De Novo Generation of Pulmonary Ionocytes from Normal and Cystic Fibrosis Human Induced Pluripotent Stem Cells

To the Editor:

The advent of single-cell RNA sequencing (scRNA-Seq) has resulted in new lung cell atlases and the recognition of previously undescribed lung cell types (1–3). Particularly relevant for epithelial biologists are recent reports of a rare population of CFTR (cystic fibrosis transmembrane conductance regulator)-rich airway epithelial cells, marked by the expression of transcription factors FOXI1 (forkhead box I1) and ASCL3 (achaete-scute homolog 3) (1, 2). Given their similarity to ionocytes previously described in other organisms or tissues, such as in Xenopus larva skin (4, 5), these cells were termed “pulmonary ionocytes.” Pulmonary ionocytes are present in both mouse and human airways, representing ∼1% of all airway epithelial cells. Although a growing literature shows that ionocytes express the highest levels of CFTR mRNA in the respiratory tree, their functional relevance remains controversial (6), and new human model systems are needed to examine their developmental origins and biological roles. Here we report the de novo derivation of pulmonary ionocytes through the developmental directed differentiation of human induced pluripotent stem cells (iPSCs). We recently published methods for the differentiation of iPSCs into cells expressing a molecular and functional phenotype reminiscent of airway basal cells (7). As with their in vivo correlates, iPSC-derived airway basal-like cells (iBCs) exhibit functional features of tissue-resident airway stem cells, given their ability to indefinitely self-renew or undergo multipotent differentiation into airway secretory or multiciliated cells, both in vitro and in vivo, in tracheal xenograft models (7). In our original publication, we did not detect the presence of cells resembling ionocytes (7). Given prior studies demonstrating that primary basal cells in vivo can serve as precursors for pulmonary ionocytes (2), we have reexamined the differentiation repertoire of iBCs after a modification of our differentiation protocol (Figure 1A). We report here that after additional expansion of iBCs including two rounds of sequential sorting on the cell surface protein, NGFR (nerve growth factor receptor), iBCs can give rise to ionocytes in addition to our previously reported trilineage airway epithelial differentiation repertoire. For example, after two rounds of NGFR sorting, iBCs maintained in three-dimensional cultures in our previously published basal cell medium (7) can be transferred to two-dimensional air–liquid interface (ALI) culture (in our previously published ALI protocol [7]), giving rise after 3 weeks to basal, secretory, multiciliated, and ionocyte progeny (Figure 1B).

Figure 1.

Directed differentiation of normal and cystic fibrosis induced pluripotent stem cells (iPSCs) into putative ionocytes. (A) Schematic of the six-stage iPSC-derived airway differentiation protocol. (B) UMAP with Louvain clustering (clusters 0–5) of five combined datasets of iPSC-derived differentiations after air–liquid interface (ALI) culture (cell lines 1,567 [F508del homozygous cystic fibrosis (CF)], 1,566 [1,567 CFTR (cystic fibrosis transmembrane conductance regulator)-corrected], 1,565 [F508del homozygous CF] repeated twice, 1,564 [1,565 CFTR-corrected]) cultured at ALI. (C) Right: Heatmap of top 20 differentially expressed genes (DEGs) in each cluster (FDR, <0.05; ranked by logFC per cluster). Left: Mini-heatmap of the expression of the top 20 ionocyte DEGs in the iPSC ionocyte and iPSC basal cells. (D) Cell lineage frequencies across all five combined datasets from B. (E) Violin plots comparing expression of ionocyte markers across cell clusters from B. (F) FOXI1 (forkhead box I1) nuclear protein immunostaining (green) (top/bottom) and MUC5B (red) (bottom) of iPSC-derived airway after ALI culture (iPSC line 1,566; gray, DNA; scale bar, 20 μm). (G) Distribution of CFTR⁺ cells by airway cell type (cluster). FDR = false discovery rate; logFC = log fold change; UMAP = Uniform Manifold Approximation and Projection.

Using this modified protocol, we performed duplicate differentiations of four iPSC clones into ALI cultures (n = 8 total differentiations of two iPSC clones from two distinct cystic fibrosis donors carrying homozygous F508del CFTR mutations and two clones consisting of their progeny after mutation correction by CRISPR gene editing). We profiled the resulting cells by scRNA-Seq, generating eight datasets representing the global transcriptomes of 19,255 cells (Gene Expression Omnibus accession numbers GSE203006, GSE203007, and GSE203008). All four iPSC lines in every differentiation gave rise to secretory, basal, and multiciliated cells, and in five of eight differentiations, these four lines also gave rise to a transcriptionally discrete cluster of FOXI1⁺/CFTR^hi cells that coexpressed a diversity of ionocyte marker transcripts (Figures 1B–1E and E1D in the online supplemental). We visualized the transcriptomic programs of all cells derived in these five differentiations (Uniform Manifold Approximation and Projection; Figure 1B), and we identified six cell clusters (0–5) through Louvain clustering. We annotated each cluster on the basis of canonical airway lineage markers as defined in published lung cell atlases (1, 7, 8) (Figures 1B–1D). We observed a distinct cluster 5, composed of 182 cells (average frequency, 0.95% of all cells in each differentiation; range, 0.34–1.67%), enriched in ionocyte markers, including FOXI1, ASCL3, SEC11C, TMPRSS11E, TMEM61, and AKR1B1, subunits of the vacuolar ATPase such as ATP6V0B and ATP6V0D2 (Figures 1C, 1E, and E1A), alternative chloride channel CLCNKB, calcium-activated potassium channel KCNMA1, and CFTR (Figure 1E). The ∼1% frequency of iPSC-derived putative ionocytes is consistent with published frequencies of primary ionocytes in mouse and human large airways (1, 2) and did not differ between CFTR mutant and corrected samples (Figure E1D). To validate FOXI1 expression on a protein level, we performed immunostaining, identifying rare cells expressing nuclear FOXI1 protein (Figure 1F).

As expected, we found more frequent representation of secretory, ciliated, and basal cell lineages in our profiles (Figure 1D), and we, therefore, assessed the expression of CFTR across these lineages relative to the rarer ionocytes. We found that 10% of all cells in our iBC-derived ALI cultures expressed detectable CFTR transcripts. CFTR expression levels were highest in iPSC-derived ionocytes per cell compared with other cell types, which is consistent with prior observations in primary cells (Figures 1E, E1A, and E1B) (1, 2). However, despite the high CFTR expression level in iPSC ionocytes, CFTR was expressed more frequently in other airway lineages, albeit at lower levels, with secretory cells composing the vast majority of all CFTR⁺ iPSC-derived airway epithelia (Figures 1E, 1G, and E1A). We found no consistent statistically significant differences in CFTR expression (levels or frequencies) between cystic fibrosis and CFTR-corrected iPSC airways (Figure E1C). Specifically, counting all CFTR⁺ cells, only 7.5% were ionocytes, whereas 63% were secretory cells, 20% were basal secretory transition cells, 3.9% were basal cells, 4% were immature ciliated cells, and 1% were mature ciliated cells (Figures 1E and 1G). Our data validate previous findings in primary airways that ionocytes express high levels of CFTR and make up only a small portion of CFTR⁺ cells (1, 2, 6).

To compare iPSC-derived ionocytes with primary ionocytes, we quantified the expression levels and frequencies of ionocyte marker transcripts both in our cells and in annotated, published benchmark scRNA-Seq profiles of fresh (uncultured) human lung airway epithelia (9) and primary human bronchial epithelial cells after ALI culture (7). We found similar frequencies and expression levels of most ionocyte markers in iPSC-derived versus primary ionocytes, including FOXI1, ASCL3, IGFBP5, TMEM61, TMPRSS11E, SEC11C, CLNK, and ATP6V0B (Figure 2A). IGF1 (insulin-like growth factor 1) expression was lower and less frequent in cultured human bronchial epithelial cell–derived ionocytes, but it was similar between fresh and iPSC-derived ionocytes. Within the top 50 differentially expressed genes (DEGs) enriched in iPSC-derived ionocytes compared with all other iPSC-derived cell clusters, 70% (35 of 50) of genes were also found within the top 50 DEGs of at least one of these two primary datasets (data not shown). Furthermore, an unbiased comparison of iPSC-derived ionocyte transcriptomes (top 50 DEGs) with all available datasets from all tissues and lineages (Enrichr CellMarker Augmented 2021 database) indicated the most closely related transcriptome to be primary lung ionocytes (false discovery rate, 1 × 10⁻⁸⁹; Figure E1E).

Figure 2.

Comparison of primary and induced pluripotent stem cells (iPSC)-derived ionocytes and impact of FOXI1 (forkhead box I1) forced overexpression. (A) Dot plot comparing expression of each indicated ionocyte marker transcript across iPSC-derived airway cell types versus published uncultured (9) and cultured primary human bronchial epithelial cell (HBEC) (7) single-cell RNA sequencing datasets. HBECs after air–liquid interface (ALI) culture. (B) Immunofluorescence microscopy of two-dimensional cultures of iPSC-derived airway cells after transduction with lentiviral FOXI1:GFP (CFTR [cystic fibrosis transmembrane conductance regulator]-corrected line 1,566). Left: Low (left upper) and high (left lower; confocal microscopy) magnifications are shown of nuclear FOXI1⁺ cells (purple). CFTR-corrected line 1,566; scale bars, 200 μm (left upper) and 10 μm (left lower). Blue = DAPI-stained cell nuclei. Right: Transverse sections of cryo-embedded iPSC airway (KOLF2-C1 line [13]; scale bars, 25 μm). (C and D) Gene expression quantitation by RT-qPCR of each indicated transcript in iPSC-derived airway cells transduced with either FOXI1:GFP or control GFP lentiviruses versus uninfected controls. FC expression compared with uninfected [2^−ΔΔCt] after 18S normalization is shown. n = 3 differentiations of CFTR-corrected line 1,566. (E) Immunofluorescence confocal microscopy of two-dimensional cultures (in ALI for 20 d) of iPSC airway cells after transduction with lentiviral FOXI1:GFP, showing immunostaining for GFP (upper left), FOXI1 (upper middle), and CFTR (upper right). Lower panels indicate colabeling of GFP/CFTR (lower left), FOXI1/CFTR (lower middle), and merged GFP/FOXI1/CFTR (lower right). Blue = DAPI-stained cell nuclei (CF line 007CF, generous gift of Amy Ryan; scale bars, 5 μm). FC = fold change. *P ⩽ 0.05; **P ⩽ 0.01; ***P ⩽ 0.001; ****P ⩽ 0.0001.

Because ionocyte differentiation has been demonstrated to require Notch signaling (1), we investigated whether activating this pathway might be used to augment ionocyte differentiation and presence. We used an iPSC line with doxycycline-inducible NICD1 (NOTCH intracellular domain 1) targeted to the safe harbor AAVS1 human locus (Dox-NICD) (10) and differentiated the Dox-NICD iPSCs into airway epithelium (Figures E2A and E2B). Activation of NOTCH signaling in iBCs either before or during ALI differentiation led to increased basal cell differentiation into secretory cells at the expense of ciliated differentiation (Figures E2A and E2B), validating previous studies (11, 12). However, it did not lead to increased ionocytes or augmented FOXI1 expression (Figures E2A and E2B). Therefore, we concluded that in our system, NOTCH signaling is likely not sufficient for human ionocyte differentiation.

Finally, we tested the effect of forced overexpression of FOXI on ionocyte frequency in iPSC-derived cells, given that this approach generates more ionocytes in primary cell cultures, as demonstrated by Plasschaert and colleagues (1). We transduced iBCs in Transwells with a lentiviral vector encoding a FOXI1:GFP fusion cassette versus GFP-only control (OriGene, catalog no. RC218102L2V; multiplicity of infection, 1) (Figure 2B). FOXI1 overexpression led to significantly increased expression of ASCL3, FOXI1, and CFTR (Figures 2C and 2D), and expression was enriched in GFP⁺ cells sorted for analysis after FOXI1:GFP lentiviral transduction compared with GFP⁻ cells or compared with GFP⁺ cells transduced with the control GFP lentivirus (Figure 2D). At the protein level, we found colocalization of GFP, FOXI1, and CFTR by immunostaining (confocal immunofluorescence microscopy; Figure 2E).

Overall, our results indicate the successful directed differentiation of normal and cystic fibrosis iPSCs via iBCs into a human ionocyte-like population, providing a source of cells with relevance for basic studies and potential future applications for regenerative medicine.