Airway epithelial cell identity and plasticity are constrained by Sox2 during lung homeostasis, tissue regeneration, and in human disease

Abstract

Maintenance of the cellular boundary between airway and alveolar compartments during homeostasis and after injury is essential to prohibit pathological plasticity which can reduce respiratory function. Lung injury and disease can induce either functional alveolar epithelial regeneration or dysplastic formation of keratinized epithelium which does not efficiently contribute to gas exchange. Here we show that Sox2 preserves airway cell identity and prevents fate changes into either functional alveolar tissue or pathological keratinization following lung injury. Loss of Sox2 in airway epithelium leads to a loss of airway epithelial identity with a commensurate gain in alveolar and basal cell identity, in part due to activation of Wnt signaling in secretory cells and increased Trp63 expression in intrapulmonary basal-like progenitors. In idiopathic pulmonary fibrosis, loss of SOX2 expression correlates with increased WNT signaling activity in dysplastic keratinized epithelium. SOX2-deficient dysplastic epithelial cells are also observed in COVID-19 damaged lungs. Thus, Sox2 provides a molecular barrier that suppresses airway epithelial plasticity to prevent acquisition of alveolar or basal cell identity after injury and help guide proper epithelial fate and regeneration.

Fig. 1. Deletion of Sox2 disrupts the maintenance of airway epithelial cell identity.

a Positive Sox2 expression distinguishes airway epithelial cells from alveolar epithelial cells. Nkx2-1⁺ Alveolar cells do not express Sox2. AW denotes airway, and Alv indicates alveoli. b SCGB3A2⁺ SCGB1A1^high and SCGB3A2⁺ SCGB1A1^low human secretory cells express SOX2 (n = 4 healthy human samples) but SFTPC⁺ AT2 cells do not. c–e Efficient lineage tracing and Sox2 deletion using Sox2^CreERT2/Flox; R26R^EYFP mice. Sox2^CreERT2/wt; R26R^EYFP (control) and Sox2^CreERT2/Flox; R26R^EYFP (Sox2^cKO) mice were given tamoxifen and analyzed 7, 14, and 21 days later (c). Quantification for lineage-trace efficiency and Sox2 deletion in control and Sox2^cKO mice (d). Representative images showing deletion of Sox2 in Sox2^cKO mice (e). Each dot represents an individual mouse, and error bars indicate mean with s.d. N = 5 mice for control and Sox2^cKO. f Secretory cell marker Scgb1a1 and ciliated cell marker β-Tubulin IV are downregulated in Sox2^cKO mice. Images obtained from mice 21 days after tamoxifen administration. Scale bars: a, b (bottom) 100 μm; b (top), e, f 25 μm.

Fig. 2. scRNA-seq analysis reveals acquisition of alveolar signature in Sox2-deficient airway cells.

a Airway cells from control and Sox2^cKO mice form distinct cell populations on the UMAP plot. Individual sample was plotted in the joint UMAP. b Dotplot showing canonical cell marker gene expressions in each cell population. Marker genes were extracted from CellCards. Both control and Sox2-deficient cells were used for plotting. c Cell proportions in control and Sox2^cKO mice. Heatmaps showing the top 50 differentially expressed genes for secretory cells (d) and BASCs (e). f Trajectory analysis shows lineage relationships between Sox2^cKO secretory cells and BASCs. g Heatmap showing expression of genes defining trajectory. h Scgb1a1 was downregulated and Sftpd was upregulated in secretory cells and BASCs of the Sox2^cKO mice. AT2 score, generated using AT2 cell marker genes, was higher in secretory cells and BASCs. i Ctnnb1 was upregulated in secretory cells and BASCs of the Sox2^cKO mice. Wnt activation score, generated using Wnt downstream genes, was higher in Sox2^cKO mice. Wnt downstream genes were obtained from “the Wnt homepage”.

Fig. 3. Sox2 deletion upregulates basal signature in basal-like progenitors.

a Trp63 expression is upregulated in Trp63⁺ intrapulmonary basal-like progenitor cells of Sox2^cKO. b, c p63 is upregulated in basal-like progenitors of Sox2^cKO mice (b). The number of Intrapulmonary basal-like progenitors increase in Sox2^cKO mice (n = 6 mice) compared to control mice (n = 4 mice). Quantification was performed in intrapulmonary airways within 1-2 bifurcations of the right or left intrapulmonary bronchus (c). Scale bars: 25 μm. d–f Basal cell genes reported to have p63 ChIP-seq peaks were upregulated (d) and basal cell genes without p63 ChIP-seq peaks were downregulated (e) in p63⁺ intrapulmonary basal-like progenitor cells from Sox2^cKO mice. ChIP tracks showing p63 peaks for Itgb4 and Itga6 (f). ChIP-seq data were obtained from Weiner et al. . ***P < 0.001 by two-tailed t-test. Each dot represents an individual mouse, and error bars indicate mean with s.d.

Fig. 4. Sox2 deletion promotes airway cell contribution to alveolar regeneration.

a Influenza virus was intranasally administered to control and Sox2^cKO mice and the mice were analyzed 14 days after the injury. b, c Sox2 deletion promotes airway cell transition to Sftpc⁺ AT2 cells after influenza lung injury. EYFP⁺ lineage-traced airway cells transition to Sftpc⁺ AT2 cells in control (b top), but more so in the Sox2^cKO mice (b bottom) after injury. Quantification for the percentage of lineage-traced EYFP⁺ Sftpc⁺ cells after influenza lung injury (n = 9 control and n = 8 Sox2^cKO mice from 3 independent experiments) (c). d, e Sox2 deletion promotes airway cell transition to dysplastic keratinized pods after influenza lung injury. EYFP⁺ lineage-traced airway cells transition to Krt5⁺ dysplastic keratinized cells in control (d top), but more so in the Sox2^cKO mice (d bottom) after injury. Quantification for the area size of the Krt5⁺ region after influenza lung injury (n = 11 control and n = 8 Sox2^cKO mice from 3 independent experiments) (e). f Joint UMAP showing lineage traced populations of EYFP⁺ cells from control and Sox2^cKO mice 28 days after influenza infection. g Dotplot showing canonical cell marker gene expressions in each cell population. h Correlation analysis shows that AT2 and basal cells derived from either control or Sox2^cKO airway cells were similar. i Expression of basal cell genes and alveolar epithelial marker genes in control and Sox2^cKO mice 28 days after influenza infection. ***P < 0.001 and *P < 0.05 by two-tailed t-test. N.S. not significant. Each dot represents an individual mouse, and error bars indicate mean with s.d. Scale bars: b, d 100 μm.

Fig. 5. Sox2 deletion leads to airway cell contribution to alveolar regeneration in coordination with Wnt signaling.

a Expression of nuclear β-catenin is more prominent in airways of Sox2^cKO/Ctnnb1^GOF mice than Ctnnb1^GOF or control wild-type mice. b Intrapulmonary basal-like progenitors from Sox2^cKO/Ctnnb1^GOF mice express more p63 than Ctnnb1^GOF mice. c The number of intrapulmonary basal-like progenitors increase in Sox2^cKO/Ctnnb1^GOF mice (n = 4 mice) compared to Ctnnb1^GOF mice (n = 4 mice). Quantification was performed in intrapulmonary airways within 1-2 bifurcations of the right or left intrapulmonary bronchus. The results from Fig. 3c are reused for comparison. d Influenza virus was intranasally administered to control Ctnnb1^GOF and Sox2^cKO/Ctnnb1^GOF mice and the mice were analyzed 14 days after the injury. e, f Sox2 deletion promotes airway cell transition to Sftpc⁺ AT2 cells after influenza lung injury in Ctnnb1^GOF mice. EYFP⁺ lineage-traced airway cells transition to Sftpc⁺ AT2 cells in control Ctnnb1^GOF (e top), but more so in the Sox2^cKO/Ctnnb1^GOF mice (e bottom) after injury. Quantification for the percentage of lineage-traced EYFP⁺ Sftpc⁺ cells after influenza lung injury (n = 6 control and n = 7 Sox2^cKO/Ctnnb1^GOF mice from 2 independent experiments) (f). The results from Fig. 4c, which are from different experiments, are reused for comparison. g, h Sox2 deletion promotes airway cell transition to dysplastic keratinized pods after influenza lung injury in Ctnnb1^GOF mice. Krt5⁺ pods were not formed in control (g top), but the pods were observed in the Sox2^cKO/Ctnnb1^GOF mice (g bottom) after injury. Quantification for the area size of the Krt5⁺ region after influenza lung injury (n = 6 control and n = 5 Sox2^cKO/Ctnnb1^GOF mice from 2 independent experiments) (h). The results from Fig. 4e, which are from different experiments, are reused for comparison. i Model describing the loss of Sox2 and airway fate decisions after lung injury. ***P < 0.001 and **P < 0.01 by one-way ANOVA. Each dot represents an individual mouse, and error bars indicate mean with s.d. Scale bars: e, g 100 μm; a, b 25 μm.

Fig. 6. Identification of SOX2 deficient, WNT active, KRT5^–/KRT17⁺ dysplastic epithelial cells in IPF.

a UMAP plot showing epithelial cell clusters in patients with IPF. scRNA-seq Data obtained from Habermann et al.. b, c KRT5^–/KRT17⁺ dysplastic cells are TP63⁺ and SOX2^–. UMAP plot (b) and Dotplot (c). d Immunohistochemistry analysis of human IPF showing negative SOX2 in KRT5^–/KRT17⁺ dysplastic cells (n = 3 samples). e Immunohistochemistry analysis of human IPF showing positive TP63 in KRT5^–/KRT17⁺ dysplastic cells (n = 2 samples). f, g Wnt activation score is higher in KRT5^–/KRT17⁺ cells. Wnt downstream genes obtained from “the Wnt homepage” were used for scoring. UMAP plot (f) and Dotplot (g) are shown. h Immunohistochemistry analysis of human IPF showing nuclear β-catenin signal in the epithelial cells and spindle-shaped fibroblasts of fibroblastic foci. Nuclear β-catenin signal was not prominent in healthy human samples (n = 2 samples). White arrows denote epithelial cells and yellow arrows denote fibroblasts (n = 4 IPF samples). Scale bars: d, e 100 μm; h 25 μm.

Fig. 7. KRT17⁺ dysplastic epithelial cells in post-COVID-19 pulmonary fibrosis patients are SOX2 deficient.

a Immunohistochemistry analysis of human post-COVID19 pulmonary fibrosis showing negative SOX2 in KRT5^–/KRT17⁺ dysplastic cells (n = 3 samples). b Model describing negative expression of SOX2 in dysplastic KRT5⁻/KRT17⁺ cells in human pulmonary fibrosis. Scale bars: 100 μm.