A self-assembled protein nanocage as a universal influenza vaccine induces enhanced broadly cross-reactive immunity

Abstract

Proteins are an effective platform for vaccine design by enabling multivalent antigen presentation and stabilizing antigen structures to enhance immunogenicity. This study investigates the molecular design and ability to position antigens on self-assembled protein nanocages (SAPNs) as a universal influenza vaccine, incorporating highly conserved nucleoprotein peptides (NP55–69 and NP147–158) as CD4⁺ and CD8⁺ T cell antigens alongside hemagglutinin stalk (HrHA) and matrix protein 2 ectodomain (4M2e). The SAPNs display HrHA and 4M2e on their external surface while embedding NP antigens internally. The engineered SAPNs feature a modular design, enabling precise antigen placement and enhanced accessibility through optimized linker length. This nanostructure elicited robust cellular and humoral immune responses in mice, including cross-reactive antibodies and antigen-specific T cell activation. These findings highlight the potential of SAPNs as a versatile platform for universal influenza vaccine development and the role that protein design plays in the spatial organization of self-assembled protein materials and its effect on biomedical function.

1. Introduction

Protein nanocages serve as an effective vaccine platform due to their ability to incorporate multiple antigens while preserving the native antigen structures [1–6]. Remarkable advances in chemical and biological conjugation and particle synthesis have further enhanced the performance of protein nanoparticles for prophylactic immunization [7–9]. Due to their ability to present multivalent arrays of antigens to immune cells and trigger immune responses against infectious diseases, self-assembled protein nanoparticles have progressed to clinical trials. For example, self-assembled ferritin nanoparticles displaying conserved HA stalks have entered Phase I trials (NCT05155319 and NCT03814720), while OVX836, a recombinant protein vaccine featuring an oligomerized nucleoprotein, is in Phase II (NCT05569239). Furthermore, protein nanoparticles such as virus-like particles, which mimic viral structures without containing genetic material, have shown significant protection against influenza. Notably, quadrivalent virus-like particles displaying HA antigens have demonstrated substantial efficacy in adults during a Phase III clinical trial (NCT03301051 and NCT03739112) [10]. Additionally, computational design of self-assembling protein nanocages has facilitated the development of de novo vaccine nanoparticles [11–14].

Despite these advances, there remain challenges in designing and developing vaccine nanoparticles that can confer complete protection against heterogeneous influenza viruses. Influenza viruses are zoonotic and rapidly mutate, making them difficult to vaccinate against [15]. Hemagglutinin (HA) comprises most of the surface of influenza and has a high instance of seasonal mutation [15]. While the head region of HA constantly mutates, the stalk remains relatively conserved [16,17], presenting an opportunity for vaccination. Current strategies for influenza vaccine nanoparticles include presenting highly conserved, yet weakly immunogenic domains of influenza antigens with the goal of inducing sufficiently strong cross-reactive immune responses [18–22]. Adjuvants are commonly used to achieve this and to enhance the efficacy of subunit vaccine nanoparticles generally [19–21,23]. However, their application must be carefully evaluated to ensure safety. Heterologous sequential immunization with different vaccine nanoparticle platforms has also been explored as a strategy to improve cross-reactive immune responses [24]. Specifically, a combination of intramuscular priming with mRNA lipid nanoparticles and intranasal booster with protein nanoparticles has been shown to modulate Th1 and Th2 antibody bias and influence immunodominance hierarchies, thereby enhancing cross-reactive immunity. Mosaic nanoparticle immunogens utilizing I53 nanocages, which co-display heterosubtypic HA from multiple strains, have emerged as a promising approach to broaden immune responses against influenza [1]. By leveraging avidity to preferentially engage cross-reactive B cells, this strategy shifts the immune focus from the variable HA head to the conserved HA stalk. Although potent stalk-directed antibody responses have been observed in macaques, even in the presence of strong pre-existing immunity against the immunodominant HA head, complete avoidance of anti-HA head responses remains challenging. This issue may be addressed by masking immunodominant sites using PEGylation or glycosylation [25–27]. However, multiple studies have demonstrated that while this approach does not impact the absolute magnitude of antibodies against unmasked antigens, it does increase their proportion, raising questions about its benefit in the context of antigen display for humoral immune responses [28–31]. An alternative approach involves presenting HA stalks on multivalent nanoparticles. However, scaffold-specific immune responses, which were observed with nanoparticles such as ferritin [32, 33] or I53 nanocages [30], can interfere with weakly immunogenic antigen-specific responses. To minimize non-specific immune activation and enhance immunity against weakly immunogenic antigens, it may be beneficial to selectively incorporate conserved antigens onto scaffolds that do not elicit immune responses themselves.

Previously, we developed self-assembled protein nanocages (SAPNs) without any detected immunity directed toward the scaffold for application as a universal influenza vaccine [34]. Two sets of coiled-coil peptides drive the self-assembly of SAPNs [35,36], tGCN4, a truncated version of GCN4pII [37,38], is the homotrimeric coiled coil building block and forms disulfide bonds with either component of the heterodimeric coiled coil pair, tZE or tZR, which are truncated forms of ZE and ZR leucine zippers [39], respectively. The high affinity between tZE and tZR facilitates the assembly of the cage. The design of the previous SAPN had a high valency of head-removed HA stalk (HrHA), genetically fused to tGCN4 to preserve trimeric HA conformation, that elicited broadly cross-reactive humoral immune responses against different HA subtypes [34]. Additionally, matrix protein 2 ectodomain chimera peptide (4M2e), with epitopes derived from human, fowl, avian, and swine viruses, was genetically fused to tZE and was presented on the inner surface of the cage. Whilst T cell responses play an integral role in cross-reactive prophylactic activity [40–43], significant anti-M2e and anti-HA cellular immune responses were not observed in mice immunized with the HrHA/4M2e SAPNs [34]. A previous study demonstrated that cross-reactive protection against influenza infections strongly correlates with broadly reactive T cell responses, independent of neutralizing humoral immunity [44]. Consequently, incorporating strong, highly conserved T cell epitopes into vaccine formulations could substantially enhance the potency of a universal influenza vaccine.

In this work, we redesigned the components of the SAPN vaccine by incorporating highly conserved nucleoprotein peptides with amino acid sequence 55–69 (NP55–69) and 147–158 (NP147–158) as CD4⁺ and CD8⁺ T cell antigens [45–49], respectively, on the interior of the cages fused to tZR and moved 4M2e antigen to the exterior surface fused to tZE. This design maintained the high trimeric valency of HrHA on the external surface of SAPNs, thereby inducing strong anti-NP cellular and cross-reactive anti-HrHA humoral immune against influenza. Interestingly, significantly improved M2e-specific T cell responses were seen in mice vaccinated with the new SAPN design, but anti-M2e humoral immunity was not observed. Given the ease of recombinant protein design, 4M2e was extended from tZE on the external surface of SAPNs by a longer linker to elicit anti-M2e antibodies. Hence, by applying molecular design to control the spatial presentation of antigens, HrHA/4M2e/NP SAPNs substantially improved humoral and cellular immune responses against influenza antigens.

2. Results

2.1. Synthesis and characterization of HrHA/4M2e/NP SAPNs

SAPNs were functionalized with HrHA (A/Puerto Rico/8/1934) and 4M2e externally and with three tandem copies of NP55–69 and NP147–158 (NP55-NP147) (Fig. 1a) by synthesis with using tGCN4, tZE, and tZR recombinantly fused antigens. Three copies were selected based on prior work showing potent anti-NP cellular responses [50]. As demonstrated in our previous report [34], HrHA was genetically fused to the N-terminus of tGCN4 (HrHA-tGCN4) to induce trimerization of HrHA (Fig. 1b). To move 4M2e from the internal to external SAPN surface, it was fused to the N-terminus of tZE (4M2e-tZE) instead of the C-terminus (Fig. 1b). In contrast to HrHA and 4M2e, a tandem copy of three repeats of NP55–69 and NP147–158 was fused to the C-terminus of tZR (tZR-NP55-NP147) to be incorporated inside the nanocage. The expression of 4M2e-tZE (MW~17 kDa) and tZR-NP55-NP147 (MW~14 kDa) monomers was confirmed by SDS-PAGE (Fig. 1c). When 4M2e-tZE and tZR-NP55-NP147 were mixed at an equimolar ratio, bands near ~14–17 kDa and 31 kDa were detected, indicating a mixture of heterodimer and monomeric protein due to disruption of association by SDS-PAGE. However, no monomer was detected and only a band near ~31 kDa was observed when proteins were stabilized by 1 % glutaraldehyde, representing heterodimerization of 4M2e-tZE and tZR-NP55-NP147 in solution. Trimeric HrHA was successfully expressed from a stable Expi293F cell line as verified by SDS-PAGE (Fig. 1d) and its trimeric conformation proven in our previous work [34].

Fig. 1.

Characterization of SAPN. a, Schematic representation of protein nanocage assembly. b, Diagram of constructs for recombinant proteins HrHA-tGCN4, 4M2e-tZE, and tZR-NP55-NP147. c,d, SDS-PAGE analysis of purified 4M2e-tZE, tZR-NP55-NP147 (c), and HrHA-tGCN4 (d). Interaction between 4M2e-tZE and tZR-NP55-NP147 stabilized by 1 % GTA resulted in the formation of heterodimers. HrHA-tGCN4 was expressed from a stable Expi293F cell line and purified using 50 mM and 300 mM imidazole buffers for washing and elution, respectively. e, Size distribution, PDI, and surface charge of 100 % HrHA/4M2e/NP SAPNs were measured by DLS zetasizer. f, Morphology and surface feature of 100 % HrHA/4M2e/NP SAPNs were imaged by negative staining TEM. g-i, Antibody binding assay against HrHA (g), 4M2e (h), and NP55-NP147 (i) was conducted to examine antigen presentation in 100 % HrHA/4M2e/NP SAPNs compared with BSA as a control and soluble HrHA-tGCN4, 4M2e-tZE, and tZR-NP55-NP147. Statistics (n = 3) were analyzed via one-way ANOVA with Turkey’s post hoc multiple comparison analysis. Significance is reported as follows (*) for p ≤ 0.05, (**) for p ≤ 0.01, (***) for p ≤ 0.001, and (****) for p ≤ 0.0001.

Multivalency plays a crucial role in inducing strong humoral immune responses by facilitating B cell receptor clustering, which enhances B cell activation [51,52]. Consequently, nanocages with high valency are desirable for the synthesis vaccine nanoparticles. To achieve this, SAPNs were constructed using HrHA-tGCN4 trimers, along with 4M2e-tZE and tZR-NP55-NP147 heterodimers, ensuring full antigen decoration of the SAPNs (100 % HrHA/4M2e/NP SAPNs). Before assembly, free thiol groups of tGCN4 were activated with aldrithiol-2 to facilitate asymmetric covalent bonding with tZE (HrHA-tGCN4/4M2e-tZE) or tZR (HrHA-tGCN4/tZR-NP55-NP147). The resulting blocks were then mixed at an equimolar ratio to form 100 % HrHA/4M2e/NP SAPNs. The assembly of most 100 % HrHA/4M2e/NP SAPNs with a hydrodynamic average diameter of 336 nm (77 %) was accomplished within 2 h (Supplementary Fig. 1). Yet, the populations were dynamic and there was further assembly between nanocages to yield 87 % of SAPN population with an average hydrodynamic diameter of ~296 nm over the next 3 h. No further significant changes in the SAPN population were observed with longer assembly time. To improve the monodispersity of nanocages, a fraction of small nanocages and the soluble proteins were removed from fully assembled nanocages by centrifugation at 14,000×g and collection of pelleted nanocages. As a result, final assembled 100 % HrHA/4M2e/NP SAPNs showed an average hydrodynamic diameter of 301 ± 25 nm, PDI of 0.359 ± 0.008, and a negatively charged surface (−18 ± 1 mV) (Fig. 1e), displaying appropriate properties for vaccine application. As imaged by TEM, spike-like structures were observed on the surface of nanocages, indicating multivalency of protein antigens protruding from the external surface (Fig. 1f). The observed size reduction of nanocages during TEM imaging highlights the impact of sample preparation on structural measurements. Drying processes often lead to shrinkage, especially in hollow and protein-based nanostructures. Our prior findings with SAPNs exhibited similar trends in size between hydrodynamic solution measurements and dried measurements [34]. This points to the importance of complementary characterization methods such as DLS for accurate structural evaluation of protein nanomaterials.

The accessibility of antigens on and in 100 % HrHA/4M2e/NP SAPNs was assessed by ELISA. Compared to bovine serum albumin as a control, both 100 % HrHA/4M2e/NP SAPNs and soluble HrHA-tGCN4 showed similar significant levels of anti-HrHA antibody recognition, as expected (Fig. 1g). In parallel, 4M2e antigens presented on the external surface of 100 % HrHA/4M2e/NP SAPNs and soluble 4M2e-tZE were recognized by anti-M2e antibodies at a greater level than bovine serum albumin negative control (Fig. 1h). These data demonstrate that HrHA and 4M2e antigens were displayed on the surface of nanocages and exhibit good accessibility to antibody. In contrast, anti-NP55–69 and anti-NP147–158 antibody interactions with 100 % HrHA/4M2e/NP SAPNs were very low relative to soluble tZR-NP55-NP147 protein, supporting their internal location (Fig. 1i). Altogether, the ELISA data demonstrate the fidelity of exterior vs. interior antigen presentation induced by the assembly of SAPNs and support that the TEM images reflect nanocages with structures consistent with the design.

To estimate antigen density estimation, the geometric parameters of SAPNs were calculated by running molecular dynamics (MD) simulations of the HrHA-tGCN4-tZE/tZR-tGCN4 protein unit assembly. MD simulations of the trimeric building unit indicated that the length of each edge in the resulting polyhedral lattice is approximately 4 nm (Supplementary Fig. 2). Based on this parameter, the total number of antigens displayed on a single SAPN nanocage was estimated using a geometric model. The SAPN was approximated as a spherical polyhedral cage composed primarily of hexagonal faces, consistent with Euler’s theorem for fullerene-like structures, which requires the presence of 12 pentagonal faces, with the remaining faces being hexagonal. Assuming an overall particle diameter of approximately 300 nm (Fig. 1e), the total surface area of the nanocage was calculated as 4πr². The number of hexagonal faces was then estimated by subtracting the surface area contribution of the 12 pentagons and dividing by the area of a single regular hexagon with side length a = 4 nm:

$$\begin{gathered} \#\mathit{\text{of}}\mathit{\text{hexagons}}\approx \frac{A_{\mathit{\text{sphere}}}-12A_{\mathit{\text{pentagon}}}}{A_{\mathit{\text{hexagon}}}}=\frac{4\pi r^2-12\left(\frac{1}{4}\sqrt{5\left(5+2\sqrt{5}\right)}{a}^2\right)}{\frac{3\sqrt{3}}{2}{a}^2} \\ \approx \mathrm{6,794}\text{hexagons} \end{gathered}$$

Each HrHA trimer is positioned at a vertex of the polyhedral nanocage. Therefore, the total number of HrHA antigens was estimated from the total number of vertices, accounting for the shared vertices among faces:

$$\begin{gathered} \#\mathit{\text{of}}\mathit{\text{vertices}}=\frac{5\times \left(12\mathit{\text{pentagons}}\right)+6\times \left(\#\mathit{\text{of}}\mathit{\text{hexagons}}\right)}{3} \\ \approx \mathrm{13,608}\mathit{\text{HrHA}}\text{antigens} \end{gathered}$$

Similarly, 4M2e or NP antigens are positioned along the edges of the nanocage. The total number of these antigens was estimated from the number of edges:

$$\begin{gathered} \#\mathit{\text{of}}\mathit{\text{edges}}=\frac{5\times \left(12\mathit{\text{pentagons}}\right)+6\times (\#\mathit{\text{of}}\mathit{\text{hexagons}})}{2} \\ \approx 20,3994M2eorNP\mathit{\text{antigens}} \end{gathered}$$

Based on these geometric considerations, a single SAPN with a diameter of approximately 300 nm is estimated to display ~13,600 HrHA trimers and ~20,400 each of 4M2e and NP antigens.

2.2. In vitro activation of bone marrow dendritic cells by HrHA/4M2e/NP SAPNs

Dendritic cells (DCs) play a pivotal role in the onset of adaptive immune responses by presenting antigens to T cells for their antigen-specific activation [51,53,54]. Prior to the initation of antigen presentation and T cell activation, DCs must first internalize and process antigens and be activated. To assess potential for DC maturation, bone marrow DCs (BMDCs) derived from BALB/C mice were stimulated with 100 % HrHA/4M2e/NP SAPNs. As indicated by upregulation of the maturation markers CD86 (Fig. 2a) and MHC-II (Figs. 2b), 100 % HrHA/4M2e NP SAPNs significantly promoted the activation of BMDCs compared with PBS and soluble HrHA/4M2e/NP55-NP147, likely due to the increased internalization of antigens in nanoparticle form compared to soluble [55,56]. Whereas small soluble antigens are primarily taken up by antigen presenting cells through fluid-phase micropinocytosis, which is less efficient for antigen uptake, particulate antigens are preferentially internalized through endocytic and phagocytic pathways [51, 55,57,58]. This results in improved intracellular processing and thus DC activation. Another possible mechanism is multivalency-driven receptor clustering on dendritic cells. Antigens displayed in a multivalent, repetitive arrangement on nanoparticles can simultaneously engage multiple antigen uptake receptors and pattern recornition receptors, promoting stronger activation signals for DCs [50,57,59].

Fig. 2. In vitro assessment of BMDC activation.

a,b, The population of CD11c + CD86⁺ (a) and CD11c + MHC-II+ (b) BMDCs were analyzed by flow cytometry after being incubated with PBS, soluble HrHA/4M2e/NP55-NP147 mixture, and 100 % HrHA/4M2e/NP SAPNs. c, Cytokine profiles (TNFα, IL-12p70, IL-4, and IL-10) of activated BMDCs were measured via ELISA. Increased concentration of both pro-inflammatory TNF-α, IL-12p70, and IL-4 and suppressor IL-10 cytokines were observed from BMDCs activated by 100 % HrHA/4M2e/NP SAPNs. Statistics (n = 3) were analyzed via one-way ANOVA with Tukey’s post hoc multiple comparison analysis. Significance is reported as follows (*) for p ≤ 0.05, (**) for p ≤ 0.01, (***) for p ≤ 0.001, and (****) for p ≤ 0.0001.

The strong activation of BMDCs is also accompanied by improved secretion of pro-inflammatory cytokines (Fig. 2c). Although IL-12 cytokine levels secreted from BMDCs stimulated by SAPNs and soluble mixture of antigens were comparable, 100 % HrHA/4M2e/NP SAPNs induced significantly higher levels of TNF-α and IL-4 than soluble HrHA/4M2e/NP55-NP147 and PBS. Elevated concentration of suppressor IL-10 cytokines was also observed from SAPN stimulated BMDCs (Fig. 2c). Concomitant to pro-inflammatory cytokine secretion, the production of IL-10 can be triggered in DCs in response to stimulation with antigens, counteracting pro-inflammaotry responses at the same time [60–63]. As such, IL-10 is also an important cytokine to balance immune responses to avoid overly intense pro-inflammatory responses. In general, the results suggest that SAPNs can induce strong activation of DCs and, therefore, can potentially promote antigen-specific adaptive immune responses.

2.3. HrHA/4M2e/NP SAPNs induce cross-reactive humoral immune responses

Following SAPN characterization, mice were vaccinated with 100 % HrHA/4M2e NP SAPNs, a soluble mixture of the same antigens, or saline according to the vaccine schedule in Fig. 3a. The same batch of SAPNs was used for both prime and boost, stored at 4 °C. Mouse body weight measurements in the week following prime and boost indicate there is no gross toxicity of the formulations (Supplementary Fig. 3). To assess cross-reactive humoral immune responses, we measured IgG, Th1 associated IgG2a, and Th2 associated IgG1 titers against HrHA (A/Puerto Rico/8/1934H1N1) from vaccinated mice. Blood was collected 21 days after a single dose of vaccine was given intramuscularly (Fig. 3a). Mice immunized with the prime dose of 100 % HrHA/4M2e NP SAPNs displayed significantly enhanced anti-HrHA IgG and IgG1 responses compared to those administered PBS and soluble HrHA/4M2e/NP55-NP147 mixture (Supplementary Fig. 4a). This may be partly due to the presentation of NP55-NP147, strong T cell epitopes, as a single shot with the previous SAPN vaccine, which did not incorporate NP55-NP147 antigens, could not elicit robust humoral immune responses [34].

Fig. 3. Anti-HrHA and HA serotype humoral immune responses elicited by 100 % HrHA/4M2e/NP SAPN.

a, Animal study strategy for vaccination and blood collection. b-e, Anti-HrHA (A/Puerto Rico/08/1934), H1N1 HA (A/California/04/2009), H3N2 HA (A/Aichi/2/1968), and H5N1 HA (A/Vietnam/1194/2004) IgG (a), IgG2a (b), and IgG1 (c) titers in sera collected from immunized mice at week 7. Statistics (n = 5) were analyzed via one-way ANOVA with Turkey’s post hoc multiple comparison analysis. Significance is reported as follows (*) for p ≤ 0.05, (**) for p ≤ 0.01, (***) for p ≤ 0.001, and (****) for p ≤ 0.0001.

A subsequent dose of 100 % HrHA/4M2e/NP SAPN 4 weeks after the prime elevated the humoral immune responses against HrHA (Fig. 3b). In contrast to a single dose of 100 % HrHA/4M2e/NP SAPN (Supplementary Fig. 4a), booster immunization with 100 % HrHA/4M2e/NP SAPN induced higher anti-HrHA IgG2a titer (Th1 response) in mice than that with PBS and soluble antigens (HrHA/4M2e/NP55-NP147) (Fig. 3b). Boosting with 100 % HrHA/4M2e/NP SAPN also elicited strong cross-reactive humoral immune responses. 100 % HrHA/4M2e/NP SAPN raised greater IgG (Fig. 3c) titers against H1N1 HA (A/California/04/2009), H3N2 HA (A/Aichi/2/1968), and H5N1 HA (A/Vietnam/1194/2004) than vaccination with PBS or a mixture of soluble HrHA/4M2e/NP proteins. Additionally, 100 % HrHA/4M2e/NP SAPN promoted cross-reactive IgG2a (Fig. 3d) and IgG1 (Fig. 3e) responses against different HA subtypes including H3N2 HA, despite H3N2 being phylogenetically distant. While IgG2a responses against H5N1 HA were not seen in animals vaccinated with soluble protein, IgG2a titers against H5N1 HA were substantially increased by 100 % HrHA/4M2e/NP SAPN vaccination, indicating augmented Th1 associated antiviral antibody responses.

2.4. HrHA/4M2e/NP SAPNs induce strong cellular immune responses

In the current SAPN design, NP55–69 and NP147–158 were incorporated at the interior SAPN surface to promote CD4⁺ and CD8⁺ T cell responses. To investigate cellular immune responses induced by 100 % HrHA/4M2e/NP SAPN, we harvested spleens from mice 28 days after the booster shot and measured IFN-γ⁺ and IL-4⁺ CD4⁺ and CD8⁺ T cell populations by flow cytometry (Supplementary Fig. 5). When splenocytes were restimulated with NP55–69 and NP147–158, substantially increased IFN-γ⁺ CD8⁺ and IFN-γ⁺ CD4⁺ CD3⁺ splenocytes were detected in the 100 % HrHA/4M2e/NP SAPN group (Fig. 4a and Supplementary Fig. 6). While 100 % HrHA/4M2e/NP SAPN group also displayed enhanced percentage of IL-4⁺ CD8⁺ and IL-4⁺ CD4⁺ CD3⁺ populations, the nanocage generally induced IFN-γ⁺ biased cellular responses against NP55–69 and NP147–158.

Fig. 4. Cellular immune responses elicited by SAPN.

Percent populations of antigen-specific IFN-γ or IL-4 expressing CD8⁺ CD3⁺ (cytotoxic T cells) and IFN-γ or IL-4 expressing CD4⁺ CD3⁺ (helper T cells) responsive to NP55–69 and NP147–158 peptides (a), a mixture of consensus M2e peptides derived from human, swine, avian, and fowl (b), and a pool of H1N1 HA peptides (c). Statistics (n = 5) were analyzed via one-way ANOVA with Tukey’s post hoc multiple comparison analysis. Significance is reported as follows (*) for p ≤ 0.05, (**) for p ≤ 0.01, (***) for p ≤ 0.001, and (****) for p ≤ 0.0001.

Additionally, 100 % HrHA/4M2e/NP SAPN greatly improved M2e-specific cellular immune responses (Fig. 4b and Supplementary Fig. 6) in contrast to the previous SAPN vaccine [34], which did not include NP55–69 and NP147–158 epitopes. These results suggest that NP55–69 and NP147–158 epitopes in the nanocage not only strongly activated NP55–69 and NP147–158 responsive T cells but also improved M2e-specific T cell responses. In contrast to NP55–69 and NP147–158-specific T cell responses, 100 % HrHA/4M2e/NP SAPN immunized mice exhibited IL-4⁺ or Th2-biased M2e-specific T cell responses, balancing the Th1-biased NP55–69 and NP147–158-specific T cell responses.

Similar to the previous SAPN vaccine design [34], HrHA-specific T cell responses were weakly induced, with only IL4⁺ CD4⁺ CD3⁺ T cell responses showing noticeable improvement compared to PBS and soluble antigen groups (Fig. 4c and Supplementary Fig. 6). Therefore, there are important implications based on these results: (1) NP55–69 and NP147–158 can strongly activate T cells when incorporated in nanoparticles [64,65]. (2) The immunogenicity of M2e peptides can be significantly enhanced by presenting them in nanoparticles in concert with strong T cell epitopes such as NP55–69 and NP147–158. (3) HA stalks are not good candidates for inducing cellular immune responses but do induce strong cross-reactive humoral immune responses.

2.5. Influence of epitope accessibility on humoral immune responses induced by HrHA/4M2e/NP SAPNs

The accessibility of epitopes displayed on SAPNs can significantly influence and shape humoral immune responses. For instance, antigens localized within the interior of SAPNs likely fail to induce humoral responses since they are not accessible to B cells and, therefore, cannot be recognized. As demonstrated in Fig. 1i, NP55-NP147 was not recognized by anti-NP55-NP147 antibodies. As expected, both prime and booster shots with 100 % HrHA/4M2e/NP SAPN did not generate appreciable levels of anti-NP55-NP147 IgG, IgG2a, and IgG1 (Fig. 5a and Supplementary Fig. 4c) due to the shielded presentation of NP55-NP147 inside the nanocage. However, although 4M2e antigens externally presented on SAPNs were recognized by antibodies in the binding assay (Fig. 1h), vaccination with 100 % HrHA/4M2e/NP SAPN did not improve anti-4M2e humoral immunity over saline (Fig. 5b and Supplementary Fig. 4b). This implies that 4M2e antigens on SAPNs were likely not accessible to, and therefore not recognized by, B cell receptors. This may be due to the insufficient length of the linker between SAPNs and 4M2e, which could have led to steric hindrance from neighboring, much larger HrHA antigens, thereby limiting accessibility to B cell receptors.

Fig. 5. Anti-NP and 4M2e humoral immune responses.

a,b, Anti-NP55-NP147 (a) and 4M2e (b) IgG, IgG2a, and IgG1 titers in sera collected on week 7 from immunized mice vaccinated with 100 % HrHA/4M2e/NP SAPN. c,d, Anti-NP55-NP147 (c) and 4M2e (d) IgG, IgG2a, and IgG1 titers in sera collected on week 7 from immunized mice vaccinated with 100 % HrHA/4M2e/NP SAPN-XL. Statistics were calculated via one-way ANOVA with Tukey’s post hoc multiple comparison analysis for (a,b) and the Mann-Whitney test for (c,d). Significance is reported as follows ns for not significant, (*) for p ≤ 0.05, and (**) for p ≤ 0.01.

To address the weak anti-4M2e humoral immune responses, we extended the length of the linker between 4M2e and tZE coiled coil from (GGGGS)₃ to (GGGGS)₈. Based on AlphaFold prediction of HrHA-tGCN4 [66] and analysis via ChimeraX [67], the height of the HrHA antigen, not including tGCN4, is about 10 nm. The current flexible (GGGGS)₃ linker is estimated to have a length of about 5.7 nm [68,69], and may not allow for the complete 4M2e antigen to be accessible to B cell receptors due to steric hindrance from surrounding HrHA trimers. We lengthened the linker to (GGGGS)₈ to create the new fusion protein 4M2e-XL-tZE, with an estimated linker length of about 15.2 nm (Supplementary Fig. 7). The hydrodynamic average diameter of the SAPN with the extended linker (100 % HrHA/4M2e/NP SAPN-XL) is 187 ± 2 nm with PDI of 0.300 ± 0.020 (Supplementary Fig. 8). The decrease in size from 100 % HrHA/4M2e/NP is potentially due to steric hindrance of the larger 4M2e-XL-tZE protein incorporated into the cage, which increases the curvature to accommodate. A similar trend was noted in previous SAPNs, where the SAPN size decreased when the number of large ovalbumin antigens on the surface was increased [34]. In a vaccination study comparing 100 % HrHA/4M2e/NP SAPN-XL-vaccinated mice to saline controls, following the immunization schedule in Fig. 3a, significant IgG, IgG1, and IgG2a antibody reponses were induced against 4M2e (Fig. 5d). This is in constrast to the vaccination with 100 % HrHA/4M2e/NP SAPN with the shorter linker between 4M2e and tZE, which did not elicit increased anti-4M2e titers over saline vaccination (Fig. 5b, Supplementary Fig. 4b). Importantly, SAPNs with 4M2e-XL-tZE proteins also retained their ability to induce anti-HrHA IgG, IgG1, and IgG2a (Supplementary Fig. 9). As expected, no humoral response towards NP55-NP147 was observed (Fig. 5c), consistent with 100 % HrHA/4M2e/NP with short linker. Therefore, SAPNs with a longer linker between 4M2e and tZE coiled coil improved anti-4M2e humoral immune responses while maintaining strong anti-HrHA humoral immune responses.

3. Conclusions

In this study, we demonstrated that the potency of SAPNs as a universal influenza vaccine can be significantly improved through the incorporation new antigens, optimization of antigen placement, and adjustment of the linker length between antigen and nanocage. This rational design of nanocages enables the incorporation of antigens in a repetitive array with controlled orientation, displaying highly conserved trimeric HA stalk and 4M2e on the external surface, while embedding NP T cell epitopes within the nanocage interior. The presented results demonstrate that these SAPNs elicit cross-reactive antibodies against multiple HA serotypes and particularly strong Th1 cell responses, which were absent in the previous SAPN formulations containing only HA stalks and 4M2e [34]. Given that T cell activation generally depends on antigen presentation by antigen presenting cells rather than on the accessibility of antigens on the surface of nanoparticles [70–72], these results suggest a supportive role for NP-specific T cell responses in broadening cellular immunity to 4M2e-specific T cells. The cellular data presented here shows that 3 repeats of each NP antigen was effective in eliciting a strong cellular response, as in previous literature [50]. However, density and ratio of T cell epitopes can have a strong impact on the nature of T cell responses, including tolerance [73–75]. Investigation into the number and ratio of NP55–69 and NP147–158 antigens in SAPNs may reveal important levers to optimize the T cell response.

However, the enhancement in anti-4M2e humoral immune responses does not appear to be due to the inclusion of NP epitopes. This is evident from our observation that external display of 4M2e on SAPNs containing NP epitopes without a sufficiently long spacer did not result in a marked increase in anti-4M2e antibody responses over the saline control. Instead, the improved response is likely due to the increased accessibility of 4M2e to B cells, achieved by positioning the antigen on the external surface of the nanoparticle with a longer spacer between 4M2e and the scaffold. This configuration may facilitate more efficient recognition and activation of B cells, leading to a stronger humoral response. These findings emphasize that antigen accessibility and spatial arrangement are critical parameters in vaccine design, particularly for optimizing B cell-mediated immunity. Beyond antigen accessibility, nanoparticle size can also influence the mechanisms of immune activation, as it can affect the uptake and processing of nanoparticles by immune cells. In particular, the size of SAPNs (~300 nm) may facilitate their internalization by macrophages, while still allowing efficient uptake and activation of DCs, likely through clathrin-independent endocytic pathways such as macropinocytosis or phagocytosis, which are known to favor the uptake of nanoparticles larger than 200 nm [76–82]. Although both DCs and macrophages function as antigen-presenting cells, DCs are considered the most potent, given their ability to migrate to lymphoid organs and effectively prime T cells. However, the inflammatory responses initiated by macrophages can play a critical role in recruiting other immune cells, including DCs, to the vaccine injection site [83–85]. Thus, macrophage uptake may synergistically enhance DC antigen uptake and activation, contributing to a stronger adaptive immune response. Furthermore, the proinflammatory cytokines released by macrophages following SAPN internalization can mediate and promote effective T cell activation, thereby amplifying the overall immune response [83,86]. Recently, it was also reported that macrophages can even direct location-dependent recall of B cells into germinal centers upon vaccination [87]. While this work demonstrated significant activation of primary DCs in vitro by SAPNs, further investigation into the role of macrophages may provide valuable insights into the full mechanism of SAPN-induced adaptive immune activation.

Although the durability of immune responses elicited by SAPNs was not examined in this study, our previous work demonstrated that desolvated protein nanoparticles formulated with HrHA and 4M2e induced humoral responses lasting up to 4 months, despite lacking control over antigen presentation and stability [88]. More recently, we showed that nanoparticle formulation and immunization route both influence immune response durability, with intranasal delivery yielding the most potent mucosal and systemic immunity [89]. Therefore, assessing the durability of SAPN-induced immunity via different vaccination routes, especially intranasal delivery, will be an important direction for future study.

In addition to exploring different routes of vaccine delivery, heterologous sequential vaccination can be an effective strategy to enhance the immunogenicity of SAPNs. As demonstrated in a recent study, sequential immunization with mRNA and influenza protein vaccines elicited strong cross-reactive immune responses, particularly by shaping immunodominance and Th1 responses [24]. Specifically, priming with an influenza mRNA vaccine followed by boosting with influenza protein nanoparticle vaccines promoted Th1-biased responses and markedly enhanced cross-protective immunity. Therefore, incorporating diverse approaches such as alternative delivery routes and heterologous sequential vaccination could be a useful step toward improving cross-reactive immune responses.

Although further studies are needed to examine the vaccine’s prophylactic efficacy against influenza A viruses, this work clearly demonstrates the advantages of SAPNs within the context of vaccine modality design and cross-reactive immunity. SAPNs self-assemble without the need for crosslinkers and allow precise antigen presentation on either the exterior or interior of the nanocage, illustrating the versatility and modularity of the SAPN platform for antigen presentation. Furthermore, we previously showed that the SAPN does not induce antibodies to the tGCN4, tZE or tZR scaffold building blocks [34], thus avoiding unintended immune responses. Moreover, the incorporation of new NP antigens or making M2e more accessible enhances overall immunogenicity without any observed epitope competition. While the safety profile has not been examined in depth, protein subunit vaccines are generally considered a very safe platform and tend to exhibit minimal off-target responses as they utilize purified protein components [90, 91]. In this study, we successfully enhanced immunogenicity without the use of an adjuvant, which can otherwise lead to side effects such as reactogenicity [92,93]. Nevertheless, despite the favorable safety profile, the markedly enhanced immune responses observed here raise the possibility that regulatory immunity, particularly the role of Tregs, may be affected. It will be important in future studies to investigate whether SAPNs modulate immune tolerance or regulatory networks, especially in the context of repeated immunization.

4. Experiment section

4.1. Expression and purification of recombinant proteins

4M2e-tZE, 4M2e-XL-tZE, and tZR-NP55-NP147 were expressed from E.coli and buffer exchanged against phosphate-buffered saline (PBS) after purification. Briefly, 4M2e-tZE and tZR-NP55-NP147 were cloned into pET17b plasmids with codon optimization (GenScript). Protein was expressed in E.coli BL21 Star (DE3) in 1 L Lysogeny Broth (LB) medium with 100 μg/mL ampicillin at 37 °C at 120–130 rpm. Protein expression was induced by adding isopropyl-β-d-1-thiogalactopyranoside to a final concentration of 1 mM at an OD600 of 0.4–0.6 for 5 h. For protein extraction, bacteria were sonicated on ice in lysis buffer containing 8 M urea, 10 mM Tris-Cl, and 100 mM NaH₂PO₄ with pH 8.0. The extract was then incubated with Ni-NTA resin (Qiagen) and loaded into a column. After washing with 8 M urea buffer at pH 6.3, proteins were eluted with 8 M urea buffers at pH 5.9 and 4.5 from the column.

HrHA-tGCN4 and H1N1 (A/California/04/2009) HA recombinant proteins were obtained from stable Expi293F cell lines; H1N1 HA recombinant proteins were used for measuring anti-H1N1 HA titers via ELISA. In brief, 0.7 × 10⁶ Expi293F cells/well in 1.5 mL OptiMEM in a flat-bottom 6-well plate (Thermo Fisher Scientific) were transfected by adding a mixture of 3 μg/well of DNA encoding HrHA-tGCN4 or H1N1 HA, 380 μL/well of Opti-MEM (Thermo Fisher Scientific), 6 μL/well of Lipofectamine (Thermo Fisher Scientific), and 6 μL of p3000 reagent (Thermo Fisher Scientific) followed by incubation for 6 h at 37 °C, 8 % CO₂. Expi293F cells were then resuspended and incubated in Dulbecco’s Modified Eagle Medium (DMEM, Corning) supplemented with 10 % fetal bovine serum (FBS, Avantor) for 48 h. Afterwards, the media was replenished with DMEM supplemented with 10 % FBS and geneticin (G418 Sulfate, Thermo Fisher Scientific) at a final concentration of 300 μg/mL every 3 days until 80–90 % confluency was achieved. After monoclonal selection, the monoclonal stable cells were progressively scaled up and adapted to suspension growth in Expi293 Expression Medium in baffled polycarbonate vented Erlenmeyer Flasks. Following incubation for 5 days in suspension, proteins were harvested by lysing the stable cells in lysis buffer containing 20 mM imidazole, 300 mM NaCl, and 50 mM NaH₂PO₄ by sonication. His-tagged HrHA-tGCN4 and H1N1 HA were then purified from the cell lysate through Ni-NTA agarose affinity chromatography with 300 mM imidazole buffer after washing with 50 mM imidazole buffer.

Recombinant proteins were buffer exchanged against PBS by centrifuging through a 3 kDa molecular weight cut-off Amicon Ultra-4 Centrifugal Filter Unit (Millipore Sigma). To minimize or avoid inadvertent immune responses, recombinant proteins were subjected to endotoxin removal using Pierce High-Capacity Endotoxin Removal Spin Columns (Thermo Fisher Scientific) as per the manufacturer’s instructions. ToxinSensor^™ Chromogenic LAL Endotoxin Assay Kit (GenScript) was used to ensure that the endotoxin levels remained below the endotoxin limit of 15 EU/mg [94]. All protein sequences and molecular weights are in Table S1.

4.2. Analysis of protein expression and oligomerization

SDS-PAGE was performed to analyze protein size and purity. Soluble HrHA-tGCN4, 4M2e-tZE (and XL), and tZR-NP55-NP147 were mixed with Laemmeli buffer solution (Biorad) with dithiothreitol (DTT, Thermo Fisher Scientific) and incubated at 95 °C for 5 min. For the analysis of heterodimerization between 4M2e-tZE and tZR-NP55-NP147, the soluble proteins were stabilized by 1 % glutaraldehyde crosslinker. The protein samples were then loaded into 12 % SDS-PAGE gel with Laemmeli buffer without DTT at 95 °C for 5 min. 150 V was applied for 80 min to elute the proteins through the gels in tris-glycine SDS gel electrophoresis buffer. The proteins in the gel were stained with Coomassie Blue R-250 for 2 h and destained with DI water for overnight. All the stained gels were imaged by Gel Doc X + Gel Documentation System (Biorad).

4.3. Synthesis of influenza vaccine nanocage

100 % HrHA/4M2e/NP SAPNs were synthesized as previously described for 100 % HrHA/4M2e SAPNs [34]. Briefly, aldrithiol-2 (MilliporeSigma) treated HrHA-tGCN4 protein was mixed with 4M2e-tZE(-XL) or tZR-NP55-NP147 in equimolar ratio in PBS to make building blocks via the formation of a disulfide bond. Then, 150 μL of 0.8225 μg/μL HrHA-tGCN4 (25 μM) was mixed with 150 μL of 0.4245 μg/μL of reduced 4M2e-tZE (25 μM) or 150 μL of 0.3545 μg/μL reduced tZR-NP55-NP147 (25 μM) to synthesize 25 μM HrHA-tGCN4/4M2e-tZE and 25 μM HrHA-tGCN4/tZR-NP55-NP147 oligomers, respectively. The two building blocks were then mixed and incubated for 2 or 5 h at 25 °C. Following incubation, SAPNs were collected by centrifugation at 14, 000×g for 10–15 min at 4 °C and resuspended in 100 μL PBS.

4.4. Characterization of influenza vaccine nanocage

Dynamic light scattering (DLS) with a Malvern Zetasizer Nano ZS was used to evaluate the size distribution of SAPNs. The size measurement was taken at a scattering angle of 173° with a beam wavelength at 633 nm. A refractive index of 1.33 with viscosity of 0.8882 cP and a refractive index of 1.45 were used for protein and PBS, respectively. Concentration of all the purified proteins and synthesized SAPNs was measured by performing a Pierce bicinchoninic acid (BCA) protein assay according to the manufacturer’s instructions (Thermo Fisher Scientific). SAPNs were further characterized by TEM at an acceleration voltage of 100 kV using a JEOL 100 CX-II TEM. For the TEM sample preparation, 5 μL of SAPNs was loaded onto a 300-meshed carbon film supported copper grid (MilliporeSigma) and washed with deionized water. After drying with a Kimwipe, the sample was stained with 5 μL of 1 % phosphotungstic acid solution for 20 s. The sample was then washed with deionized water and allowed to dry overnight.

4.5. Molecular dynamics simulation and calculation for antigen density

HrHA-tGCN4-tZE/tZR-tGCN4 protein model was prepared as described previously [34]. GROMACS was used to simulate the dynamics and folding of trimeric HrHA-tGCN4-tZE protein interacting with three tGCN4/tZR blocks with the Amber99SB-ildn force field. The system was solvated with TIP3 water molecules and neutralized with 150 mM NaCl. Energy minimization was performed to relax the system with the simple steepest descent minimization and the Particle Mesh Ewald method for a maximum of 50000 steps. Afterwards, the energy of the system was stabilized by NVT and NPT equilibration at 298 K. Coordinate restraints were applied on the heavy atoms of protein and hydrogen bonds during equilibration. After the equilibration, the molecular dynamics were simulated for 1000 ps in the NPT ensemble. The simulated protein structure was visualized by ChimeraX.

4.6. Antibody binding assay for influenza vaccine nanocage

Maxisorp 96 well immune assay plates (Nunc) were coated with BSA as a control, soluble HrHA, 100 % HrHA/4M2e/NP SAPNs with 10-fold serial dilutions ranged from 10⁻⁵ ng/μL to 10 ng/μL. Following overnight incubation at 25 °C, the plates were washed with buffer containing PBS and 0.1 % Tween-20. Each well was blocked with 2 % BSA in PBS for 2 h at 25 °C. Afterwards, the plates were washed again with wash buffer and incubated with 1:10,000 dilution of anti-HrHA, 1:4500 dilution of anti-4M2e, and 1:6500 dilution of anti-NP55-NP147 serum antibodies for 1 h at 25 °C. The serum antibodies were collected from mice administered with 2 doses of 20 μg HrHA-tGCN4, 20 μg tetrameric 4M2e [88,89], and 20 μg tZR-NP55-NP147. Consecutively, the plates were washed with washing buffer and incubated with 1:5000 dilution of HRP-conjugated goat anti-mouse IgG (H + L) (SouthernBiotech, cat: 1036–03) for 1 h at 25 °C. After washing, the color of each well was developed by incubating with 1-Step Ultra TMB substrate solution (Thermo Fisher Scientific) for 15 min. The reaction was stopped by adding 2 N H₂SO₄ solution (Thermo Fisher Scientific). Absorbance (450 nm–540 nm) was measured on a BioTek Synergy H4 Micro plate reader.

4.7. Animal immunization

Five mice (BALB/C strain, 6–8 weeks of age, male) were intramuscularly administered with 50 μL PBS, a soluble mixture of 6.8 μg HrHA-tGCN4, 1.7 μg 4M2e-tZE, and 1.5 μg tZR-NP55-NP147, or 10 μg of 100 % HrHA/4M2e/NP SAPNs (containing 6.8 μg HrHA-tGCN4, 1.7 μg 4M2e-XL-tZE, and 1.5 μg tZR-NP55-NP147. An identical injection was given as a booster 4 weeks post prime. Blood sera were collected from the submandibular vein of mice with anesthesia 3 weeks after each injection. Mice were euthanized at week 8 by CO₂ asphyxiation to harvest spleens. All animal experiments were performed in accordance with guidelines and regulations approved by Georgia Institute of Technology Institutional Animal Care and Use Committee (IACUC) under approved protocol A100576.

4.8. Enzyme-linked immunosorbent assay for antibody titers

The anti-HrHA, anti-4M2e, anti-NP55-NP147, and anti-HA IgG, IgG2a, and IgG1 titers from sera were measured by enzyme-linked immunosorbent assay (ELISA). 1 μg/mL of H1N1 HrHA (derived from A/Puerto Rico/8/1934), 4M2e-tZE, tZR-NP55-NP147, H1N1 HA (A/California/04/2009, produced in house as described above), H3N2 HA (A/Aichi/2/1968) (SinoBiological), and H5N1 HA (A/Vietnam/1194/2004) (SinoBiological) in PBS were coated onto Maxisorp 96 well immune assay plates overnight at 25 °C. The plates were washed with PBS containing 0.1 % Tween-20, blocked with 2 % BSA for 2 h at 25 °C, and incubated with serially 5-fold diluted sera for 1 h at 25 °C. After washing, the plates were further incubated with HRP-conjugated goat anti-mouse IgG (H + L) (SouthernBiotech, cat: 1036–03), IgG1 (South-ernBiotech, cat: 1071–05), or IgG2a (SouthernBiotech, cat: 1081–05) secondary antibodies at 1:5000 dilution for 1 h at 25 °C and washed. The plates were incubated with 1-Step Ultra TMB substrate solution (Thermo Fisher Scientific), and the reaction was stopped by 2 N H₂SO₄ solution. Absorbance was read at 450 nm with the correction wavelength set to 540 nm using a BioTek Synergy H4 Micro plate reader.

4.9. Intracellular cytokine staining for cellular immune responses

Splenocytes collected from euthanized mice were analyzed by performing intracellular cytokine staining. After spleens were harvested, they were mechanically disrupted and strained through 70 μm strainers, followed by resuspension of the single cells in complete RPMI 1640 medium supplemented with HEPES, L-glutamine (Thermo Fisher Scientific), and 10 % FBS. Red blood cells were removed by incubation with ACK lysing buffer (Thermo Fisher Scientific) at 25 °C for 9 min. Splenocytes at a cell density of 1×10⁶/well (1×10⁷ cells/mL) of complete RPMI 1640 medium were then seeded in a round-bottom 96 well plate and stimulated with 2.5 μg/mL per peptide of consensus M2e peptides (Synpeptide, Seq: SLLTEVETPIRNEWGSRSN, SLLTEVETPTRSEWESRSS, SLLTEVETPTRNGWESKSS, and SLLTEVETPTRNGWESNSS for human, swine, avian, and fowl, respectively), 2.5 μg/mL per peptide of NP55–69 (Synpeptide, Seq: RLIQNSLTIERMVLS) and NP147–158 peptides (Synpeptide, Seq: TYQRTRALV), or 2 μL of reconstituted PepTivator Influenza A HA stock (Miltenyi Biotec). At the same time, 1×10⁶ cells/well for mock restimulation were incubated with 0.5 μL of 2 μg/mL phorbol 12-myristate 13-acetate and 0.5 μL of 100 μg/mL calcium ionophore. Following 2 h incubation at 37 °C, 5 % CO, 1×10⁶ cells/well were incubated with 2 μL mixture of 50x brefeldin A and 50x monensin (BioLegend, cat: 420701) for 4 h. Restimulated splenocytes were then stained with Zombie Violet fixable viability dye (BioLegend, cat: 423113) as per manufacturer’s instructions. After washing, splenocytes were blocked with 5 μL/10⁶ cells Trustain FcX (BioLegend, 101319), followed by surface staining with 1 μL/10⁶ cells PerCP anti-CD3 (BioLegend, cat:100236), 1 μL/10⁶ cells FITC anti-CD8 (BioLegend, cat: 100706), and 1 μL/10⁶ cells APC/Cy7 anti-CD4 (BioLegend, cat: 100525). After surface staining, restimulated splenocytes were fixed with 3.7 % formaldehyde and stained with 1.5 μL/10⁶ cells PE anti-–IFN–γ (BioLegend, cat:505808) and 5 μL/10⁶ cells PE/Cy7 anti-IL-4 (BioLegend, cat: 504118) in permeabilization buffer (Thermo Fisher Scientific). Cells were resuspended in 1 % BSA prior to flow cytometry analysis. Stained cells were detected and analyzed by Cytek Aurora flow cytometry (Cytek Biosciences). The gating strategy used for analysis is shown in Supplementary Fig. 7.

4.10. Statistical analysis

All data comparing more than 2 groups used one-way ANOVA to calculate p values for statistical analysis using GraphPad Prism 9, except where noted. Tukey’s post-hoc multiple comparison analysis was applied to determine statistical significance as follows: (*) for p ≤ 0.05, (**) for p ≤ 0.01, (***) for p ≤ 0.001, (****) for p ≤ 0.0001. All data plotted with error bars represent means with standard deviation.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.biomaterials.2026.124025.

Data availability

Data will be made available on request.