Induction of mucosal immune responses against SARS-CoV-2: a heterologous intramuscular mRNA-LNP prime/pulmonary recombinant subunit pull vaccination strategy

Abstract

Approved mRNA vaccines administered intramuscularly (i.m.) induce strong systemic immune responses, but provide limited protection at the respiratory mucosa, where many infections are initiated. Designing safe and efficacious mucosal vaccines is challenging because it requires vaccine administration at the mucosa that is equipped with protective barriers and characterized by tolerogenic predominance. Here we show that i.m. prime immunization of mice with lipid nanoparticles (LNPs) loaded with mRNA encoding the SARS-CoV-2 spike protein, followed by pulmonary pull immunization with either mRNA-LNPs or spike protein adjuvanted with cationic adjuvant formulation (CAF)01 induce high systemic immune responses and virus-neutralizing spike-specific antibody responses in the lungs. However, only pulmonary pull immunization with CAF01-adjuvanted spike protein induces spike-specific mucosal antibody and lung-resident T-cell responses in the respiratory tract. This suggests that i.m. priming of strong immune responses with mRNA-LNPs combined with mucosal pull immunization with a subunit vaccine can re-direct the immune response to the respiratory mucosa.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-026-60179-5.

Introduction

The development of lipid nanoparticle (LNP)-based messenger RNA (mRNA) vaccines for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) represents a major milestone in the field of mRNA therapeutics^1–3. Currently approved corona virus disease 2019 (COVID-19) vaccines, administered parenterally for both prime and booster doses, elicit robust systemic humoral and cellular immune responses, and are highly effective at preventing severe disease and death^1,4. However, breakthrough infections and ongoing viral transmission among fully vaccinated individuals highlight the inability of these vaccines to prevent virus transmission and induce airway mucosal immunity^5–10. This limitation has sparked significant interest in developing vaccines capable of inducing strong, cross-protective mucosal immunity, in addition to systemic immunity¹¹. While intramuscular (i.m.) vaccination is less effective at generating mucosal immune responses^8,12, both mucosal vaccination and natural infection have demonstrated the ability to induce robust mucosal immunity^13–16.

For respiratory viruses like SARS-CoV-2, inducing protective mucosal immune responses, i.e. secretory antibodies and tissue-resident T cells, is crucial to block initial infection and to prevent viral dissemination to the lower respiratory tract, which is required for controlling pandemic spread^15,17. Hybrid immunity, which is generated when vaccine-induced immunity is boosted by a subsequent infection, has been shown to provide enhanced protection against reinfection, e.g., with antigenically drifted variants^18,19. Hence, a systemic prime/mucosal pull vaccination regimen is promising by leveraging mucosal vaccination to mimic the effects of natural infection and bolster vaccine-induced immunity within the respiratory tract^13,14.

Real-world experience with the currently approved COVID-19 and respiratory syncytial virus mRNA vaccines has shown that vaccines based on mRNA-LNPs are effective in preventing severe disease caused by infectious respiratory pathogens^20–23. Importantly, mRNA vaccines are manufactured by in vitro transcription in an enzymatic and cell-free process, which allows for rapid and scalable production²⁴. In addition, multiple antigens can be encoded within a single mRNA vaccine that enables targeting of several viral variants with one formulation^25,26. LNPs used for mRNA delivery typically contain four components: an ionizable lipid, cholesterol, a helper phospholipid, and a polyethylene glycol (PEG)-conjugated lipid²⁷. The ionizable cationic lipid plays a crucial role in: (i) enabling payload encapsulation, (ii) interaction with cellular membranes, and (iii) endosomal escape, facilitating the uptake of mRNA-LNPs into cells and subsequent endosomal escape^28–30. Ionizable lipids are positively charged in acidic environments and neutral at physiological pH³⁰. During mRNA-LNP formation at acidic pH (~ 4), below the apparent pK_a of the ionizable lipid (~ 6.5), the ionizable lipid is positively charged, facilitating mRNA encapsulation via attractive electrostatic interactions with the negatively charged phosphate backbone. Cholesterol primarily governs the rigidity and fluidity of LNPs^31,32. Helper lipids, e.g., 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC) and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), influence the lipid packing of LNPs, thereby affecting their stability and transfection efficiency³³. The PEG-lipid primarily functions as a steric barrier to prevent LNP aggregation during manufacture and storage³⁴. Moreover, PEG has been shown to facilitate the permeation of the mucus barrier in the respiratory tract^35,36.

It is crucial to design an efficient immunization regime, optimize the administration route and dose, and account for dose-dependent toxicity, while investigating the design criteria of the delivery system to induce mucosal immunity. Currently, several studies have focused on various prime and pull immunization regimens, utilizing approaches such as polymer- and lipid-based delivery systems, or subunit-based prime/boost strategies to elicit mucosal immune responses^15,16,37. These approaches often involve parenteral immunization to elicit a systemic response, combined with mucosal immunization, which has been shown to enhance mucosal immunity and induce neutralizing antibodies^15,16.

This study aimed to assess the potential of mRNA-LNPs encoding for the spike protein of SARS-CoV-2 to induce mucosal immune responses in the respiratory tract. We used LNPs based on C12-200, which in the present study served as a model ionizable lipid⁴¹. For comparison, we also evaluated recombinant spike protein adjuvanted with cationic liposomal formulation (CAF)01, which is a liposomal adjuvant system composed of the cationic surfactant dimethyldioctadecylammonium (DDA) bromide and the Mincle receptor agonist trehalose 6,6-dibehenate (TDB)³⁸. First, we investigated local nasal and pulmonary administration of mRNA-LNPs to mice using a homologous prime-boost immunization scheme to evaluate mucosal immunity by assessing spike-specific T-cell and antibody responses in the respiratory mucosa. Second, we hypothesized that mucosal pull immunization is critical for inducing a mucosal immune response. Therefore, we employed a parenteral i.m. prime using spike mRNA-LNPs, followed by mucosal pull immunization using spike mRNA-LNPs or spike protein adjuvanted with CAF01 dosed by nasal and pulmonary administration, respectively.

Materials and methods

Materials

SARS-CoV-2 (E484K; N501Y) spike mRNA produced by in vitro transcription and fully substituted with N¹-methylpseudouridine (m1Ψ) to reduce innate immune responses (1.0 mg/mL in 1 mM sodium citrate, pH 6.4) was purchased from Oz Biosciences (catalogue number MRNA43-1000, Marseille, France). The integrity of the mRNA was verified by agarose gel mobility and fragment analyzer by the manufacturer. The mRNA is stabilized at the 5’ end by modified nucleotides capping (Cap1) and contains a poly(A) tail at the 3’ end. According to the manufacturer´s specifications, the content of double stranded RNA is < 0.5%. Recombinant SARS-CoV-2 prefusion-stabilized S ectodomain HexaPro trimer was produced by transient expression in FreeStyle human embryonic kidney 293-F cells and purified from the filtered supernatant using a HiTrap column (Cytiva) (90% purity by SDS-PAGE) as described previously^39,40. C12-200 was synthesized, purified, and characterized as reported previously^41,42. The lipids DOPE and 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (DMPE-PEG₂₀₀₀) were purchased from Avanti Polar Lipids (Alabaster, AL, USA). Cholesterol was obtained from Sigma-Aldrich (St. Louis, MO, USA). RNase-free water was used for procedures pertaining to mRNA-LNP preparation.

Preparation of mRNA-LNPs and physicochemical characterization

LNPs composed of C12-200 (35.0 mol%), DOPE (16.0 mol%), cholesterol (47.5 mol%), and DMPE-PEG₂₀₀₀ (1.5 mol%) loaded with mRNA were prepared by microfluidic mixing using the NanoAssemblr Ignite microfluidic mixer system equipped with NxGen cartridges (Precision Nanosystems, Vancouver, Canada/Cytiva, Marlborough, MA, USA), essentially as described previously⁴³. In brief, the aqueous phase consisted of mRNA dissolved in sterile-filtered citrate buffer (50 mM, pH 4). The organic solvent phase consisted of C12-200, DOPE, cholesterol, and DMPE-PEG₂₀₀₀ solubilized in absolute ethanol. The two phases were mixed at an aqueous:organic flow rate ratio of 3:1 using a total flow rate of 12 mL/min. The weight ratio of C12-200 to mRNA was 10:1. The mRNA-LNP dispersions were dialyzed for 3.5 h against RNase-free Dulbecco’s phosphate-buffered saline (D-PBS, pH of 7.4, Thermo Fisher Scientific) to exchange the solvent and neutralize the pH using Slide-A-Lyzer G3 dialysis cassettes with a molecular weight cutoff of 3.5 kDa (Thermo Fisher Scientific), and concentrated using an Amicon Ultra-4 centrifugal filter with a molecular weight cutoff of 10 kDa (EMD Millipore, Darmstadt, Germany) by centrifugation for 45 min at 2500 × g at 4 °C. The resulting formulations were characterized with respect to: (i) average intensity-based hydrodynamic diameter (z-average) and polydispersity index (PDI) by dynamic light scattering using the photon correlation spectroscopy technique, and (ii) mRNA encapsulation efficiency (EE%) and recovery using the Quant-iT RiboGreen reagent (Molecular Probes, Invitrogen, Paisley, UK), as previously described⁴³.

For the second immunogenicity study, the mRNA-LNP dispersions were frozen before testing in mice, as described previously⁴⁴. For freezing the formulations, the mRNA-LNP dispersions were prepared essentially as described above, but the mRNA-LNP dispersions were dialyzed and concentrated using 20 mM Tris-HCl buffer [containing 8.7% (w/v) sucrose, final concentration], pH 7.4. The nanoparticles were washed with at least four volumes of buffer during the concentration step, and the final volume of the mRNA-LNP dispersions was adjusted to 1200 µL. The tubes containing the nanoparticle dispersions were plunged into liquid nitrogen for 5 min. Following this, they were immediately transferred to -80 ◦C and stored until use. Freezing and thawing of mRNA-LNPs using this procedure only slightly reduces the mRNA encapsulation efficiency and does not affect the in vivo transfection efficiency⁴⁵.

Spike HexaPro trimer was adjuvanted with CAF01 as reported previously⁴⁶, and CAF01 was manufactured by Statens Serum Institut (SSI), Denmark. CAF01 (250 mg DDA/50 mg TDB) in 10 mM Tris buffer containing 2.2% glycerol (pH 7.0) was prepared as previously described⁴⁷. The final concentration of CAF01 was adjusted to 20/4 mg/mL DDA/TDB, corresponding to a DDA:TDB molar ratio of 89:11, as previously described⁴⁸. The spike HexaPro trimer (in 20 mM Tris buffer with 200 mM NaCl, pH 8.0) was mixed with CAF01 liposome dispersion (20/4 mg/mL of DDA/TDB) by vortexing for 5 s, and the final volume was adjusted to 50 µL with Tris buffer for pulmonary administration. The spike HexaPro trimer was then allowed to adsorb to the surface of CAF01 by incubation for 30 min at room temperature, and the lag time between the 30 min adsorption and administration was 5 min.

Immunization

Female C57BL/6 mice aged 7–9 weeks and CB6F1 mice aged 6 weeks were purchased from ENVIGO (Limburg, Netherlands) and Charles River Laboratories (Laval, Canada), respectively, and acclimatized for one week before the experiments. Food and water were supplied ad libitum. The experimental work of the first immunization experiment was approved by the Danish National Experiment Inspectorate under permit 2022-15-0201-01221 and performed in accordance with the directive 2010/63/EU of the European Parliament and of the Council on the protection of animals used for scientific purposes. The heterologous parenteral prime-mucosal pull immunization study was approved by the University Animal Care Committee and Animal Research Ethics Board at the University of Saskatchewan (Animal Use Protocol number 20220078) and was conducted in accordance with the Canadian Council of Animal Care guidelines. All procedures as well as the methods description in this article follow the ARRIVE Guidelines 2.0⁴⁹. Mice (3–7/group) were immunized once by i.m., nasal, or pulmonary homologous prime and boost immunization after an interval of 2 weeks, or by i.m. priming followed by pulmonary pull immunization after an interval of 3 weeks for the heterologous immunization regimen. For prime immunization i.m., a dose volume of 50 µL mRNA-LNP dispersion containing 5 µg spike mRNA was injected into the right thigh muscle of conscious mice (Supplementary Table S1). For nasal vaccination, mice were anesthetized with 2% isoflurane prior to nasal dispensation of 10 µg spike mRNA-loaded LNPs, and 10 µg spike HexaPro trimer surface-adsorbed to CAF01, respectively, in a total dose volume of 30 µL by pipetting 15 µL into each nostril (Supplementary Table S1). To investigate the effect of pull immunization on systemic and mucosal immune responses, 5 µg spike mRNA-loaded LNPs, 10 µg spike HexaPro trimer surface-adsorbed to CAF01, and unadjuvanted spike HexaPro trimer, respectively (Supplementary Table S1), were dosed via the pulmonary administration route in a dose volume of 50 µL, as described previously⁴³. All vaccines were adjusted to the correct dose volume with Tris buffer (20 mM, pH 7.4). Mice dosed by pulmonary administration with 50 µL isotonic Tris buffer served as negative controls. Prior to pulmonary administration by intratracheal instillation, the mice were anesthetized with a mixture of xylazine and ketamine (9 mg/kg and 100 mg/kg, respectively) by intraperitoneal injection. The mice were then placed in the supine position on a rodent tilting work stand at a 45° angle by the upper incisors, and the tongue was rolled out of the oral cavity. A laryngoscope fitted with 41 mm intubation specula was slowly advanced into the oral cavity, past the hard and soft palates. Once the tracheal opening was visible, the laryngoscope with the specula was taken out, the PennCentury MicroSprayer needle (Wyndmoor, Pennsylvania, USA) was inserted with the tip over the tracheal bifurcation, and the vaccine was sprayed into the lungs using the PennCentury MicroSprayer, as previously described⁴³.

Organ collection and single cell preparation

For the heterologous prime/boost immunization scheme, mice were dosed intravenously (i.v.) with 2.5 µg FITC-labeled anti-CD45.2 monoclonal antibody (10 µg/mL, BD Biosciences, New Jersey, USA) in 250 µL via the tail vein 3 min before euthanasia to distinguish T cells present in the vasculature from T cells present in tissue, as described previously⁵⁰. Five weeks after prime immunization, the mice were euthanized with an overdose of isoflurane followed by terminal cardiac puncture, and the lungs and spleen were harvested, as previously described⁵¹. Nasal washes were collected by flushing the nose of mice with 350 µL PBS containing 0.05% bovine serum albumin (BSA). Harvested organs were placed in RPMI 1640 (Merck, Darmstadt, Germany) at 4 °C.

Lungs were transferred to gentleMACS C tubes (Miltenyi Biotec Norden, Lund, Sweden) containing 2 mL RPMI 1640, 5% (v/v) fetal calf serum (FCS, Gibco Thermo Fisher, Hvidovre, Denmark), and 0.8 mg/mL collagenase type IV (Merck), and dissociated by using the gentleMACS dissociator (Miltenyi Biotec, Gladbach, Germany) for initially 7 s to dissociate the lungs. After termination of the first run, the C Tube was detached from the gentleMACS Dissociator, and the samples were incubated at 37 °C for 30 min. After incubation, lung pieces were dissociated again by using the gentleMACS dissociator for 37 s. After the second homogenization step, the samples were centrifuged at 300 × g for 7 min, and the supernatants of the lung samples were carefully collected for ELISA-based antibody detection. Lung cell pellets were transferred to a 70 μm cell-strainer (Falcon, Durham, NC, United States) and forced to obtain single cell suspensions, followed by two washing steps with medium including RPMI 1640 and 2% (v/v) FCS.

Single cell suspensions of splenocytes were obtained by homogenizing spleens through a 70 μm cell strainer and dispensed into 6-well plates. Cell suspensions were centrifuged at 300 × g for 3 min and washed twice with PBS. Spleen cells were incubated with 1 mL lysis buffer for 5 min at 4 °C to remove red blood cells between the two washing steps.

In accordance with the previously outlined protocol⁵¹, lung and spleen cell suspensions were adjusted to 1 × 10⁷ cells/mL with RPMI 1640 supplemented with 5 × 10^− 5 M 2-mercaptoethanol (Gibco Thermo Fisher), 1% (v/v) sodium pyruvate (Sigma-Aldrich), 1% (v/v) penicillin streptomycin (Gibco Thermo Fisher), 1% (v/v) HEPES (Gibco Thermo Fisher), and 10% (v/v) FCS (Gibco Thermo Fisher), and 100 µL cell suspension/well (adjusted to 1–2 × 10⁶ cells/well) was transferred to 96-well, U-bottomed plates.

Flow cytometry

Tetramer staining was performed to determine the percentages of spike-specific CD4⁺ and CD8⁺ T cells in the spleen and the lungs. Single-cell suspensions of spleen and lung tissues were washed with RPMI 1640 medium (for the lung samples) or PBS (for the spleen samples) and resuspended in FACS buffer [PBS supplemented with 1% (v/v) FCS and 0.1% (w/v) sodium azide]. A 1:1000 dilution of the fixable viability dyes FVS 510 or FVS 780 (BD Biosciences) was used to exclude dead cells. The Spike_539–546 (VNFNFNGL)-specific major histocompatibility complex (MHC) class I tetramer was obtained from the NIH tetramer core facility. The MHC class II spike tetramer Spike_62–76 (VTWFHAIHVSGTNGT) was custom made, using the corresponding peptide (JPT) and provided from the NIH tetramer core facility. For tetramer staining, the spike peptide-MHC tetramers were diluted in FACS buffer (VNFNFNGL, 1:150; VTWFHAIHVSGTNGT, 1:230) supplemented with F_c block (anti-CD16/32; BD Biosciences), and cells were incubated with the tetramers for 70 min at 25 °C in the dark. Following tetramer staining, the cells were washed with FACS buffer and then stained with 50 µL/well of a mixture of anti-CD44-PE-CF594 (eBioscience, Thermo Fisher Scientific, catalogue number 11-0441-85), anti-CD8-PerCp-Cy5.5 (eBioscience, Thermo Fisher Scientific, catalogue number 45-0081-82), anti-CD19-PE-Cy7, anti-CD4-APC-eFluor780 (eBioscience, Thermo Fisher Scientific, catalogue number 47-0042-82), and anti-CD62L-BV421) diluted 1:200 in FACS buffer in the dark for 30 min at 4 °C.

For the intracellular staining, lung and spleen cells were stimulated with recombinant spike protein minimal CD4 epitope (VTWFHAIHVSGTNGT, 1 µg/mL) and the spike protein minimal CD8 epitope (VNFNFNGL, 1 µg/mL), respectively, together with anti-CD28 (37.51, BD Biosciences) and anti-CD49d (9C10, BD Biosciences) (both at 1 µg /mL), at 37 °C overnight. As negative and positive controls, cRPMI + 10% FBS alone, and a mixture of PMA (0.05 µg/mL, Sigma-Aldrich) and ionomycin (1 µg/mL, Sigma-Aldrich), respectively, were used. Brefeldin A (10 µg/mL, Sigma-Aldrich) was added before an additional 5 h of incubation at 37 °C using a heat block, which was then programmed to cool down to 4 °C for the next day.

For detection of cytokine-producing CD8⁺ T cells, a protocol was used that combines surface CD4, CD8, and CD44 staining and intracellular cytokine staining. Following stimulation, cells were washed with PBS + 1% FBS and stained with 50 µL per well of a mixture of anti-CD44-CF594 (1:200, eBioscience, Thermo Fisher Scientific, catalogue number 11-0441-85), anti-CD8-PerCp-Cy5.5 (1:200, eBioscience, Thermo Fisher Scientific, catalogue number 45-0081-82), and anti-CD4-APC-eFluor780 (1:200, eBioscience, Thermo Fisher Scientific, catalogue number 47-0042-82) in the dark for 20 min at 4 °C. The cells were then washed twice, fixed, and permeabilized with 100 µL per well of Cytofix/Cytoperm (BD Biosciences) in the dark for 20 min at 4 °C. The cells were washed twice with Perm/wash buffer (BD Biosciences), and 50 µL per well of a mix of anti-IFN-γ-PE-Cy7 (1:200, eBioscience, Thermo Fisher Scientific, catalogue number 25-7311-82), anti-TNF-α-PE (1:200, BD Biosciences, catalogue number 554418), and anti-IL-2-APC (1:200, eBioscience, Thermo Fisher Scientific, catalogue number 17-7021-82) were added to the cells. The cells were incubated in the dark for 20 min at 4 °C.

Cells were washed twice, resuspended in FACS buffer, and analyzed using an LSR-Fortessa (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) and CytoFLEX flow cytometer (Beckman Coulter Inc., Brea, CA, USA). To determine background fluorescence, compensation controls were prepared. The gating for surface markers was based on fluorescence minus one controls (FMO) (Supplementary Table S2). The gating strategy (Supplementary Fig. S1 and S2) employed in the identification of distinct cell populations in the lungs and spleen is based on previous reports⁴⁸.

Antibody detection

Spike protein-specific IgG1, IgG2a, and IgA responses were measured in serum and lung homogenates by ELISA, as previously described⁴⁶. Briefly, Nunc Maxisorp plates (Merck) were coated with 0.5 µg/mL spike HexaPro trimer in carbonate-bicarbonate buffer (pH 9.6). Serum samples were serially diluted three-fold over 11 dilutions, starting from a 1:30 dilution in bicarbonate buffer. Lung homogenate samples were serially diluted three times from an initial 1:10 dilution in bicarbonate buffer. IgG1 and IgA specific for spike protein were detected with horse radish peroxidase (HRP)-conjugated secondary antibodies, i.e., goat anti-mouse IgG1 (Southern Biotech, 1:20000 dilution), and IgA (ThermoFisher, 1:5000 dilution). The substrate 3,3′,5,5′-tetramethylbenzidine Plus2 (Kem-En-Tec, Taastrup, Denmark) was utilized for the enzymatic reaction, which was terminated by addition of 0.2 M H₂SO₄, as previously described⁵². Subsequently, absorbance was measured at 450 nm using a FLUOstar Optima plate reader (BMG Labtech, Ortenberg, Germany). Non-linear regression analysis of the serum optical density values was performed to calculate ELISA titers, and the areas under the curve (AUC) were calculated and used for statistical analysis⁵³.

Plaque reduction neutralization (PRNT) assay

The virus neutralization titers in lung homogenates were measured as described previously⁵⁴. The SARS-CoV-2 original Wuhan and Beta variant (B.1.351, GISAID accession no. EPI_ISL_678615) were kindly provided by Professor Alex Sigal from the African Health Research Institute, South Africa. The virus was propagated in A549 cells expressing human ACE-2, kindly provided by Benjamin tenOever from New York University. Supernatants containing the virus progeny were harvested 72 h post-infection and centrifuged at 3000 × g for 10 min to remove cellular debris. The virus-containing supernatants were then filtered through a 0.45 μm filter and concentrated by centrifugal filtration at 4000 × g for 30 min using centrifugal filters (Millipore Sigma, Burlington, MA, USA). The virus titers were determined using the TCID₅₀ method (titer required to reduce 50% of the plaques) and calculated using the Reed-Muench method⁵⁵.

Statistical analysis

The results are presented as mean values ± standard deviation (SD). Graphing and statistical analyses were conducted using GraphPad Prism v9 (GraphPad Software Inc., La Jolla, CA, USA). Immune responses were compared using one-way ANOVA for cytokine-producing T cells and antibody responses, at a 0.05 significance level. Differences between groups were analyzed by one-way ANOVA, using mice primed only with spike mRNA-LNPs or naïve mice as reference, with Dunnett’s correction for multiple comparisons. The normality of the data in each group was evaluated using the Shapiro-Wilk test. All datasets were consistent with a normal distribution (P > 0.05), indicating that the assumptions for one-way ANOVA were met. For antibody responses, a data set was generated using serial dilutions of concentration. A curve was fitted to each replicate in log₁₀ scale using non-linear regression. The AUC) was then calculated for each replicate based on the fitted curve. Statistical analysis of AUC values was performed using one-way ANOVA with Dunnett’s correction for multiple comparisons at a 0.05 significance level, using mice primed only with spike mRNA-LNPs as reference. All comparisons with no significant difference are not indicated. The specific statistical tests and comparisons used for data analysis are stated in the figure captions.

Results

Physicochemical characterization of spike mRNA-loaded LNPs

The spike mRNA-loaded LNPs displayed an average hydrodynamic diameter of 132 ± 0 and an average PDI value of 0.07 ± 0.02 (mean value ± SD, n = 3). The mRNA encapsulation efficiency of the mRNA-LNPs was 95.7 ± 0.5% (mean value ± SD, n = 3), and the mRNA recovery was 96.8 ± 0.3% (mean value ± SD, n = 3), as reported previously⁴³.

Mucosal homologous prime-boost immunization in the airways with mRNA-LNPs does not induce spike-specific mucosal IgA responses

To investigate strategies for inducing mucosal immune responses using mRNA-LNP formulations, two distinct prime/boost immunization regimens were assessed. These regimens aimed to evaluate the most effective approach to activate mucosal immune responses, including the generation of SARS-CoV-2-specific IgA antibodies and tissue-resident T-cells in the respiratory tract. The first approach solely involved local dosing via pulmonary (Fig. 1A, group E) and nasal (Fig. 1A, group F) administration, respectively, of spike mRNA-loaded C12-200-based LNPs using the same administration route for both priming and boosting (Fig. 1A). This immunization regimen is based on our previous finding that pulmonary and nasal administration of LNPs loaded with mRNA encoding the model protein Firefly luciferase (FLuc) induces FLuc expression in lung tissue⁴³, and it was tested in a smaller pilot study (n = 3). Naïve mice (Fig. 1A, group A) and mice vaccinated with spike mRNA-LNPs by i.m. prime and boost immunization (Fig. 1A, group D) served as negative and positive control groups, respectively. Pulmonary (Fig. 1A, group B) and nasal (Fig. 1A, group C) vaccination with CAF01-adjuvanted spike HexaPro trimer were also included as positive controls in the study, because previous studies have shown that vaccines based on CAF01-adjuvanted protein antigens induce mucosal immune responses upon nasal and pulmonary administration^46,56. Hence, the study included in total six different experimental groups (group A-F, Fig. 1A).

Fig. 1.

Homologous prime and boost immunization of C57BL/6 mice. (A) Immunization scheme. Spike-specific IgG1 in serum (B, C) and spike-specific IgA in nasal washes (D) and serum (E). Optical density at 450 nm values (B, D) and area under the curve (AUC) (C, E) were determined 2 weeks after the boost immunization (n = 3 mice/group with data represented as mean values ± SD with individual replicates shown). Statistical significance relative to naïve mice was determined using one-way analysis of variance (ANOVA) and Dunnett’s post-test (P ≤ 0.05). Non-significant comparisons are not shown.

As expected, immunization with spike mRNA-loaded LNPs using the i.m./i.m. prime/boost regimen induced the highest spike protein-specific IgG1 responses in serum (Fig. 1B, C). In addition, pulmonary prime/boost vaccination with CAF01-adjuvanted spike HexaPro trimer elicited significantly higher IgG1 levels in serum compared to the levels in the serum of naïve mice (P = 0.0005) (Fig. 1C). Pulmonary administration of CAF01-adjuvanted spike HexaPro trimer and spike mRNA-LNPs resulted in higher spike-specific IgG1 levels in the serum compared to nasal administration of these formulations (P < 0.0001), suggesting that the pulmonary route of administration may be more effective than the nasal route of administration for induction of systemic antibody responses. In contrast, pulmonary and nasal prime/boost vaccination with spike mRNA-LNPs did not induce an IgA response in the nasal washes (Fig. 1D, E). Moreover, pulmonary and nasal administration of CAF01-adjuvanted spike HexaPro trimer did not induce an IgA response in nasal washes (Fig. 1D, E). The results of this homologous immunization regimen demonstrate that vaccination with spike mRNA-loaded C12-200-based LNPs using i.m./i.m. administration induces a circulating antibody response but does not induce IgA responses. In addition, homologous vaccination by local pulmonary and nasal administration does not induce nasal IgA responses.

Homologous parenteral (i.m.) prime and boost immunization with mRNA-LNPs induce CD4⁺ and CD8⁺ T-cell activation in the spleen

The induction of Th1-associated IgG subclasses and IgA antibody titers relies on efficient class switching within the germinal centers, which depends on induction of robust CD4⁺ T-cell responses⁵⁷. Additionally, CD4⁺ T cells play a vital role in supporting the induction of CD8⁺ T cell-mediated antiviral responses, which are crucial for protection against respiratory pathogens like SARS-CoV-2^58,59. Therefore, we further examined vaccine-induced CD4⁺ and CD8⁺ T-cell responses in the spleen and the lungs of mice vaccinated by homologous immunization. Homologous i.m. prime/boost vaccination with mRNA-LNPs elicited spike-specific CD8⁺ T-cell activation in the spleen, whereas no activation was observed upon nasal and pulmonary administration of mRNA-LNPs and CAF01-adjuvanted spike HexaPro trimer, respectively (Fig. 2A). These results demonstrate that a homologous prime/boost immunization regimen using either mRNA-LNPs or CAF01-adjuvanted spike HexaPro trimer, administered locally via pulmonary or nasal routes, does not contribute to the activation of circulating CD8⁺ T cells. However, the frequencies of CD4⁺ T-cell activation in the spleen were significantly higher in mice dosed with mRNA-LNPs (i.m.) and CAF01-adjuvanted spike HexaPro trimer (pulmonary) compared to naïve mice (P < 0.0001 and P < 0.0061, respectively) (Fig. 2B). This indicates that both i.m. immunization with mRNA-LNPs and pulmonary prime/boost vaccination with CAF01-adjuvanted spike HexaPro trimer can enhance circulating CD4⁺ T-cell responses. Moreover, the induction of CD4⁺ T cells in the spleen after pulmonary administration of CAF01-adjuvanted spike HexaPro trimer correlates with higher levels of IgG (Fig. 1B, C). This finding underscores the critical role of CD4⁺ T cells in facilitating class switching, ultimately leading to the production of IgG and IgA. CD8⁺ T cell activation in the spleen, producing IFN-γ (Fig. 2C) and TNF-α (Fig. 2D), was induced by parenteral prime-boost immunization with mRNA-LNPs (P < 0.0001 and P < 0.0012, respectively). Overall, these results demonstrate that C12-200-based LNPs loaded with spike mRNA administered i.m. effectively induces systemic T-cell and B-cell responses, resulting in high levels of CD8⁺ and CD4⁺ T-cell activation in the spleen, as well as strong IgG1 responses.

Fig. 2.

Spike-specific CD8⁺ and CD4⁺ T-cell levels in the spleen following homologous prime/boost immunization of C57BL/6 mice. Percentages (%) of spike tetramer (VNFNFNGL)-positive CD8⁺CD44⁺ T cells (A) and spike tetramer (VTWFHAIHVSGTNGT)-positive CD4⁺CD44⁺ T cells (B) in the spleen were quantified. Percentages of IFN-ɣ⁺ (C) and TNF-α⁺ (D) producing CD8⁺CD44⁺ T cells in the spleen. (n = 3 mice/group with data represented as mean values ± SD with individual replicates shown). Statistical significance relative to naïve mice was determined using one-was analysis of variance (ANOVA) and Dunnett’s post-test (P ≤ 0.05). Non-significant comparisons are not shown.

CD4⁺ and CD8⁺ T-cell activation in the lungs was also assessed following homologous prime and boost immunization (Supplementary Fig. S3). Parenteral (i.m.) prime/boost immunization with spike mRNA-LNPs resulted in induction of higher CD8⁺ T-cell activation in the lungs compared to local spike mRNA-LNPs vaccine administration (P < 0.0001). However, this result reflects the total frequency of both circulating and lung-resident spike-specific CD8⁺ T cells. To better evaluate mucosal immune responses, it is crucial to specifically examine tissue-resident T cells, as they are a more relevant marker. Therefore, it is essential to distinguish between circulating and resident T cells using markers like i.v. CD45 labeling before euthanizing the mice.

Parenteral prime-mucosal pull immunization with mRNA-LNPs induces IgG responses in the serum and lungs, but does not elicit IgA in the respiratory mucosa

Based on the results obtained from the first immunization scheme, the second immunization approach was evaluated in a larger study. This study was designed as a combined strategy of parenteral (i.m.) prime immunization followed by local pull immunization by pulmonary administration to investigate if the high immune response induced systemically by priming with spike mRNA-LNPs can be redirected to the respiratory mucosa by pulmonary pull immunization. Based on the above results, we chose the pulmonary administration route for the pull immunization. Mice were immunized i.m. with a prime dose of spike mRNA-loaded C12-200-LNPs (5 µg mRNA), followed by pulmonary pull immunization with different vaccines (Fig. 3A): (i) spike mRNA-loaded LNPs, (ii) recombinant spike HexaPro trimer alone (unadjuvanted), and (iii) CAF01-adjuvanted spike HexaPro trimer, respectively.

Fig. 3.

Homologous and heterologous prime and pull immunization of CB6F1 mice. (A) Immunization scheme. (B, C) Spike-specific IgG1 optical density at 450 nm values with corresponding area under the curve (AUC) values, respectively, in the serum and (D, E) lungs measured 3 weeks post-pull immunization. (F, G) Spike-specific IgG2a optical density at 450 nm values with corresponding AUC values in the lung supernatant, respectively, (H, I) and spike-specific IgA optical density at 450 nm and corresponding AUC values in the lungs measured 3 weeks post-pull immunization. Neutralizing activity of lung homogenates from immunized mice at week 6 after both prime/pull immunizations was measured using pseudoviruses for the SARS-CoV-2, (J) original Wuhan (Wuhan-Hu-1), and (K) South African beta (B.1.351) variants (n = 6–7 mice/group with data represented as mean values ± SD with individual replicates shown). Statistical significance relative to mice receiving an i.m. prime with spike mRNA-LNPs only was determined using one-way analysis of variance (ANOVA) and Dunnet’s post-test (P ≤ 0.05). Non-significant comparisons are not shown.

All vaccines induced spike-specific IgG1 responses in serum (Fig. 3B, C). Pulmonary pull immunization induced significantly higher spike-specific IgG1 titers than i.m. prime immunization with mRNA-LNPs alone (P < 0.0001; P = 0.0036), independently of the vaccine type used for the pull immunization. These results suggest that local pull immunization contributes to systemic IgG1 responses. The relatively low dose of spike mRNA used for pulmonary pull immunization (2.5 µg) is apparently sufficient to enhance the spike-specific IgG1 titers in the serum (Fig. 3B, C), compared to i.m. prime immunization alone. Pulmonary pull vaccination with CAF01-adjuvanted spike HexaPro trimer induced significantly higher spike-specific IgG1 titers than spike mRNA-loaded LNPs (P = 0.0087). Pull immunization with spike HexaPro trimer alone was as efficient as pull immunization with CAF01-adjuvanted spike HexaPro trimer (P = 0.6076), which suggests that CAF01 does not enhance the IgG1 response in the serum. Importantly, the spike-specific IgG1 titers induced in serum by the prime-boost immunization regimen (Fig. 1C) were lower compared to the titers induced by the prime-pull immunization regimen (Fig. 3C). However, the interval between the first and the second immunization was longer for the prime-pull regimen (21 days) than the prime-boost regimen (14 days), which may have provided more time for the development of a higher response.

Antibodies of the IgG class are abundantly present in blood, but are found in much lower concentrations in mucosal tissues, where they primarily originate from passive leakage from the blood circulation^60,61. The spike-specific IgG1 levels in the lungs were significantly higher upon vaccination compared to the IgG1 titers in naïve mice (P < 0.0001), even after one i.m. vaccination with the spike mRNA-loaded LNPs only (Fig. 3D, E). This suggests that IgG1 antibodies generated after the prime dosing circulate to the lung tissue. However, pulmonary pull immunization induced significantly higher IgG1 levels in the lungs, compared to i.m. prime immunization alone (P < 0.0001), demonstrating that local pull immunization enhances the IgG1 levels in the lung mucosa. Pull immunization with CAF01-adjuvanted spike HexaPro trimer resulted in the highest IgG1 levels in the lungs (Fig. 3D, E), which was consistent with the highest levels observed for serum IgG1 and lung IgA. Parenteral prime immunization alone did not induce IgG2a responses in the lungs. In contrast, mucosal pull immunization induced IgG2a responses in the lungs (Fig. 3F, G).

Pulmonary pull vaccination with spike mRNA-loaded LNPs did not elicit IgA responses in the lungs. Only CAF01-adjuvanted spike HexaPro trimer induced an IgA response in the lungs upon pulmonary pull immunization (Fig. 3H, I). When compared to the homologous immunization regimen, this finding highlights the importance of systemic prime dosing for induction of mucosal IgA responses.

Lung homogenates from vaccinated mice that received an i.m. prime with spike mRNA-LNPs, followed by pulmonary pull immunization with CAF-adjuvanted spike HexaPro trimer, showed significant differences in PRNT₅₀ neutralizing antibody titers against SARS-CoV-2 original Wuhan (Fig. 3J) and South African beta (Fig. 3K) variants ( P < 0.0001; P = 0.0007, respectively), compared to mice that received only the i.m. mRNA-LNPs prime. In contrast, no neutralizing antibody responses were detected in lung homogenates from mice that were vaccinated by pull immunization with unadjuvanted spike HexaPro trimer and spike mRNA-LNPs, respectively.

The weight of the animals was measured at three timepoints: (i) before the prime immunization, (ii) before the pull immunization, and (iii) just prior to euthanasia (Supplementary Fig. S4). Especially for mucosal administration of vaccines, it is crucial to monitor the health of the mice, because dose-related toxicity represents a risk. No weight loss was observed at the examined timepoints, and the survival rate for animals vaccinated using the heterologous immunization scheme was 100%. However, we cannot exclude that transient changes in body weight may have occurred more proximal to immunization.

Parenteral (i.m.) immunization with mRNA-LNPs combined with pulmonary pull vaccination with CAF-adjuvanted and unadjuvanted spike HexaPro trimer induces spike-specific CD8⁺ T-cell responses in the lungs

We further explored whether different mucosal pull immunization strategies can redirect the existing T-cell activation and antibody production to the respiratory mucosa upon a parenteral (i.m.) prime with mRNA-LNPs. To distinguish circulating T cells from tissue-resident T cells, mice were dosed with a FITC-labeled anti-CD45.2 monoclonal antibody administered i.v. before euthanasia to label the circulating T cells⁵⁰. Intramuscular priming with mRNA-LNPs and subsequent pull immunization by pulmonary administration of CAF-adjuvanted (P = 0.0441) and unadjuvanted (P = 0.0110) spike HexaPro trimer, respectively, induced significantly higher percentages of spike-specific, tissue-resident CD8⁺ T cells in the lungs, as compared to prime only vaccination (Fig. 4A). However, in the spleen, only the unadjuvanted spike HexaPro trimer (P = 0.0037) induced significantly higher percentages of spike-specific CD8⁺ T cells, as compared to prime only vaccination (Fig. 4B). Also, this vaccination regimen did not induce spike-specific CD4⁺ T cells, neither in the lungs nor in the spleen (Fig. 4C, D).

Fig. 4.

Spike-specific CD8⁺ and CD4⁺ T-cell levels in the lungs and the spleen following heterologous prime/boost immunization of CB6F1 mice. Frequencies (%) of spike tetramer (VNFNFNGL)-positive CD8⁺CD44⁺ T cells and spike tetramer (VTWFHAIHVSGTNGT)-positive CD4⁺CD44⁺ T cells in the lungs (A,C), and the spleen (B,D), respectively, were quantified (n = 6–7 mice/group, the bars represent mean values ± SD with individual replicates shown). Resident lung cells were distinguished from circulating T cells by using the CD45 FITC-A i.v marker. Statistical significance relative to mice receiving an i.m. prime with spike mRNA-LNPs only was determined using one-way analysis of variance (ANOVA) and Dunnet’s post-test (P ≤ 0.05). Non-significant comparisons are not shown.

Prime vaccination with spike mRNA-LNPs initiated priming events, but pulmonary pull immunization with different vaccine types (mRNA-LNPs, unadjuvanted spike HexaPro trimer, and CAF01-adjuvanted spike HexaPro trimer) did not induce distinct spike-specific cytokine profiles in the examined organs. There were no significant differences in the production of IFN-γ, TNF-α, or IL-2 between the groups (Supplementary Fig. S5-S6).

Discussion

During the SARS-CoV-2 pandemic, the ability of mRNA-LNP vaccines to induce systemic immunity upon parenteral administration has been extensively studied, whereas data on mucosal immune responses induced by mRNA-LNPs are scarce. In this study, we evaluated the ability of mRNA-LNPs dosed via local administration routes (nasal and pulmonary) to induce mucosal immune responses and explored how different immunization schemes may influence the induction of mucosal immune responses.

In the prime/boost immunization scheme, mice dosed i.m. with spike mRNA-LNPs elicited the highest systemic immune response, characterized by the highest IgG1 levels in serum (Fig. 1C), as reported in the literature^62–64. Vaccination by pulmonary administration of spike mRNA-LNPs and CAF01-adjuvanted spike HexaPro trimer, respectively, resulted in higher serum IgG1 levels in mice compared to vaccination by nasal administration (Fig. 1C). Our previous study of the protein expression kinetics after pulmonary and nasal administration of LNPs loaded with mRNA encoding FLuc supports these findings⁴³: Protein expression in the lungs of mice upon nasal administration decreases more rapidly compared to protein expression after pulmonary administration, even though the mRNA dose used for nasal administration (10 µg) was higher than the dose used for pulmonary administration (5 µg). These results suggest that pulmonary administration may be more efficient than nasal administration for antigen expression and subsequent induction of immune responses. This could be attributed to the different properties of the nasal and pulmonary barriers. Of note, the nasal epithelium in particular is equipped with rapid clearance mechanisms, and nasal administration also implies a risk of swallowing or sneezing out the vaccine. Although nasal administration has been extensively studied to assess mucosal immunity for distinct types of SARS-CoV-2 vaccines, particularly in preclinical animal models, it remains unclear whether pulmonary or nasal delivery is the more efficient route for inducing mucosal immunity. The adenovirus-based vaccine Convidecia Air (CanSino Biologics) has been approved in China for pulmonary administration^65,66, while another adenovirus-based SARS-CoV-2 vaccine (iNCOVACC, Bharat Biotech) has been approved in India for nasal administration⁶⁷. The observed differences between pulmonary and nasal administration may be explained by differences in the systemic absorption of the antigen. Both pulmonary and nasal vaccine delivery can induce local and systemic immune responses via absorption into the systemic circulation. However, pulmonary vaccination leads to higher systemic antigen absorption, compared to nasal delivery, because the lungs display a large and highly vascularized mucosal surface (surface area: 70–140 m²), compared to the smaller nasal mucosa surface (surface area approx. 180 cm²)⁶⁸.

In the present study, the serum IgG1 titers were slightly lower after parenteral prime-boost immunization (i.m./i.m.) in the first animal experiment (Fig. 1C), compared to the parenteral (i.m.) prime-mucosal (pulmonary) pull strategy in the second animal experiment (Fig. 3C). While higher IgG1 levels are typically expected after parenteral immunization, the time interval between the prime and the boost doses may explain the observed difference. A longer interval between the two immunizations may contribute to increased IgG1 titers. Similarly, it was reported that a longer interval between the prime (with spike mRNA-LNPs) and the boost (with spike protein via nasal administration) immunization resulted in elevated mucosal IgA and IgG levels in bronchoalveolar lavage fluid (BALF), as well as increased serum IgA and IgG titers, measured 56 days after the boost¹⁶.

The IgG1 levels in the lungs of mice following the i.m. prime and pulmonary mucosal pull immunization demonstrated that mucosal pull immunization enhances the IgG1 levels in the lungs (Fig. 3E). The response observed in mice that received only an i.m. prime dose confirmed that circulating IgG1 can transudate from the bloodstream to the lung mucosa (Fig. 3E). A recent study also demonstrated significantly increased BALF IgG levels in mice after a parenteral spike mRNA-LNP prime (i.m.) followed by a nasal spike protein boost vaccination. These levels were also higher compared to those observed with parenteral mRNA-LNP prime/boost immunization¹⁶. Another study demonstrated that mucosal immunization with an mRNA-poly(-amine-co-ester)-based vaccine via the nasal route (prime/boost) induced both serum and BALF IgG, but not IgA³⁷. However, it is important to emphasize that induction of mucosal sIgA is critical for protection, because IgA possesses a higher virus-neutralizing capacity than IgG⁶⁹.

In the present prime/boost immunization studies, none of the vaccines induced a significant spike-specific IgA response in the lungs (Fig. 1E). However, i.m. prime combined with pulmonary pull immunization with CAF01-adjuvanted spike HexaPro trimer induced IgA responses in the lungs (Fig. 3I). This finding highlights the importance of systemic prime dosing for inducing mucosal IgA responses. An appropriate local boost immunization can further direct the existing systemic immune response towards the local mucosal tissues, as shown in a recent study¹⁵, where mice were immunized with an i.m. prime using Comirnaty, followed by a heterologous nasal boost with a nanoemulsion system alone or combined with an RNA-based RIG-I agonist. The results demonstrate that i.m. priming with mRNA-LNPs, followed by heterologous nasal boosting with nanoemulsion/RNA-based RIG-I agonist-adjuvanted recombinant antigen induce robust mucosal and systemic antibody responses. This immunization approach also enhances antigen-specific T-cell responses in the mucosa-draining lymph nodes compared to i.m./i.m. and nasal/nasal prime-boost regimens. Specifically for mRNA-based vaccines, an additional mucosal booster dose of mRNA-LNPs, or extending the interval between immunizations, could be an effective strategy to improve mucosal immunity.

Parenteral administration of spike mRNA-LNPs using the i.m./i.m. prime-boost immunization strategy activated both CD8⁺ and CD4⁺ T cells in the spleen (Fig. 2A, B), which is consistent with findings reported in previous studies^70–73. Additionally, mRNA-LNPs administered via the parenteral route stimulated the production of IFN-γ and TNF-α in the spleen (Fig. 2C, D), which are critical cytokines for elimination of virus-infected cells⁷⁴. A recent immunogenicity study also demonstrated that high levels of IFN-γ were induced in the spleen following i.m. vaccination with Comirnaty, boosted either i.m. with an mRNA vaccine or nasally with nanoemulsion-adjuvanted spike protein¹⁵.

In this study, we show that pulmonary pull immunization with unadjuvanted and CAF01-adjuvanted spike HexaPro trimer after i.m. boost immunization induces spike-specific CD8⁺ T-cell responses in the lungs (Fig. 4A), whereas no CD4⁺ T-cell activation is observed (Fig. 4C). Another study also demonstrated that nasal boosting with stabilized unadjuvanted spike protein, following i.m. priming with mRNA-LNPs, induced activation of resident CD8⁺ and CD4⁺ T cells in the lungs¹⁶. However, there was a remarkable 100-fold increase in CD8⁺ T-cell activation compared to i.m. prime immunization alone, whereas CD4⁺ T-cell activation was less pronounced, exhibiting less than a 10-fold increase relative to the mice that received only an i.m. prime dose. In contrast, only pull immunization with unadjuvanted spike protein significantly increases the spike-specific CD8⁺ T-cell responses in the spleen (Fig. 4B), and no CD4⁺ T-cell activation is observed in the spleen (Fig. 4D). This may reflect differences in antigen distribution following pulmonary administration. Pulmonary administration of unadjuvanted antigen has been shown to result in rapid clearance (within 6 h) of antigen from lung tissue via the systemic circulation, whereas pulmonary administration of CAF01-adjuvanted antigen results in longer retention of the antigen in the lungs, thereby preferentially boosting local immune responses⁴⁸.

Only mice that received an i.m. prime with spike mRNA-LNPs followed by a pulmonary pull immunization with CAF-adjuvanted spike HexaPro trimer developed neutralizing antibody titers in the lungs against the original Wuhan SARS-CoV-2 strain (Fig. 3J) and the Beta variant (Fig. 3K). The lack of neutralizing activity in lung homogenates after homologous mRNA-LNP boosting may reflect limited antigen availability within the respiratory mucosa, which could reduce local germinal center and T follicular helper (Tfh) recall responses that are important for the induction of high-affinity neutralizing antibodies. A recent study reported that nasal administration of recombinant spike HexaPro protein formulated with novel cationic liposomes induced neutralizing antibodies against a Wuhan-Hu-1–like strain and protected K18-hACE2 mice from a homologous SARS-CoV-2 challenge, preventing both weight loss and viral replication in the upper and lower airways.⁷⁵. Furthermore, in the same study, vaccination of Syrian hamsters prevented weight loss and significantly reduced the viral load following challenge with the homologous strain as well as the Omicron BA.5 variant.

In the present study, pulmonary boosting following an i.m. prime dose of spike mRNA-LNPs did not induce distinct spike-specific cytokine profiles (IFN-γ, TNF-α, and IL-2) in the analyzed organs (Fig. S5-6). This may be attributed to the dose used in the study, which might have been insufficient to elicit detectable cytokine responses. A recent study highlighted that mice should be immunized with higher doses of mRNA and protein to better observe differences in cytokine-producing CD8⁺ and CD4⁺ T cells across vaccine regimens, while lower doses were used for antibody detection¹⁵.

We found that mucosal administration of spike mRNA-loaded LNPs did not result in the induction of spike-specific mucosal IgA (Fig. 3I) and T-cell responses (Fig. 4A, C). An explanation may be that the dose of spike mRNA is too low for sufficient spike protein expression in situ to effectively induce mucosal immune responses. The chemical instability of mRNA may lead to insufficient production of spike protein in lung tissue, limiting vaccine efficacy. Innate immune activation by mRNA may also result in rapid clearance. Chemical modification of mRNA might be required for administration specifically to lung tissue. A study using mRNA complexed with the polymer polyethyleneimine for pulmonary delivery revealed that substituting 25% of uridine and cytidine with 2-thiouridine and 5-methyl-cytidine significantly reduced mRNA binding to pattern-recognition receptors, minimized innate immune activation, and improved stability⁷⁶. This modification enabled sustained protein expression in more than 80% of human and mouse alveolar epithelial cells.

Dose-related material toxicity is a significant bottleneck for mucosal delivery. For LNPs consisting of an ionizable cationic lipid (proprietary to Acuitas Therapeutics), phosphatidylcholine, cholesterol, and PEG-lipid at a molar ratio of 50:10:38.5:1.5, high inflammatory responses were observed in the lungs following nasal administration⁷⁷. In dose-response experiments, 80% of the mice treated with 10 µg mRNA-LNPs died within 24 h. A dose of 5 µg resulted in 20% mortality within the same period, while all mice treated with a dose of 2.5 µg (50 µg lipid) survived without weight loss. The ionizable lipid C12-200 used in this study is a multi-tail ionizable lipid with a cone-shaped structure, which enhances endosomal disruption, compared to two-tail ionizable lipids such as DLin-MC3-DMA (MC3)⁷⁸. However, the use of multi-tail ionizable lipids in LNPs generally have been associated with toxicity and immune stimulation due to their stable backbones, which are not easily degradable. Although C12-200 has been characterized as pro-inflammatory and exhibiting low tolerability in vivo, it has been reported that modifying its stereochemistry can reduce these undesirable effects while also improving the mRNA delivery efficiency⁷⁹. In our previous study, we demonstrated that protein expression in the lungs of BALB/cOlaHsd mice, measured using an IVIS bioluminescence imaging system, was 1000 times higher than the background level 24 h after pulmonary administration of 5 µg mRNA⁴³. Based on these results, the same mRNA dose was used in the pilot immunogenicity study in which mice of the C57BL/6 strain were used. However, this dose reduced the survival rate of mice to 50%. The mRNA dose for pulmonary administration of spike mRNA-LNPs in the immunogenicity study was 2.5 µg. For mRNA-based therapeutics, a certain protein expression level at the target site is required to achieve therapeutic efficacy⁸⁰. The lack of T-cell responses in the lungs with the absence of antigen-specific IgA responses may be attributable to an insufficient dose to induce mucosal immunity, despite the observation of protein expression in the lungs at a higher dose. Designing a delivery system for mRNA that is both safe and highly tolerable for mucosal tissues is critical. Recently, various delivery systems for mRNA have been explored in clinical trials. For instance, a nanoemulsion system, currently in phase 1 trials (NCT04148118, NCT01354379), has demonstrated a favorable safety profile for nasal administration⁸¹. Polymeric nanoparticles based on biodegradable polymers like P76 (Tether Therapeutics) have proven to be highly effective for delivering various mRNAs to the lungs for vaccination against respiratory infectious diseases. These nanoparticles have also shown good tolerability with low toxicity in a variety of preclinical models from small rodents to larger mammals, including non-human primates (NHPs)⁸². Ethris has developed the Stabilized NanoParticle (SNaP) lipid nanoparticle delivery system, which is currently assessed in Phase I clinical trials. This system is designed to deliver mRNA encoding type III interferon via a nasal spray. Initial Phase I data indicate that the formulation is safe and well-tolerated⁸³.

The physiological and chemical barriers in the lungs, including the mucus layer, mucociliary clearance, pulmonary surfactant, and enzymatic degradation, can affect the uptake of nanoparticles⁸⁴. Rapid clearance may result in reduced uptake of mRNA-LNPs, thereby hindering their delivery to target cells and resulting in suboptimal mucosal immune responses. Incorporating mucoadhesive agents, e.g., cyclodextrin⁸⁵ and chitosan⁸⁶, could enhance mucosal mRNA delivery by prolonging nanoparticle residence time within the mucus layer, thereby improving mRNA uptake. The lower pH of normal airway surface liquid, compared to physiological pH, may also affect protein expression⁸⁷. A study evaluating mRNA-LNP delivery in lung fibrosis investigated mRNA release at both physiological and lower pH levels, finding that lower pH accelerates mRNA release⁸⁸. This rapid release could be problematic in the lungs; if mRNA is released before reaching the target cells, it may be degraded in the enzymatic lung environment, affecting mRNA bioavailability and reducing proteins translation. Hence, the ionizable lipid used for pulmonary delivery should be designed specifically with this slightly more acidic environment in mind, in particular with respect to the pK_a of the ionizable headgroup. C12-200 is expected to display a partially ionized state in the acidic lung tissue environment relative to muscle tissue. This ionization could lead to suboptimal transfection efficiency due to reduced lipid-membrane fusion or endosomal escape in the slightly acidic lung environment.

For effective mucosal vaccination, it is crucial to target mRNA to antigen-presenting cells (APCs), because they process and present antigens to other immune cells, thereby enhancing vaccine efficacy⁸⁹. However, there is a possibility that the mRNA-LNPs may be taken up by non-APCs at the target site. This off-target uptake could result in reduced efficacy. Therefore, investigating which cells in the lungs take up mRNA-LNPs and identifying the specific immune cell subsets responsible for inducing mucosal immune responses will provide valuable insights for optimizing mucosal mRNA vaccines. This aspect remains a key area that is not yet fully explored.

Conclusion

This study demonstrates that prime-boost immunization by mucosal administration of spike mRNA-loaded C12-200-based LNPs to the upper (via nasal administration) or lower (via pulmonary administration) respiratory tract alone is insufficient to induce spike-specific mucosal IgA responses. In contrast, prime-pull immunization by parenteral administration (i.m.) of spike mRNA-loaded LNPs effectively induces high levels of circulating spike-specific CD8⁺ T cells and IgG titers in serum. The heterologous immunization strategy including parenteral (i.m.) prime immunization with spike mRNA-loaded LNPs, combined with mucosal pull immunization by pulmonary administration of CAF01-adjuvanted spike HexaPro trimer, improves the spike-specific mucosal IgA response in the lungs, whereas homologous pulmonary pull immunization with spike mRNA-loaded LNPs does neither induce spike-specific mucosal IgA nor resident T-cell responses in the lungs.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

MO, STS, AGL, AT, SRP and CF conceived and designed the experiments. MO, NSK, DP, TCR, AGL, HF, SRP, and AT performed the experiments. MO, STS, and AT analyzed the data. AT and CF aquired funding. STS, AGL, TR, AT, and CF supervised the project. MO and CF wrote the manuscript. All authors read and approved the final manuscript.

Funding

We acknowledge the support from the Study Abroad Program of the Ministry of National Education, The Republic of Türkiye (Grant EABGA6ATKILK4Y8 to MO), the Independent Research Fund Denmark (Grant DFF-9041-00198B to CF), the Novo Nordisk Foundation (NNF22OC0076808 and NNF22SA0081890 to CF), and the Canadian Institutes of Health Research (CIHR) under the Project Grant - Priority Announcement: Pandemic Preparedness and Health Emergencies Research (PPE-196071 to AT). VIDO receives operational funding from the Government of Saskatchewan through Innovation Saskatchewan and the Ministry of Agriculture and from the Canada Foundation for Innovation through the Major Science Initiatives Fund.

Data availability

Data will be made available by the corresponding authors (C.F. and A.T.) on request.

Declarations

Competing interests

STS is employed by Statens Serum Institut, a non-profit government research facility. CF consults for Lonza and Sanofi. The other authors have no financial conflicts of interest.