Generation of Alveolar Epithelial Spheroids via Isolated Progenitor Cells from Human Pluripotent Stem Cells

1Type II alveolar epithelial cells (AECs) are a major cellular component of the distal lung epithelium, where they secrete pulmonary surfactant and generate type I AECs that cover most of the surface area of the alveoli. The stepwise differentiation of human pluripotent stem cells (hPSCs), including human embryonic stem cells (hESCs) and induced pluripotent stem cells (hiPSCs), into lung epithelial cells would help to elucidate the etiologies of human lung diseases and create novel treatments, and has been reported in both proximal airway cells and distal lung epithelial cells. Currently, however, there are no surface markers that can be used to purify human NKX2-1+ “ventralized” anterior foregut endoderm cells (VAFECs) as alveolar epithelial progenitor cells (AEPCs), although NKX2-1 is an early marker of lung and thyroid development. Here, we report the efficacy of carboxypeptidase M (CPM) as a surface marker of AEPCs for generating type II AECs.

2We hypothesized that identifying a surface marker for NKX2-1+ VAFECs would be helpful for isolating a homogeneous population of AEPCs without establishing NKX2-1 reporter cell lines. We constructed a stepwise protocol to induce hPSCs to AECs (Figure 1A). On day 0, previously established hPSCs were seeded following single-cell enzymatic dissociation, resulting in definitive endodermal cells (DECs) at an efficiency of ≥80% (Figure S1A available online). In step 2, the DECs were differentiated to anterior foregut endodermal cells (AFECs) at an efficiency of ≥88% (Figure S1B). In step 3, the concentrations of all-trans retinoic acid, CHIR99021, and BMP4 were optimized for seven hPSC lines for differentiation into NKX2-1+FOXA2+ cells, attaining an efficiency of 57.0%–77.5% (Figures 1C and 1D; Supplemental Experimental Procedures). In step 4, cells were cultured in medium containing FGF10 for 7 days. In step 5, the cells were differentiated in medium containing dexamethasone, 8-Br-cAMP, 3-isobutyl-1-methylxanthine, and KGF. We confirmed induction of AECs by detecting SFTPB and SFTPC using RT-PCR and double staining SFTPC and SFTPB with NKX2-1 (Figures S1C and S1D). Transcription factors were analyzed by quantitative RT-PCR (qRT-PCR; Figure 1B). SOX17, FOXA2, GATA6, and SOX2 were compatibly changed on day 6 and day 10 as previously described. On day 14, NKX2-1, GATA6, ID2, SOX9, and HOPX levels simultaneously increased. Interestingly, NKX2-1, GATA6, and HOPX levels decreased on day 21 and then increased again on day 25. The levels of other organ lineage markers were found to be limited from day 0 to day 25 (Figure S1E).

3In order to identify candidate markers of VAFECs, we performed a microarray analysis to compare the global gene-expression patterns of AFECs (day 10) and VAFECs (day 14) in 201B7 hiPSCs. CPM and NKX2-1 were remarkably upregulated on day 14 (Figures 1E and S1F). In immunofluorescence (IF) staining, CPM and NKX2-1 increased from day 10 to day 14 (Figure 1F), whereas EPCAM and FOXA2 did not appear to change (Figure S1G). Although CPM was reported to be a marker of type I AECs, only CPM drastically increased on day 14 in a similar pattern to NKX2-1, in contrast to other markers of type I AECs (AQP5 and CAV1) (Figure S1H). On day 25, the various airway markers, including distal lung transcription factors (SOX9 and NKX2-1), type II AEC markers (SFTPB and SFTPC), and a club cell marker (SCGB3A2), were expressed in the CPM+ cells. KRT5, a marker of basal cells, was not expressed in the CPM+ cells (Figure 1G).

4In fetal human lung at 18.5 weeks of gestation, SFTPC and T1α were expressed separately (Figure S1I), while CPM was expressed in NKX2-1+, SFTPC+, and T1α+ cells (Figure 1H), thus indicating that CPM is expressed in both type I and II AECs in the fetus. The sequential expression of CPM was confirmed in NKX2-1+ cells of fetal murine lungs at embryonic day 12.5 (E12.5), E15.5, and E17.5 (Figure 1I). For thyroid lineage cells, which differentiated from NKX2-1+ VAFECs, CPM was found to be negative in both NKX2-1+ cells and PAX8+ epithelial cells, but weakly positive in PECAM+ endothelial cells in the adult human samples (Figure S1J). In E17.5 fetal and adult murine thyroids, CPM was also negative in NKX2-1+ and PAX8+ cells (Figure S1K), suggesting that CPM is a lung-lineage marker of VAFECs.

5As CPM is a membrane-bound surface protein, we performed flow cytometry with anti-EPCAM and anti-CPM antibodies after dissociating VAFECs on day 14 (Figures 2A and S2A). We then sorted EPCAM+CPM+ and EPCAM+CPM− cells and examined the global gene-expression patterns of these two populations using microarrays. We screened 560 probes with a false discovery rate (FDR)-adjusted p value of <0.05 among 54,675 probes. Gene clustering was performed in 336 probes that differed between the EPCAM+CPM+ and EPCAM+CPM− cells with a fold change (FC) cutoff value of 2.0 (Figures 2B and S2B). Of the clustered genes with the highest expression in the EPCAM+CPM+ cells, CPM ranked among the top five probes with a log FC of >6, as expected. Importantly, the log FCs of two probes for NKX2-1 were 4.89 and 4.82, respectively. FOXA1, FOXA2, HOPX, and GATA6 were also included in the list of upregulated genes with log FCs of 3.79, 3.06, 3.61, and 3.29, respectively. Next we sorted the CPM+ cells using a magnet-activated cell sorting (MACS) system to increase the yield, as almost all of the CPM+ cells were EPCAM+ cells (96.7% ± 2.1% of CPM+ cells; Figure 2A). After MACS-based sorting, the proportion of CPM+ cells in three populations (presorting, positive selection, and negative selection) was 63.4% ± 5.8%, 98.8% ± 0.4%, and 34.0% ± 7.8%, respectively, by flow cytometry (Figure 2C). We then evaluated the proportion of positive NKX2-1+ cells among the MACS-sorted CPM+ and CPM− cells using IF staining (93.0% ± 1.0% versus 29.0% ± 1.0%; Figure S2C) and flow cytometry (92.3% ± 0.7% versus 22.2% ± 2.3%; Figure S2D). Because a portion of the CPM+ cells appeared to be sorted according to MACS-based CPM negative selection, we investigated the average proportion of NKX2-1+ cells among the fluorescence-activated cell sorting (FACS)-sorted CPM+ and CPM− cells using IF staining (89.9% ± 0.4% versus 4.5% ± 1.7%; Figure 2D). Following CPM-based purification on day 14, CPM increased significantly from 0.74-fold ± 0.12-fold to 4.94-fold ± 0.51-fold of that observed in the fetal human lung (n = 5), while NKX2-1 increased from 0.41-fold ± 0.10-fold to 1.95-fold ± 0.36-fold (n = 5) on qRT-PCR (Figure 2E). We then applied this method to purify AECs on day 25. CPM, NKX2-1, SFTPA2, SFTPB, SFTPC, DCLAMP, SCGB1A1, and SCGB3A2 were significantly increased in the CPM+ cells (n = 5); however, the level of SFTPC was extremely low compared with that observed in the fetal lung. NGFR, a marker of proximal airway basal stem cells, was significantly decreased in the CPM+ cells (n = 5; Figure 2F).

6In order to investigate whether CPM is a potential surface marker of AEPCs, we generated SFTPC-GFP knockin reporter hPSC lines from 201B7 hiPSCs using BAC-based homologous recombination methods (; Figure 3A; Supplemental Experimental Procedures), as SFTPC is the most specific marker of type II AECs. Following electroporation of the targeting vectors, 12 of 55 G418-resistant clones were found to have a heterozygous deletion of the genomic endogenous SFTPC-coding region (Figure 3B). The pgk-Neo cassette was removed via electroporation of the Cre-expression vector (Figure 3C), and normal karyotypes of the A17-14 and B2-3 clones were confirmed (Figure S3). The genomic copy number was calculated as previously described. The parental 201B7 (data not shown), A17-14, and B2-3 clones have two copies of the SFTPC gene loci, in contrast to the A17-13 clone, in which random transgenic integration is supposed to have occurred, as indicated by three copies of the loci (Figure 3D). No copy-number variation was detected for the B2-3 clone, whereas a copy-number loss at chromosome 16 q23.3 and gain at chromosome 20 p13 were detected for the A17-14 clone (data not shown). Both SFTPC-GFP reporter hPSCs were then differentiated to the end of step 5 and GFP+ and GFP− cells were obtained by FACS after the CPM+ cells were sorted using MACS (Figure 3E). We confirmed the correlation between GFP and SFTPC on RT-PCR (Figure 3F). GFP was detected in SFTPC+, SFTPB+, and NKX2-1+ cells for both clones (Figure 3G).

7We attempted 2D differentiation, reseeding the CPM+ SFTPC-GFP reporter hPSCs purified from VAFECs on day 14 onto Matrigel-coated, 96-well plates. After 14 days of differentiation in step 5 medium, SFTPB became positive in the reseeded CPM+ cells (Figure S2E); however, SFTPC was almost negative (Figure S4D, condition b). We obtained similar results when we sorted and reseeded CPM+ cells on day 23 (Figure S2F). The discrepancy between the expression of SFTPB and SFTPC in developing human lungs was previously reported. Therefore, we hypothesized that some missing factors are important for the coexpression of SFTPB and SFTPC. We then adopted a 3D coculture with fetal human lung fibroblasts (FHLFs) obtained at 17.5 weeks of gestation (Figure 4A). CPM+ cells purified from VAFECs on day 14 and FHLFs were mixed at a ratio of 1:50 and reseeded onto cell inserts. After 10 days of differentiation in step 5 medium, GFP became positive in some spheroids (Figure 4B). The spheroids were subsequently examined with a transmission electron microscope and lamellar-body-like structures were noted (Figure 4C). On hematoxylin-and-eosin staining, cyst-like spheroids consisting of pseudostratified, columnar, or cuboidal cells with dark pink cytoplasm were observed in the CPM+ cell-derived spheroids, whereas small pieces of spheroids consisting of cuboidal cells with clear cytoplasm were noted in the CPM− cell-derived spheroids (Figure S4A). On IF staining, CPM and NKX2-1 were double positive in most CPM+ cell-derived spheroids, while GFP and SFTPC were double positive in some spheroids (Figure 4D). In the CPM− cell-derived spheroids, EPCAM was positive, whereas no CPM+ or NKX2-1+ cells were identified (Figure S4B). SFTPA, SFTPB, SFTPC, and SFTPD (representative markers of type II AECs) were positive in the CPM+ cell-derived spheroids (Figure S4C). AQP5+ cells were adjacent to SFTPC+ cells in some spheroids (Figure 4D). ID2 and SOX9 (markers of differentiation into the distal lung-lineage fate) were positive in some NKX2-1+ and CPM+ cells, respectively (Figure S4C). Next, we trypsinized the cells in 3D structures and determined the proportion of SFTPC-GFP+ cells, detecting 3.82% ± 0.50% cells obtained from the CPM+ cell-derived 3D structures and 0.29% ± 0.03% cells obtained from the CPM− cell-derived structures including fibroblasts (Figure 4E). Excluding the fibroblasts, the ratio of the number of SFTPC-GFP+ cells to that of EPCAM+ cells was calculated to be 9.81% ± 1.81% in the CPM+ cell-derived spheroids and 1.07% ± 0.16% in the CPM− cell-derived spheroids. Almost all of the GFP+ cells sorted by FACS were SFTPC+, whereas the GFP− cells were SFTPC− (Figure 4F). The levels of alveolar markers (SFTPB and SFTPC), rather than club cell markers (SCGB1A1 and SCGB3A2), were significantly elevated following the 3D coculture differentiation of CPM+ cells derived from three hPSC lines (H9 hESCs and parental 201B7 and 604A1 hiPSCs) compared with the 2D differentiation employing the three protocols separately starting on day 14 (Figure 1A; ) and the 3D coculture differentiation of CPM− cells (Figure 4G). Interestingly, the levels of SFTPB and SFTPC were quite low for 585A1 hiPSCs, suggesting that the concentration of retinoic acid required to induce NKX2-1+ VAFECs in step 3 is less important for subsequent differentiation into AECs than the difference in the cell lines or donors. Moreover, the expression of SFTPB and SFTPC was small for the 2D and 3D differentiation of CPM+ cells alone or FHLFs alone (Figure S4D). Finally, other cell-type markers (AQP5 [type I AECs], FOXJ1 [ciliated cells], and AGR2 [goblet cells]) appeared to be elevated in the CPM+ cell-derived structures rather than in the CPM− cell-derived structures, suggesting that cell-type markers other than club-cell markers were expressed in the CPM+ cell-derived spheroids. KRT5 (a basal cell marker, possibly including both airway and esophageal basal cells) was exclusively expressed in the CPM− cell-derived structures. In addition, PAX8 (a thyroid marker), PAX6 (a neuron marker), and the other foregut endodermal lineage cells (FOXN1, ALB, and PDX1) were only minimally or slightly induced following 3D coculture differentiation (Figure S4E).

8In this work, we identified CPM as a surface marker that is expressed in NKX2-1+ VAFECs, including AEPCs, and demonstrated that the CPM+ cell-derived spheroids obtained via 3D coculture differentiation with FHLFs enabled more efficient differentiation to AECs than did 2D differentiation. The gene-expression pattern of CPM in developing lungs has not received significant attention, although in situ hybridization of Cpm in anterior DECs as early as E7.5 in mice has been reported. Our data from IF staining of murine fetal lungs (Figure S1I) also suggest that lineage-tracing studies may provide answers to the following questions: Is Cpm a possible “specific” marker of lung-lineage progenitor cells such as Shh, Id2, and Nkx2-1? What is the relationship between CPM+ cells and bipotent cells that are capable of generating type I and type II AECs? Do CPM+ cells differentiate into type II AECs directly or indirectly via SFTPC+SCGB1A1+ cells? Furthermore, the present study suggests that a 3D microenvironment and coculture with FHLFs are important factors in the differentiation of progenitor cells into AECs rather than club cells. Although maintaining type II AECs in 2D conditions is often difficult, 3D conditions have recently been applied with better outcomes. Therefore, our 3D differentiation protocol appears to be a reasonable approach for maintaining differentiated type II AECs, although methods for expanding such cells for longer periods should be established in the next step.

9The limitations of the present study include the fact that we were unable to demonstrate whether CPM is a more appropriate marker for lung-lineage cells than NKX2-1. Future studies focusing on the possible contribution of NKX2-1−CPM+ cells and/or NKX2-1+CPM− cells to the differentiation of lung epithelial cells may resolve this issue, although we found only two isolatable populations of NKX2-1+CPM+ and NKX2-1−CPM− cells using the present protocol. In addition, we were unable to demonstrate the highest induction efficiency of AECs, as recently described, although we employed a different method for evaluating efficiency using SFTPC-GFP reporter hPSCs. Another limitation is that the functions of the induced AECs remain to be elucidated.

10Nevertheless, the methods applied in the present study to induce and isolate AEPCs using CPM and consequently generate alveolar epithelial spheroids in a stepwise fashion may help to elucidate the complicated differentiation of human AECs and open the door for the development of new strategies for in vitro toxicology and cell replacement therapy, as well as screening for therapeutic drug compounds, in the future.

11CHIR99021 (Axon Medchem), an activator of canonical Wnt signaling, was substituted for WNT3A. For details regarding the protocols used for each differentiation medium, see the Supplemental Experimental Procedures.

12The protocol for the 3D culture was modified from a previous report. For further details, see the Supplemental Experimental Procedures.

13The use of H9 hESCs was approved by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan. Human ethics approval was obtained from the Institutional Review Board and Ethics Committee of Kyoto University Graduate School and Faculty of Medicine. Animal ethics approval was obtained from the Animal Ethics and Research Committee of Kyoto University.

14Values are expressed as the mean ± SEM and “n” stands for the number of independent experiments. Two-tailed Student’s t test was performed to identify significant differences between two conditions of qRT-PCR.

15S.G., I.I., and K.O. designed the study. S.G., T.N., Y.Y., S.K., Y.K., and A.S.-O. performed the experiments. S.G., I.I., T.N., Y.Y., S.K., A.S.-O., S.O., and K.O. analyzed the data. S.G. and I.I. wrote the manuscript through fruitful discussions with and supervision by H.M., S.M., T.H., S.O., K.O., and M.M. M.F., S.-I.M., T.T., and K.O. provided the method for inducing definitive endoderm and advised on the methods used for vector construction and other basic techniques.