Fraction of MHCII and EpCAM expression characterizes distal lung epithelial cells for alveolar type 2 cell isolation

Nine-week-old C57BL/6 J mice (Charles River, Japan), BALB/c mice (SLC, Japan), FVB/N mice (CLEA, Japan), and A/J mice (SLC, Japan) were purchased for use in this study. Double-transgenic Scgb1a1-rtTA (Line 1)/(tetO)⁷CMV-Cre mice (a gift from Dr. Machiko Ikegami and Jeffrey A. Whitsett) [22] were bred with ROSA^mT/mG mice (the Jackson Laboratory) to generate triple-transgenic Scgb1a1-rtTA/(tetO)⁷CMV-Cre/ROSA^mT/mG mice. Doxycycline (600 ppm) was added to the chow starting from 5 weeks old to 8 weeks old to activate cre-mediated recombination in the triple-transgenic mice.

An overview flowchart of cell isolation protocol with estimated time and the yield of cells are depicted in Additional file 1: Figure S1. Single lung cells were obtained using Corti’s protocol [13] with some modifications. Briefly, mice were anesthetized via intraperitoneal injection of pentobarbital with 10 U of heparin. The abdominal cavity was opened, and mice were exsanguinated via transection of the abdominal aorta. The trachea was exposed and intubated with a 20-gauge catheter (Terumo, Tokyo, Japan). The thoracic cavity was opened, and the lungs were perfused with 10 mL of cold 0.9% saline from the right ventricle. In total, 1.5 mL of dispase solution (Corning, Corning, NY) was instilled into the lungs from the intubated catheter, followed by 0.2 mL of 1% low melting point (LMP) agarose (Wako Pure Chemical Industries, Osaka, Japan). The lungs were immediately covered with ice and incubated for 2 min. After incubation, the lungs were isolated from the thoracic cavity and incubated for 45 min in 5 mL of Hank’s Balanced Salt Solution (HBSS) (Wako Pure Chemical Industries) at room temperature (RT). Then, the lungs were minced using scissors and a transfer pipette in Dulbecco’s Modified Eagle Medium (DMEM) (Nacalai Tesque, Kyoto, Japan) containing 25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) (Life Technologies, Gaithersburg, MD), antibiotics, antimycotic solution (Life Technologies). After mincing, 100 μL of 1 mg/mL DNase I (Sigma-Aldrich, St. Louis, MO) was added to the sample and incubated for 4 min at RT. The cells were sequentially filtered through 70-μm and 40-μm filters (BD Biosciences, San Jose, CA). Then, the cell suspension was centrifuged for 7 min at 340×g and washed once with Phosphate buffered saline (PBS) (Nacalai Tesque). Next, the cell pellet was suspended with 3 mL of lysis buffer (eBioscience, San Diego, CA), and incubated for 2 min at RT, followed by washing with PBS. The cell pellet was resuspended in PBS containing 3% fetal bovine serum (FBS) for cell isolation or in staining buffer (PBS/0.1% bovine serum albumin (BSA)/0.02% NaN₃/0.5 mM ethylenediaminetetraacetic acid (EDTA)) for cell analysis. The average number of cells in the single-cell lung suspensions was 9.0 ± 0.9 × 10⁶ /lung, and the viability was 83.0 ± 1.2% (n = 8). For the 3-week-old mice, 22-gauge catheters were used for the tracheal intubation, and the instilled volume of dispase solution and 1% LMP agarose were reduced to 1 mL and 0.1 mL, respectively.

A list of antibodies used for flow cytometry is shown in Additional file 2: Table S1. For cell-surface antigen staining, the samples were incubated with antibodies for 20 min at 4 °C, washed and resuspended in PBS containing 3% FBS. To exclude dead cells, 7-amino actinomycin D (7-AAD, BD Biosciences) was added to the samples before sorting. We sorted live, single cells using a FACSAria III Cell Sorter with FACSDiva ver. 8.0.1 (BD Biosciences). Sorted cells were collected in DMEM containing 10% FBS, antibiotics, antimycotic solution, and 25 mM HEPES for further analyses. For proSP-C staining, single cell suspension of the lung or the sorted cells were incubated with fixable viability dye eFluor780 (eBioscience) and surface antigens, fixed, and permeabilized with fixation and permeabilization buffer (eBioscience) according to the manufacturer’s instructions. Permeabilized cells were incubated with anti-mouse proSP-C antibody or rabbit IgG (Santa Cruz Biotechnology, Santa Cruz, CA) and labeled with PE-conjugated F(ab’)2 donkey anti-rabbit IgG (1:200; eBioscience). Fixed cells were analyzed using a BD LSR Fortessa (BD Biosciences). Appropriate isotype control samples were utilized for all FACS analyses. The data were analyzed using FlowJo software (ver. 7.6.5, Tree Star, San Carlos, CA).

Mice were anesthetized with isoflurane and hung upright at a 45-degree angle. LPS from E. coli (Sigma-Aldrich, St. Louis, MO) (1 mg/kg body weight in 100 μL of PBS) or PBS (control) was aspirated intratracheally as reported previously [23]. The mice were sacrificed at 24 h after intratracheal instillation for further analyses.

Methods for immunofluorescence and RT-PCR analyses are provided in the online Data Supplement.

The values are expressed as the means ± SEM. Statistical analyses were performed using JMP ver. 10 (SAS Institute, Cary, NC). Comparisons between two groups were performed using the Wilcoxon rank sum test.

To demonstrate the localization of MHCII in adult murine lungs, we analyzed MHCII expression by immunofluorescence. As shown in Fig. 1a, proSP-C⁺ AT2 cells also expressed MHCII, while AT1 cells were negative for MHCII. In the alveoli, alveolar macrophages were also positive for MHCII expression. All proSP-C⁺ cells were positive for MHCII expression.

To investigate MHCII expression in AT2 cells further, we performed FACS analysis of component cells of murine lungs. Single-cell suspensions obtained from enzymatically digested lungs were stained for surface antigens, fixed, permeabilized, and then stained for proSP-C. Using FACS analysis, CD45⁻CD31⁻EpCAM⁺ cells (henceforth, EpCAM⁺ cells) were analyzed for proSP-C expression (Fig. 1b). Among EpCAM⁺ cells, 90.4% ± 1.7% were positive for proSP-C expression, and almost all proSP-C⁺ cells expressed MHCII (99.0 ± 0.2%) (Fig. 1c). In contrast, the majority of proSP-C⁻ EpCAM⁺ cells were negative for MHCII expression (Fig. 1d). This observation suggests that EpCAM⁺ cells from enzymatically digested murine lungs primarily consist of AT2 cells but also contain a substantial amount of proSP-C⁻ epithelial cells. Thus, MHCII could be a useful surface marker for classifying lung epithelial cells to identify AT2 cells.

In the two-dimensional plot of EpCAM and MHCII, EpCAM⁺ cells were classified into 3 different subpopulations based on EpCAM and MHCII expression: EpCAM^medMHCII⁺ cells (Population 1; P1 cells), EpCAM^hiMHCII⁻ cells (P2 cells), and EpCAM^lowMHCII⁻ cells (P3 cells) (Fig. 1e). Most P1 cells were positive for proSP-C expression (97.8 ± 0.4%), while P2 and P3 cells were negative for proSP-C expression (Fig. 1f and g). To evaluate the efficiency of the gating strategy for AT2 cell identification, proSP-C⁺ cells were back-gated in the plot of EpCAM and MHCII, demonstrating that 97.6 ± 0.3% of the cells were in the P1 gate (Fig. 1h).

During AT2 cell isolation, we evaluated whether P1 cell isolation was superior to EpCAM⁺ cell isolation in terms of purity. Live single cells from enzymatically digested lungs were stained for surface antigens. EpCAM⁺ cells were classified into 3 subpopulations as shown in the fixed cell analysis (P1, P2, and P3) (Fig. 2a and Additional file 3: Figure S2). The average yield of P1 cells was 6.2 ± 0.7 × 10⁵ /lung (n = 8), and the viability was 89.0 ± 1.3% when assessed with fixable viability dye staining. EpCAM⁺ cells were also sorted (Fig. 2b), and the purity of both sorted cells was evaluated by immunofluorescence and FACS. The average yield of P2 and P3 cells were 1.8 ± 0.1 × 10⁴ and 2.3 ± 0.2 × 10⁴ /lung, respectively (n = 3).

Immunofluorescence analysis of cytospin preparations of sorted cells demonstrated that the proportion of proSP-C⁺ cells in sorted P1 cells was 99.0 ± 0.3%, which was significantly higher than that in sorted EpCAM⁺ cells (94.0 ± 0.8%) (Fig. 2c and e).

For FACS analysis of proSP-C expression, sorted cells were fixed, permeabilized and stained for proSP-C. The purity of sorted P1 cells was 98.0 ± 0.2%, whereas that of sorted EpCAM⁺ cells was 95.4 ± 1.5% (Fig. 2d and f), demonstrating again that the purity of P1 cells was significantly higher.

To characterize P1 and EpCAM⁺ cells sorted by surface markers, we performed mRNA expression analysis to confirm cell specific gene expression. Sftpc (Surfactant Protein C, AT2 cell specific surfactant protein) expression levels were 4.4-fold higher in P1 cells compared to whole lung cells, while the expression of other epithelial lineage markers including Scgb1a1 (Secretoglobin Family 1A Member 1, club cell specific protein), Pdpn (Podoplanin, AT1 cell specific protein among lung epithelial cells), and Foxj1 (Forkhead Box J1, ciliated cell specific nuclear protein) was minimal (Fig. 2g). In contrast, sorted EpCAM⁺ cells displayed Pdpn and Foxj1 expression.

To characterize P2 and P3 cells, we sorted these subpopulations and performed immunofluorescence and mRNA expression analyses. The majority of P2 cells were positive for acetylated tubulin (92.3 ± 2.0%, n = 3), with sparse SCGB1A1⁺ cells and unspecified cells as determined by immunofluorescence analysis (Fig. 2h). Foxj1 expression in sorted P2 cells was 116-fold higher compared to that in whole lung cells (Fig. 2i), suggesting that P2 cells are enriched with ciliated cells. Approximately half of P3 cells, which were negative for major epithelial cell markers including proSP-C, SCGB1A1, acetylated tubulin, and T1α as determined by immunofluorescence analysis, were positive for AQP5 (Fig. 2j). mRNA analysis revealed 3.9-fold higher Pdpn expression (Fig. 2k), suggesting that AT1 cells are the major cell type among P3 cells.

FACS analysis with fluorescent antibody staining is a useful method to explore antigen expression. However, this approach is inevitably accompanied by possible non-specific antibody binding, which we verified using isotype controls. To evaluate the consistency of our gating strategy, we employed AT2 cells that have intrinsic GFP expression. We bred double-transgenic Scgb1a1-rtTA/(tetO)⁷CMV-Cre mice [22] with ROSA^mT/mG mice to generate triple-transgenic Scgb1a1-rtTA/(tetO)⁷CMV-Cre/ROSA^mT/mG mice (Fig. 3a). In these triple-transgenic mice, most AT2 cells and a subset of club cells display enhanced GFP expression following the addition of doxycycline to the chow (Fig. 3b), starting from 5 weeks old to 8 weeks old to minimize GFP labeling in AT1 cells and ciliated cells due to lung maturation after birth [24].

Using single lung cells obtained from these triple-transgenic mice, we analyzed GFP⁺ cells for EpCAM and MHCII expression and found that 97.4 ± 0.3% (n = 4) of GFP⁺ cells were gated with P1 cells (Fig. 3c). Then, we sorted GFP⁺ cells (5.8 ± 0.5 × 10⁵ cells /lung), analyzed proSP-C expression by FACS; the results indicated that most sorted GFP⁺ cells were positive for proSP-C (96.0 ± 0.6%) (Fig. 3d).

Because CD74 has been reported to be a potential surface marker of AT2 cells [25, 26], we evaluated whether CD74 can be utilized for the positive selection of AT2 cells. Live-cell analysis demonstrated that only a small fraction of EpCAM⁺ cells expressed CD74 (Fig. 4a). We further explored CD74 expression among EpCAM⁺ cells using fixed and permeabilized cells stained for proSP-C. When CD74 staining was performed before cell fixation and permeabilization (surface CD74 analysis), proSP-C⁺ cells were found to express CD74 weakly as a whole, with a small fraction of cells showing higher CD74 expression (Fig. 4b), while proSP-C⁻ cells did not express CD74. In contrast, when CD74 staining was performed after fixation and permeabilization (intracellular CD74 analysis), proSP-C⁺ cells displayed much higher CD74 expression compared with isotype controls (Fig. 4c). These observations suggest that epithelial CD74 expression is limited to AT2 cells, although CD74 is not suitable for the positive selection of AT2 cells due to its predominantly intracellular expression. Although why most of the epithelial cells were negative for CD74 in the live-cell analysis is unclear, surface CD74 may be unstable and may be transformed into a soluble form [27] without cell fixation.

To investigate the applicability of our newly developed AT2 cell isolation strategy, we evaluated EpCAM, MHCII, and proSP-C expression levels in single lung cells from different strains and ages of mice (Table 1, Additional file 4: Figure S3).

Mouse MHCII is highly polymorphic [28]. We performed FACS analysis using BALB/c, FVB/N, and A/J mice. In BALB/c and FVB/N mice, EpCAM⁺ cells were similarly classified into 3 subpopulations, and proSP-C⁺ cells were enriched in P1 cells (Table 1) in these mice and in C57BL6/J mice. However, in A/J mice, MHCII expression among proSP-C⁺ cells was negative, resulting in the failed classification of EpCAM⁺ cells, although non-EpCAM⁺ cells expressed MHCII, suggesting that AT2 cells in A/J mice lack MHCII expression.

Next, we evaluated young (3-week-old) and old (1-year-old) mice. In both groups, almost all proSP-C⁺ cells expressed MHCII, and P1 cells identified based on EpCAM and MHCII expression were enriched with proSP-C⁺ cells. These results suggested that MHCII expression does not alter as a function of age or mouse strain, except for A/J mice.

To evaluate whether we can identify subpopulations of distal lung epithelial cells (P1, P2, and P3) during lung injury, we applied our isolation strategy to an LPS-induced lung injury model. LPS stimulation of the lung induced inflammatory cell infiltration in alveolar spaces (Fig. 5a). Single-cell suspensions of LPS model lungs contained higher numbers of cells (2.2 ± 0.2 × 10⁷ per sample) with the same viability (91.6 ± 1.0% by trypan blue exclusion) as control samples, which were prepared in the same manner (8.3 ± 0.8 × 10⁶ per sample and 89.7 ± 1.2%, respectively). FACS analysis indicated that EpCAM and MHCII expression levels were not significantly altered in distal lung epithelial cells following LPS instillation, which resulted in the successful identification of P1 cells in addition to P2 and P3 cells (Fig. 5b, c, and Additional file 5 : Table S2). The yield of P1 cells in LPS model was 4.2 ± 0.3 × 10⁵, which was slightly smaller than that of control samples (5.4 ± 0.3 × 10⁵) due to the lower sorting efficiency in the LPS model, in which the cell suspension contained a greater number of non-AT2 cells. In the LPS model, sorted P1 cells showed 15.6-fold higher Sftpc expression compared with whole-lung cells. In P1 cells, Sftpc expression was downregulated in response to LPS stimulation (Fig. 5d). Cxcl1 and Tnf were upregulated to higher levels in P1 cells following LPS stimulation (Fig. 5e and f) than those observed in whole-lung cell analyses. In P1 cells, Mmp9 was upregulated in response to LPS (Fig. 5h), while Mmp2 and Mmp12 expression was undetected in both control and LPS model cells (Fig. 5g and i). MMP12 is a metalloprotease that is expressed by macrophages, and the lack of MMP12 expression in sorted cells further supports the successful depletion of hematopoietic cells from the inflammatory lung injury model.