Single-cell transcriptomics reveals lasting changes in the lung cellular landscape into adulthood after neonatal hyperoxic exposure

Abstract

Ventilatory support, such as supplemental oxygen, used to save premature infants impairs the growth of the pulmonary microvasculature and distal alveoli, leading to bronchopulmonary dysplasia (BPD). Although lung cellular composition changes with exposure to hyperoxia in neonatal mice, most human BPD survivors are weaned off oxygen within the first weeks to months of life, yet they may have persistent lung injury and pulmonary dysfunction as adults. We hypothesized that early-life hyperoxia alters the cellular landscape in later life and predicts long-term lung injury. Using single-cell RNA sequencing, we mapped lung cell subpopulations at postnatal day (pnd)7 and pnd60 in mice exposed to hyperoxia (95% O₂) for 3 days as neonates. We interrogated over 10,000 cells and identified a total of 45 clusters within 32 cell states. Neonatal hyperoxia caused persistent compositional changes in later life (pnd60) in all five type II cell states with unique signatures and function. Premature infants requiring mechanical ventilation with different durations also showed similar alterations in these unique signatures of type II cell states. Pathologically, neonatal hyperoxic exposure caused alveolar simplification in adult mice. We conclude that neonatal hyperoxia alters the lung cellular landscape in later life, uncovering neonatal programing of adult lung dysfunction.

Keywords: Alveolar type I and Alveolar type II cells; Bronchopulmonary dysplasia; Hyperoxic lung injury; Lipid and matrix homeostasis; Progenitor cells.

Graphical abstract

Fig. 1

Experimental design. Whole lung single-cell suspensions from neonatal mice exposed to air or hyperoxia for 3 days. The samples collected after air recovery at both pnd7 and pnd60 were analyzed using the Drop-seq workflow.

Fig. 2

Drop-seq analysis identifies a diversity of cell types in mouse lungs. (A) t-SNE plot representing the integration of the four conditions (Air/pnd7, O₂/pnd7, Air/pnd60 and O₂/pnd60). Single cells are colored by cluster identity. Forty-five clusters were detected across the four conditions. (B) The dot plot shows (1) the percentage of cells expressing respective selected marker gene using dot color and (2) the average expression level of each gene based on unique molecular identifier (UMI) counts. Rows represent clustered cell types, demonstrating similarities of transcriptional profiles. (C) t-SNE plot representing the integration of the four conditions (Air/pnd7, O₂/pnd7, Air/pnd60 and O₂/pnd60). Single cells are colored by cell type identity. Thirty-two cell types are detected across the four conditions. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 3

t-SNE plots of lung cell subpopulations. t-SNE plots represent 45 clusters in four integrated conditions: Air/pnd7, O₂/pnd7, Air/pnd60 and O₂/pnd60.

Fig. 4

Percentage of cell states detected in each condition. Percentage of each cell status in Air/pn7, O₂/pnd7, Air/pnd60 and O₂/pnd60 groups.

Fig. 5

Effect of hyperoxia on B cells, Club cells, ciliated cells, AT1 and AT2 cell heterogeneity. (A) Fold change in percentage of B cells, ciliated cells and Club cells in O₂/pnd7, Air/pnd60 and O₂/pnd60 groups compared to Air/pnd7. (B) Numbers of cells detected in each cell population per condition. (C) Fold change in percentage of clusters 1, 2, 4, 6, 9, and 21 in B cells from O₂/pnd7, Air/pnd60 and O₂/pnd60 groups compared to Air/pnd7. (D) Numbers of cells detected in each cluster per condition in B cells. (E) Fold change in percentage of AT1 and AT2 clusters in O₂/pnd7, Air/pnd60 and O₂/pnd60 compared to Air/pnd7. (F) Numbers of cells detected in each cell population per condition. (G) Fold change in percentage of clusters 3, 11, 13, 22, 25 and 27 in O₂/pnd7, Air/pnd60 and O₂/pnd60 groups compared to Air/pnd7. (H) Numbers of cells detected in each of these clusters per condition.

Fig. 6

tSNE plots showing AT2 cell subpopulations and their unique gene expression. (A) tSNE plots show AT2 cell subpopulations in clusters 3, 11, 13, 22 and 27. (B) tSNE plots overlaid with SCTransform-normalized expression values of Cftr, Malat1, Tkt, Spock2 and Alas2 in AT2 cells among Air/pnd7, O₂/pnd7, Air/pnd60 and O₂/pnd60 groups.

Fig. 7

Expression of Alas2, Cftr, Spock2 and Tkt in AT2 cells from mice exposed to hyperoxia. C57BL/6J mice (<12 h old) were exposed to hyperoxia for 3 days, and allowed to recover in room air until pnd7 and pnd60. (A) Double immunofluorescence was performed to detect co-localization of Alas2, Cftr, Spcok2 or Tkt with pro-Spc (Spc) in lungs, and representative images are shown for each group. Arrows denote co-localization of Alas2, Cftr, Spcok2 or Tkt with pro-SPC. (B) Numbers of co-localized cells were counted in three randomly selected high-power fields (HPF) in each mouse. This was normalized to Dapi⁺ cells. Bar size: 50 μm. N = 3 mice including 2 males and 1 female per group. (C) Immunofluorescence was performed to detect Cftr⁺/pro-SPC⁺ cells in lungs of premature infants requiring mechanical ventilation. Numbers of Cftr⁺/pro-SPC⁺ cells were counted in three randomly selected HPF per sample. This was normalized to Dapi⁺ cells. Bar size: 50 μm. Data are expressed as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 vs respective air groups or control group. ^†P < 0.05, ^††P < 0.01, ^†††P < 0.001 vs respective pnd7 groups.

Fig. 8

Increased Malat1 expression in AT2 cells of mice exposed to hyperoxia as neonates and in premature infants requiring mechanical ventilation. (A, B) C57BL/6J mice (<12 h old) were exposed to hyperoxia for 3 days, and allowed to recover in room air until pnd7 and pnd60. RNAscope and immunofluorescence were performed to detect co-localization of Malat1 with pro-Spc in lungs. Representative images are shown for each group. Arrows denote co-localization of Malat1 with pro-SPC. Numbers of cells with co-localized Malat1 and pro-Spc were counted in three randomly selected high-power fields (HPF) for each mouse. This was normalized into Dapi⁺/cells. Bar size: 50 μm. N = 3 mice including 2 males and 1 female per group. (C, D) Immunofluorescence was performed to detect Malat1⁺/pro-SPC⁺ cells in lungs of premature infants requiring mechanical ventilation. Numbers of Malat1⁺/pro-SPC⁺ cells were counted in three randomly selected HPF per sample. This was normalized to Dapi⁺ cells. Bar size: 50 μm. N = 3 per group. Data are expressed as mean ± SEM. *P < 0.05, ***P < 0.001 vs respective air groups (B) or control subjects (D). ^†P < 0.05 vs respective pnd7 groups (B) or short-term ventilation group (D).

Fig. 9

Neonatal hyperoxia causes persistent lung injury in mice. (A) C57BL/6J neonatal mice (<12 h old) were exposed to air or hyperoxia (95% O₂) for 3 days, and then allowed for recover in room air until pnd7 and pnd60. H&E staining was performed to measure mean linear intercept (Lm) and radical alveolar count (RAC) in mouse lungs. Bar size: 200 μm. Data are expressed as mean ± SEM. N = 5 mice including 3 males and 2 females per group. **P < 0.01, ***P < 0.001 vs respective air group. (B) Schematic figure showing hyperoxia-induced long-term changes in AT2 cell subpopulations and functions, leading to alveolar simplification.