against heterologous H3N2 and H7N9 viruses were limited.

Mantus et al. [106] characterized the HA stalk-specific humoral response elicited by three group 2 immunogens from past clinical trials: H10ssF from Casazza et al[105], H3 quadrivalent influenza vaccine (H3 QIV, clinical trial NCT01132859), and H7 monovalent inactivated vaccine (H7 MIV, clinical trial NCT02206464). The group 2 anti-stalk antibodies predominantly targeted the central epitope or lower epitope of HA stalk (Figure 8A, E). It was revealed that H7 MIV vaccination elicited B cell response primarily towards the central epitope and generated cross-group binding and neutralization; H10ssF or H3 QIV vaccination could bias the B cell response towards the lower epitope (Figure 8A, E) and induced narrower cross-binding but more potent binding and neutralization of recent H3N2 strains. However, there was some variability in the biases of H10ss-immunized individuals with some participants biased toward the central epitope (Figure 8B–D). In general, however, the lower epitope antibodies correlated with stronger neutralization and protection in mice against H3N2 compared to the central epitope binding antibodies, while central epitope antibodies were correlated with cross-group breadth in binding and neutralization.

Figure 8. Epitope mapping of human sera after vaccination with group 2 HA subtypes.

A) Percentage of mAbs isolated from vaccinated individuals binding the central epitope or lower (anchor) epitope on HA stalk. B) Epitope-biased binding profiles of plasmablasts for two H10ss-vaccinated participants. C) Electron microscopy polyclonal epitope mapping (EMPEM) of sera from participants in (B). D and E) Percent reduction in binding to HA stalk after depletion of central-epitope targeting or lower-epitope targeting antibodies for (D) individual participants in (B) or (E) for participants in all three vaccine regimens. Reproduced from Mantus et al. Science Translational Medicine, 10.1126/scitranslmed.adr8373 2025, AAAS [106].

An alternative strategy to stalk-only immunogens on VLPs is to invert the orientation of full-length HA on VLPs, thereby exposing the stalk domain outward and improving its accessibility to immune cells while limiting the accessibility of the head domain. This approach avoids the challenges of engineering full-length stalk constructs while overcoming the immunodominance of the HA head. Frey et al. [93] first developed an inverted H1 HA vaccine using MS2 bacteriophage VLPs covered with streptavidin linkers and H28-D14 antibody fab fragments. The Fabs on VLPs bind to the apex of HA head, displaying the HA in an inverted orientation (Figure 9A). Compared to VLPs displaying HA in a regular orientation, the inverted HA VLPs elicited significantly stronger anti-stalk responses in mice, inducing 10-fold higher anti-stalk antibodies after prime and 5-fold higher after boost. This enhanced HA stalk-focused immunogenicity translated into broader sera antibody binding against both influenza A and B HAs and superior protection against a cH6/1 challenge virus, a chimeric influenza virus with HA head from H6 and stalk from H1 (Figure 9B).

Figure 9. Assembly and protective efficacy of VLP-HA_Regular and VLP-HA_Inverted.

A) Schematic of assembly of VLP-HA_Regular and VLP-HA_Inverted. B) Protective efficacy of VLP-HA_Regular and VLP-HA_Inverted against a cH6/1N5 challenge in various doses. Adapted with permission from Frey et al. (Advanced Healthcare Materials 2023). Copyright © 2023 Wiley-VCH GmbH. [93].

Xu et al. [94] inverted the display of H2 HA on alum adjuvant by inserting stretches of charged aspartate residues on the head domain of HA. The aspartate residues anchored the HA head to the alum, exposing the stalk domain outwards. The inverted H2 HA on alum (reoH2HA) was further adjuvanted with CpG and was compared with H2 HA without an adjuvant. Mouse immunization studies showed that reoH2HA and H2 HA sera had comparable levels of antigen-specific antibody titers, but reoH2HA elicited 10-fold higher H1 HA stalk directed titers. reoH2HA vaccinated sera also elicited antibody titers with significantly greater cross-binding against both group 1 and group 2 HAs compared to the H2 HA sera.

5. Nanovaccines that display other antigens along with HA

Hemagglutinin is not the only influenza membrane protein that has been used as a vaccine antigen; neuraminidase (NA) and matrix 2 protein (M2) are two other membrane proteins that have also been employed in vaccine candidates to provide non-HA directed protection. As such, vaccines containing one or more of these antigens in addition to HA may be able to induce multiple mechanisms of protection from a single vaccine formulation. This section will review nanoparticle-based vaccines that co-display HA with NA, M2, or both.

HA- and NA-displaying enveloped VLPs have been employed by several research groups as a vaccine platform. Kang et al. used an M1 VLP displaying H5 HA (HA VLP), N1 NA (NA VLP), or both (HANA VLP) to compare the protection provided by each component [107]. In a challenge with a vaccine-matched H5N1 in BALB/c mice, mice immunized with the HANA VLP provided 100% protection at 11 days post-infection compared to only 60% survival at 11 days post-infection for HA VLP-immunized or NA VLP-immunized mice. The same research group also produced similar vaccines with antigens from other neuraminidase subtypes[108] or from influenza B[109]. Liu et al. designed M1 VLPs displaying computationally designed antigens on H1 and N1 VLPs (VLP-1-1) or H3 and N2 VLPs (VLP-3-2) and compared them to commercially available quadrivalent inactivated influenza vaccines (QIV) for the 2020-2021 or 2021-2022 seasons [110]. VLP-1-1 and VLP-3-2 provided complete protection in BALB/c mice against A/Puerto Rico/8/1934 (H1N1) and A/Aichi/2/1968 (X-31) (H3N2) challenges at 14 days post-infection while QIV-immunized mice had all succumbed by day 9 for both viruses. Sun et al. developed mosaic influenza VLPs with bovine immunodeficiency virus Gag protein forming the VLP core instead of the M1 matrix protein[111]. Mosaic Gag VLPs displaying both H3 and H9 hemagglutinins together with N2 neuraminidase protected chickens against both H3N2 and H9N2 challenge to a comparable degree as commercial virus-matched homotypic vaccines.

CoPoP-functionalized liposomes were also employed to co-display HA and NA. In one study, liposomes displaying eight different HAs, with seven different influenza A HAs and one influenza B HA, and two neuraminidases were compared to a homotypic liposome displaying the H5 hemagglutinin from the challenge strain [57]. While the H5-displaying liposome protected 100% of mice from the lethal challenge, only 40% of mosaic-vaccinated mice survived, although this survival rate was significantly increased compared to an alum-adjuvanted soluble mix of the eight HAs and two neuraminidases in which all mice succumbed by six days post-infection. In a second study, these CoPoP liposomes were used to create “hexaplex” nanovaccines that display three hemagglutinins and three neuraminidases on the same liposomes [112]. These hexaplex vaccines provided 100% protection in BALB/c mice against mouse-adapted H1N1 and influenza B challenges but only provided 50% protection against mouse-adapted H3N2 challenge. Meanwhile, an H3-only liposomal vaccine provided 100% protection against the H3N2 challenge. Thus, for the liposome vaccines, vaccine-matched homotypic vaccines appear to provide better protection in mice than the mosaic vaccines, though the breadth of protection afforded by the homotypic vaccines was not evaluated.

Various vaccines have co-displayed the M2 protein or just its ectodomain (M2e) together with HA on nanoparticles. Sheng et al. conjugated a heterobifunctional linker to lysine residues on horse spleen apoferritin using an NHS ester while M2e and/or reduced H1 hemagglutinin antigens were coupled via free cysteines to the maleimide group on the opposing end of the linker[113]. The dual M2e and HA-displaying apoferritin vaccine completely protected mice from a homologous A/Puerto Rico/8/1934 (H1N1) challenge while providing partial protection against a heterologous H1N1 challenge. Zhao et al. created three different H. pylori ferritin-displayed nanovaccines consisting of an H1 HA displayed on ferritin (HA-f), a genetic fusion of H1 and three repeats of M2e displayed on ferritin (HM-f), and the H1-M2e genetic fusion also fused to an M cell targeting ligand Co4B, all displayed on ferritin (CHM-f) [114]. Two doses of CHM-f or HM-f adjuvanted with CpG and delivered intranasally provided complete protection in mice against H1N1, H3N2, H5N8, and H9N2 challenges, while adjuvanted HA-f vaccines provided only partial protection. The coat protein of the T4 phage have also been used as a vaccine platform to elicit strong humoral and cellular immune responses while imparting protection against viral challenge[115].

Layer-by-layer (LBL) HA-4M2e NPs are nanoparticle vaccines that consist of nanoclusters of four M2e consensus sequences (4M2e) coated with HA from A/California/04/2009 (H1N1) and layered in a chitosan-CpG-chitosan adjuvant coating[116]. The vaccine demonstrated strong anti-HA and anti-M2e humoral responses, especially when administered intranasally in a prime-boost manner. Intranasal delivery of LBL HA-4M2e NPs also significantly improve HA-specific CD4⁺ and CD8⁺ T cells in the lungs and spleen. In lethal virus challenge, intramuscular immunization resulted in no protection for most mice. In contrast, intranasally immunized groups demonstrated high survival rates and low weight loss. While intranasal immunization with HA-4M2e demonstrated higher survival rates in mice than LBL HA-4M2e (100% vs. 80%), lung viral titers 10.5 weeks after immunization were most significantly reduced in the LBL HA-4M2e group relative to HA-4M2e.

Some vaccines co-displayed just the HA stalk with M2e antigens. Park et al. [117] developed a self-assembling protein nanocage (SAPN) using homotrimeric and heterodimeric coiled-coil motifs to display the H1 head-removed HA stalk (HrHA) on the outside of the nanoparticle and 4M2e on the inside. Vaccination with HrHA-SAPNs elicited higher IgG2a titers against homologous, heterologous, and heterosubtypic HAs compared to soluble HrHA, suggesting enhanced Fc-mediated cellular responses such as antibody dependent cellular cytotoxicity (ADCC). Notably, no antibody response was detected against the empty SAPN scaffold, mitigating concerns of scaffold-directed immunity associated with VLP platforms. In addition, HrHA-SAPNs promoted dendritic cell activation and interestingly increased IL-4⁺ CD4⁺ and CD8⁺ T cell populations, indicative of a Th2-biased response. This contrasted with the increased IgG2a isotype profile, which is associated with Th1-biased immunity. The authors interpret these findings as evidence that HrHA-SAPNs induce a balanced humoral and cellular response, engaging both Th1- and Th2-associated pathways.

Heinimäki et al. displayed an H1 hemagglutinin HA2 subunit and consensus M2e on the surface of norovirus VP1 protein nanoparticles using SpyTag/SpyCatcher conjugation that induced high IgG serum binding against H1 in mice but only minimal binding to M2e and neutralization of H1N1 virus[118]. Shi et al designed M1 VLPs co-displaying the stalk domain of H3[119] with M2e that induced high serum and mucosal antibody binding titers against both H3 and M2e in serum and nasal washes and provided 100% protection against H7N9 and H3N2 challenges [120]. Zykova et al. developed fusion proteins of a self-assembling protein domain, group 1 hemagglutinin consensus HA2, and M2e that formed nanoparticles of 30-50 nm that also protected 90% of immunized BALB/c mice from challenge with H1N1 and 100% of immunized mice from challenge with H3N2[121].

Finally, some vaccines co-displayed both NA and M2e with hemagglutinin on nanoparticles. Mao et al. designed M1 VLP vaccines to compare the co-display of H3 hemagglutinin with N1 and N2 neuraminidases to the co-display of H3 with N1 neuraminidase and a repeat of human, swine, and avian M2e (M2e5x) in mice [122]. Both groups displayed high serum binding after three doses to H1N1, H3N2, and H5N1 viruses though the M2e-containing vaccine group did show higher binding compared to the N2 group against H1N1 and H5N1 viruses. While both groups fully protected mice from challenges with H1N1, H5N1, and H3N2, the M2e-containing vaccine group showed significantly less virus in the lungs compared to the N2-containing group, indicating the utility of including a third different antigen rather than two different NA subtypes. Imagawa et al. [123] and Nerome et al. [124] developed VLPs produced in silkworm pupae that co-expressed hemagglutinin with a “chimeric cytokine” (CC) fusion of M2, the NA stalk, and interleukin 12 (IL-12), but found that a mixture of CC-displaying and HA-displaying VLPs provided better protection against H1N1 challenge than the co-display of CC and HA on the same particle [123].

6. Conclusion

Nanoparticle-based vaccines displaying hemagglutinin (HA) have advanced from preclinical studies to clinical trials, demonstrating their capacity to elicit immune response against HA. These various nanoparticle strategies have elicited robust neutralizing antibodies, Fc-mediated effector functions, T cell responses, and more, translating into enhanced protection against homologous, heterologous, and heterosubtypic viral challenges in animal models.

Several of these nanovaccine platforms – in particular, the stalk-directed vaccines – show promise as universal influenza vaccine candidates. Ferritin nanoparticle display of full-length H2 or stalk-only HA antigens appear closest to clinical translation with encouraging results in phase I trials. Nanoparticle vaccines that control the orientation of HA to elicit a stalk-directed response also show promise in preclinical animal models, but the ability to manufacture these vaccines at scale has yet to be explored. The mosaic display of full-length hemagglutinin is another possible approach to generating a broadly protective vaccine, but its ability to induce better outcomes in preclinical models when compared to an equivalent admixture of nanovaccines remains currently unclear. While the mosaic approach would in theory preferentially stimulate B cells capable of binding to multiple subtypes, the humoral responses between mosaic-immunized and admixture-immunized mice have been similar, even against vaccine-mismatched viral strains. However, the benefit to cellular immunity of having multiple distinct hemagglutinins physically linked to each other as a mechanism for overcoming T cell biases in individuals is a promising result and is worthy of additional exploration.

By presenting HA in multivalent arrays, nanoparticles not only enhance the immunogenicity of HA but also enable various strategies to focus immunity on conserved HA epitopes or present multiple influenza antigens simultaneously. These various nanoparticle strategies have elicited robust neutralizing antibodies, Fc-mediated effector functions, T cell responses, and more, translating into enhanced protection against homologous, heterologous, and heterosubtypic viral challenges in animal models. Further, encouraging results in the clinical trials for these vaccines such as the ferritin nanoparticles displaying HA or HA stalk demonstrate that these nanoparticle platforms could be applicable for humans as well.

Together, these trends highlight the potential of HA-displaying nanovaccines as a foundation for the next-generation influenza vaccines, capable of providing durable, broad, and cross-protective immunity against both seasonal stains and potential pandemic strains of influenza.

7. Future Perspectives

Looking ahead, although the portfolio of HA-displaying nanovaccines is highly diverse and promising, several limitations must be addressed before these candidates can advance towards the clinic. A major challenge lies in scalability and cost, which will be critical for enabling large-scale manufacturing and broad accessibility of these candidates. Even after efficacy is demonstrated in clinical trials, developing robust and cost-effective strategies for antigen and nanoparticle production remains essential. Another consideration is the impact of pre-existing immunity. Clinical trials with HA stalk-based nanovaccines have highlighted how prior influenza exposure can significantly influence humoral response profiles. Therefore, future vaccine strategies must account for these effects and be tailored to diverse populations, especially for older adults with different influenza exposure histories and greater risk of severe disease. Additionally, anti-scaffold immunity, which occurs from the immune response against the nanoparticle itself, is a potential concern. Antibody responses against the scaffold could divert the antigen-specific response, reduce the effectiveness of the boost due to the pre-existing immunity towards the scaffold, and complicate the reuse of the same nanoparticle platform for different viral antigens. Although strategies such as PEGylation, PASylation, and glycosylation have been explored to mask the scaffold, these modifications can weaken antigen-specific responses and pose significant challenges for large-scale implementation. Lastly, alternative vaccination platforms such as intranasal delivery would need to be further explored. Most studies in this review have relied on the traditional intramuscular injection, but intranasal delivery remains underrepresented despite its distinct advantages. By eliciting mucosal immunity at the respiratory tract, the primary site of influenza infection, this approach could complement IgG responses and provide additional protection at the site of infection.

Table 1:

Summary of HA-displaying nanovaccines published since 2020.

Platform	Antigens	Adjuvants	Animal Models	Clinical Stage	Challenge outcome	Immune outcome	References
HA homotypic nanovaccines
Saponin nanoparticle	H1, H3, H5 influenza B HA	Matrix-M	NHPs	Phase III	n/a	High HA-binding, HAI, and neutralizing Ab titers; Strong CD4+ T cell response	[30–31, 34–35]
Protein nanoparticle	H1, H3, H5	AddaVax, alum, flagellin, CpG, Quil-A, monophosphoryl lipid A	Mice	n/a	Protection in mice against H1N1 with H1 antigen, against H3N2 with H3 antigen, and against H5N1 with H5 antigen	High HA-binding, HAI, and neutralizing Ab titers, Strong CD4+ and CD 8+ T cell responses	[37–38, 40–44, 54–56, 66]
Liposome	H1, H3	AddaVax, phosphorylated hexaacyl disaccharide, ISA720, alum	Mice, ferrets	n/a	Protection in mice against H1N1 with H1 antigen and against H3N2 with H3 antigen; protection in ferrets against H3N2 with H3 antigen	High HA binding and HAI titers, Strong CD4+ T cell response	[45, 57–58]
Influenza Ml VLP	H1	GM-CSF, IL-12	Mice	n/a	Protection in mice against H1N1 with H1N1 antigen	High HA-binding, HAI, . Strong CD4+ and CD8+ T cell responses	[46–47, 62]
Plant VLP	H1, H3, influenza B HA	None	Mice	Phase III	Protection in mice against H1N1 with H1 antigen	Strong CD4+ and CD8+ T cell responses; Th1 bias	[48–51]
Polymer nanoparticle	H1	AddaVax	Mice	n/a	Protection in mice against H1N1 with H1 antigen	High binding and HAI titers; strong CD4+and CD8+ T cell responses	[52]
HA mosaic nanovaccines
Protein nanoparticle	H1, H2, H3, H4, H5, H7, H9, H10, influenza B HA	AddaVax, Sigma Adjuvant System	Mice, Ferrets, NHPs, Human organoid	n/a	Protection in mice against H5N1 and H7N9 with H1 + H3 + Influenza B HA antigens; protection in ferrets against H5N1 and H7N9 with H1 + H3 + Influenza B HA antigens Protection in mice against influenza B HA antigens	High HA-binding, HAI and neutralization titers. Strong CD4+ Tfh response	[60–61, 65, 67]
Influenza Ml VLP	H1, H3, H5	None	Mice, chickens	n/a	Protection in mice against influenza B with H1 + H3 antigens; protection in chickens against H5N1 with H5 antigens	High HA-binding antibody titers. Elicited ADCC- activating antibodies. Strong CD4+ T cell response.	[63–64]
Computationally-designed HA nanovaccines
Protein nanoparticle	H1, H3, H5	AddaVax	Mice	n/a	Protection in mice against H1N1 with H1 antigen.	High binding and neutralizing antibody titers; Strong CD4+ T cell response	[68,70]
Influenza Ml VLP	Influenza B	None	Mice	n/a	Protection in mice against influenza B with influenza B HA antigens	High binding antibody titers; Strong CD4+ response; Th1 bias	[69]
HIV Gag VLP	H1, H3, H7, influenza B HA	AddaVax	Mice, ferrets	n/a	Protection in mice against H1N1 with H1 antigen and against H7N9 with H7 antigen; protection in ferrets against H3N2 with H3 antigen and against influenza B with influenza B HA antigen	High HA-binding, HAI, and neutralizing antibody titers. Strong CD4+ response	[71–76]
Stalk-directed HA nanovaccines
Full HA	H2	None	n/a	Phase I	n/a	High HA-binding and neutralizing antibody titers; High stalk binding antibody titers; Elicited ADCC- activating antibodies	[95]
Stalk-only HA	H1, H3, H7 H10	Sepivac SWE, AF03, Sigma Adjuvant System, AddaVax	Mice, Ferrets, NHPs	Phase I	Protection in mice against H1N1 with H1 antigen, against H10N8 with H3 antigen; against H3N2 and H10N8 with H7 antigen; against H3N2, H7N9, and H10N8 with H10 antigen	High HA-binding, stalk-binding, and neutralizing antibody titers; Elicited ADCC-activating antibodies	[96–99,101–106]
Inverted HA	H1, H2	AddaVax,, Alum, CpG	Mice	n/a	Protection in mice against cH6/1N5 with H1 antigen	High HA-binding and stalk-binding antibody titers	[93–94]
Multi-antigen HA nanovaccines
Protein nanoparticle	H1, H3, group 1 HA consensus, M2e	Co4B, CpG	Mice	n/a	Protection in mice against H1N1, H3N2, H5N8, and H9N2 with various antigen combinations	High HA-binding, stalk-binding, HAI, and neutralizing antibody titers; Strong CD4+ and CD8+ T cell responses	[113–115, 117–118, 121]
Composite nanoparticle	H1, M2e	Chitosan, CpG	Mice	n/a	Protection in mice against H1N1 with H1 + M2e antigens	High HA-binding antibody titers; Strong CD4+ and CD8+ T cell responses	[116]
Influenza Ml VLP	H1, H3, H5, influenza B HA, N1, N2, N6, N8, influenza B NA M2e	None	Mice	n/a	Protection in mice against H1N1, H3N2, H5N1, H7N9, and influenza B with various antigen combinations	High HA-binding and HAI antibody titers; Elicited ADCC-activating antibodies; Strong CD4+ and CD8+ T cell responses	[107–110, 120,122]
BIV Gag VLP	H3, H9, N2	Alum	Chickens	n/a	Protection in chickens against H3N2 and H9N2 with H3 + H9 + N2 antigens	High HA-binding, HAI, and neutralizing antibody titers	[111]
Liposome	H1, H2, H3, H5, H7, H9, H13, influenza B HA, N2, N3, influenza B NA	Phosphorylated hexaacyl disaccharide, alum, QS21	Mice	n/a	Limited protection in mice against H5N1 with 8 HA + 2 NA antigens; protection in mice against H1N1, H3N2, and influenza B with 3 HA + 3 NA antigens	High HA binding, HAI and neutralizing antibody titers	[57, 112]

Article Highlights.

• Nanoparticle display of hemagglutinin can enhance the immunogenicity of hemagglutinin when compared to soluble hemagglutinin alone in murine and non-human primate models.

• Mosaic nanoparticles displaying multiple hemagglutinin subtypes are capable of protecting mice against mismatched challenges.

• Inverted HA on nanoparticles drive increased anti-stalk humoral immunity compared to regular orientation HA

• Stalk-only HA antigens displayed on ferritin nanoparticles have entered clinical trials and can elicit antibodies capable of binding across multiple HA subtypes.

• Nanoparticle platforms have also been used to co-display additional antigens such as neuraminidase (NA) or matrix protein 2 (M2) on the same particle as hemagglutinin.