Intranasal S-2P and lentinan formulation confers broad protection against SARS-CoV-2 VOCs via IFN-γ-dominant mechanisms

Abstract

Effective respiratory mucosal vaccines remain urgently needed to mitigate the rapid mutation and transmission of SARS-CoV-2. Here, we demonstrated that the spike protein (S-2P) of ancestral SARS-CoV-2 acted as a self-adjuvanted antigen for intranasal immunization, inducing robust systemic and mucosal immunity via integrin- and STING-dependent pathways. In contrast, H1N1 influenza hemagglutinin (HA) failed to generate measurable serum IgG or mucosal IgA following intranasal immunization. In mice, intranasal S-2P vaccination conferred complete protection against lethal ancestral SARS-CoV-2 challenge and partial cross-protection against heterologous Omicron variants, with both effects being IFN-γ- and CD8 + T cell-dependent. Co-administration of S-2P with the clinical immunomodulator lentinan (LNT) achieved complete protection against Omicron variants, mediated by IFN-γ but largely independent of CD8 + T cells. These findings establish S-2P + LNT as a safe, broad-spectrum mucosal vaccine candidate against emerging SARS-CoV-2 variants and reveal novel protection mechanisms beyond neutralizing antibodies and T cell immunity.

Introduction

The coronavirus disease 2019 (COVID-19) pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has led to over 778 million infections and 7 million deaths worldwide¹. The rapid development and clinical application of SARS-CoV-2 vaccines have substantially reduced morbidity and mortality; however, the continuous emergence of immune-evading variants poses ongoing challenges to control the outbreak². SARS-CoV-2 enters host cells through a process mediated by the spike glycoprotein. This process occurs in two distinct stages: first, viral attachment to host cell membranes facilitated by the S1 subunit, followed by membrane fusion orchestrated by the S2 subunit³. Both the full-length spike protein and its receptor-binding domain (RBD), which directly interacts with the angiotensin-converting enzyme 2 (ACE2) receptor, serve as primary targets for vaccine development⁴. The majority of currently available vaccines are delivered through intramuscular injection. While this approach effectively generates systemic humoral and cellular immunity, it typically elicits only weak respiratory mucosal immune responses. This limitation compromises protection against both initial infection and subsequent viral transmission^5,6. Furthermore, inadequate respiratory mucosal immunity may not only allow breakthrough infections but could also potentially drive the selection of more transmissible and immune-evasive viral variants⁷. Thus, there is an urgent need to develop effective respiratory mucosal vaccines that can induce broad-spectrum immune protection against SARS-CoV-2 variants and control recurrent outbreaks.

The mucosa serves not only as a critical physical barrier against pathogen invasion, but also plays an essential role in adaptive immunity due to its specialized immune structures, including regional mucosa-draining lymph nodes and organized mucosa-associated lymphoid tissues (MALTs)⁸. Mucosal vaccination holds great potential for conferring protection against respiratory infections^5,6. In general, intranasal delivery of vaccines can elicit local mucosal tissue-resident immune responses (mucosal secretory IgA antibodies and resident memory T cells), which provide the earliest barrier against pathogen entry⁶. Many mucosal vaccines are currently being investigated in clinical trials, and most of these vaccines are delivered by viral vectors, such as adenovirus vectors and modified vaccinia virus vectors⁹. Although intranasal vaccination with viral vector vaccines may have advantages in eliciting mucosal immune responses, some challenges and potential risks still need to be addressed. The pre-existing vector-specific immune responses produced from natural exposure or prior vaccinations hamper vaccine effectiveness or limit the application of booster immunization¹⁰. Additionally, the application of virus-vectored vaccines has potential safety concerns because the virus can enter the central nervous system (CNS) via the olfactory bulb, which lies next to the nasal cavity^10,11. If vaccine-vector virus spreads to the CNS, it could infect vulnerable neurons, potentially causing direct cytopathic effects and secondary neuroinflammation via the microglial activation.

Recombinant protein vaccines have the advantages of good accessibility, safety and stability, and their efficacy can be enhanced by booster vaccination¹². Adjuvants serve as crucial components of recombinant vaccines, orchestrating both antigen presentation to immune cells and delivery of immunostimulatory signals required for effective immune priming¹³. Considerable research has focused on developing experimental mucosal adjuvants for use with spike or RBD proteins. However, due to stringent safety requirements for intranasal vaccines, no respiratory mucosal adjuvants have yet been approved for human use, necessitating extensive preclinical and clinical evaluation¹⁴. Although adjuvant-free spike or RBD proteins have been administered intranasally to induce mucosal immunity, most current strategies employ heterologous prime-boost regimens—typically intramuscular mRNA priming followed by intranasal protein boosting^15,16. Few studies have systematically evaluated the immunogenicity and protective efficacy of adjuvant-free spike protein delivered intranasally using homologous prime-boost regimen. Notably, many intranasal spike or RBD vaccine studies employing mucosal adjuvants fail to include critical adjuvant-free control groups. This oversight likely stems from the prevailing assumption that spike protein exhibits poor immunogenicity when administered intranasally without adjuvants^17,18. Accumulating evidence demonstrates that the spike protein interacts with multiple immune receptors, including C-type lectins¹⁹, integrins^20,21, and toll-like receptors (TLR2 and TLR4)^22,23. These findings warrant further investigation into whether intranasal delivery of spike protein can bridge innate and adaptive immunity to induce protective mucosal immune responses.

In this study, the full-length recombinant prefusion-stabilized spike protein (S-2P) of ancestral SARS-CoV-2 was used as the immunogen to evaluate its immunogenicity and protective efficacy following intranasal delivery. We demonstrate that S-2P exhibits self-adjuvant activity characterized by the activation of innate immunity and that unadjuvanted S-2P is capable of inducing robust mucosal and systemic immune responses, as evidenced by respiratory tract IgA antibodies and T cell responses that conferred complete protection against lethal challenge with both ancestral SARS-CoV-2 and Omicron variants in mice. Moreover, formulation of S-2P with lentinan (LNT), a clinically approved immunomodulator functioning as a mucosal adjuvant, significantly augmented protection against Omicron variants. Finally, LNT was shown to provide complementary protective mechanisms against Omicron variants, beyond CD8 + T cell-mediated immunity.

Results

Intranasal vaccination with S-2P demonstrates favorable immunogenicity

To compare the immunogenicity of S-2P via intranasal or intramuscular vaccination, C57BL/6 mice were administered 10 μg of CHO cell-expressed ancestral SARS-CoV-2 S-2P protein via intranasal or intramuscular administration at 3-week intervals. Serum and bronchoalveolar lavage fluid (BALF) samples were collected at the indicated time points (Fig. 1A), and RBD-specific antibodies were measured by ELISA. The analysis revealed that intranasal delivery generated 12-fold and 21.6-fold higher anti-RBD serum IgG titers than intramuscular vaccination following prime and boost immunizations, respectively (Fig. 1B). Notably, anti-RBD IgA antibodies were exclusively detected in BALF from intranasally immunized mice (Fig. 1C). The neutralizing activity against ancestral SARS-CoV-2 was assessed via authentic virus neutralization assay. At day 35 post-immunization, intranasal S-2P vaccination induced significantly higher serum neutralizing antibody titers compared to intramuscular delivery (Fig. 1D). Neutralizing activity in BALF was exclusively observed in intranasally immunized mice (Fig. 1E). Comparable antibody responses were generated following intranasal administration of HEK293 cell-produced S-2P from different commercial sources (Fig. S1A).

Fig. 1. Humoral immune responses following intranasal immunization with S-2P.

A Immunization scheme and timeline (Created with BioRender.com). C57BL/6 mice were immunized with 10 μg of specified protein via intranasal (i.n.) or intramuscular (i.m.) routes in prime-boost regimen (3-week interval; n = 4 mice/group). RBD-specific antibody responses after i.n. or i.m. S-2P immunization. Serum IgG titers at days 14 and 35 (B) and BALF IgG/IgA titers at day 35 (C) (n = 4 mice/group). Neutralizing antibody titers (NT50) against authentic SARS-CoV-2 in serum (D) or BALF (E) at day 35 (n = 4 mice/group) RBD-specific antibody responses after i.n. immunization with S-2P, S1 or RBD protein: serum IgG titers at days 14 and 35 (F) and BALF IgG/IgA titers at day 35 (G) (n = 4 mice/group). All titers represent reciprocal endpoint dilutions, and the dashed lines represent the lower limits of detection (LOD). In B, D and F, the LOD is 100-fold dilution and samples below detection were assigned a titer of 100. In C, E and G, the LOD is 10-fold dilution and samples below detection were assigned a titer of 10. The data are presented as the means ± SEMs and are representative of at least three independent experiments. Statistical analysis: one-way ANOVA (B, D, E, F and G) or Kruskal-Wallis test (C) (*P < 0.05, **P < 0.01, ***P < 0.001, ****P ; ns, not significant).

Building upon the immunogenicity observed with intranasal S-2P vaccination, we examined whether envelope glycoproteins from other respiratory viruses share this property. Using the same protocol as for S-2P administration, we evaluated antibody responses generated by influenza H1N1 hemagglutinin (HA) protein administered via intranasal or intramuscular routes in C57BL/6 mice. Notably, HEK293-expressed HA protein alone failed to induce detectable antibody responses via either administration route, although intramuscular delivery of HA adjuvanted with aluminum hydroxide elicited serum antibody responses (Fig. S1B, C).

Furthermore, we assessed the immunogenicity of the RBD and S1 subunit (both expressed in HEK293 cells) via intranasal vaccination. Compared with S-2P, S1 protein also induced serum anti-RBD IgG antibodies and detectable IgA antibodies in the BALF, although at lower levels (Fig. 1F, G). In contrast, intranasal immunization with RBD failed to induce detectable IgG and IgA antibodies in either serum or BALF (Fig. 1F, G). Authentic virus neutralization assays revealed that intranasal immunization with S-2P induced the strongest neutralizing antibody responses in both serum (Fig. S1D) and BALF (Fig. S1E) at day 35. Notably, insect cell-expressed full-length spike extracellular domain failed to induce serum anti-RBD IgG antibodies via intranasal immunization (Fig. S1F).

These findings demonstrate that mammalian cell-expressed SARS-CoV-2 spike proteins possess superior immunogenic properties when administered intranasally.

Intranasal immunization with S-2P induces potent systemic immune responses, and LNT acts as an effective mucosal adjuvant

Adjuvants play a vital role in enhancing immune responses to recombinant or inactivated vaccines. Considering the special requirements for the safety of respiratory mucosal adjuvants, together with our clinical findings that intranasal instillation of LNT effectively reduced viral shedding time in SARS-CoV-2-infected individuals with favorable safety in a randomized controlled trial²⁴, we speculated that LNT could serve as a safe and effective mucosal adjuvant for human use. To assess its adjuvant activity as a respiratory mucosal adjuvant, we compared serum RBD-specific IgG responses following intranasal immunization with S-2P plus either LNT or various pattern recognition receptor (PRR) agonists, including resiquimod, flagellin (FLIC), poly(I:C), or CpG. S-2P plus LNT induced comparable or superior anti-RBD serum IgG titers to those with S-2P plus PRR agonists (Fig. S2A). Furthermore, intranasal instillation of LNT 6 h prior to S-2P administration increased antibody titers (Fig. S2B).

To elucidate the mechanisms underlying LNT’s mucosal adjuvant activity, we performed transcriptome sequencing of mouse lungs collected 6 h after intranasal LNT administration. LNT treatment significantly upregulated genes associated with the type I interferon (IFN) signaling pathway (Cxcl10, Irf7, Isg15, Tnf, Ifit3, Ifit1, Ifit2, Oas3 and Oasl1) and Toll-like receptor (TLR) pathway (Tlr4, Tlr2 and Tlr6) (Fig. S2C), demonstrating its potent activation of innate immunity. KEGG analysis further confirmed that upregulated pathways were strongly linked to innate immunity (Fig. S2D). Additionally, histological examination revealed that the nasal and lung tissues of mice receiving intranasal S-2P or S-2P plus LNT maintained normal architecture, comparable to those of PBS-treated controls (Fig. S2E, F).

To comprehensively evaluate the immunogenicity of S-2P as a mucosal immunogen and the effect of LNT as a mucosal adjuvant, C57BL/6 mice were intranasally immunized with 10 μg S-2P (S-2P i.n.) or 10 μg S-2P plus 200 μg LNT (S-2P + LNT i.n.) or intramuscularly immunized with 10 μg S-2P plus 100 μg aluminum adjuvant (S-2P + AL i.m.) on days 0, 21 and 112. The third-dose (booster) strategy follows current practices to address the rapid decline in antibodies, a major factor in breakthrough infections. Hence, post- and pre-booster assays respectively evaluate the booster’s efficacy and the primary intranasal vaccination’s immunogenicity and durability. Serum, spleen, lung, BALF, nasal lavage fluid (NLF) and vaginal lavage fluid (VLF) samples were collected at the indicated time points (Fig. 2A). By day 14, mice immunized with adjuvanted S-2P (S-2P + LNT i.n. or S-2P + AL i.m.) exhibited significantly higher anti-RBD serum IgG titers than those receiving unadjuvanted S-2P (10.6-fold and 12.3-fold higher, respectively). This difference decreased by day 35 (5.7-fold and 3.8-fold, respectively) (Fig. 2B). By day 35, intranasal immunization with S-2P alone (S-2P i.n.) induced lower anti-RBD IgG titers than S-2P + LNT i.n., though titers were comparable to those induced by S-2P + AL i.m. (Fig. 2B). The serum antibody kinetics between days 35 and 105 demonstrated that LNT significantly slowed antibody waning (Fig. 2C). Following the 3^rd immunization, S-2P i.n.-immunized mice exhibited a rapid increase in anti-RBD IgG titers, reaching levels comparable to the S-2P + LNT i.n. group by day 126 (Fig. 2B). The booster dose elicited 12.3-fold, 5.6-fold, and 1.7-fold increases in anti-RBD IgG titers for the S-2P i.n., S-2P + LNT i.n., and S-2P + AL i.m. groups, respectively.

Fig. 2. Systemic immune responses to intranasal S-2P immunization and the adjuvant activity of LNT.

A Immunization and sampling scheme (Created with BioRender.com). C57BL/6 mice were immunized with 10 μg S-2P ± LNT via intranasal (i.n.) or 10 μg S-2P + AL via intramuscular (i.m.) routes at days 0, 21, and 112. Serum was collected at days 14, 35, 105, and 126 post-immunization. B RBD-specific serum IgG titers measured by ELISA at days 14, 35, and 126 (n = 6 mice/group at each time point). C RBD-specific IgG titers at days 35 and 105 (n = 6 mice/group at each time point). D Neutralizing antibody titers (NT50) against authentic SARS-CoV-2 at day 35 (n = 6 mice/group). ELISpot analysis of IFN-γ and IL-17A-secreting splenocytes following stimulation with S1 (E) or S2 (F) protein (day 35; n = 4 mice/group). For antibody titers (B–D), values represent the reciprocal of the endpoint dilution and the dashed lines represent the lower limits of detection (LOD). The LOD is 100-fold dilution and samples below detection were assigned a titer of 100. The data are presented as the means ± SEMs and are representative of at least three independent experiments. Statistical analysis: one-way ANOVA (B, D, E and F); paired Student’s test (C) (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant).

We evaluated serum-neutralizing activity against SARS-CoV-2 variants of concern (VOCs) using authentic virus neutralization assays. At day 35, sera from S-2P i.n.-immunized mice showed variant-dependent neutralization profiles, with differential activity against ancestral and Omicron strains. Neutralizing antibody titers in the S-2P i.n. group were consistently lower than those in the S-2P + LNT i.n. group but comparable to the S-2P + AL i.m. group (Fig. 2D). All groups demonstrated the highest neutralization potency against ancestral SARS-CoV-2, with significantly reduced activity against Omicron variants (Fig. 2D). Notably, the S-2P + LNT i.n. group exhibited superior neutralizing activity, particularly for the Omicron variants (Fig. 2D).

On day 35 post-immunization, systemic T cell responses to the spike protein were assessed by ex vivo recall stimulation of splenocytes with the S1 or S2 protein, followed by IFN-γ and IL-17A ELISpot assays. These cytokines were selected as key readouts for assessing vaccine-induced T cell responses based on their distinct functions: (1) IFN-γ serves as a canonical marker of Th1-type T cell activation²⁵, while (2) IL-17A represents a signature cytokine of Th17 cells that plays an essential role in mucosal immunity²⁶. At day 35, both intranasal groups (S-2P i.n. and S-2P + LNT i.n.) exhibited significantly stronger spike-specific IFN-γ + T cell responses compared to intramuscular group (Fig. 2E, F and Fig. S3A, B). IL-17A-producing T cell responses, however, were negligible in all immunization groups (Fig. 2E, F and Fig. S3A, B).

These findings demonstrate that intranasal vaccination with S-2P effectively induces systemic immunity, with LNT functioning as a potent mucosal adjuvant that promotes durable and high-titer antibody responses with cross-neutralizing capacity.

Intranasal immunization with S-2P induces potent respiratory mucosal immune responses

We next assessed respiratory mucosal immune responses following intranasal immunization with S-2P. Consistent with the immunization route, intranasal- but not intramuscular- S-2P administration induced anti-RBD IgA responses in both BALF and NLF, with no detectable IgA observed in the intramuscular group (Fig. 3A, D). While anti-RBD IgG antibodies were detectable in all immunization groups, the S-2P + LNT i.n. group showed significantly elevated titers in BALF compared to other groups (Fig. 3B, E).

Fig. 3. Mucosal immune responses following intranasal immunization with S-2P.

Mice were immunized and samples were collected as depicted in Fig. 2B. RBD-specific IgA (A) and IgG (B) titers in bronchoalveolar lavage fluid (BALF) at days 35 and 126 post-immunization (n = 4 mice/group at each time point). C Neutralizing antibody titers (NT50) against authentic SARS-CoV-2 in BALF at day 35 (n = 4 mice/group at each time point). RBD-specific IgA (D) and IgG (E) titers in nasal lavage fluid (NLF) at days 35 and 126 (n = 4 mice/group at each time point). Antigen-specific T cell responses in lung cells isolated at day 35 and stimulated with S1 (F) or S2 (G) protein, quantified by ELISpot for IFN-γ and IL-17A-secreting cells (n = 4 mice/group). For all antibody titers, values represent the reciprocal of the endpoint dilution and the dashed lines represent the lower limits of detection (LOD). The LOD is 10-fold dilution and samples below detection were assigned a titer of 10. The data are presented as the means ± SEMs and are representative of at least three independent experiments. Statistical analysis: Kruskal-Wallis test (A, C, D and E); one-way ANOVA (B, F and G) (*P < 0.05, **P < 0.01; ns, not significant).

The mucosal immune system operates as an interconnected network, enabling respiratory tract-initiated immune responses to extend to distal mucosal sites. Therefore, vaginal IgA antibodies may serve as valid distal biomarkers for intranasal vaccine-induced mucosal immunity. Indeed, we detected low but measurable IgA levels in VLF at day 126 from intranasally immunized mice (Fig. S4B, C), supporting this immunological paradigm.

At day 35, BALF and NLF samples were evaluated for neutralizing activity. The S-2P + LNT i.n. group exhibited the highest neutralizing titers in BALF against both ancestral SARS-CoV-2 and Omicron variants (Fig. 3C). Notably, only NLF from the S-2P + LNT i.n. group showed detectable neutralization against ancestral SARS-CoV-2 (Fig. S4A).

On day 35 post-immunization, lung-resident T cell responses were evaluated by ELISpot following ex vivo stimulation with SARS-CoV-2 S1 or S2 proteins. Intranasal immunization groups (S-2P i.n. and S-2P + LNT i.n.) showed significantly enhanced S1- and S2-specific IFN-γ and IL-17A responses compared to intramuscular immunization group (Fig. 3F, G and Fig. S4D, E). These results demonstrate the superior capacity of intranasal S-2P vaccination to elicit robust respiratory mucosal immunity.

Intranasal immunization with S-2P provides complete protection against lethal challenge with ancestral SARS-CoV-2 in K18-hACE2 mice

To assess protective efficacy, K18-hACE2 transgenic mice were immunized following the regimen in Fig. 4A. Notably, unlike observations in C57BL/6 mice, all three immunization protocols elicited robust serum anti-RBD IgG antibody responses in K18-hACE2 mice by day 14. S-2P i.n. alone induced anti-RBD IgG titers comparable to S-2P + LNT i.n. (Fig. 4B). However, the S-2P + LNT i.n. group showed superior neutralizing activity against both ancestral SARS-CoV-2 and Omicron BA.2 (Fig. 4C).

Fig. 4. Protective efficacy of intranasal S-2P immunization against ancestral SARS-CoV-2 challenge in K18-hACE2 mice.

A Experimental design (Created with BioRender.com). K18-hACE2 mice received either intranasal (i.n.) immunization with 10 μg S-2P ( ± LNT) or intramuscular (i.m.) immunization with 10 μg S-2P ( + AL) in a prime-boost regimen with 3-week intervals (n = 10 mice/group). On day 42, mice were challenged intranasally with 10 PFU ancestral SARS-CoV-2. Five mice/group were sacrificed at 4 days post-infection (dpi) for viral load and histopathology analysis; remaining mice were monitored for clinical outcomes. B RBD-specific serum IgG titers at days 14 and 35 post-immunization. C Serum neutralizing antibody titers (NT50) against authentic SARS-CoV-2 at day 35. Clinical monitoring showing body weight changes (D) and survival rates (E) post-challenge. Viral RNA copies quantified by RT-qPCR for SARS-CoV-2 N protein in nasal turbinates (F) and lung tissues (G) at 4 dpi (n = 5 mice/group). H Representative H&E-stained lung sections at 4 dpi (scale bars = 100 μm), the blank control group consisted of parallel-housed mice receiving neither immunization nor virus challenge. For B–C, titers represent reciprocal endpoint dilutions and dashed lines represent the lower limits of detection (LOD). The LOD is 100-fold dilution and samples below detection were assigned a titer of 100. The data are presented as the means ± SEMs. Statistical analysis: Kruskal–Wallis test (B, C, F, G); log-rank (Mantel–Cox) test (E) (*P < 0.05, **P < 0.01; ns=not significant).

Having confirmed the antibody response after a standard two-dose schedule, the mice were subsequently challenged with SARS-CoV-2. At 21 days post-boost, the mice were intranasally challenged with 10 PFU ancestral SARS-CoV-2. Each challenge group comprised 10 mice, with 5 mice monitored for clinical progression and the remaining 5 mice euthanized at 4 days post-infection (dpi) for viral load quantification and histopathological examination. PBS-control mice exhibited rapid weight loss and 100% mortality within 6 days post-infection. Both intranasal regimens (S-2P i.n. and S-2P + LNT i.n.) conferred complete protection (5/5), with all mice surviving without weight loss (Fig. 4D, E). In contrast, S-2P + AL i.m. provided only 40% protection (2/5). Viral load analysis revealed significant suppression of replication in nasal turbinates and lungs for intranasal immunization groups, but not for intramuscular immunization group (Fig. 4F, G).

Histopathological examination was conducted at 4 days post-challenge. Lung tissues of PBS-control mice exhibited thickened alveolar walls and inflammatory cell infiltration. In contrast, intranasally immunized mice (either S-2P i.n. or S-2P + LNT i.n.) displayed significantly attenuated pulmonary inflammation (Fig. 4H). However, prominent inflammation was observed in S-2P + AL i.m. mice (Fig. 4H). Pathological scoring further confirmed that PBS-control mice had significantly higher scores for alveolar wall thickening, impaired pulmonary expansion, intraalveolar hemorrhage, and inflammatory infiltration compared to intranasally immunized groups (Fig. S5A).

Intranasal S-2P immunization confers protection against lethal SARS-CoV-2 Omicron variant challenge

Cross-protective capacity is an important indicator for evaluating the efficacy of COVID-19 vaccines. To assess protection against Omicron variants, K18-hACE2 mice were immunized intranasally with S-2P in prime-boost regimen (3-week interval) and subsequently challenged with Omicron BA.5 or EG.5 variants 21 days post-boost. Each challenge group comprised 10 mice, with 5 mice monitored for clinical progression and the remaining 5 mice euthanized at 4 dpi for viral load quantification and histopathological examination. PBS-control mice exhibited 100% mortality within 8 days post-infection. S-2P i.n. provided 60% protection (3/5) against BA.5 and 40% protection (2/5) EG.5 (Fig. 5B, D). Strikingly, S-2P + LNT i.n. conferred complete protection against both variants, as evidenced by sustained weight stability and 100% survival (5/5) (Fig. 5A–D). Conversely, S-2P + AL i.m. failed to protect against either Omicron variant (Fig. 5A–D).

Fig. 5. Protective efficacy of intranasal S-2P immunization against SARS-CoV-2 Omicron variants in K18-hACE2 mice.

K18-hACE2 mice received i.n. immunization with 10 μg S-2P ± LNT adjuvant or i.m. immunization with 10 μg S-2P + AL adjuvant in prime-boost regimen (3-week interval; n = 10 mice /group). At day 21 post-boost, mice were challenged intranasally with 1000 PFU Omicron BA.5 or 10 PFU EG.5. Five mice/group were sacrificed at 4 dpi for virological/histopathological analysis; remaining mice were monitored for clinical outcomes. Clinical outcomes following Omicron challenge: body weight changes (A, C) and survival rates (parentheses, B, D) after BA.5 (A–B) or EG.5 (C–D) infection. Representative H&E-stained lung sections at 4 days post-challenge with BA.5 (E) or EG.5 (F), the blank control group consisted of parallel-housed mice receiving neither immunization nor virus challenge (scale bars=100 μm). Viral RNA copies of SARS-CoV-2 N gene in nasal turbinates (G) and lung tissues (H) at 4 days post-challenge, quantified by RT-qPCR (n = 5 mice/group). The data are presented as the means ± SEMs. Statistics: log-rank (Mantel–Cox) test (B, D); Kruskal–Wallis test (G); one-way ANOVA (H) (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant).

Following Omicron BA.5 (Fig. 5E) or EG.5 (Fig. 5F) challenge, S-2P + LNT i.n. group exhibited significantly attenuated pulmonary inflammation. S-2P i.n. group similarly showed markedly reduced lung inflammation after BA.5 challenge (Fig. 5E) and moderately reduced pathology following EG.5 challenge (Fig. 5F). In contrast, the S-2P + AL i.m. group demonstrated severe pulmonary inflammation comparable to PBS controls (Fig. 5E, F). Pathological scoring of lung tissues corroborated these observations (Fig. S5B, C). Intranasal immunization significantly reduced viral loads for both BA.5 and EG.5 variants in nasal turbinates and lung tissues, while intramuscular immunization showed no significant reduction (Fig. 5G, H).

Together, these results demonstrate that intranasal immunization with ancestral S-2P confers significant cross-protection against Omicron variant challenge, with LNT adjuvant further enhancing protective efficacy. Although intranasal immunization with S-2P induced only modest levels of cross-reactive neutralizing antibodies against heterologous Omicron variants, it nevertheless provided substantial protection against virus challenge. This observation suggests that additional immune mechanisms beyond humoral responses contribute to S-2P-mediated protection against Omicron strains following intranasal immunization.

The immune protection against variants challenge is mediated by IFN-γ, and LNT adjuvant activates effector cells beyond CD8 + T cells

Given the remarkable antibody evasion capacity of Omicron variants, we next examined the contribution of CD4+ and CD8 + T cells to S-2P intranasal immunization-mediated protection against Omicron BA.5. Immunized K18-hACE2 mice were subjected to CD4 + T cell, CD8 + T cell, or combined depletion before and after BA.5 challenge (Fig. 6A). Peripheral blood samples were collected and analyzed by flow cytometry to verify depletion efficiency. Consistent with previous reports²⁷, the antibodies effectively eliminated target cell populations from peripheral circulation (Fig. 6B–E and Fig. S6A).

Fig. 6. Intranasal immunization with S-2P requires IFN-γ and CD8 + T cells for protection, whereas LNT adjuvant activates additional effector cells.

A Experimental design (Created with BioRender.com). The K18-hACE2 mice received i.n. immunization with 10 μg S-2P ± LNT in prime-boost regimen (3-week interval). Depleting antibodies (anti-CD4, anti-CD8, or combination) were administered intravenously on days 38, 40, and 45; anti-IFN-γ neutralizing antibody was given on days 42 and 45 (n = 5 mice/group). Mice were challenged intranasally with 1000 PFU Omicron BA.5 on day 42. Flow cytometry analysis of CD8+ (B, D) and CD4+ (C, E) T cell frequencies in peripheral blood of S-2P (B, C) or S-2P + LNT (D, E) immunized mice at days 36 and 41. Clinical outcomes showing body weight changes (F, H) and survival rates (parentheses, G, I) of S-2P (F, G) or S-2P + LNT (H, I) immunized mice post-challenge. Statistics: one-way ANOVA (B–E); log-rank (Mantel–Cox) test (G, I) (*P < 0.05, **P < 0.01, ****P < 0.0001; ns, not significant).

Consistent with above findings, S-2P i.n. achieved 60% protection (3/5), while S-2P + LNT i.n. provided complete protection (5/5) (Fig. 6F-I). CD4 + T cell depletion minimally affected protective efficacy in both immunization groups (Fig. 6F-I). Notably, while CD8 + T cell ablation abolished all protection in S-2P i.n.-immunized mice (0% survival) (Fig. 6F, G)., S-2P + LNT i.n.-immunized mice retained partial protection (60% survival, 3/5) with sustained weight maintenance post-depletion (Fig. 6H, I and Fig. S6C). Additionally, although both groups showed comparable 60% survival (3/5) against BA.5 challenge, S-2P + LNT i.n.-immunized mice following CD8 + T cell depletion exhibited significantly milder weight loss than S-2P i.n.-immunized mice (Fig. S6B, C).

To delineate the protective mechanisms, we evaluated IFN-γ, a pivotal cytokine orchestrating antiviral cellular immunity. K18-hACE2 transgenic mice were administered anti-IFN-γ antibody via intravenous injection at 4 h pre-challenge and 3 days post-challenge with Omicron BA.5 (Fig. 6A). IFN-γ neutralization substantially abrogated vaccine efficacy, with treated mice demonstrating comparable disease progression to unimmunized PBS controls, as evidenced by similar weight loss kinetics and mortality rates.

These findings demonstrate that vaccine-mediated protection against heterologous SARS-CoV-2 variants is IFN-γ-dependent. The results further reveal distinct mechanistic profiles: (1) CD8 + T cells serve as the principal effectors mediating protection following intranasal immunization with S-2P, while (2) LNT adjuvant engages additional immune effectors beyond CD8 + T cells to enhance protection.

STING-dependent innate immune activation by S-2P

The observed protection against both homologous and heterologous SARS-CoV-2 challenges following intranasal immunization with S-2P prompted mechanistic investigation. Transcriptomic analysis of cervical lymph nodes (CLNs) from C57BL/6 mice 24 h post-intranasal administration of 10 μg S-2P or HA (n = 3) revealed 789 differentially expressed genes (DEGs; fold change ≥2, P < 0.05) in S-2P-treated mice. Notably, S-2P specifically upregulated type I IFN-related genes (Oas3, Ifit3, Ifit1), which remained unchanged in HA control (Fig. 7A). While Ifnb1 and Cxcl10 showed upregulation in S-2P group, high inter-individual variability yielded non-significant differences.

Fig. 7. Innate immune activation and self-adjuvant activity of intranasal S-2P immunization.

A Transcriptomic analysis of cervical lymph nodes (CLNs) from C57BL/6 mice 24 h post intranasal S-2P administration (n = 3 mice/group) revealed upregulated type I interferon (IFN)-related genes. B Bone marrow-derived macrophages (BMDMs) stimulated with S-2P or HA protein (n = 3) for 24 h showed selective induction of type I IFN pathway genes. C RBD-specific serum IgG titers (days 14 and 35) in wild-type C57BL/6 and knockout mice (Sting-/-, Tlr4-/-, Tlr2-/-, and Cd209-/-) following intranasal S-2P prime-boost immunization (10 μg/dose, n = 4 mice/group). D RBD-specific serum IgG titers at days 14 and 35 post-immunization in mice pretreated with intranasal Diprotein A (control peptide), ATN161, or RGDS 6 h before S-2P immunization (n = 4 mice/group). C57BL/6 mice following intranasal HA/S-2P prime-boost immunization (10 μg each of HA and spike protein per dose, n = 4 mice/group): HA-specific (E) and RBD-specific (F) serum IgG titers at days 14 and 35, and their correlation at day 35 (G) (n = 4 mice /group). For antibody titers (C–F), values represent reciprocal endpoint dilutions and the dashed lines represent the lower limits of detection (LOD). The LOD is 100-fold dilution and samples below detection were assigned a titer of 100. The data are presented as the means ± SEMs. Statistical analysis: Kruskal–Wallis test (A, C, E and F); one-way ANOVA (B and D)(*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns=not significant). Correlation in G analyzed by Pearson’s test (r = 0.9961, P = 0.0039).

KEGG analysis identified multiple S-2P-modulated signaling pathways, with the cytosolic DNA-sensing pathway showing particularly strong association (Fig. S7A). To further characterize S-2P’s immunostimulatory properties, we performed transcriptomic profiling on bone marrow-derived dendritic cells (BMDCs) and macrophages (BMDMs) stimulated with 5 μg/mL S-2P or HA control for 24 h. S-2P treatment significantly upregulated type I IFN-related genes (Cxcl10, Irf7, Isg15, Ifit3, Oasl1, Oas3) compared to HA control (Fig. 7B, S7B). Remarkably, the gene expression profiles were highly concordant between BMDCs and BMDMs.

Consistent with observations in CLNs, the expression of Sod2 and Trex1, genes linked to the cytosolic DNA-sensing pathway, was upregulated by S-2P. This strongly implicates the activation of cGAS-STING signaling by S-2P. Therefore, we utilized Sting-/- mice to investigate the role of the cGAS-STING signaling pathway in immune responses elicited by intranasal vaccination with S-2P (Fig. 7C). At day 14, anti-RBD serum IgG titers were undetectable in Sting-/- mice, and the antibody titers were significantly lower than those in C57BL/6 mice at day 35, indicating that cGAS-STING signaling plays an important role in inducing humoral responses to intranasal immunization with S-2P. Previous studies have identified three PRRs, including CD209, TLR2 and TLR4, capable of binding SARS-CoV-2 spike protein^19,22,23. We found comparable anti-RBD serum IgG titers in S-2P intranasally vaccinated Cd209-/-, Tlr2-/- and Tlr4-/- mice versus wild-type C57BL/6 mice. Therefore, these three PRRs are unlikely to critically participate in the induction of immune responses to intranasal administration of S-2P. Integrins are widely reported to be involved in SARS-CoV-2 entry via direct binding the spike protein^20,21 and are abundantly expressed on respiratory epithelial cells. We utilized small molecule integrin inhibitors (ATN161 and RGDS) to determine whether integrins are involved in immune responses to intranasal immunization with S-2P. The results revealed that the serum RBD-specific IgG titers significantly decreased in the integrin inhibitor-treated mice (Fig. 7D), demonstrating the important role of integrins in mediating immune responses following intranasal S-2P immunization.

Based on S-2P’s ability to activate innate immunity via intranasal immunization, we assessed whether S-2P could enhance HA immunogenicity by co-administering both antigens intranasally. At day 35, serum anti-HA IgG titers were detectable in the HA + S-2P i.n. group but undetectable in the HA i.n. group (Fig. 7E). Meanwhile, Serum anti-RBD antibody production appeared unaffected by HA co-administration (Fig. 7F). Furthermore, anti-HA IgG titers showed a strong positive correlation with anti-RBD IgG titer (r = 0.9961, P = 0.0039) (Fig. 7G). These results demonstrate that S-2P exhibits potential as a self-adjuvant, capable of enhancing immune responses to a co-delivered antigen.

Discussion

While heterologous prime-boost strategies (intramuscular priming followed by intranasal unadjuvanted spike boosting) have shown protective efficacy^15,16, our study demonstrates that homologous intranasal vaccination with adjuvant-free ancestral SARS-CoV-2 spike protein efficiently elicits systemic and mucosal immunity, conferring protection against lethal challenge by both ancestral SARS-CoV-2 and Omicron variants. Furthermore, the addition of LNT, a clinically used immunomodulator, significantly enhanced immune responses and protective efficacy, conferring complete protection against Omicron variants. Our findings underscore that engineering vaccine antigens with intrinsic immunostimulatory properties represents a promising strategy for the development of effective mucosal vaccines.

Accumulating clinical evidence demonstrates that respiratory mucosal IgA antibodies play a pivotal role in preventing SARS-CoV-2 respiratory tract infection, with antibody levels showing an inverse correlation with breakthrough infection risk^28,29. However, intramuscular vaccination fails to elicit mucosal IgA antibodies, and naturally acquired mucosal IgA responses cannot be boosted through subsequent vaccination^30,31. Notably, intranasal delivery of unadjuvanted spike protein alone effectively elicits robust respiratory IgA responses, highlighting its unique immunogenic properties when administered via intranasal route.

The HEK293 cell-expressed S1 subunit similarly elicited serum RBD-specific IgG and BALF IgA responses, albeit at lower magnitudes than S-2P, likely attributable to S1’s reduced molecular weight and monomeric structure versus S-2P’s stable homotrimeric conformation (incorporating full-length S1 and S2 extracellular domains). Strikingly, RBD (containing the primary ACE2 receptor-binding motif), despite its apparent immunological paradox, failed to elicit detectable antibodies—an observation consistent with prior reports^32,33. Even intranasal immunization with a triple RBD scaffold protein (HEK293-expressed) proved ineffective at inducing serum IgG or mucosal IgA responses³³, suggesting that non-RBD regions within S1 play essential roles in mucosal immunity induction. Notably, insect cell-expressed full-length spike ectodomain failed to induce detectable serum IgG or BALF IgA antibodies, consistent with reported failures of Drosophila cell-expressed trimeric S-2P to elicit neutralizing antibodies or protection after multiple immunizations³⁴. The observed immunogenicity differences are likely attributable to distinct glycosylation patterns between insect cell-derived and mammalian cell-derived spike proteins.

In contrast to S-2P protein, intranasal immunization with influenza HA protein (HEK293-expressed) failed to induce detectable antibody responses, prompting comparative immunogenicity analysis of these viral glycoproteins. Notably, S-2P, but not HA, induced significant upregulation of type I IFN-related genes in vitro (BMDMs and BMDCs) and in vivo (CLNs). Mechanistic studies revealed that integrins and cGAS-STING signaling mediated the immune response to intranasal S-2P immunization, whereas pattern recognition receptors (PRRs) known to interact with spike protein, including CD209, TLR2, and TLR4, were dispensable for this response. Substantial evidence demonstrates that SARS-CoV-2 spike protein interacts with multiple integrin subtypes through distinct binding domains^20,21. Crucially, these integrin interactions participate in two key immunological processes: First, integrins expressed on mucosal microfold (M) cells mediate antigen transcytosis across the epithelial barrier to submucosal lymphoid tissues³⁵. Second, as established by studies, integrin engagement triggers focal adhesion kinase (FAK)-mediated mitochondrial dysfunction, leading to STING pathway activation^36,37. We therefore propose that integrins play dual roles in mucosal immunity: (1) integrins localized on M cells facilitate spike protein transport to lymphoid tissues, while (2) integrins expressed on antigen-presenting cells (APCs) initiate STING signaling cascades. The limited immunogenicity of intranasally administered influenza HA likely results from its inability to traverse the epithelial barrier and lack of PRR binding motifs, though impaired epithelial transcytosis appears to be the predominant factor. These functional limitations may originate from unique structural features or glycosylation profiles intrinsic to the HA protein.

Type I IFNs serve dual immunological functions, mediating both viral clearance and adaptive immune regulation^38,39. Some inducers of type I IFNs, such as STING agonists, have been reported to significantly enhance immune responses as adjuvants of recombinant vaccines⁴⁰. Moreover, direct fusion of IFN-β to the SARS-CoV-2 RBD efficiently enhances the immunogenicity of RBD for sterilizing immunity⁴¹. Notably, natural infection with SARS-CoV-2 is inefficient in inducing effective and durable respiratory mucosal immunity, which may be highly correlated with the suppression of type I IFN signaling during viral infection⁴². In this study, S-2P co-administration significantly enhanced HA-induced antibody responses following intranasal vaccination, with serum anti-HA titers positively correlating with anti-RBD titers. This demonstrates both the self-adjuvant properties of spike protein and provides proof-of-concept for spike-based intranasal immunization strategies.

LNT is a type of β-glucan, purified from Shiitake mushrooms, that consists of a backbone of β-(1, 3)-linked β-D-glucopyranosyl units with β-(1, 6)-linked side chains. β-glucans are recognized by the mammalian innate immune system via a number of cell surface receptors, including Dectin-1 (a C-type lectin), TLR2, and complement receptor type 3 (CR3)⁴³. LNT has been shown to be capable of inducing the expression of interferon-stimulated genes (ISGs) and evoking antiviral immunity in several viral infection models⁴⁴. In our randomized, placebo-controlled trial involving 24 COVID-19 patients (12 receiving escalating doses [50, 75, or 100 µg/day] and 12 controls), lentinan nasal drops demonstrated safety, with the 100 µg/day dose significantly reducing viral shedding time²⁴, which inspired us to investigate whether LNT possesses mucosal adjuvant activity. Our study demonstrates that intranasal administration of LNT significantly upregulates type I IFN-related gene expression in lung tissues, underlying its mechanistic basis as a potent mucosal adjuvant. When co-administered with S-2P, LNT enhanced both the magnitude and durability of immune responses, enabling complete protection against lethal challenge with heterologous Omicron BA.5 and EG.5 variants—a marked improvement over S-2P alone.

Although conserved neutralizing epitopes are distributed in the spike protein, including in the N-terminal domain (NTD) of S1 and S2 and even in highly variable RBD regions, the corresponding antibodies are not effectively induced by natural infection or vaccination, and their neutralizing activities are relatively inferior⁴⁵. As shown in this study, the neutralizing activity against Omicron variants were significantly compromised in mouse serum, especially for the EG.5 variant, which dropped to the detection limit. Unlike neutralizing antibody epitopes, T cell epitopes are fairly conserved. A previous study revealed that the majority of T cell responses to the spike protein induced by vaccination or infection cross-recognized Omicron variants⁴⁶. Several studies reported that durable and broadly reactive T cell responses to the spike protein were detected in most individuals recovered from COVID-19^47–49. Both natural infection and vaccination have been shown to elicit antibody responses that provide sufficient protection against homologous SARS-CoV-2 strains⁵⁰. Moreover, the CD8 + T cell responses are sufficient to clear not only homologous but also heterologous SARS-CoV-2^27,50,51. In this study, intranasal vaccination with either S-2P or S-2P plus LNT induced robust systemic and respiratory mucosal T cell responses. Beyond inducing IFN-γ-producing T cells, intranasal immunization with S-2P triggered robust IL-17A-producing T cell responses in mouse lung tissues. While IL-17A mediates crucial mucosal defense functions, its dysregulation may potentially drive immunopathology²⁶. Our results demonstrate that for vaccines against severe respiratory pathogens, the protective advantages of IL-17A induction substantially outweigh its potential detrimental effects.

In contrast to intranasal administration, S-2P + AL i.m. failed to induce apparent T cell responses both in lungs and spleens. Given the comparable serum neutralizing antibody titers against ancestral SARS-CoV-2, Omicron BA.5 or EG.5 variants across vaccination groups and the overwhelming neutralizing antibody evasion ability of Omicron variants, the substantial protection differences against Omicron variants likely reflect mechanisms beyond humoral immunity. Whereas S-2P + AL i.m. mediated partial protection against the ancestral strain principally via neutralizing antibodies, the enhanced protective efficacy of intranasal vaccination appears to be largely dependent on T cell mediated immunity.

In this study, depletion of CD4 + T cells showed negligible effect on the protective efficacy against Omicron BA.5 challenge, suggesting a minor or complementary contribution of CD4 + T cells in the cellular immunity component to viral clearance. Consistent with prevailing perspective, depletion of CD8 + T cells completely abolished the protective effect of intranasal immunization with S-2P. However, depletion of CD8 + T cells only partially compromised the effect of intranasal immunization with S-2P plus LNT. As described previously, the procedure almost eliminates CD8 + T cells from peripheral blood and eliminates more than 99.9% of CD8 + T cells from lungs²⁷. Nevertheless, 60% of the mice survived without showing significant physical signs of disease or weight loss. Several independent studies have shown that CD4 + T cells play a minor role in viral clearance, but the CD8 + T cells are essential for combating infections in the absence of matched neutralizing antibodies^27,50–52. Thus, the immune protection of S-2P + LNT i.n. involves additional effector cells other than CD8 + T cells.

IFN-γ, the only type-II interferon, is an important antiviral cytokine and an activator of macrophages²⁵. The potent direct antiviral function of IFN-γ has been demonstrated against a variety of viruses⁵³, including SARS-CoV-2⁵⁴. Our findings demonstrate the essential role of IFN-γ in ancestral spike vaccine-induced immune protection against heterologous variants. IFN-γ is mainly produced by activated Th1 cells, CD8+ cells and NK cells. Emerging data provide evidence that NK cells, like macrophages, can also establish trained immunity following primary antigen exposure and mediate antigen-specific immunological memory against subsequent infectious agents, including HIV, influenza virus, cytomegalovirus and hepatitis B virus^55–57. Furthermore, β-glucans have been characterized as potent inducers of trained immunity, training innate cells with enhanced responsiveness^58,59.

Building on these findings, we will pursue two complementary avenues: defining the minimal immunogenic determinants of the spike protein and elucidating lung-resident NK cell responses against Omicron variants. These investigations may yield fundamental insights for next-generation mucosal vaccine development, offering a dual advancement in structural antigen engineering and adjuvant mechanism understanding.

Our study demonstrates that mammalian cell-derived SARS-CoV-2 spike protein exhibits intrinsic self-adjuvant activity through STING-dependent type I interferon signaling following intranasal vaccination. Furthermore, LNT serves as a potent and safe mucosal adjuvant that engages complementary effector mechanisms beyond CD8 + T cell responses. The combined use of spike protein and LNT constitutes a promising broad-spectrum mucosal vaccine platform capable of addressing the persistent evolution and transmission of SARS-CoV-2. These findings not only advance the development of respiratory mucosal vaccines against emerging pathogens but also reveal new mechanisms of vaccine-induced protective immunity.

Methods

Mice and ethics

Wild-type C57BL/6 mice were purchased from Shanghai Jihui Experimental Animal Breeding Co., Ltd. Transgenic mice expressing human ACE2 via the human cytokeratin 18 promoter (K18-hACE2), Tlr4-/-, and Tlr2-/- mice were purchased from Jiangsu GemPharmatech Co., Ltd. Cd209-/- mice were purchased from Shanghai Model Organisms Center, Inc. Sting-/- mice were kindly provided by Prof. Haipeng Liu (Shanghai Pulmonary Hospital affiliated to Tongji University). All mice were housed in individually ventilated cages (IVCs) in ABSL-2 for noninfectious studies. The infection studies were performed in the ABSL-3 facility at Naval Medical University. The animal experiments were approved by the Institutional Committee for Animal Care and Biosafety of Naval Medical University and conducted according to the Regulations for the Administration of Affairs Concerning Experimental Animals in China.

Cells, proteins, lentinan and viruses

Vero E6 cells (kindly provided by Dr. Rong Zhang of Fudan University) were maintained in complete DMEM (Gibco) supplemented with 10% fetal bovine serum (Gibco), 1% penicillin-streptomycin (Thermo Fisher Scientific), 1% L-glutamine (Thermo Fisher Scientific), and 1% nonessential amino acids (Thermo Fisher Scientific), and incubated at 37 °C under 5% CO₂. Suspended CHO (CHO-S) cells (Invitrogen) were cultured in EX-CELL™ 325 PF CHO serum-free medium (Sigma-Aldrich) at 37 °C with 5% CO₂ and shaken at 130 rpm.

The spike protein trimers expressed in HEK293 cells were purchased from Novoprotein Scientific Inc. (Novoprotein Scientific, DRA49) and Sino Biological Inc. (Sino Biological, 40589-V08H4). The full-length ectodomain of the spike protein expressed in baculovirus-insect cells (Sino Biological, 40589-V08B1), the RBD protein expressed in HEK293 cells (Sino Biological, 40592-V08H), the S1 protein expressed in 293 cells (Sino Biological, 40591-V08H), the S2 protein expressed in HEK293 cells (Sino Biological, 40590-V08H1) and the HA protein of the influenza virus expressed in HEK293 cells (Sino Biological, 40668-V08H) were purchased from Sino Biological, Inc. Lentinan was purchased from MCE (MedChemExpress), Catalog Number: HY-N6653, with a polysaccharide content of 51.15%.

All the SARS-CoV-2 viruses were isolated from nasopharyngeal swab samples harvested from nucleic acid-positive cases and were propagated and titrated in Vero E6 cells as described previously⁶⁰. The viral genomes were sequenced via next-generation sequencing. The viruses used in this study included the ancestral SARS-CoV-2 (GenBank accession no. MT622319.1), Omicron BA.2 (GenBank accession no. MT627325.1), Omicron BA.5 (GenBank accession no. PQ213095.1), Omicron BF.7 (GenBank accession no. PQ351184), Omicron EG.5 (GenBank accession no. PQ215937).

S-2P preparation

The mammalian cell codon-optimized cDNA expressing the full-length ectodomain of the spike protein of the ancestral SARS-CoV-2 was synthesized by Shanghai Generay Biotech Co., Ltd. as described previously⁶¹. A trimeric foldon of the T4 fibritin was inserted at the C-terminal domain of the ectodomain of the spike protein (GenBank: NC_045512.2), in which the furin cleavage site RRAR was replaced with GSAS and 2 proline mutations were introduced at residues 986 and 987 to stabilize the prefusion conformation. A polyhistidine sequence was inserted at the C-terminus for protein purification. The cDNA was inserted into the modified CHO cell expression vector pCI-GS⁶² and transfected into CHO cells via electroporation. The cells stably expressing the spike protein were selected in a serum-free medium devoid of L-glutamine and supplemented with 25 μM methionine sulfoximine. The stable cells were gradually amplified to a volume of 200 mL per shake flask with a capacity of 1 L. The supernatants were subsequently collected and purified via high-speed centrifugation, ultrafiltration, nickel-immobilized metal affinity chromatography and desalination. The purified protein, designated S-2P, was identified via SDS-PAGE and Coomassie blue staining and confirmed < 0.1 EU per μg protein as determined by the Limulus Amebocyte Lysate (LAL) method. The antigenicity of S-2P was confirmed by western blot using rabbit anti-RBD polyclonal antibody (Sino Biological, 40592-T62) and mouse anti-S2 monoclonal antibody (Sino Biological, 40590-MM05), and its concentration was determined via bicinchoninic acid (BCA) assay.

Vaccination and virus challenge

Given that recombinant protein vaccines are conventionally administered via intramuscular injection with adjuvants, we compared the immune responses and protective efficacy induced by intranasal versus intramuscular delivery of spike protein in 6-to-8-week-old female C57BL/6 or K18-hACE2 mice. Prior to immunization, mice were anesthetized with isoflurane. For intranasal delivery, 10 μg S-2P or 10 μg S-2P plus 200 μg LNT was administered in a 50 μL volume. For intramuscular immunization, 10 μg S-2P with 100 μg aluminum adjuvant (Croda, Denmark) in a 100 μL volume was injected into the tibialis anterior muscles. Unless otherwise specified, mice were vaccinated twice at 3-week intervals.

For virus challenge experiment, mice were inoculated intranasally with 50 μL medium containing either 10 PFU ancestral SARS-CoV-2, 1000 PFU Omicron BA.5 or 10 PFU Omicron EG.5. Mouse body weights and survival rates were monitored daily for 14 days post-challenge. On day 4 post-challenge, lungs and nasal turbinates were collected for viral load quantification and histopathological analysis.

Sample collection and histology

The mice were euthanized via isoflurane asphyxiation. Blood was harvested from the orbital venous plexus of each mouse, and serum was obtained by centrifugation (7000 rpm for 10 min). Bronchoalveolar lavage fluid (BALF) and nasal lavage fluid (NLF) were obtained via tracheal intubation by irrigating with 1 mL of PBS, and collecting the fluid. Vaginal lavage fluid (VLF) was obtained by irrigating the vagina with 1 mL of PBS and collecting the fluid. For lung tissues, the lobes were diced and digested for 45 min at 37 °C with collagenase D (Roche) and DNase I (Roche), and then filtered through 70-μm sieves to obtain single-cell suspensions. Spleens and cervical lymph nodes were pressed through 70-μm sieves, and red blood cell lysis buffer (Invitrogen) was subsequently added to lyse erythrocytes to obtain single-cell suspensions of mononuclear cells.

Nasal turbinates or lungs of the mice were collected and fixed in 4% paraformaldehyde for 1 week at room temperature for hematoxylin and eosin (H&E) staining. Whole-slide imaging was performed using a Pannoramic 250FLASH slide scanner (3DHISTECH). Histopathological changes were evaluated based on four key parameters: alveolar wall thickening, impaired pulmonary expansion, intraalveolar hemorrhage, and inflammatory cell infiltration. Each parameter was graded on a 0-4 severity scale, with total pathology scores representing the cumulative sum across all categories per tissue section.

Enzyme-linked immunosorbent assay (ELISA)

Antibody responses were assessed via ELISA. In general, 96-well Maxisorp plates (Thermo Fisher Scientific) were coated with the purified recombinant protein (1 μg/mL) in carbonate-bicarbonate buffer overnight at 4 °C. The coating buffers were aspirated, and the wells were blocked with 180 μL of 1% bovine serum albumin (BSA) plus 0.05% Tween 20 (blocking buffer) at room temperature for 2 h. Next, serially diluted samples in blocking buffer were added to the plates for 3 h at room temperature. After washing four times with PBS plus 0.05% Tween 20 (PBST), goat anti-mouse IgG-HRP (Invitrogen) or goat anti-mouse IgA-HRP antibodies (Southern Biotech) were added to the plates for 2 h at room temperature. The plates were washed four times with PBST, and 100 μL of substrate (1-Step^TM Slow TMB, Thermo Fisher Scientific) was added. Reactions were stopped with 50 μL of ELISA stop solution (Solarbio) after 25-30 min incubation. The absorbance of each well at 450 nm was determined via a microplate reader (BioTek). Endpoint titers for IgG and IgA were calculated as the reciprocal of the highest serum dilution yielding an optical density (450 nm) value 2.1-fold greater than the negative control. The dashed line in figures indicates the lower limit of detection (LOD, defined as 100 for serum and 10 for BALF, NLF, or VLF samples); samples below the detection threshold were assigned the minimum values.

Enzyme-linked immunospot assay (ELISpot)

Capture antibodies (IFN-γ or IL-17A; Mabtech) were diluted with sterile PBS following the manufacturer’s instructions and coated onto MSIP 96-well plates (Mabtech) overnight at 4 °C. After washing 5 times with sterile PBS, the plates were blocked with RPMI 1640 medium for 1 h at room temperature. Splenocytes or lung cells isolated from mice were seeded into plates (2 × 10⁵ cells/well) in serum-free medium (Dakewe) and stimulated with S1 or S2 protein at 10 μg/ml for 36 h at 37 °C with 5% CO₂. Detection antibodies were diluted with PBS-0.5% FCS and incubated for 2 h at room temperature. Streptavidin-ALP (Mabtech) was diluted with PBS-0.5% FCS and incubated for 1 h at room temperature. Substrate (BCIP/NBT-plus, Mabtech) was added to develop spots in the dark for 30 min. Spot-forming units (SFU) were imaged via an S6 ultra-immunoscan reader (Cellular Technology Ltd.) and enumerated using S6 Ultra-ImmunoSpot reader with ImmunoSpot 5.1.36 software (Cellular Technology Ltd.). Spot-forming units (SFUs) were quantified and reported per 2 × 10⁵ splenocytes or lung cells. To express results per 10⁶ cells, SFU counts were multiplied by 5. Results represent mean SFUs ± SEM from triplicate wells, with each spot corresponding to a single cytokine-secreting cell.

Neutralization assay

Serial threefold dilutions (1:10 to 1:10240) of samples were incubated with 0.1 MOI of ancestral SARS-CoV-2, Omicron BA.2, BA.5, BF.7, or EG.5 for 45 min at 37 °C. Antibody-virus complexes were added to Vero E6 cell monolayers in 96-well plates and incubated for 18-20 h at 37 °C with 5% CO₂. Then, cells were fixed with methanol for 30 min at −20 °C and blocked with 3% BSA for 2 h at room temperature. The SARS-CoV-2 nucleocapsid antibodies (Sino Biological) were added to the plates for 3 h at room temperature. After washing with PBS three times, 100 μL of Alexa Fluor 488-conjugated goat anti-rabbit IgG antibody (Invitrogen) was added to the plates, which were incubated for 2 h at room temperature. The plates were washed three times with PBS, and 50 μL of diamidino-phenyl-indole (DAPI) was added to the plates for 15 min at room temperature. After washing, the fluorescence of each well was read with a microplate reader (Bio Tek). The neutralization titer was defined as the reciprocal of the serum dilution at which neutralizing antibodies caused 50% inhibition of viral infection compared with the virus control wells and was analyzed by GraphPad Prism 7 software.

RNA extraction and viral load quantification by RT-qPCR

Total RNA was extracted from tissues using TRIzol^TM Reagent (Invitrogen). For cDNA synthesis, 2 μg RNA was reverse transcribed with PrimeScript^TM RT Master Mix (TaKaRa). Quantitative PCR was subsequently performed with TB Green^® Premix Ex Taq^TM II (TaKaRa). β-actin expression was used for normalization. The sequences of the primers used were as follows: SARS-CoV-2 N gene forward: 5′-AAGGCGTTCCAATTAACACCA-3′; SARS-CoV-2 N gene reverse: 5′-TGCCGTCTTTGTTAGCACCA-3′; β-actin forward: 5′-GGCTGTATTCCCCTCCATCG-3′; and β-actin reverse: 5′-GCACAGGGTGCTCCTCAG-3′. Gene expression data were normalized to β-actin expression and quantified via the 2^−ΔΔCt method.

T cell depletion

and IFN-γ neutralization

T cell depletion and IFN-γ neutralization were performed as previously described²⁷. CD8+ or CD4 + T cell depletion was achieved through intravenous administration of 200 μg anti-CD8α (clone 53-6.7; BioXCell) or anti-CD4 (clone YTS 191; BioXCell) antibodies 4 days (day 38) and 2 days (day 40) before virus challenge and 3 days later (day 45). Control mice received anti-trinitrophenol isotype antibodies (clone 2A3; BioXCell). Depletion efficiency was verified by flow cytometry using FITC-conjugated anti-CD3ε, PerCP/Cyanine5.5-conjugated anti-CD4, and APC/Fire™ 750-conjugated anti-CD8α antibodies (BioLegend) on a CytoFLEX platform (Beckman Coulter), with data analyzed using CytExpert software. For IFN-γ neutralization, mice received 250 μg anti-IFN-γ (clone XMG1.2; BioXCell) intravenously at 4 h pre-challenge and 3 days post-challenge (day 45).

Preparation of bone marrow-derived dendritic cells (BMDCs) and bone marrow-derived macrophages (BMDMs)

Primary cultures of BMDMs and BMDCs were prepared as previously described^63,64. Briefly, bone marrow cells were harvested from the femurs and tibiae of mice. Erythrocytes were removed from cells by using a hypotonic solution. Cells were cultured for 7 days in DMEM containing 10% fetal bovine serum, 1% penicillin-streptomycin, and 40 ng/mL M-CSF. Following removal of nonadherent cells through medium replacement, cultures were maintained with passaging twice weekly. Flow cytometric analysis confirmed differentiation into mature macrophages, with >85% of cells co-expressing CD11b and F4/80 surface markers.

Bone marrow cells isolated from femurs and tibiae were cultured in RPMI 1640 medium supplemented with 10% FBS and 20 ng/mL recombinant murine GM-CSF. Cells were seeded at 0.5 × 10⁶ cells/well in 12-well plates, with 50% medium replacement using fresh GM-CSF-containing medium on days 3 and 6. On day 8, non-adherent cells were collected and characterized by flow cytometry, demonstrating >80% purity for CD11c⁺ dendritic cells.

RNA sequencing

Total RNA was extracted from tissue or cell samples with the Universal RNA Extraction CZ Kit (RNC643, ONREW) according to the manufacturer’s instructions. The RNA quantity was analyzed via Qubit 4.0 (Invitrogen), and the quality was examined via electrophoresis on a denaturing agarose gel. The RNA libraries were prepared via the VVAHTS^® Universal V8 RNA-seq Library Prep Kit for Illumina (NR605-0, Vazyme), followed by sequencing via the Illumina NovaSeq 6000 platform via the 150 paired-end sequencing strategy. mRNA enrichment, library construction, sequencing and data analysis were performed by Shanghai Xu Ran Biotechnology Co., Ltd. (http://www.xurangene.com).

The raw data were processed via Skewer v0.2.2 (https://sourceforge.net/projects/skewer/files/?source=navbar), and the data quality was checked via FastQC (v0.11.2) (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The read length was 2 × 150 bp. Clean reads were aligned to the mouse genome (mm10) from Ensembl via STAR (https://github.com/alexdobin/STAR), with one mismatch allowed. StringTie (v1.3.1c) was used to generate gene expression data, and differential gene expression was analyzed via DESeq2 (v1.16.1) (https://bioconductor.org/packages/release/bioc/html/DESeq2.html). The thresholds for determining DEGs were P < 0.05 and absolute fold change ≥ 2. The DEGs were subsequently chosen for function and signaling pathway enrichment analysis via the TopGO (https://www.bioconductor.org/packages/release/bioc/html/topGO.html) and KEGG (https://www.genome.jp/kegg/pathway.html) databases. The significantly enriched pathways were determined when P < 0.05. KEGG (Kyoto Encyclopedia of Genes and Genomes) is a widely used bioinformatics database that integrates genomic, biochemical, and systemic functional information, particularly for pathway analysis and molecular interaction networks.

Statistical analysis

Statistical analysis was performed via GraphPad Prism 7 software. The significance of differences between the two experimental groups was calculated via the two-tailed Student’s t-test or the Mann-Whitney test. Multiple experimental groups were analyzed via one-way ANOVA or the Kruskal-Wallis test, followed by Tukey’s or Dunn’s multiple comparison test. Significance values are indicated as follows: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; ns, not significant. P < 0.05 was considered significant.

Author contributions

P.Z., W.W., Z.Q. and Z.P. designed and supervised this study. Z.P., X.Z. and L.J. developed the methodology. Z.P., X.Z., L.J., C.D., Y.L., H.P., Y.L., Y.H., W.T., C.Z., D.W. performed the experiment. P.Z., Z.P., X.Z. and J.X. processed the data. Z.P., P.Z. and W.W. wrote the original draft. All the authors read and approved the manuscript.

Data availability

The datasets generated and analyzed in this study are presented in the manuscript and the supplementary files. Original raw data are also available from the corresponding authors upon request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41541-026-01383-2.