Directed Induction of Functional Multi-ciliated Cells in Proximal Airway Epithelial Spheroids from Human Pluripotent Stem Cells

1Proximal airway epithelial cells (PAECs) play a pivotal role in the host defense in the respiratory tract via mucociliary clearance organized by multi-ciliated airway cells (MCACs) and secretory cells. An abnormal function of MCACs is associated with various lung diseases such as primary ciliary dyskinesia (PCD) and cystic fibrosis (CF). It has been reported that PAECs could be generated from human pluripotent stem cells (hPSCs) involving human embryonic stem cells (hESCs) and induced pluripotent stem cells (hiPSCs). The ciliary movement of hPSC-derived MCACs has not yet been reported, although that of murine embryonic stem cell-derived MCACs has been reported. In our previous study, we identified carboxypeptidase M (CPM) as a surface marker of NKX2-1+ “ventralized” anterior foregut endoderm cells (VAFECs) and demonstrated the potency of CPM+ VAFECs to differentiate into alveolar type II cells. We hypothesized that PAECs could also be induced from CPM+ VAFECs, as all lung epithelial lineage cells have been reported to be differentiated from NKX2-1+ VAFECs. We herein report a method of directed differentiation of hPSCs into MCACs and pulmonary neuroendocrine cells (PNECs) and functional analyses of the ciliary movement of hPSC-derived MCACs.

2Because proximal airways develop as 3D branching structures in vivo, we adopted 3D differentiation from CPM+ VAFECs to proximal airway epithelial progenitor cells (PAEPCs) (Figure 1A). Undifferentiated hPSCs consisting of H9 hESCs, 201B7, 585A1, and 604A1 hiPSCs, were stepwise differentiated into NKX2-1+FOXA2+ VAFECs as previously reported, with the exception of the dose of BMP4 used in Step 3. We identified the minimal and sufficient dose of BMP4 to be 20 ng/ml for each hPSC line (Figure 1B). Interestingly, NKX2-1 was downregulated in the presence of Noggin, which inactivates BMP signaling according to a quantitative RT-PCR (qRT-PCR) analysis. On day 14, CPM+ VAFECs were isolated and 3D culture was started in a similar manner as demonstrated in a tracheosphere assay using primary cells (; Supplemental Experimental Procedures). In the hope of generating MCACs at the last step, the optimal medium conditions for proliferating spheroids and inducing FOXJ1, a representative marker of MCACs, were screened by combining FGF10, CHIR99021 (a WNT agonist), KGF, and DAPT (a γ-secretase inhibitor that blocks the Notch pathway), which have been considered to be important (Figure S1A). The growth of the spheroids and NKX2-1, SOX2, and FOXJ1 levels were compared on day 28 (Figures S1B and S1C), and the medium condition of 3 μM CHIR99021 and 100 ng/ml FGF10 was chosen. Under all conditions, SOX9 was only slightly detected by qRT-PCR (Figure S1C). In Step 4, the spheroids grew larger and some of them began to fuse by day 28 (Figure 1C). Importantly, confocal immunofluorescence (CIF) imaging studies showed that nearly all the cells forming spheroids were SOX2+NKX2-1+ cells (Figure 1D), whereas SOX9 was not detected (data not shown), indicating that these cells were of PAEC lineage.

3At the end of Step 4, no MCACs were observed, which prompted us to hypothesize that there might be another step for inducing MCACs. Therefore, we switched the medium to Step 5 medium based on PneumaCult-ALI medium (P-ALI) (Stemcell Technologies), a medium for primary bronchial epithelial cells (Figure 2A). On day 42, clusters of MCACs were observed by H&E staining (Figure 2B). CIF imaging revealed acetylated tubulin (Ac-Tub)+FOXJ1+ cells and closely aligned Ac-Tub+ cells and MUC5AC+ cells, as observed in the fetal human lung (FHL), while secreted MUC5AC markedly accumulated in the closed lumen of the hPSC-derived spheroids (Figure 2C). A small number of SCGB1A1+cells (club cells), KRT5+cells (basal cells) and chromogranin A (CHGA)+ and synaptophysin (SYP)+cells (PNECs) were also found (Figure 2D). Nearly all the hPSC-derived PAECs expressed NKX2-1 (Figures 2D and S2A), consistent with the previous reports and CIF imaging of the FHL (Figure S2A). By triple immunostaining, each representative marker of MCACs, club cells and basal cells was expressed in the different cells (Figure 2E). FOXJ1+ cells did not overlap with the CHGA+ or SYP+ cells as in the FHL (Figure 2C). PGP9.5, another PNEC marker, was confirmed to be expressed in both CHGA+ and SYP+ cells (Figures S2B and S2C).

4Because FOXJ1 is reportedly expressed before multi-ciliogenesis in vitro and in vivo, SNTN, which specifically marks MCACs, was adopted to detect the suitable conditions for multi-ciliogenesis. SNTN was significantly increased on day 42 (Figure 3A, condition b), compared with day 28 (Figure 3A, condition a) in all hPSC lines (p < 0.05) (Figure 3B). In addition, each PAEC marker of MCACs (Figures 3B and S3B), club cells (Figures 3C and S3C), PNECs (Figures 3D and S3D), basal cells (Figure S3E), and mucus-producing cells (Figures 3E and S3F) increased after starting 3D culture (Figure S3A, condition a, b or both) compared with before 3D culture (days 6 and 14), while AQP5 and SFTPC (alveolar type I and II cells, respectively) were almost negative (Figure S3G). SFTPB only slightly increased in accordance with an elevation of club cell markers (Figures 3C, S3C, and S3G, protocols a and b). PAX6 (neuronal cells) and PAX8 (thyroid cells) were also negative (data not shown). Next, the 3D protocol (Figure S3A, protocol a) was compared with the two-dimensional protocol (Figure S3A, protocol f) between days 14 and 28, resulting in an increase of some PAEC markers (Figures S3C, S3D, and S3F, protocols a and f). Because the cells spontaneously detached in 3D culture after day 28, three 3D protocols after the induction of VAFECs (Figure S3A, protocols b, c, and e) were compared with the four air-liquid interface (ALI) protocols (Figure S3A, protocols g–j), which involved two protocols modified from previous reports (Figure S3A, protocols i and j) (,; Supplemental Experimental Procedures).

5DAPT was added to the media from days 28 to 42 (Figures 3A and S3A, protocol c and d) to increase hPSC-derived FOXJ1+ cells (Figure 3F). On day 42, the 3D protocols for CPM+ cells (Figures 3A and S3A, protocols b and c) appeared to induce higher gene expressions of MCAC and club cell markers than the 3D protocol for CPM− cells (Figures 3A and S3A, protocol d) and ALI protocols (Figure S3A, protocols g–j), while the 3D protocol for CPM− cells (Figures 3A and S3A protocol d) appeared to induce KRT5 (a marker of both airway and esophageal basal cells), but not NKX2-1 (Figure S3E). Importantly, SNTN increased only in the 3D protocols for CPM+ cells (Figures 3A and S3A, protocols b and c). Therefore, we concluded that the 3D protocols for CPM+ cells were beneficial for the induction of PAECs.

6Next, we extended the culture period to day 56 (Figure S3A, protocol e), which increased FOXJ1, DNAH5, and SNTN. H&E staining and CIF imaging revealed that MCACs comprised a major part of the epithelia (Figures 3G and 3H). The rate of hPSC-derived FOXJ1+ cells was quantified on day 56 and compared with that on day 42, resulting in an increase in the ratio of FOXJ1+ cells to the total number of cells up to 85.65 ± 1.59% (p = 0.043), 85.82 ± 3.35% (p = 0.030), 72.7 ± 6.6% (p = 0.105), and 87.06 ± 0.43% (p = 0.011) in each hPSC line of 201B7, 585A1 and 604A1 hiPSCs, and H9 hESCs, respectively (n = 3 independent experiments). SNTN was localized at the tips of multiple cilia on day 56 (Figure 3H), which is consistent with the qRT-PCR results (Figures 3B and S3B). Moreover, hPSC-derived CHGA+ cells and SYP+ cells on day 42 (Figure 3A, protocols b and c) increased by adding DAPT (Figure 3F), consistent with the qRT-PCR results (Figures 3D and S3D, protocols b and c). Both the CHGA+ and SYP+ cells were localized to the aligning epithelium sparing FOXJ1+ cells (Figure 3I). In all the comparisons, the vehicle control (DMSO) was added to the media under the counterpart conditions in order to exclude the effects by DMSO solvent of DAPT.

7To elucidate the role of the Notch pathway in DAPT-induced differentiation of PNECs, NOTCH1 intracellular domain (N1ICD), HES1, and PGP9.5 were triply immunostained on day 42, and N1ICD+HES1+ cells were detected among the small number of non-PNECs (Figure S4A). By qRT-PCR, DLL1 was significantly upregulated by DAPT in the H9 hESC line (p = 0.002), but not significantly in 201B7, 585A1, and 604A1 hiPSC lines (p = 0.114, 0.128, and 0.215, respectively). HES1 was significantly suppressed by DAPT in the H9 hESC line (p = 0.013), but not significantly in 201B7, 585A1, and 604A1 hiPSC lines (p = 0.063, 0.225, and 0.44, respectively). NOTCH1-3 on day 42 were unaffected, compatible with DAPT-mediated suppression of the Notch pathway (Figure S4B).

8Next, the morphology of hPSC-derived MCACs was examined using electron microscopy, demonstrating multiple cilia originating from individual basal bodies on the apical surface of columnar epithelial cells (Figures 4A and S4C) and a “9+2” structure consisting of nine doublet and a central pair of singlet microtubules with dynein arms (Figure 4A, rightmost panel), which are specific features of motile cilia.

9On light microscopy, beating cilia were easily observed in the lumen of the spheroids and recorded by a high-speed camera (Movie S1, left panel). Metachronal wave-like beating of the cilia was observed in some MCACs (Figure S4D). In order to quantify the mucociliary flow over the MCACs, we established a protocol of passaging hPSC-derived MCACs in PAEC spheroids to ALI condition (Figure 4B, 3D-ALI protocol) due to the difficulty in measuring the flow rate inside the 3D spheroids. On day 56 of the 3D-ALI protocol, ciliary beating was observed on the apical side of MCACs (Movie S1, right panel). SNTN was localized at the tips of multiple cilia (Figure 4C), and CFTR was detected in the apical surface of MCACs (Figure S4E) and FOXJ1, DNAH5, SNTN, and CFTR levels appeared to be slightly, but not significantly, lower in the 3D-ALI protocol than in the 3D protocol (Figure 4D).

10The ciliary beating frequency (CBF) was calculated by acquiring bright-field images of MCACs in the spheroids and the 3D-ALI condition based on the concepts previously described (Figure S4F; Supplemental Experimental Procedures). The CBF of each hiPSC line (201B7, 585A1, and 604A1)-derived MCACs showed 8.9 ± 0.27, 9.3 ± 0.34, and 6.5 ± 0.17 Hz in the spheroids on day 42 and 10.9 ± 0.31, 10.5 ± 0.26, and 10.0 ± 0.17 Hz in the 3D-ALI condition, respectively. A similar CBF was calculated for normal human bronchial epithelial cell (NHBEC)-derived MCACs in each condition (8.7 ± 0.30 Hz in the spheroids and 8.1 ± 0.33 Hz in the ALI condition) (Figure 4E). To measure mucociliary transport, the fluorescent beads placed on MCACs were traced (Movie S2; Figure 4F; Supplemental Experimental Procedures). The estimated flow velocity of the beads was approximately 7.4–10.1 μm/s in both hPSC- and NHBEC-derived MCACs. However, the values appeared to be affected by the lack of synchroneity of ciliary beating for generating a unidirectional flow (Movie S2; Figure 4F). Therefore, we analyzed the diffusion of the beads from their trajectories based on the concepts in a previous report. We defined the diffusion coefficient normalized to Brownian motion as the mucociliary transport index (MTI) (Supplemental Experimental Procedures). Then, the MTIs in the hPSC- and NHBEC-derived MCACs were calculated, and all the hPSC-derived MCACs showed slightly smaller MTIs compared with NHBEC-derived MCACs and significantly greater MTIs compared with Brownian motion (Figure 4G).

11We established a method of 3D differentiation without feeder cells to generate hPSC-derived PAEC spheroids via isolated progenitor cells using CPM as a surface antigen, which is reportedly a biomarker of lung diseases, such as acute pneumonia and lung cancer. It is noted that the inhibition of the Notch pathway induced not only MCACs but also PNECs from hPSCs, which is consistent with the studies of genetic murine models. PNECs have been proposed to be the origin of small-cell lung cancer, thus suggesting its future application in cancer studies.

12The ciliary function analyses of hPSC-derived MCACs, as well as induction efficiency, are important aspects of the present study. Previously, the functional analyses of hPSC-derived PAECs mostly focused on CFTR, and not on ciliary movement. In addition, the ciliary function was not shown in hPSC-derived lung organoids due to immaturity. In the ciliary function tests, the CBF of hPSC-derived MCACs in spheroids appeared to be lower than that in the 3D-ALI protocol (Figure 4E) for at least two reasons. First, mucoid secretion was trapped in the closed lumen and its increased viscosity might reduce the CBF in the spheroids (Figure 2C and Movie S1, left). Second, we had to mince the spheroids and place cover slips on the samples during image acquisition, which may have reduced the CBF in the hPSC-derived spheroids, while we could directly observe the samples in the 3D-ALI protocol. The CBF of hPSC-derived MCACs in the 3D-ALI protocol was near the normal CBF of human MCACs, which range from 10 to 14 Hz. Next, to quantify mucociliary transport, fluorescent beads were tracked as previously demonstrated in resected murine trachea. Because synchronized ciliary beating for generating a unidirectional flow appeared to be incomplete in both hPSC- and NHBEC-derived MCACs (Movies S1 and S2; Figure 4F), as was reported for NHBEC-derived MCACs, we further focused on the diffusion of the beads, demonstrating the potency of mucociliary clearance in hPSC-derived MCACs (Figures 4F and 4G). The difference between MCACs derived from hPSCs and NHBECs might be partly due to the difference in maturity. In addition, the ideal balance in the number of between MCACs and mucus-producing cells for mucociliary clearance remains to be elucidated. MUC5AC and SPDEF levels on day 56 were lower than on day 42 (Figure S3F), which might be due to differentiation and/or apoptosis. In this respect, the regulation of MUC5AC+ cells by modulating factors, such as IL-13, remains to be a future subject. In conclusion, the findings of the present study are thus considered to pave the way for future applications toward modeling airway diseases, such as PCD and CF, or developing methods of airway reconstruction such as an artificial trachea.

13To measure the CBF, movies of hPSC- and NHBEC-derived MCACs were captured on a high-speed camera (FASTCAM MC2.1; Photron) connected to an upright microscope (Zeiss Axioplan; Carl Zeiss) with ×63 objectives. To measure MTI, the flow of fluorescent beads (Fluoresbrite, 0.5 μm; Polysciences) was recorded by an Orca-ER CCD camera (Hamamatsu) connected to an upright fluorescent microscope (BX51; Olympus) with a ×20 objective. See the Supplemental Experimental Procedures.

14The use of H9 hESCs was approved by the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan. For the use of human samples, human ethics approval was obtained from the Institutional Review Board and Ethics Committee of Kyoto University Graduate School and Faculty of Medicine.

15At least three independent experiments were conducted in each study. The values are expressed as the means ± SEM. A two-tailed t test was performed to determine the statistical significance. p < 0.05 was considered to be significant.

16S.G., S.K., K.T., and S.T. designed the study. S.K., K.T., S.G., Y.Y., Y.K., and T.N., performed the experiments. S.G., S.K., K.T., T.N., and S.T. analyzed the data. S.G. and S.K. wrote the manuscript through a fruitful discussion with and supervision by K.T., H.M., S.M., T.H., I.I., S.T., and M.M.