Endothelial cell expression of a STING gain-of-function mutation initiates pulmonary lymphocytic infiltration

Abstract

Patients afflicted with Stimulator of interferon gene (STING) gain-of-function mutations frequently present with debilitating interstitial lung disease (ILD) that is recapitulated in mice expressing the STING^V154M mutation (VM). Prior radiation chimera studies revealed an unexpected and critical role for non-hematopoietic cells in initiating ILD. To identify STING-expressing non-hematopoietic cell types required for the development of ILD, we use a conditional knockin (CKI) model and direct expression of the VM allele to hematopoietic cells, fibroblasts, epithelial cells, or endothelial cells. Only endothelial cell-targeted VM expression results in enhanced recruitment of immune cells to the lung associated with elevated chemokine expression and the formation of bronchus-associated lymphoid tissue, as seen in the parental VM strain. These findings reveal the importance of endothelial cells as instigators of STING-driven lung disease and suggest that therapeutic targeting of STING inhibitors to endothelial cells could potentially mitigate inflammation in the lungs of STING-associated vasculopathy with onset in infancy (SAVI) patients or patients afflicted with other ILD-related disorders.

Keywords: CP: Immunology; SAVI; STING; autoinflammation; bronchus-associated lymphoid tissue; conditional knockin; endothelial cells; innate immunity; interstitial lung disease; stromal cell immunity.

Figure 1.. A mouse model for Cre-recombinase-dependent STING V154M expression

(A) Diagram of the STING V154M conditional knockin (CKI). (B) Tail DNA from a STING^CKI/WT mouse and a STING^CKI/WT × CMV-Cre mouse was PCR amplified using primers indicated in (A). STING WT allele gives a 596-bp fragment, STING CKI allele gives a 774-bp fragment, and upon deletion of the gene trap from the CKI allele a 636-bp fragment is generated. (C) STING expression by CD45⁺ immune cells from the blood of mice inheriting the indicated STING alleles as assessed by flow cytometry. (D–H) Eight-week-old age-, sex-, and littermate-matched CKI (n = 13–23, white) and CKI × CMV-Cre mice (n = 16–29, red) and 12-week-old age-, sex-, and littermate-matched WT (n = 6, gray) and VM mice (n = 10, pink) were evaluated by the following measures. Data shown represent at least two independent experiments. Bar graphs represent mean ± SD. (D) Representative 43 field H&E-stained lungs. Images are representative of at least two independent experiments. (E) Immunofluorescence staining for DAPI (gray), CD3 (cyan), LYVE1 (yellow), and B220 (magenta) on CKI × CMV-Cre mouse lung. (D and E) Scale vars, 200 μm. Images are representative of at least two independent experiments. (F) Percentage of EV immune cells among live CD45⁺ lung cells, and total counts of CD45⁺ lung EV cells. (G) Percentage of CD69⁺ EV CD3⁺ T cells in the lung. (H) Body weight from mice, normalized as the fold change relative to the mean body weight of sex-matched CKI and WT controls, and spleen weight. See also Figures S1 and S12. ns, not significant; p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Figure 2.. Lung stromal tissues express STING

Immunofluorescent staining of lungs from STING^KO/KO, STING^VM/WT, and STING^WT/WT mice stained for DAPI (gray), LYVE1 (yellow), PDPN (magenta), and STING (cyan). Magnified view (3×) is shown from the highlighted portion of each image and is shown to the right. Examples of LYVE1⁺PDPN⁻ blood vessels (BV), LYVE1⁺PDPN⁺ lymphatic vessels (LVs), LYVE1⁻PDPN⁻ conducting airway (CA), LYVE1⁻PDPN⁺ respiratory airway (RA), and morphologically apparent tertiary lymphoid organs (TLOs) are annotated by white text. Scale bar, 200 μm. Images represent data from one experiment.

Figure 3.. Tie2-Cre targeted expression of STING VM is sufficient to initiate immune recruitment to the lung

(A) CD45⁺ immune and CD45⁻ non-hematopoietic cells from the lungs of 12-week-old CKI × YFP (n = 8, white), CKI × Nkx2.1-Cre × YFP (n = 3, purple), CKI × PDGFRa-Cre × YFP (n = 4, blue), CKI × Tie2-Cre × YFP (n = 4, green), and CKI × CMV-Cre × YFP (n = 4, red) mice were evaluated for the percentage of YFP⁺ cells within the EPCAM⁺CD31⁻CD140a⁻ epithelial, CD140a⁺EPCAM⁻CD31⁻ fibroblast, and CD31⁺EPCAM⁻CD140a⁻ endothelial cell compartments. Data are from one experiment. (B–F) Eight-week-old CKI × Nkx2.1-Cre (n = 9, purple), CKI × PDGFRa-Cre (n = 6, blue), and CKI × Tie2-Cre (n = 8, green) mice were compared to sex- and littermate-matched control CKI mice (no additional Cre genes) (n = 3, n = 3, n = 8, respectively). Data shown represent at least two independent experiments. Bar graphs represent mean ± SD. (B) Percentage of lung EV immune cells within the total CD45⁺ lung population, and total number of CD45⁺ EV immune cells. (C) Percentage of CD69⁺ lung EV T cells. (D and E) Representative 10× field H&E stains on lungs from CKI and CKI × Tie2-Cre mice. Immunofluorescence staining of CKI × Tie2-Cre mouse lung: DAPI (gray), CD3 (cyan), LYVE1 (yellow), and B220 (magenta). Scale bars, 200 μm. Images are representative of at least two independent experiments. (F) Body weight normalized as the fold change compared to the mean body weight of sex-matched CKI control mice; spleen weight. See also Figure S2. ns, not significant; p > 0.05; **p < 0.01; ***p < 0.001.

Figure 4.. Targeted expression of STING VM in T cells and LTi induces lymphopenia and lymph node agenesis, but not ILD

(A) CD45⁺ splenic immune cells from 12 week-old CKI × YFP (n = 7, white), CKI × Rorc-Cre × YFP (n = 3, teal), and CKI × LysM-Cre × YFP (n = 3, brown) mice were evaluated for the percentage of YFP⁺ cells within B220⁺ B cells, CD3⁺ T cells, CD11b⁺ and/or CD11c⁺ myeloid cells, CD11b⁺Ly6G⁺ neutrophils, CD11b⁺Ly6G⁻Ly6C⁺ monocytes, and CD11c⁺MHCII⁺ dendritic cells. Data are from one experiment. (B–H) Eight-week-old CKI × Rorc-Cre (n = 8, teal) and CKI × LysM-Cre (n = 7, brown) mice were compared to age- and sex-matched control CKI mice (n=5, n=5, respectively). Data shown represent at least two independent experiments. Bar graphs represent mean ± SD. (B) Percentage of CD3⁺ T cells and CD11b⁺ and/or CD11c⁺ myeloid cells within the total CD45⁺ splenocyte population. (C) Percentage of CD69⁺ activated T cells. (D) Percentage of Ly6G⁺ neutrophils and Ly6C⁺Ly6G⁻ monocytes within the splenic myeloid subset. (E) Percentage of lung EV immune cells within total number of CD45⁺ lung cells and total number of EV immune cells. (F) Mean number of inguinal lymph nodes in CKI controls compared to the indicated strains. An additional cohort of CKI × CMV-Cre mice (n = 27, red) and their matched CKI controls (n = 18, white) are included. (G) Percentage of CD69⁺ and PD-1⁺ cells within EV T cell compartment. (H) Percentage of CD11b⁺Ly6C^hi inflammatory monocytes within lung EV myeloid compartment, and percentage of CD86⁺ cells within the lung EV monocyte compartment. ns, not significant; p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Figure 5.. Endothelial-specific expression of the STING VM mutation is sufficient to initiate immune recruitment to the lung

(A) CKI × YFP (n = 9, white), CKI × Cdh5-Cre^ERT2 × YFP (n = 11, yellow), and CKI × CAGG-Cre^ERTM × YFP (n = 8, orange) mice were treated with tamoxifen P0-P2 and CD31⁺ LECs and CD45⁺ lung cells were evaluated for YFP expression at 5–7 weeks of age. Eight-week-old sex- and littermate-controlled STING ^CKI × YFP (n = 4, white) and STING ^CKI × Tie2-Cre × YFP (n = 5, green) mice were similarly assessed. (B) Representative 4× field H&E histology of lung sections from CKI × Cdh5-Cre^ERT2 (two mice top row: left shows modest immune aggregate formation, right shows more extensive immune aggregate formation), CKI controls, and CKI × CAGG-Cre^ERTM mice. Images are representative of at least two independent experiments. (C) Immunofluorescence staining of mouse lungs from indicated strains: for DAPI (gray), CD3 (cyan), LYVE1 (yellow), and B220 (magenta). Images are representative of at least two independent experiments. (D) Percentage of lung EV immune cells within total CD45⁺ lung populations, and total number of EV immune cells. (E) Percentage of CD69⁺ EV T cells. (F) Body weight, normalized as the fold change compared to the mean body weight of sex-matched CKI control mice, and spleen weight. (A and D–F) Data shown represent at least two independent experiments. Bar graphs represent mean ± SD. See also Figures S1, S3, and S4. ns, not significant; p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Figure 6.. Lung inflammation is enhanced by non-endothelial expression of STING VM

(A) Six-week-old sex- and littermate-matched tamoxifen-treated CKI (n = 9), CKI × Cdh5-Cre^ERT2 (n = 11), and CKI × CAGG-Cre^ERTM (n = 8) mice were evaluated for the percentage of PD-1⁺ EV T cells and percentage of CD86⁺CD19⁺ lung EV B cells. (B) Percentage of CD11b⁺Ly6G⁻Ly6C^hi inflammatory monocytes (IMs) within the CD11b⁺ and/or CD11c⁺ EV myeloid cells, and percentage of CD86⁺CD11b⁺Ly6C⁺ lung EV monocytes. (C) Percentage of MHCII⁺ LECs and percentage of VCAM1⁺ LECs. (D) Percentage of ICAM1⁺ and VCAM1⁺ in CD31⁻CD140a⁺ lung fibroblasts. (E) Immunofluorescence staining of mouse lungs for DAPI (gray), VCAM-1 (cyan), LYVE1 (yellow), and PDPN (magenta). Twelve-week-old STING^WT/WT and STING^VM/WT are also included. Scale bar, 200 μm. Images are representative of at least two independent experiments. (A–D) Data shown represent at least two independent experiments. Bar graphs represent mean ± SD. See also Figure S5. ns, not significant; p > 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Figure 7.. Endothelial-directed VM shows a transcriptional signature in lung parenchyma and stroma significant for chemokine production

RNA-seq was performed on lung parenchymal and stromal cells from 6- to 8-week-old sex- and littermate-matched STING^WT/WT (WT, n = 5), STING^VM/WT (VM, n = 5), and tamoxifen-treated STING^CKI/WT (CKI, n = 5), STING^CKI/WT × CAGG-Cre^ERTM (CKI × CAGG, n = 5), and STING^CKI/WT × Cdh5-Cre^ERT2 (CKI × Cdh5, n = 5). Data shown are from one experiment. (A) PCA plot using the top 1,000 most varied genes. (B) Volcano plots comparing the following groups: WT vs. VM (left), CKI vs. CKI × CAGG (middle), and CKI vs. CKI × Cdh5 (right). Differentially expressed genes (DEGs) are identified as having a p_adj < 0.05 and either a fold change (FC) >2 (red) or <2 (blue). A subset of selected genes are labeled in green, which represent a basic SAVI transcriptional signature—Isg15, Cxcl9, Cxcl10, Ccl5, and H2dma. (C) A dot plot showing enrichment for GO:BP terms in upregulated DEGs across our comparison groups. Statistical significance is shown as color, and the fraction of genes in each term identified as upregulated DEGs (gene ratio) is shown as size. (D) A heatmap of genes relating to themes identified from a GSEA leading-edge analysis. Genes selected contribute to >3 signatures related to the identified theme and have an FC > 2 and p_adj < 0.1. Gene expression is shown as log2 transformed fold change (log₂FC) of individual mice expressing the VM allele normalized by the mean expression level in their littermate control group, with VM normalized by WT littermates, and CKI × CAGG and CKI × Cdh5 normalized by CKI littermates. Rows (genes) have been organized by hierarchical clustering on expression data across the three comparisons, with a dendrogram shown on the left side of each thematic plot. See also Figures S6–S11.