Induction of lung mucosal immunity by a next-generation inhaled aerosol COVID-19 vaccine: an open-label, multi-arm phase 1 clinical trial

Abstract

The current COVID-19 vaccines are suboptimal against the evolving SARS-CoV-2 variants, particularly in high-risk populations. A next-generation vaccine strategy capable of effective induction of respiratory mucosal immunity remains to be clinically developed. Here, we report an open-label, multi-arm phase 1 study (NCT05094609) to evaluate a multi-antigenic COVID-19 vaccine delivered once via inhaled aerosol to the lung of intramuscular mRNA-vaccinated humans without or with prior SARS-CoV-2 infection (uninfected vs infected). Escalating doses of a human adenoviral (HuAd)-vectored or chimpanzee Ad (ChAd)-vectored vaccine are evaluated in the uninfected cohort. A selected Ad vaccine is further evaluated in the infected cohort. The safety is assessed as a primary outcome. Ag-specific immune responses (secondary outcome) are assessed in peripheral blood and in respiratory tract via bronchoscopy at baseline and at timepoint(s) post-vaccination. Eighteen-65-year-old, healthy participants who have received at least 3 doses of mRNA COVID-19 vaccine are enrolled with those vaccinated with any Ad-vectored COVID-19 vaccine excluded. At baseline, there is minimally detectable mucosal immunity in the lung of uninfected or infected humans. While all tested doses (1 × 10⁵ to 1 × 10⁸ TCID₅₀) of HuAd and ChAd vaccines are safe, ChAd vaccine markedly outperforms the HuAd counterpart in immunogenicity. Thus, an optimal aerosol dose of ChAd vaccine induces the tripartite respiratory mucosal immunity consisting of T cell, trained innate and antibody immunity. Our study thus presents a promising next-generation aerosol COVID-19 vaccine strategy for further clinical development.

Fig. 1. Study design, solicited adverse events and respiratory function following inhaled aerosol vaccination.

a Schematic of clinical trial design indicating the number of participants, the time intervals, samples obtained, and major immune analysis performed at each visit. b Solicited systemic (top) and respiratory (bottom) adverse events that occurred within 72 h of inhaled aerosol vaccination in mRNA vaccinated, uninfected participants administered 3 × 10⁷ TCID₅₀ HuAd- or ChAd-vectored SARS-CoV-2 vaccines, and in mRNA vaccinated previously SARS-CoV-2 infected participants administered 3 × 10⁷ or 6 × 10⁷ TCID₅₀ ChAd-vectored SARS-CoV-2 vaccine. c, d Both forced expiratory volume in 1 s (FEV₁) and forced vital capacity (FVC) expressed as % predicted were used to assess pulmonary function of individuals administered escalating aerosol doses of HuAd- or ChAd-vectored SARS-CoV-2 vaccine over 12 weeks. Three participants for each vaccine arm in the uninfected cohort and in the infected 3 × 10⁷ ChAd cohort. Six in the infected 6 × 10⁷ ChAd cohort. Mean of the data is denoted by the line. Source data are provided as a Source Data file. In this figure (a) created in BioRender. Kang, A. (2025) https://BioRender.com/bt5wx05.

Fig. 2. Comparison of antigen-specific T cell responses induced in the airways following inhaled aerosol delivery of HuAd and ChAd vaccines in mRNA-vaccinated individuals without prior SARS-CoV-2 infection.

a BAL cells were stimulated with S1, NC or POL peptide pools. Three participants per dose/vaccine type. Scatter plots depict the fold-change (FC) in frequencies of IFN-γ + CD8 or CD4 T cells for each peptide pool at 4 weeks post-vaccination compared to baseline levels. b, Representative dotplots compare the frequencies (IFN-γ+ out of total CD8 T cells) of airway S1-, and NC -specific CD8 + IFN-γ + responses from participants vaccinated with 3 × 10⁷ TCID₅₀ HuAd or ChAd. The data are representative of three participants per vaccine type. c, d Scatter plots compare the frequencies of IFN-γ + CD8 or CD4 T cells specific for S1/NC, respectively. e Representative dotplots depict the multifunctional profile of S1-specific CD8 T cells. Pie charts compare mean frequencies of S1-specific CD8 T cells for 3+, 2+, and 1+ cytokines, identified by Boolean operator between HuAd and ChAd. The scatter plot compares the frequencies of multifunctional CD8 + T cells. f Representative dotplots show total CD8 T cells gated for CD103 and CD69 expression (gray dots), overlaid with IFN-γ + CD8 T cells specific for all three vaccine-encoded antigens (mega pool (MP)) (red dots) in airways of HuAd- or ChAd-vaccinated individuals. Pie charts compare mean frequencies of CD8 T cell expressing tissue resident markers using Boolean operator, between groups. Scatter plot show frequencies of antigen-specific resident memory and non-resident memory CD8 T cells. Each data point in all panels represents one participant. Three participants per dose/per vaccine type. All scatter plots show the median (horizontal line). Zero values were handled by adding a small constant (0.01) to all data points in all panels, expect for (e), where 0.001 added before transformation, as indicated by dotted lines in the scatter plots. The statistical differences between groups were determined using ordinary two-way ANOVA with Šidák multiple comparisons test on Log-transformed data for (a, c, d). Two-way ANOVA with Fisher’s LSD test on Log-transformed data was used to analyse statistical difference between groups for (e). Significant differences (p < 0.05) and p value < 0.1 are shown. Source data are provided as a Source Data file.

Fig. 3. Minimal mucosal T cell immunity in the lung of mRNA-vaccinated individuals without or with prior SARS-CoV-2 infection and its effective induction by inhaled aerosol ChAd vaccination.

a Dotplots depict CD8 + IFN-γ + responses (numerical values in the plots indicate the frequency of IFN-γ + out of total CD8 T cells) by bronchoalveolar-derived cells from mRNA-vaccinated SARS-CoV-2-uninfected (n = 3) and infected (n = 3) individuals, at baseline (0 wk) and 4 wk post-vaccination. b–d Scatter plots show frequencies of IFN-γ producing CD8 and CD4 T cells for each vaccine-encoded antigen (S1, NC or POL) in bronchoalveolar-derived cells from uninfected and infected individuals at baseline (0 wk) and 4 wk post-vaccination. e Scatter plot compares frequency of T_RM cells among total CD8 T cells boosted post-vaccination in uninfected and infected individuals. f Histograms depict proliferation assay data from individual mRNA-vaccinated, previously SARS-CoV-2-infected participants. Bronchoalveolar-derived cells were labelled with CellTrace Violet and stimulated with S1 peptide pool or left unstimulated for 6 days. The percentage in the scatter plot indicate the proliferating CD8 T cells (CellTrace low cells) in the gated total CD8 T cells. Each data point in all panels represents one participant. Three participants per with or without prior SARS-CoV-2 infection cohorts. All of the scatter plots show the median (horizontal line). Zero values were handled by adding a small constant (0.01) to all data points before transformation, as indicated by dotted lines in the scatter plots. The statistical differences between groups were determined using ordinary two-way ANOVA with Šidák multiple comparisons test on Log-transformed data for (b–d). Ordinary one-way ANOVA corrected for multiple comparison with two-stage step-up method of Benjamini, Krieger, and Yekutiieli was used to analyse statistical difference in proliferation capacity between unstimulated and S1 peptide pool stimulated cells for f. Significant differences (p < 0.05) are shown. Source data are provided as a Source Data file.

Fig. 4. High-dose (HD) aerosol ChAd vaccine induces greater magnitude and broader antigen-specific T cell responses over the LD in the airways of mRNA-vaccinated individuals with prior SARS-CoV-2 infection.

a, b Total cell counts and differential cell type frequencies in BALF from low dose (LD) and high dose (HD) ChAd vaccine cohorts. c, d Antigen-specific CD8 and CD4 responses assessed after stimulation with S1, NC, or POL peptide pools. Scatter plots compare total numbers of cytokine + (calculated using Boolean gating, which sums non-overlapping single and multiple cytokine producers) CD8 (c) and CD4 (d) T cells. e Pie charts depict mean frequencies of S1- and NC-specific CD8 and CD4 T cells for 3+, 2+, and 1+ cytokines, out of total CD8 and CD4 T cells, identified by the Boolean operator on single IFN-γ + /TNF + /IL-2 + expressing cells at 4 wk post-vaccination in the HD cohort. f Dot plots show CD8 + CD69 + T cells gated for CD103 and IFN-γ after mega pool stimulation. Scatter plot compares frequencies of tissue-resident memory CD8 T cells (CD69 + CD103 + IFN-γ +) among total IFN-γ + CD8 + T cells across cohorts. g t-SNE plots show airway VLA-4 + (CD29 + CD49d +) CD8 T cell distribution at baseline (0 wk) and 4 wk post-vaccination in the HD cohort. Dimensionality reduction performed in FlowJo on live BAL cells stimulated with the mega pool. Colours indicate T cell subset: Green: (CD8 +), orange (CD8 + VLA-4 + IFN-γ −), red: (CD8 + VLA-4 + IFN-γ +). Scatter plot compares frequencies of both antigens-specific and non-specific CD8 + VLA-4 + T cells. h Cross-reactivity of HD ChAd-induced S1-specific CD8 T cells with Omicron XBB.1.5 assessed by peptide stimulation. Dot plots and pie charts show multifunctional profiles; scatter plot compares ancestral vs XBB.1.5-specific responses. Each data point in all panels represents one participant. Three participants for LD and six for HD. Lung-homing molecules were assessed in five participants in the HD cohort, due to cell limitations. XBB.1.5 cross-reactivity was evaluated in three participants in HD cohort. All scatter plots show the median (horizontal line). The statistical differences between groups were determined using ordinary two-way ANOVA with Šidák multiple comparisons test for a and on Log-transformed data for (c, d, e, and h). Two-tailed Mann–Whitney test was used to analyze statistical difference between vaccine doses for (f) and timepoints for (g). Significant differences (p < 0.05) are shown. Source data are provided as a Source Data file.

Fig. 5. High-dose (HD) aerosol ChAd vaccine boosts mucosal antibody responses and induces trained innate immunity in alveolar macrophages in the airways of mRNA-vaccinated individuals with prior SARS-CoV-2 infection.

a, b Antigen-specific airway antibody responses to HD ChAd vaccine. Concentrated BALF collected at baseline (0 wk) and at 4 wk post-vaccination were subject to ELISAs using ancestral spike and RBD to generate reciprocal dilution curves. Area under curve (AUC) values for spike- and RBD-specific IgG compared between baseline and 4 wk. c, d AUC values for IgA specific to both spike and RBD are compared in concentrated BALF between baseline (0 wk) and 4 wk post-vaccination using ELISA. e Antibody-mediated neutralization assessed by a Surrogate Virus Neutralization Test (sVNT) against the ancestral RBD (left) and Omicron XBB.1.5 RBD (right). f Principal component analysis (PCA) of gene expression in alveolar macrophages (AM) from three individuals before (0 wk) and 4 wk after HD ChAd vaccination cultured with (WCL) or without (US) stimulation. Samples from the same individual are connected by colored lines. PCA visualization performed using rgl package in R. g Significantly enriched top 10 canonical pathways associated with uniquely upregulated and downregulated DEGs in WCL-stimulated 4 wk aerosol AM (Group 4/Group 2) when compared with WCL-stimulated baseline 0 wk AM (Group 3/Group 1). Horizontal bars: different pathways based on Z-scores. Red color: upregulated pathways; blue color indicates downregulated ones. Differential expression analysis performed using limma package and from each comparison significant genes (adjusted p value < 0.05) showing absolute regulation of at least 1.5-fold were selected. Obtained gene lists analyzed using STRING database to obtain lists of significantly over-represented pathways. h Heatmap shows differential-expressed genes associated with 3 selected pathways among top 10 pathways significantly enriched in WCL-stimulated 4 wk aerosol AM (Group 4/Group 2) over WCL-stimulated baseline 0 wk AM (Group 3/Group 1). i Heatmap shows 10 highly upregulated and downregulated genes unique to 4 wk WCL-stimulated inhaled aerosol vaccinated AM (Group 4)/Group 2 compared to WCL-stimulated baseline 0 wk AM (Group 3/Group 1). Each data-point in panels represents one participant. The statistical differences between groups were determined on Log-transformed data using a paired one-tailed t-test (a–e). Each data point represents a single participant in (c–e). Data are presented as mean values +/− SEM by the shaded area (error bands) in reciprocal dilution graphs. Source data are provided as a Source Data file.

Fig. 6. Systemic T cell responses induced by high-dose inhaled aerosol ChAd vaccine and its correlation with those in the airways.

a, b Representative dotplots of IFN-γ + CD8 T cells following HD ChAd. Scatterplots show kinetics of total cytokine+CD8 T cells, calculated using Boolean gating, which sums non-overlapping single/multiple cytokine producers. c, d Representative dotplots of IFN-γ + CD4 + T cells. Scatterplots show kinetic frequencies of total cytokine + CD4 T cells, calculated using Boolean gating, which sums non-overlapping single/multiple cytokine producers. e Simple linear regression of fold-change in S1-specific airway CD8/CD4 T cells vs. corresponding cells in blood. Each datapoint represents one participant. f Representative dotplots show AIM + out of total non-naïve CD8 T cells in S1-peptide-pool-stimulated PBMCs at 0 and 8 wk. Scatterplot compares AIM+ (CD25 + CD107a + or CD25 + CD137 + ) at 0 and 8 wk (DMSO background subtracted). Each datapoint represents one participant for two CD8-T cell independent functions. g Frequencies of memory subsets in AIM + CD8 T cells at 8 wk. CM: central memory (CCR7 + CD27 + ), EM1: effector memory (CCR7 − CD27 +), EM2/3: recently-activated (CCR7 − T cells, defined by CCR7 and CD27 CD27 − and CCR7 + CD27 −). Each datapoint represents one participant for memory phenotype of two CD8-T cell independent functions. h Representative dotplots show frequencies of proliferating IFN-γ + CD8/CD4 T cell in mega-pool (MP)-stimulated PBMCs at 0 and 12 wk. Scatterplot compares frequencies of total cytokine + proliferating-CD8/CD4 T cells (DMSO background subtracted) (n = 3 participants/timepoint). i Representative dotplots show S1-specific CD8/CD4 T cells for VLA-4 (CD29/CD49d) expression. Red: IFN-γ + VLA-4 + CD8/CD4 T cells post-S1-stimulation at 0, 2 and 4 wk; Grey: total CD8/CD4 T cells. Scatterplot shows kinetics of VLA-4 + S1-specific CD8/CD4 T cells (n = 6 participants/timepoint). j Linear regression of fold-change in airway S1-specific VLA-4 + CD8/CD4 T cells vs. corresponding cells in blood. Each point represents an individual participant. Each data-point represents one participant. All timpoints except 24/48 weeks (one dropped out) had six participants. All scatterplots show the median (horizontal line). Differences between baseline and timepoints in (b, d, and I) were determined using one-way ANOVA with Benjamini and Hochberg corrected for multiple comparison on Log-transformed data. A small constant (0.0001) was added to zero values before transformation, shown as dotted lines. Two-tailed Wilcoxon test was used to analyze differences between timepoints in (f and h). Simple linear regression analysis was used to calculate r² for (e and j). Significant differences (p < 0.05) are shown. Source data are provided as a Source Data file.

Fig. 7. Conceptual illustration showing the minimal presence of effective mucosal immunity in the lung of intramuscular mRNA-vaccinated individuals without or with prior SARS-CoV-2 infection, and its effective induction by inhaled aerosol immunization with a next-generation COVID-19 vaccine.

Top: Intramuscular mRNA vaccination in uninfected and previously infected individuals generates strong systemic antibodies against spike protein and RBD but weak antigen-specific mucosal antibodies and T cells in the lung or lower respiratory tract. Bottom: Inhaled aerosol delivery of a next-generation multiantigenic ChAd-vectored COVID-19 vaccine deposits the vaccine particles of 2–5 µm deep into the major airways, inducing tripartite respiratory mucosal immunity consisting of persisting tissue-resident memory cytotoxic CD8 T cells, trained innate immunity in airway macrophages, and virus-neutralizing antibodies. Such all-around mucosal immunity in the lung offers optimal, durable protection against SARS-CoV-2 infection, particularly in high-risk human populations. Created in BioRender. Kang, A. (2025) https://BioRender.com/bt5wx05.