Identification of a novel subset of alveolar type 2 cells enriched in PD-L1 and expanded following pneumonectomy

Abstract

Alveolar type 2 (AT2) cells are heterogeneous cells, with specialised AT2 subpopulations within this lineage exhibiting stem cell properties. However, the existence of quiescent, immature cells within the AT2 lineage that are activated during lung regeneration is unknown.

Sftpc^CreERT2/+;tdTomato^flox/flox mice were used for the labelling of AT2 cells and labelled subpopulations were analysed by flow cytometry, quantitative PCR, assay for transposase-accessible chromatin using sequencing (ATAC-seq), gene arrays, pneumonectomy and culture of precision-cut lung slices. Single-cell RNA-sequencing (scRNA-seq) data from human lungs were analysed.

In mice, we detected two distinct AT2 subpopulations, with low tdTomato level (Tom^Low) and high tdTomato level (Tom^High). Tom^Low cells express lower levels of the AT2 differentiation markers Fgfr2b and Etv5, while Tom^High, as bona fide mature AT2 cells, show higher levels of Sftpc, Sftpb, Sftpa1, Fgfr2b and Etv5 expression. ATAC-seq analysis indicates that Tom^Low and Tom^High cells constitute two distinct cell populations, with specific silencing of Sftpc, Rosa26 and cell cycle gene loci in the Tom^Low population. Upon pneumonectomy, the number of Tom^Low but not Tom^High cells increases and Tom^Low cells show upregulated expression of Fgfr2b, Etv5, Sftpc, Ccnd1 and Ccnd2 compared to Sham. Tom^Low cells overexpress programmed cell death 1 ligand 1 (PD-L1), an immune inhibitory membrane receptor ligand, which is used by flow cytometry to differentially isolate these two subpopulations. In the human lung, data mining of a recent scRNA-seq AT2 data set demonstrates the existence of a PD-L1^Pos population. Therefore, we have identified a novel population of AT2 quiescent, immature progenitor cells in mouse that expand upon pneumonectomy and we have provided evidence for the existence of such cells in human.

Short abstract

A novel population of AT2 progenitor cells enriched for PD-L1 has been identified. This normally quiescent subpopulation of AT2 cells becomes highly proliferative and differentiates into mature AT2 in response to alveolar injury. https://bit.ly/31G0IIW

Introduction

Alveolar type 2 (AT2) progenitor cells display self-renewal capabilities to maintain the AT2 cell pool and differentiate into AT1 cells following lung injury [1–5]. However, it is still debatable whether quiescent and immature AT2 progenitor cells exist in mouse and human lungs and if surface markers can be used to isolate these cells.

In rodents, AT2 cells are heterogeneous. Different subpopulations within this lineage, with different stem cell properties, have been identified based on various markers such as E-cadherin (E-CAD), AXIN2, integrin α6β4 and the dual expression of surfactant protein C (SFTPC)/secretoglobin family 1A member 1 (SCGB1A1) [6–9]. AT2 cells can be divided equally into E-CAD^Pos and E-CAD^Neg cells. E-CAD^Neg cells are more resistant to hyperoxia injury, more proliferative and display higher level of telomerase activity, while E-CAD^Pos cells are more sensitive to damage [6]. In the adult mouse lung, AXIN2^Pos cells make up 1% of the mature AT2 (SFTPC^High) cells and are distributed throughout the lung [8]. This ratio of AXIN2^Pos cells is stable over time. In vivo lineage tracing of AXIN2^Pos (SFTPC^High) cells revealed their ability to form clones of two to five cells accompanied by their expansion and differentiation into AT1 expressing podoplanin (PDPN). In addition, these cells display enhanced self-renewal capabilities in alveolosphere assays compared to the AXIN2^Neg AT2 cells, suggesting that these AXIN2^Pos cells represent a subpopulation of mature AT2 cells with enhanced stem cell capacities [8]. In a similar and independent study, AXIN2^Pos cells belonging to the mature AT2 cell population were characterised. These cells, called alveolar epithelial progenitor (AEP) cells, comprise ∼20% of adult mature AT2 cells and are enriched for WNT target genes and developmental genes such as Fgfr2, Nkx2.1, Id2, Etv4, Etv5 and Foxa1. These cells also display increased self-renewal ability in alveolospheres compared to total AT2 cells [9].

AT2 cell heterogeneity is not limited to the adult lung. During the alveologenesis phase of lung development, AXIN2^Pos AT2 cells display enhanced proliferation compared to total AT2 cells [10]. A subset of SFTPC^Neg, laminin receptor integrin α6β4^Pos cells, located at the bronchoalveolar duct junctions (BADJ) and within the alveoli, regenerate AT2 SFTPC^Pos cells in the alveoli after severe lung injury [7]. Finally, bronchoalveolar stem cells, positive for the alveolar marker SFTPC and the club cell marker SCGB1A1, are located at the BADJ and are amplified upon injury to give rise to either alveolar epithelial cells or club cells [11–13]. AT2 cell heterogeneity was equally shown in the context of diphtheria toxin injury, with a subpopulation of lineage-labelled AT2 cells displaying relatively more resistance to diphtheria toxin toxicity able to repopulate the AT2 pool following injury. The identity of these survivor cells, however, is not clear [3].

In the present study, we identified a novel population of AT2 quiescent and immature progenitor cells characterised by low Sftpc level and high expression of the surface marker programmed cell death 1 ligand 1(PD-L1) (also known as CD274), an immune checkpoint protein expressed in cancer stem cells and mediating anti-tumour suppression response [14–16]. We deployed a series of in vivo approaches in mice to label different subpopulations of AT2 cells, examined their epigenetic and transcriptomic characteristics, and tested their respective response in vivo during lung regeneration following pneumonectomy as well as in vitro using the culture of precision-cut lung slices (PCLS). We also validated the use of PD-L1 antibodies to differentially sort epithelial subpopulations and carried out data mining of recently published AT2 single-cell RNA-sequencing (scRNA-seq) experiments. Our work opens the way for an in-depth characterisation of this novel quiescent and immature AT2 stem cell population.

Materials and methods

Animal experiments

All animal studies were performed according to protocols approved by the Animal Ethics Committee of the Regierungspraesidium Giessen (permit numbers: G7/2017–No.844-GP and G11/2019–No. 931-GP).

Pneumonectomy

See supplementary material.

Lung dissociation and fluorescence-activated cell sorting

Single-cell suspension was generated from adult lungs and stained with anti-EPCAM (APC-Cy7-conjugated, Biolegend, 1:50), CD49F (APC-conjugated, Biolegend, 1:50), anti-PDPN (FITC-conjugated, Biolegend, 1:20) and anti-PD-L1 (unconjugated, Thermo Fisher, 1:100) antibodies for 20 min on ice in the dark, followed by washing. The cells were then stained for goat anti-rabbit secondary antibody Alexa fluor 488 (Invitrogen, 1:500) for 20 min on ice in the dark. Next, cells were washed and stained with SYTOX (Invitrogen), a live/dead cell stain, according to the manufacturer's instructions. Flow cytometry data acquisition and cell sorting were carried out using a FACSAria III cell sorter (BD Biosciences). Data were analysed using FlowJo software version X (FlowJo, LLC).

See supplementary material for the mouse lung dissociation.

RNA extraction and quantitative real-time PCR

See supplementary material.

Immunofluorescence staining

See supplementary material.

Alveolosphere assay

See supplementary material.

Microarray

See supplementary material.

ATAC-seq

See supplementary material.

Mining of scRNA-seq data set for human AT2 cells

For the analysis, we used the synapse ID (syn21041850) data set deposited at https://www.synapse.org/#!Synapse:syn21041850/wiki/600865 and published by Travaglini et al. [17]. The Seurat objects were updated to Seurat v3 [18] and AT2 cells were extracted using the subset’ function. The data set was reanalysed starting from raw counts using the SCTransform wrapper in Seurat. Data imputation was then performed using a low-rank approximation with the package ALRA [19].

Quantification and statistical analysis

For quantification of immunofluorescence, cells were counted in 10 independent 63× fields per sample. Statistical analysis and graph assembly were carried out using GraphPad Prism 6 (GraphPad Prism Software). Significance was determined by unpaired two-tailed t-tests. Data are presented as mean±sem. Values of p<0.05 were considered significant. The number of biological samples (n) for each group is stated in the corresponding figure legends.

Results

Identification of two AT2 subpopulations with different Sftpc and Fgfr2b levels

Sftpc^CreERT2/+;tdTomato^flox/flox mice were used to lineage label AT2 cells in the adult lung. In our experimental approach, tamoxifen was delivered in the water for 1 week. Two distinct subpopulations of tdTomato^Pos (Tom^Pos) cells, that we called tdTomato^Low (Tom^Low) and tdTomato^High (Tom^High), were identified using flow cytometry analysis (figure 1a–c). Of note, a similar observation, which was not followed up, was previously reported using the Sftpc^Cre-ER mouse line generated by Finn et al. [20]. On average, Tom^Low cells represented 9.93±1.73% of the overall EPCAM^Pos cells (n=4) and Tom^High cells represented 42.75±1.22% of the overall EPCAM^Pos cells (n=4). Out of the total Tom^Pos cells, on average 17.58±2.52% (n=4) were Tom^Low and 81.93±2.3% (n=4) were Tom^High. We also determined the extent of recombination of the LoxP-STOP-LoxP-tdTomato cassette in the isolated Tom^Low and Tom^High cells. Our results indicate that complete recombination was observed in Tom^High while incomplete recombination was observed in Tom^Low, indicated by the presence of the 795 bp stop codon band (figure 1d). This result suggested either contamination of the Tom^Low pool by non-AT2 cells or the lack of access of Cre to the LoxP-STOP-LoxP-tdTomato cassette in the Rosa26 locus of Tom^Low cells. The presence of the two subpopulations based on Tomato intensity could also be explained by improper recombination of either one or two copies of the LoxP-STOP-LoxP-tdTomato cassette in tdTomato^flox/flox mice. To test these different possibilities, we carried out a similar experiment with Sftpc^CreERT2/+;tdTomato^flox/+ mice, which exhibit only one copy of the LoxP-STOP-LoxP-tdTomato cassette (figure 1e–h). Similarly, we found two subpopulations of Tom^Pos cells and, importantly, complete recombination of the LoxP-STOP-LoxP-tdTomato cassette in Tom^Low, indicating the absence of contamination by non-AT2 cells in the Tom^Low pool. This conclusion is supported by our quantitative PCR (qPCR) data for markers of endothelial, hematopoietic, macrophage and fibroblast cells (supplementary figure S1). Overall, our data support the existence of two AT2 cell subpopulations in Sftpc^CreERT2/+;tdTomato^flox/flox lungs, based on different levels of Tomato expression. This is due to the differential expression of Tomato per se from the Rosa26 promoter in Tom^High versus Tom^Low cells. In addition, in Sftpc^CreERT2/+;tdTomato^flox/flox lungs, this is also associated with the incomplete recombination of the LoxP-STOP-LoxP cassette, which is also located in the Rosa26 locus.

FIGURE 1.

Identification of two subpopulations of AT2-lineage-labelled cells, named tdTomato^Low (Tom^Low) and tdTomato^High (Tom^High). a) Timeline of tamoxifen treatment of Sftpc^CreERT2/+;tdTomato^flox/flox mice (n=4). d: day. b) Representative flow cytometry of EPCAM^Pos population selection and identification of Tom^Low (8.5%) and Tom^High (44.2%) populations based on the tdTomato level. c) Percentage of Tom^Low (16.3%) and Tom^High (83.7%) in total tdTomato^Pos cells. d) PCR on genomic DNA isolated from fluorescence-activated cell sorting (FACS)-based sorted Tom^Low and Tom^High cells from Sftpc^CreERT2/+;tdTomato^flox/flox mice for Stop codon. e) Timeline of tamoxifen treatment of Sftpc^CreERT2/+;tdTomato^flox/+mice and f) representative flow cytometry analysis of Tom^Low (7.3%) and Tom^High (38.7%). g) The pie chart shows the percentage of Tom^Low (15.3%) and Tom^High (84.7%) in total tdTomato^Pos cells. h) PCR on genomic DNA isolated from FACS-based sorted Tom^Low and Tom^High cells from Sftpc^CreERT2/+;tdTomato^flox/+ mice. i) Quantitative PCR analysis of FACS-based sorted Tom^Low and Tom^High cells for the expression of Tomato, Fgfr2b, Etv5, Sftpc, Sftpb and Sftpa1. dCT: delta cycle threshold. Data are presented as mean±sem. *: p<0.05; **: p<0.01; ***: p<0.001.

We also confirmed that epithelial cells were specifically labelled in this mouse line and the labeling mostly targeted AT2 SFTPC^Pos cells (supplementary figure S2). Very few BACS cells, identified through their localisation at the BADJ, were labelled (0.1% of the total Tom^Pos cells, data not shown), consistent with the literature [7]. Our labelling efficiency of AT2 cells was 77±5.40% (n=4) (supplementary figure S2b). In this mouse line, 4.5% of Tom^Pos/total cells were labelled in mice exposed to normal water, indicating leakiness. This percentage increased to 20.5% of Tom^Pos/total cells in mice exposed to tamoxifen water for 1 week (supplementary figure S3), indicating robust induction of labelling. We also carried out labelling at 8 weeks of age by exposing the mice to tamoxifen water for 1 week followed by a 6- and 9-months chaseperiod. Our data indicate that both Tom^Low and Tom^High subpopulations were detected at these two time points, and were stable populations (supplementary figure S4).

Next, qPCR was performed on fluorescence-activated cell sorting (FACS)-isolated Tom^Low and Tom^High cells. Our results show that Fgfr2b and its associated downstream target Etv5, as well as the differentiation markers Sftpc, Sftpb and Sftpa1, were significantly enriched in Tom^High versus Tom^Low cells (figure 1i). Thus, we conclude that Tom^High represents the bona fide mature AT2 cells and we hypothesised that the lineage-related Tom^Low cells represent immature AT2 cells.

ATAC-seq analysis indicates that Tom^Low cells and Tom^High cells are distinct cell populations

To carry out genome-wide profiling of the epigenomic landscape, an assay for transposase-accessible chromatin using sequencing (ATAC-seq) was performed on Tom^Low and Tom^High subpopulations (figure 2a). Common and distinct peaks were identified for Tom^Low and Tom^High cells (figure 2b–d).

FIGURE 2.

Assay for transposase-accessible chromatin using sequencing (ATAC-seq) analysis on fluorescence-activated cell sorting (FACS)-based sorted tdTomato^Low (Tom^Low) and tdTomato^High (Tom^High) populations. a) Timeline of tamoxifen treatment of Sftpc^CreERT2/+;tdTomato^flox/flox mice (n=3). ATAC-seq was carried out on FACS-based sorted Tom^Low and Tom^High subpopulations. d: day. b, c) Coverage heat maps of Tom^Low and Tom^High, displaying genome-wide regions of differential open chromatin peaks in Tom^Low versus Tom^High. Tom^Low chromatin is less open and transcriptionally less active compared to Tom^High. Ctrl-low shows the peaks related to open chromatin regions in Tom^Low. Ctrl-high shows the peaks related to open chromatin regions in Tom^High. Common shows peaks detected in both Tom^Low and Tom^High subpopulations. d) ATAC-seq analysis of peaks based on the cut-offs shows 3605 upregulated in Tom^High (false discovery rate (FDR)<0.05, log₂(FC)>0.585, base mean>20), 3512 upregulated in Tom^Low (FDR<0.05, log₂(FC)>0.585, base mean>20) and 3878 non-regulated (base mean>20, FDR>0.5, log₂(FC) between −0.15 and 0.15), which means 32% and 32.2% of the genome is differently accessible in Tom^Low and Tom^High, respectively. e) ATAC-seq histogram of average read coverage at the Rosa26 locus shows distinct ATAC-seq peaks at the promoter and denser chromatin in Tom^Low compared to Tom^High in this locus. Representative peaks of Tom^Low and Tom^High are the average of three independent samples. f) Quantification of peaks at the Rosa26 locus. g) ATAC-seq profile at Sftpc locus shows distinct ATAC-seq peaks at the promoter and denser chromatin in Tom^Low compared to Tom^High. Representative peaks of Tom^Low and Tom^High are averages of three independent samples. h) Quantification of peak regions of Sftpc locus. The ATAC-seq data have been normalised for sequencing depth and the scale on the y-axis was chosen for optimal visualisation of peaks. i) ATAC-Heat-pVAI Z-score, Pearson, average heat map based on the ATAC-seq data of top cell cycle-related genes differentially regulated in Tom^Low and Tom^High. FDR: the significance of results by Benjamini–Hochberg correction of multiple tests (n=3).

Further analysis of the ATAC-seq data using the Reactome database indicated that the chromatin in loci of genes relating to metabolism, cholesterol metabolism, surfactant metabolism and triglyceride biosynthesis was more open in Tom^High cells. These data are in agreement with the known role of mature AT2 cells in surfactant production. By contrast, Tom^Low cells exhibit more accessibility for genes relating to the adaptive and innate immune system as well as genes involved in extracellular matrix (ECM) organisation, ECM proteoglycans and degradation of the ECM. These results suggest a new and important function for the Tom^Low cells in interactions with the immune system, potentially also displaying migratory capabilities through ECM degradation (supplementary figure S5a).

Our ATAC-seq data also support the observed decrease in Tomato (expressed from the Rosa26 promoter) and Sftpc expression at the mRNA level in Tom^Low versus Tom^High. In the Rosa26 and Sftpc loci, the peaks corresponding to the open chromatin configuration were detected at much higher levels in Tom^High than in Tom^Low, and this difference was confirmed to be statistically significant upon quantification (figure 2e–h). Interestingly, the analysis of ATAC-seq data also suggested that Tom^Low cells have reduced expression of cell cycle genes compared to Tom^High cells (figure 2i). This decrease in cell cycle genes was confirmed by the analysis of our gene array data between Tom^Low and Tom^High (supplementary figure S5b). Overall our data indicate that in homeostatic conditions the Tom^Low cells fit the profile of a quiescent population.

AT2 Tom^Low and Tom^High display different colony-forming capabilities

To compare the self-renewal capacity of Tom^Low and Tom^High, FACS-based sorted cells were co-cultured with CD31^NegCD45^NegEPCAM^NegSCA1^Pos resident lung mesenchymal cells according to a previously published protocol (figure 3a). Tom^High behaved as bona fide AT2 cells, forming alveolospheres with the expected colony-forming efficiency (figure 3b, d, e) [3, 9, 21], whereas Tom^Low displayed very weak organoid-forming capabilities, which is in line with their proposed quiescent status. Both populations transdifferentiated into receptor for advanced glycation end-products (RAGE)-positive AT1 cells after 14 days in culture (figure 3c).

FIGURE 3.

Different colony-forming capabilities of tdTomato^Low (Tom^Low) and tdTomato^High (Tom^High) cells. a) Representative flow cytometry shows the gating strategy of CD31^Neg CD45^Neg EPCAM^Neg population and further selection of SCA1^Pos resident mesenchymal cells from C57BL/6 lungs (upper plot), as well as the selection of Tom^Low and Tom^High from the EPCAM^Pos population from Sftpc^CreERT2/+;tdTomato^flox/flox(lower plot). Mesenchymal cells were co-cultured with Tom^Low and Tom^High separately (n=3). d: day. b) Representative alveolospheres from Tom^Low and Tom^High (n=3). Scale bar: 100 μm. c) Representative SFTPC and RAGE immunofluorescence staining of alveolospheres after 14 days in culture. Scale bar: 50 μm. d) Quantification of colony-forming efficiency (CFE) and e) alveolospheres size in Tom^High and Tom^Low (n=3). Data are presented as mean±sem. *: p<0.05.

Expansion of Tom^Low population following pneumonectomy

A critical question regarding these two subpopulations of AT2 cells is whether they are differentially engaged in the context of lung regeneration. Therefore, we used the mouse pneumonectomy (PNX) model to trigger lung regeneration, a process that is tightly associated with the proliferation of AT2 cells [22, 23]. Sftpc^CreERT2/+;tdTomato^flox/flox mice (n=4) were treated with tamoxifen water for 1 week to label the AT2 lineage. The mice were then put on normal water for 2 weeks to ensure enough time for tamoxifen clearance. Unilateral left lung pneumonectomy was then performed to induce the process of compensatory growth in the remaining right lobes. Control mice (Sham) underwent the same process but without removing the left lobe (figure 4a). The animals were killed on day 7 post-surgery and the right lobes were processed for FACS. Because the accessory lobe has been shown to respond the most to PNX [24], our results are likely underestimating the response of Tom^Low cells in the context of PNX. The quantification of the abundance of Tom^Low and Tom^High cells over the total number of EPCAM^Pos cells in Sham and PNX mice indicates that the ratio of Tom^High cells/total number of EPCAM^Pos cells remained unchanged between two conditions, while the ratio of Tom^Low cells was significantly increased in the context of PNX versus Sham (figure 4b–d). This suggests that it is the Tom^Low cells, rather than the previously thought Tom^High cells, that are contributing to the process of lung regeneration. These two AT2 populations were isolated by FACS for further analysis by qPCR. Upon PNX, the expression of Fgfr2b, Etv5, Sftpc, Cyclin D1 (Ccnd1) and Cyclin D2 (Ccnd2) was significantly upregulated in Tom^Low. A trend towards an increase was also observed in Ki67 expression (figure 4e). These results are consistent with FGFR2b signalling activation and proliferation in Tom^Low cells in the context of lung regeneration. By contrast, Tom^High cells showed no difference in Fgfr2b, Etv5 or Sftpc between PNX and Sham conditions (figure 4f). Surprisingly, we noticed an increase in Ki67, Ccnd1 and Ccnd2 (figure 4f) but because the number of Tom^High cells was not increased in PNX versus Sham mice, the significance of these results is not clear. We also carried out the labelling of proliferative cells using a single dose of intraperitoneal 5-ethynyl-2-deoxyuridine (EdU) (0.1 mg·g⁻¹ of mouse) at day 6 following PNX. We analysed the lungs at day 7 (1-day chase period, n=2 mice) and day 14 (8-day chase period, n=3 mice). AT2 PD-L1^Pos (equivalent to the Tom^Low) and AT2 PD-L1^Neg (equivalent to the Tom^High) were isolated by FACS. We performed cytospins with these isolated cells and carried out immunofluorescence for EdU. While it was easy to observe numerous AT2 PD-L1^Neg cells, the number of AT2 PD-L1^Pos cells on the slides was very low, making it difficult to analyse EdU incorporation in these cells. Our results indicated that 2.3±0.8% (n=2 independent mice, 300 cells counted per mice) of the AT2 PD-L1^Neg were EdU^Pos at day 7. However, at day 14, 12.2±0.4% were EdU^Pos (n=3 mice, 300 cells per mouse). The increase in the number of labelled cells between these two time points, after an 8-day chase period, suggests that the EdU^Pos cells in the AT2 PD-L1^Neg pool are coming from the AT2 PD-L1^Pos cells (data not shown).

FIGURE 4.

Expansion of tdTomato^Low (Tom^Low) but not tdTomato^High (Tom^High) following pneumonectomy (PNX). a) Schematic representation of the PNX and Sham models. b) Representative flow cytometry analysis of Tom^Low and Tom^High populations 7 days after PNX and Sham, and quantification of c) Tom^High and e) Tom^Low percentages in EPCAM^Pos population between Sham and PNX groups (n=4). f) Quantitative PCR (qPCR) analysis of fluorescence-activated cell sorting (FACS)-based sorted Tom^Low population for expression of Fgfr2b, Etv5, Sftpc, Ki67, Ccnd1 and Ccnd2. g) qPCR analysis of FACS-based sorted Tom^High population for expression of Fgfr2b, Etv5, Sftpc, Ki67, Ccnd1 and Ccnd2. Data are presented as mean±sem. dCT: delta cycle threshold. *: p<0.05; **: p<0.01; ***: p<0.001.

Loss of Tom^High cells leads to expansion of Tom^Low cells

We made use of PCLS to follow the fate of Tom^Pos cells over time in vitro. Our flow cytometry results indicate that this approach leads to the drastic loss of Tom^High cells (figure 5a, b). Therefore, we took advantage of this system to monitor the fate of Tom^Low cells over time after a massive loss of Tom^High. Over a culture period of 96 h, we observed the formation of cell clusters and a significant increase in tdTomato intensity (figure 5c), suggesting the differentiation of Tom^Low cells towards mature (Tom^High) AT2 cells. In addition, we observed the presence of clusters of Tom^Pos cells at 240 h (figure 5c). We also performed video-fluorescence microscopy to track individual Tom^Low cells (n=6) in PCLS over a period of 45 h. Our results indicated that the intensity of the tdTomato signal in these cells progressively increased with time, thereby confirming the transition from Tom^Low to Tom^High status (data not shown). Finally, we analysed apoptosis and proliferation in these PCLS over time. Our results indicated a very significant increase in tdTomato^PosTUNEL^Pos over total tdTomato^Pos cells (figure 5d–k) in PCLS at 20 h, which is in line with the lack of detection of Tom^High cells by FACS at that time point (figure 5b). In addition, proliferation analysis indicated a robust increase in the number of tdTomato^PosEdU^Pos over total tdTomato^Pos cells in PCLS between 20 h and 96 h (figure 5h–k), supporting our previous conclusion in the context of PNX that the Tom^Low cells are proliferating after injury.

FIGURE 5.

Characterisation of the fate of tdTomato^Low (Tom^Low) cells in precision-cut lung slices (PCLS). a, b) Representative flow cytometry analysis of Tom^Low and tdTomato^High (Tom^High) before processing the lungs for PCLS in freshly generated PCLS (a) and PCLS after 20 h culture (b). c) Visualisation of the Tom^Low in PCLS at t=0, t=96 h and t=240 h. Arrows indicate the formation of cell clusters. Scale bars: 250 μm (low magnification) and 50 μm (high magnification). d) Analysis of apoptosis with representative terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) immunofluorescence staining on the PCLS (scale bar: 50 μm) and quantification of e) tdTomato^Pos TUNEL^Pos cells out of total cells, f) TUNEL^Pos cells out of total cells and g) tdTomato^Pos TUNEL^Pos cells out of tdTomato^Pos cells (n=4). Data are presented as mean±sem. h) Analysis of proliferation with representative EdU immunofluorescence staining on the PCLS (scale bar: 50 μm) and quantification of i) tdTomato^Pos EdU^Pos cells out of total cells, j) EdU^Pos cells out of total cells and k) tdTomato^Pos EdU^Pos cells out of tdTomato^Pos cells (n=4). Data are presented as mean±sem. *: p<0.05; **: p<0.01; ***: p<0.001; ****: p<0.0001.

PD-L1 is a specific surface marker enriched in Tom^Low

To identify markers differentially expressed between Tom^Low and Tom^High, we performed a gene array on FACS-isolated Tom^Low and Tom^High cells (figure 6a, supplementary figure S5b). Among the 100 top genes differentially expressed in Tom^Low versus Tom^High cells, three cell surface markers, Cd33, Cd300lf and Cd274 (also known as Pd-l1), were identified as being enriched in Tom^Low compared to Tom^High. qPCR analysis confirmed the significantly higher expression of Cd33 and Pd-l1 in Tom^Low compared to Tom^High (figure 6b). PD-L1 is an interesting marker because it is an immune inhibitory receptor ligand and its expression is highly increased in adenocarcinoma [14, 25]. The use of this marker is also relevant because our ATAC-seq analysis revealed that Tom^Low cells are enriched in genes belonging to the immune system (supplementary figure S5a). The use of PD-L1 as a surface marker enriched in Tom^Low cells was further validated using PD-L1 immunofluorescence staining and flow cytometry. To this end, PD-L1 immunofluorescence staining on cytospins confirmed a higher level of protein at the plasma membrane in Tom^Low than in Tom^High cells (figure 6c). Moreover, flow cytometry analysis of Tom^Low and Tom^High populations separately showed that 46.9% of Tom^Low cells were PD-L1^Pos while only 0.77% of Tom^High cells were PD-L1^Pos (figure 6d). In addition, PD-L1 immunofluorescence staining on lung sections localise tdTomato^Pos PD-L1^Pos cells in the alveoli (figure 6e). In conclusion, these results suggest that PD-L1 antibodies could be instrumental in differentially isolating the equivalent of Tom^Low versus Tom^High in wild-type lungs.

FIGURE 6.

PD-L1 is a specific surface marker enriched in tdTomato^Low (Tom^Low) cells. a) Validation of gene array data by quantitative PCR (qPCR). d: day. b, c) Quantification of the expression levels of Cd33 (b) and Pd-l1 (c) in Tom^Low compared to tdTomato^High (Tom^High). Data are presented as mean±sem. dCT: delta cycle threshold. ***: p<0.001. d) Representative PD-L1 immunofluorescence staining on Tom^Low and Tom^High cytospin cells. Scale bar: 50 μm. e) Representative flow cytometry analysis of PD-L1^Pos population in Tom^Low and Tom^High. f) Representative PD-L1 immunofluorescence staining on the lung sections. Scale bar: 10 μm.

Identification of AT2 PD-L1^Pos SFTPC^Low in the lungs of wild-type mice

To address whether PD-L1 could be used to isolate the equivalent of Tom^Low and Tom^High without the need for a lineage tracing approach, FACS-based analysis was performed on isolated C57BL/6 lungs. The PD-L1 antibody was initially validated to show its expression in alveolar macrophages by immunofluorescence staining (in a lower magnification image of figure 6e) and flow cytometry (supplementary figure S6c). AT2 cell selection was based on the gating of EPCAM^Pos, CD49f^Intermediate (to label the alveolar epithelial cells), PDPN^Neg population (to exclude the AT1 cells), from which the percentages of PD-L1 positive and negative cells were analysed (figure 7a). On average, 8.99±0.51% (n=4) of these AT2 cells were PD-L1^Pos and 90.23±0.58% (n=4) were PD-L1^Neg (figure 7a). This ratio is consistent with the finding that most (80%) of the lineage-traced AT2 cells were composed of Tom^High (PD-L1^Neg) cells (figure 1c). qPCR analysis on sorted PD-L1^Pos and PD-L1^Neg AT2 cells indicated a higher level of Fgfr2b, Etv5 and Sftpc in PD-L1^Neg than PD-L1^Pos cells (figure 7b–d). This result is also in line with PD-L1^Neg cells corresponding to Tom^High cells, while the expression profile of PD-L1^High fits with these cells being the equivalent of the Tom^Low. Finally, flow cytometry analysis of alveolar epithelial cells (EPCAM^Pos, CD49f^inter) stained with SFTPC and PD-L1 identified a subpopulation of AT2 cells (6.8%) displaying a low level of SFTPC and a high level of PD-L1 (figure 7e). This SFTPC^Low PD-L1^High cluster likely contains the Tom^Low cells.

FIGURE 7.

Identification of AT2 PD-L1^Pos cells in the lungs of wild-type mice. a) Representative flow cytometry analysis of wild-type mice lungs shows the gating strategy of EPCAM^Pos CD49f^Inter population, followed by negative selection of AT2 cells with the exclusion of AT1 PDPN^Pos cells. The third plot is a representative flow cytometry analysis of AT2 cells based on the use of the PD-L1 marker. b–d) Quantitative PCR analysis of fluorescence-activated cell sorting (FACS)-based sorted PD-L1^Pos and PD-L1^Neg cells for expression of Fgfr2b (b), Etv5 (c) and Sftpc (d) (n=4). Data are presented as mean±sem. **: p<0.01; ***: p<0.001; ****: p<0.0001. e) Representative flow cytometry analysis of SFTPC and PD-L1 co-staining. FSC: forward scatter; dCT: delta cycle threshold.

qPCR analysis was carried out to validate the potential contamination of isolated AT2 PD-L1^Pos and AT2 PD-L1^Neg cells. Our results confirm the lack of contamination of these two subpopulations by endothelial, hematopoietic, macrophage and fibroblast cells (supplementary figure S6). In addition, co-culturing these cells with resident SCA1^Pos mesenchymal cells in alveolosphere assay indicated that only AT2 PD-L1^Neg cells (the equivalent of Tom^High cells) display self-renewal capabilities (supplementary figure S7). Finally, there were also higher numbers of AT2 PD-L1^Pos cells in PNX versus Sham (supplementary figure S8). Altogether, our data indicate that the AT2 PD-L1^Pos and AT2 PD-L1^Neg cells are the functional equivalent of Tom^Low and Tom^High, respectively.

Identification of PD-L1^Pos cells in a human AT2 scRNA-seq data set

The scRNA-seq results from a recent study reporting the global analysis of human lung cells [17] were used to further examine the presence of PD-L1^Pos cells in the human AT2 pool. Bioinformatic analysis using Uniform Manifold Approximation and Projection (UMAP) plots was carried out to improve the initial AT2 cell pool clustering. Our results allowed the initial AT2 cluster composed of 3198 cells to be separated into five sub-clusters (figure 8a). Analysis of PD-L1 expression distinguished two positive clusters, named AT2 PD-L1^High and AT2-proliferative, displaying high levels of PD-L1 (figure 8b, green and lavender arrows, respectively). Note that the AT2-proliferative cluster had higher expression of MIK67 and PCNA (supplementary figure S9). However, because the AT2-proliferative cluster comprised only three cells, its relevance to our Tom^Low population identified in mice is not clear. The other sub-cluster (AT2 PD-L1^High) comprised 186 cells (5.82% of the initial AT2 pool). An analysis of gene expression is represented using a violin plot (figure 8b–d). Compared to the AT2 sub-cluster (labelled in red), PD-L1 expression in AT2 PD-L1^High (labelled in green) was clearly increased (figure 8b). In addition, FGFR2b expression was not changed while SFTPC was only slightly decreased. However, lower expression of AXIN2, SFTPA1 and ETV5 was observed (figure 8d). It is therefore likely that the AT2 PD-L1^High sub-cluster is the human equivalent of the AT2 Tom^Low that we identified in mice. Interestingly, compared to the AT2 sub-cluster, the AT2-proliferative sub-cluster also displayed lower levels of ETV5, SFTPC and SFTPA1 but a higher level of AXIN2. The genes enriched in AT2 PD-L1^High were also analysed (supplementary figure S10). Transmembrane 4L6 family member 1 (TM4SF1), a gene encoding a membrane protein, was highly enriched in AT2 PD-L1^High cells (figure 8d). Interestingly, this marker was used previously to isolate AEP cells, which are proposed to express high level of AXIN2 in the human lung [9]. It is clear that this marker is also expressed in AT2 AXIN2^Low PD-L1^High human cells (as well as in Tom^Low versus Tom^High in mice, data not shown), illustrating that more work is needed to characterise human and corresponding mouse progenitor cells and to identify more markers. Finally, we used the online resource of the IPF Cell Atlas to add further evidence for the presence of AT2 PD-L1^Pos cells in human lungs (data not shown). CD274^Pos (PD-L1^Pos) cells were observed in all four human data sets analysed (Banovitch/Kropski, Kaminski/Rosas, Lafyatis and Misharin). The level of expression of ETV5, FGFR2 and SFTPC in AT2 PD-L1^Pos cells in these data sets fits with either their quiescence (low expression level) or activation (high expression level) (data not shown).

FIGURE 8.

Identification of PD-L1^Pos cells in human AT2 single-cell RNA-sequencing (scRNA-seq) data set. a) UMAP plot of the initial AT2 cell cluster indicates the presence of five different AT2 sub-clusters. b) UMAP and c) violin plots showing PD-L1 enrichment in the AT2-PD-L1^High sub-cluster. d) UMAP and violin plots showing that the AT2-PD-L1^High sub-cluster displays low levels of ETV5, SFTPC and AXIN2 and high levels of TM4SF1. Green arrows indicate AT2 PD-L1^High and lavender arrows indicate AT2-proliferative. e) Model for tdTomato^Low (Tom^Low) and tdTomato^High (Tom^High) cells in homeostasis and after pneumonectomy/injury. Tom^Low cells are quiescent in the adult lung during homeostasis but are activated after injury. They acquire proliferative capabilities and differentiate into mature AT2 cells.

Discussion

We have identified a subpopulation of AT2 progenitor cells that is distinct from already identified mature AT2 progenitor subpopulations. Tom^Low cells are quiescent and immature cells in the steady state and, unlike AT2 AXIN2^Pos cells, express low levels of AT2 differentiation markers. Moreover, Tom^Low cells express significantly lower levels of Axin2 than Tom^High cells in homeostasis, and Axin2 expression in Tom^Low cells showed no change after PNX compared to the Sham condition (supplementary figure S11a, b). Zacharias et al. [9] showed that AEPs express higher level of genes related to lung and tube development than mature AT2s. However, based on our gene array data, these genes are expressed at lower levels in Tom^Low compared with Tom^High cells (data not shown). This indicates that AEP cells are part of the Tom^High population. The newly identified Tom^Low cells are also different from the integrin α6β4 population because these cells are negative for Sftpc and, therefore, the Sftpc^CreERT2 driver line cannot lineage trace this population [7, 26]. Finally, the Tom^Low population most likely does not contain bronchoalveolar stem cells (BASCs) for several reasons. First, most of the Tom^Low cells are located in the respiratory epithelium and do not display high levels of Scgb1a1 (data not shown). Second, scRNA-seq data indicate that, compared to AT2 cells, BASCs express a similar level of Sftpc and Fgfr2 [13], suggesting that BASCs are likely contained in the Tom^High population. Interestingly, lineage tracing of BASCs upon injury indicates that these cells are not the sole contributors to newly formed bronchial airway cells after conducting airway injury or AT2/AT1 cells after alveolar injury, demonstrating that other resident stem cells such as the AT2 Tom^Low or the AT2 could contribute to the repair process. Furthermore, PD-L1 is highly expressed in Tom^Low. PD-L1 expression is also increased in adenocarcinoma and appears to be correlated with elevated tumour proliferation and aggressiveness [14, 16, 25, 27, 28]. Further investigations are required to elucidate PD-L1 function in Tom^Low cells and their potential interaction with the immune system as well as their contribution to cancer. In the future, it will be important to design dual lineage tracing strategies based on the expression of Sftpc and Pd-l1 to label the Tom^Low cells specifically.

A scRNA-seq data set examining the status of lung cells in adult mice at different time points following bleomycin injury was recently published [29]. Data mining of these scRNA-seq results indicated that AT2 Pd-l1^Pos cells could not be detected either in control (saline) or experimental (bleomycin) conditions. Because the design of the experiment was to capture all the cells in the lung, the corresponding epithelial portion was clearly underpowered because not enough subpopulations of AT2 cells were available for a robust analysis. This lack of detection of AT2 Pd-l1^Pos cells was also likely due to the trauma associated with the enzymatic digestion of the lung before the scRNA-seq procedure and is supported by our observation that Tom^Low cells had much lower viability compared to the Tom^High cells (data not shown). Preliminary experiments on bleomycin-treated (n=2 mice) versus saline-treated (n=3 mice) wild-type mice indicated a robust increase in AT2 PD-L1^Pos cells compared to AT2 PD-L1^Neg cells (21.2% versus 8.1%; data not shown). We, therefore, conclude that AT2 PD-L1^Pos (Tom^Low) cells appear to respond to at least three types of injury (PNX, bleomycin and PCLS).

An important question that we attempted to answer in this study is related to the potential role played by Tom^Low cells as progenitors for mature AT2 cells. The arguments for such a role are as follow: 1) Tom^Low cells represent a stable population over time (supplementary figure S4); 2) analysis of apoptosis and proliferation of the Tomato labelled cells in the PCLS model (figure 5) shows that Tom^High cells are massively dying while Tom^Low cells are proliferating; 3) there was an increase in the Tom^Low (PD-L1^Pos) pool in the context of PNX versus Sham (figure 4 and supplementary figure S8) without a significant change in the number of Tom^High cells; 4) while Tom^Low cells from non-injured lung indicate that these cells are not capable of self-renewal and differentiation (figure 3), our preliminary data (not shown) indicate that Tom^Low cells from Sftpc^CreERT2/+;Fgfr2b^flox/flox;Tomato^flox/flox lungs display a significant increase in alveolosphere formation, indicating that Tom^Low cells acquire self-renewal and differentiation capabilities.

In conclusion, we have identified a novel population of quiescent and immature AT2 progenitor cells with a different gene expression profile from mature AT2 cells that proliferate after PNX and are enriched in Pd-l1 expression. The equivalent of this population was also identified in human lungs. Further characterisation of these cells in homeostatic and repair/regeneration conditions will allow the signalling pathways activated in these cells to be identified, with the ultimate goal of enhancing repair after injury.

Supplementary material

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Supplementary methods and tables ERJ-04168-2020.Supplement ^{(112.6KB, pdf)}

Supplementary figure S1. Absence of contaminations in Tom^Low and Tom^High subpopulations. a) Timeline of tamoxifen treatment of Sftpc^CreERT2/+; tdTomato^flox/flox mice. Representative flow cytometry analysis of EPCAM^Pos and CD3^Pos, CD45^Pos populations selection and further gating of Tom^Neg Tom^Low and Tom^High cells. b) qPCR analysis of FACS-based sorted CD31^Pos, CD45^Pos, Tom^Neg, Tom^Low, and Tom^High cells for the expression of Pecam1, Ptprc, Adgre1 and Vim. dCT: delta cycle threshold. Data are presented as mean±SEM. *: p<0.05; **: p<0.01; ***: p<0.001. ERJ-04168-2020.Figure_S1 ^{(315.9KB, jpg)}

Supplementary figure S2. Validation of Sftpc^CreERT2/+; tdTomato^flox/flox mice. a) The pie chart represents the percentage of tdTomato^Pos cells in total cells (20.5%) and the percentage of epithelial (96.6%), mesenchymal cells (0.8%), and CD31^PosCD45^Pos cells (2.6%) of the labeled cells. b) Representative SFTPC immunofluorescence staining and quantification of tdTomato^PosSFTPC^Pos of total tdTomato^Pos as well as quantification of tdTomato^PosSFTPC^Pos of total SFTPC^Pos (n=4). Data are presented as mean±SEM. c) Representative flow cytometry analysis of EPCAM^Pos and display of EPCAM and tdTomato in a single FACS plot. ERJ-04168-2020.Figure_S2 ^{(241.4KB, jpg)}

Supplementary figure S3. AT2 cells labeling in the absence or in the presence of tamoxifen treatment. a) Timeline of tamoxifen treatment. One group of Sftpc^CreERT2/+; tdTomato^flox/flox mice were treated with tamoxifen in drinking water for 7 days and another group with no tamoxifen added to the water. b) Representative flow cytometry analysis of single-cell selection and further analysis of EPCAM^Pos population in both groups. c) Representative flow cytometry analysis of Tom^Low (4.5%) and Tom^High (6.9%) in mice in the absence of tamoxifen. The pie chart shows the percentage of Tom^Low (39.2%) and Tom^High (60.8%) in total tdTomato positive cells. d) Representative flow cytometry analysis of Tom^Low (10.3%) and Tom^High (44.1%) in mice treated with tamoxifen. Pie chart shows the percentage of Tom^Low (18.9%) and Tom^High (81.1%) in total tdTomato positive cells. ERJ-04168-2020.Figure_S3 ^{(322.4KB, jpg)}

Supplementary figure S4. Tom^Low and Tom^High AT2 subpopulations in 6- and 9-month old mice. Timeline of tamoxifen treatment of Sftpc^CreERT2/+; tdTomato^flox/flox mice and lung harvest. Flow cytometry analysis of EPCAM^Pos cells. Quantification of the percentile of Tom^Low and Tom^High (over total EPCAM^Pos cells) in 6 and 9 months old mice. ERJ-04168-2020.Figure_S4 ^{(219.1KB, jpg)}

Supplementary figure S5. Gene set enrichment for accessible regions in Tom^High versus Tom^Low. a) Analysis of peaks obtained in the ATAC-seq experiment for Tom^High and Tom^Low cells using Kobas for the Reactome database. Peaks overlapping gene body or close to the transcription starting site of genes were annotated to the corresponding genes. All annotated peaks were split into lists of genes that display more open chromatin in Tom^High or Tom^Low cells using DESeq2 on unified peak regions. Observed significance was adjusted by Benjamini-Hochberg correction for multiple tests (FDR). The resulting lists were used as input for Kobas to search for enriched terms in different databases. Top 10 terms were chosen by significance (FDR <0.2). Results indicate that the term immune system is highly enriched in Tom^Low cells, indicating that the chromatin of Tom^Low cells is more accessible in loci of genes (gene body or promoter) associated with the immune system. Higher accessibility is associated with more transcriptional activity. Numbers in brackets display number of identified genes/total the number of genes for a term in the database. DEG: differentially expressed genes. Between brackets: Genes found/total genes in the term. b) Heatmap for differentially expressed genes following a gene array using Tom^High and Tom^Low cells isolated from Sftpc^CreERT2/+; tdTomato^flox/flox lungs at homeostasis. ERJ-04168-2020.Figure_S5 ^{(316.7KB, jpg)}

Supplementary figure S6. Absence of contaminations in PD-L1^Pos AT2 cells. a) Representative flow cytometry analysis of CD31^PosCD45^Pos (endothelial and haematopoietic) and EPCAM^NegCD31^NegCD45^Neg (mesenchyme) populations as well as further gating of EPCAM^Pos cells with PDPN to isolate AT2 cells. Negative control for PD-L1 antibody staining to select PD-L1^Pos cells. Isolation of PD-L1^Pos and PD-L1^Neg out of AT2 cells. b) qPCR analysis of FACS-based sorted mesenchymal cells, CD31^PosCD45^Pos cells, PD-L1^Pos and PD-L1^Neg AT2 cells for the expression of Pecam1, Ptprc, Adgre1 and Vim. Data are presented as mean±SEM. dCT: delta cycle threshold (Ct). *: p<0.05; **: p<0.01; ***: p <0.001. c) Representative flow cytometry analysis of EPCAM^Neg and a further selection of CD31^PosCD45^Pos populations selection. Further flow cytometry analysis shows that CD31^PosCD45^Pos populations contain PD-L1^Pos cells. Negative control for PD-L1 antibody staining. ERJ-04168-2020.Figure_S6 ^{(364.3KB, jpg)}

Supplementary figure S7. Colony-forming capabilities of PD-L1^Pos and PD-L1^Neg AT2 cells. a) Representative Lysotracker immunofluorescence staining of alveolospheres from AT2 PD-L1^Pos and AT2 PD-L1^Neg cells after 14 days in culture (Scale bar: 100 μm). b) Quantification of colony-forming efficiency (CFE) of PD-L1^Pos and PD-L1^Neg cultured cells (n=3). Data are presented as mean±SEM. *: p<0.05; **: p<0.01; ***: p <0.001. ERJ-04168-2020.Figure_S7 ^{(91.6KB, jpg)}

Supplementary figure S8. Elevated percentage of PD-L1^Pos AT2 cells following pneumonectomy. a) Timeline of lung harvest after Sham or PNX. Representative flow cytometry analysis of wild-type mice lungs showing the gating strategy of EPCAM^Pos population, followed by negative selection of AT2 cells with the exclusion of AT1 PDPN^Pos cells. b) Representative flow cytometry analysis of AT2 cells based on PD-L1 marker and negative control of PD-L1 antibody staining. c) Quantification of PD-L1^Pos AT2 cells percentages between Sham and PNX groups (n=4). Data are presented as mean±SEM. *: p<0.05; **: p<0.01; ***: p <0.001. ERJ-04168-2020.Figure_S8 ^{(323.8KB, jpg)}

Supplementary figure S9. ScRNA-seq analysis of the AT2-proliferative subcluster. a) UMAP plot of the initial AT2 cell cluster indicates the presence of five different AT2 subclusters. b) UMAP plot representation showing PCNA and MKI67 enrichment in the AT2-proliferative subcluster. c) Corresponding Violin plot representation. ERJ-04168-2020.Figure_S9 ^{(767.3KB, jpg)}

Supplementary figure S10. Heat map representation of the differences in gene expression (using LFC) between the different AT2 subclusters. a) Heat map from the scRNA-seq analysis of AT2 cells in the adult human lung. b) Note the enrichment in AT2 PD-L1^High of genes involved in the immune system (CXCL1, CXCL2, CXCL3, ICAM1, ZC3H12A, IRF1, CXCL8, XBP1) as well as a gene (TM4SF1) previously described to be expressed by AEP cells. ERJ-04168-2020.Figure_S10 ^{(717.7KB, jpg)}

Supplementary figure S11. Axin2 expression analysis in Tom^Low versus Tom^High in homeostasis and in Tom^Low in PNX and Sham. a) qPCR analysis of FACS-based sorted Tom^Low and Tom^High cells for the expression of Axin2 in adult mouse lungs. b) qPCR analysis of FACS-based sorted Tom^Low cells for the expression of Axin2 from sham and PNX lungs at day 7 post-surgery. Data are presented as mean±SEM. *: p<0.05; **: p<0.01; ***: p <0.001. ERJ-04168-2020.Figure_S11 ^{(185.4KB, jpg)}

Shareable PDF

This one-page PDF can be shared freely online.

Shareable PDF ERJ-04168-2020.Shareable ^{(436.2KB, pdf)}