High-dose intravenous BCG vaccination induces enhanced immune signaling in the airways

Abstract

Intradermal Bacillus Calmette-Guérin (BCG) is the most widely administered vaccine, but it does not sufficiently protect adults against pulmonary tuberculosis. Recent studies in nonhuman primates show that intravenous BCG administration offers superior protection against Mycobacterium tuberculosis (Mtb). We used single-cell analysis of bronchoalveolar lavage cells from rhesus macaques vaccinated via different routes and doses of BCG to identify alterations in the immune ecosystem in the airway following vaccination. Our findings reveal that high-dose intravenous BCG induces an influx of polyfunctional T cells and macrophages in the airways, with alveolar macrophages from high-dose intravenous BCG displaying a basal activation state in the absence of purified protein derivative stimulation, defined in part by interferon signaling. Enhanced intercellular immune signaling and stronger T helper 1-T helper 17 transcriptional responses were observed following purified protein derivative stimulation. These results suggest that high-dose intravenous BCG vaccination creates a specialized immune environment that primes airway cells for effective Mtb clearance.

Fig. 1.. Integrated census of BAL cells displays vaccine-dependent shifts in immune cell populations.

(A) Diagrammatic overview of study workflow. O/N, overnight. (B) Visual table of macaque samples describing the vaccine route (AE, aerosol; IV, intravenous; naive, not vaccinated), log₁₀(BCG vaccine dose), and vaccine group. (C) Uniform Manifold Approximation and Projection (UMAP) embedding of all cells, colored by cell type and stimulation condition. (D) Ratio of cell fractions (as a fraction of total cells) between AMs and other immune cell types: T cells and myeloid cells. Each dot indicates a batch. Mann-Whitney U tests were performed between groups on the ratios of cell proportions per group of cell types. P values were adjusted using the Benjamini-Hochberg procedure. Adjusted P values are denoted as follows: *P < 0.05; **P < 0.01. Only comparisons significant at P < 0.05 are shown.

Fig. 2.. Recruitment and activation of myeloid cells in high-dose intravenous BCG macaques.

(A) Scaled log-normalized expression of selected genes across annotated cell states. Circles within each tile of the heatmap represent the percentage of cells with the gene detected. The top three genes by AUROC for each cell type are selected, in addition to a panel of genes encoding protein targets typically used in flow cytometry panels (PTPRC to IL3RA). (B) Odds ratio for selected myeloid cell types comparing vaccine groups to intravenous high. Odds ratios are calculated from a binomial generalized linear model analyzing the number of cells for each cell type and sample and the total number of other cells within that sample as a function of cell type and vaccine group. (C) CyTOF analysis of AMs frequency following high-dose intravenous BCG vaccination. (D) Volcano plot of unstimulated AM gene expression data comparing intravenous high to naïve animals. Pseudobulk analysis was used, combining scRNA-seq data from all cells within a sample to generate a bulk-like expression profile. Left, early; right, late. (E) Enrichment of genes up-regulated in unstimulated AMs in high-dose intravenous BCG animals using the MSigDB database. Odds ratio and −log₁₀(adjusted P value) are plotted for each term significant at P < 0.05. P values were adjusted using the Benjamini-Hochberg procedure. In the volcano plot, gray circles are not significant. NF-κB, nuclear factor κB; JAK, Janus kinase; STAT3, signal transducers and activators of transcription 3; RM, recruited myeloid cells; TP10K, normalized transcript count; KRAS, Kirsten rat sarcoma virus.

Fig. 3.. PPD-responsive T cell–myeloid interactions.

(A) Interaction strength of ligand-receptor interactions between polyfunctional, activated, and helper CD4⁺ T cells and selected myeloid cells for unstimulated cells (left) and PPD-stimulated cells (right). (B) Expression score of top 20 ligand-receptors pairs across groups. Mann-Whitney U tests were performed between vaccine groups on the total expression score. P values were adjusted using the Benjamini-Hochberg procedure. Adjusted P values are denoted as follows: *P < 0.05; **P < 0.01; ****P < 0.0001. Only comparisons significant at P < 0.05 are shown. SEMA4D, Semaphorin-4D; PLXNB2, plexin-B2; PTPRC, Protein tyrosine phosphatase receptor type C; MRC1, mannose receptor C-type 1; JAG1, Jagged-1; ANXA, annexin A; FPR1, Formyl peptide receptor 1; ADGRE5, adhesion G protein-coupled receptor E5.

Fig. 4.. Polyfunctional T cell populations are most associated with protection.

(A) UMAP embedding of T and NK cells. (B) Scaled log-normalized expression of selected genes across annotated cell states. Circles within each tile of the heatmap represent the percentage of cells with the gene detected. (C) Cell proportions of total cells for selected cell states in both unstimulated and PPD-stimulated samples across vaccine groups. Mann-Whitney U tests were performed between groups on the cell proportions of total cells for each cell state. P values were adjusted using the Benjamini-Hochberg procedure. (D) Number of differentially expressed (DE) genes between unstimulated and PPD-stimulated samples within each T cell subset. (E) T1-T17 scores across T and NK cells within stimulated samples. Mann-Whitney U tests were performed between groups. (F) T1-T17 scores across T and NK cells within all samples. Cell states are ordered by mean score from high to low. Mann-Whitney U tests were performed. (G) Volcano plot of pseudobulk PPD-stimulated T cell gene expression data comparing high-dose intravenous BCG to naïve animals. Left, early time points; right, late time points. P values were adjusted using the Benjamini-Hochberg procedure. Adjusted P values are denoted as follows: *P < 0.05; **P < 0.01; ***P < 0.001. Only comparisons significant at P < 0.05 are shown. In volcano plots, gray circles are not significant. GZMK, granzyme K.

Fig. 5.. Responses to secondary antigen challenges are associated with protection.

(A) Analytical schematic depicting the identification of genes that are variably induced upon PPD stimulation based on group. (B) Heatmap of genes induced by PPD stimulation across vaccine groups in T cells. Left, early time points; right, late time points. Color represents the gene-scaled log₂(fold change) of each gene in each group. (C) Heatmap of genes induced by PPD stimulation across vaccine groups in AMs. Left, early time points; right, late time points. Color represents the gene-scaled log(fold change) of each gene in each group. (D) Enrichment of genes whose expression is induced by PPD in group-dependent manner using the MSigDB database. Odds ratio and −log₁₀(adjusted P value) are plotted for each term significant at P < 0.05.