Single-cell landscape of bronchoalveolar immune cells in patients with immune checkpoint inhibitor-related pneumonitis

Abstract

The pathophysiology of immune checkpoint inhibitor-related pneumonitis remains incompletely understood. We conducted single-cell and T-cell receptor transcriptomic sequencing on the bronchoalveolar lavage fluid from five patients with grade ≥2 immune checkpoint inhibitor-related pneumonitis. Our analyses revealed a prominent enrichment of T cells in the bronchoalveolar lavage fluid of patients with immune checkpoint inhibitor-related pneumonitis. Within the CD4 + T cell subset, Tfh-like T cells were highly enriched and exhibited signatures associated with inflammation and clonal expansion. Regulatory T cells were also enriched and displayed enhanced inhibitory functions. Within the CD8 + T-cell subset, effector memory/tissue-resident memory T cells with an elevated cytotoxic phenotype were highly infiltrated. Among myeloid cells, alveolar macrophages were depleted, while pro-inflammatory intermediate monocytes were elevated. Dendritic cells demonstrated enhanced antigen presentation capabilities. Cytokines CXCR4, CXCL13, TNF-α, IFN-α, IFN-γ, and TWEAK were elevated. Through a comprehensive single-cell analysis, we depicted the landscape of immune checkpoint inhibitor-related pneumonitis.

Fig. 1. The immune landscape of the BALF of patients with ICP.

A Diagram of the workflow for single-cell sequencing and UMAP embedding plot displaying immune cells generated from the BALF of patients with ICP, Covid-19 pneumonia, and healthy controls. B UMAP embedding plots of immune cells colored according to the expression of canonical marker genes. C, D UMAP embedding plots of immune cells grouped into 8 major types and 23 cell subtypes. E Bar plots indicating the differences in the main cell types among different patient groups. ICP, n = 5; Covid-19, n = 8; Healthy, n = 3. Statistical testing was performed by a Kruskal-Walis test. Data were presented as mean values (central line) +/- interquartile range (border). (Created with BioRender.com).

Fig. 2. Phenotypes, quantitative and functional analyses of CD4 + T cells.

A, B UMAP embedding plots of CD4 + T cells colored according to patient groups and cell labels. C Dot plots displaying canonical marker genes of CD4 + T cell subclusters. The sizes of the dots are proportional to the fraction of the cells, and the color represents the normalized gene expression level. D Bar plots indicating the differences in the cell subtypes within the total cell population and within the CD4 + T cell populations among different patient groups. ICP, n = 5; Covid-19, n = 8; Healthy, n = 3. Statistical testing was performed by a Kruskal-Walis test. Data were presented as mean values (central line) +/- interquartile range (border). E UMAP embedding plots displaying the RNA velocities of CD4 + T cells. Cells were colored according to their cell labels and the clonal expansion ability. F Histogram showing the relative abundance of CD4 + T cells with different clonal expansion ability. Cells were grouped according to patient groups and cell subtypes. G Box plots indicating clonal expansion (upper) and developmental transition (lower) of CD4 + T cells quantified by STARTRAC for each sample divided by cell subtypes or patient groups. Two-sided Wilcoxon tests were performed to make comparisons. The center lines correspond to the median values, and the boxes correspond to the interquartile range. H Heatmap of cytokine signaling activities in CD4 + T cells among patients with ICP, patients with Covid-19 pneumonia, and healthy controls.

Fig. 3. Phenotypes, quantitative and functional analyses of regulatory and conventional T cell.

A Volcano plots displaying the DEGs of Tregs between patients with ICP, patients with Covid-19 pneumonia, and healthy controls. Statistical tests were performed by a two-sided Wilcoxon test, with adjusted P value < 0.05 and log₂ (fold change) ≥ 0.5. B Lollipop plots showing the KEGG and GO biological process enrichment results for DEGs in Tregs. C GSEA analysis indicated that genes ranked in Tregs from patients with ICP and healthy controls were enriched in the pathway “cytokine receptor interaction”. Statistical analysis was performed by permutation test. D Heatmap of core enrichment genes in the pathway “cytokine-cytokine receptor interaction” from the GSEA analysis of Tregs among patients with ICP, patients with Covid-19 pneumonia, and healthy controls. E Volcano plots displaying the DEGs of conventional T cells between patients with ICP, patients with Covid-19 pneumonia, and healthy controls. Statistical tests were performed by a two-sided Wilcoxon test, with adjusted P value < 0.05 and log₂ (fold change) ≥ 0.5. F Lollipop plots showing the KEGG and GO biological process enrichment results for DEGs in conventional T cells. G GSEA analysis indicated that genes ranked in conventional T cells from patients with ICP and healthy controls were enriched in the pathway “leukocyte trans-endothelial migration”. Statistical analysis was performed by permutation test. H Heatmap of core enrichment genes in the pathway “leukocyte trans-endothelial migration” from the GSEA analysis of conventional T cells among patients with ICP, patients with Covid-19 pneumonia, and healthy controls.

Fig. 4. Phenotypes, quantitative and functional analyses of CD8 + T cells.

A, B UMAP embedding plots of CD8 + T cells colored according to patient groups and cell labels. C Bar plots indicating the differences in the cell subtypes within the total cell population and within the CD8 + T cell populations among different patient groups. ICP, n = 5; Covid-19, n = 8; Healthy, n = 3. Statistical testing was performed by a Kruskal-Walis test. Data were presented as mean values (central line) +/- interquartile range (border). D Histogram showing the relative abundance of CD8 + T cells with different clonal expansion ability. Cells were grouped according to patient groups. E Clonal expansion (left) and developmental transition (right) of CD8 + T cells quantified by STARTRAC for each sample divided by patient groups. Two-sided Wilcoxon tests were performed to make comparisons. The center lines correspond to the median values, and the boxes correspond to the interquartile range. F Volcano plots displaying the DEGs of CD8 + T cells between patients with ICP, patients with Covid-19 pneumonia, and healthy controls. Statistical tests were performed by a two-sided Wilcoxon test, with adjusted P value < 0.05 and log₂ (fold change) ≥ 0.5. G Lollipop plots showing the KEGG and GO biological process enrichment results for DEGs in CD8 + T cells. H GSEA analysis indicated that genes ranked in CD8 + T cells from patients with ICP and healthy controls were enriched in the pathway “regulation of leukocyte migration”. Statistical analysis was performed by permutation test. I Heatmap of core enrichment genes in the pathway “regulation of leukocyte migration” from the GSEA analysis of CD8 + T cells among patients with ICP, patients with Covid-19 pneumonia, and healthy controls. J GSEA analysis indicated that genes ranked in CD8 + T cells from patients with ICP and healthy controls were enriched in the pathway “response to ketone”. Statistical analysis was performed by permutation test. K Heatmap of core enrichment genes in the pathway “response to ketone” from the GSEA analysis of CD8 + T cells among patients with ICP, patients with Covid-19 pneumonia, and healthy controls. L Heatmap of cytokine signaling activities in CD8 + T cells among patients with ICP, patients with Covid-19 pneumonia, and healthy controls.

Fig. 5. Phenotypes, quantitative and functional analyses of myeloid cells.

A–C UMAP embedding plots of myeloid cells colored according to patient groups, main labels, and cell labels. D Dot plots displaying canonical marker genes of myeloid cell subclusters. The sizes of the dots are proportional to the fraction of the cells, and the color represents the normalized gene expression level. E Bar plots indicating the differences in the cell subtypes of myeloid cells within the total cell population among different patient groups. ICP, n = 5; Covid-19, n = 8; Healthy, n = 3. Statistical testing was performed by a Kruskal-Walis test. Data were presented as mean values (central line) +/- interquartile range (border). F Lollipop plots showing the GO biological process enrichment results for DEGs in macrophages. G Heatmap of cytokine signaling activities in myeloid cells among patients with ICP, patients with Covid-19 pneumonia, and healthy controls.

Fig. 6. Phenotypes, quantitative and functional analyses of DCs.

A Bar plots indicating the differences in the cell subtypes of DCs within the total cell population among different patient groups. ICP, n = 5; Covid-19, n = 8; Healthy, n = 3. Statistical testing was performed by a Kruskal-Walis test. Data were presented as mean values (central line) +/- interquartile range (border). B Heatmap of core enrichment genes in the pathway “autoimmune thyroid disease from the GSEA analysis of DC1 among patients with ICP, patients with Covid-19 pneumonia, and healthy controls. C Heatmap of core enrichment genes in the pathway “antigen processing and presentation” from the GSEA analysis of DC2 among patients with ICP, patients with Covid-19 pneumonia, and healthy controls. D Cell communication analysis suggested an augmented antigen presentation ability of DCs in patients with ICP. E Cell communication analysis suggested a strong signaling interaction between DCs and cytotoxic T cells or Tfh-like T cells.

Fig. 7. Distinct cytokine profile characterizes the BALF of ICP.

Cytokines CXCR4, CXCL13, TNF-α, IFN-α, IFN-γ, and TWEAK were significantly higher in the BALF of patients with ICP than in healthy controls. Statistical testing was performed by unpaired t-test. Data were presented as mean values (central line) +/- interquartile range (quartile line) in the violin plots. ^**P value < 0.01; ^****P value < 0.0001; COV, covid-19 pneumonia; HC, healthy controls.