An essential function for autocrine hedgehog signaling in epithelial proliferation and differentiation in the trachea

Hedgehog (HH) signaling regulates tissue patterning and cell differentiation in most animals. In mammals, there are three HH ligands: desert hedgehog (DHH), Indian hedgehog (IHH) and sonic hedgehog (SHH). HH signaling has been investigated in depth during limb formation (Riddle et al., 1993), endodermal organ development (Sala et al., 2011; Bellusci et al., 1997) and cell differentiation in the neural tube (Ericson et al., 1995).

In the lung epithelium, SHH is a highly expressed ligand coordinating both organ formation and homeostasis (Sala et al., 2011; Bellusci et al., 1997; Peng et al., 2015). Epithelial SHH activates the transmembrane protein smoothened (SMO) in mesenchymal cells, leading to nuclear translocation and activation of GLI transcription factors. The activation of HH signaling in the signal-receiving cells depends on the negative regulators PTCH1 and HHIP, the transcription of which is in turn activated by GLI proteins, creating a tightly regulated negative-feedback loop (Wang et al., 2018; Kugler et al., 2015). During lung development, the activation balance of GLI1, GLI2 and GLI3 controls cell proliferation and differentiation of mesenchymal airway smooth muscle cells (Li et al., 2004) and endothelial cells (Li et al., 2004; Miller et al., 2004). In the mesenchyme, HH signaling activation also controls the expression of genes encoding signals such as FGF10 or BMP4 (Weaver et al., 2003), which become activated or restricted in highly localized domains to shape epithelial branching and morphogenesis (Pepicelli et al., 1998; Herriges et al., 2015). In adult lungs, HH signaling is activated upon injury (Watkins et al., 2003) and recent studies have shown that activation of HH signaling in GLI1⁺ mesenchymal cells maintains quiescence in club cells (Peng et al., 2015). In the alveolar compartment, asymmetric hedgehog signaling activation in the distal mesenchymal cells marked by Gli2 and Pdgfra expression controls alveolar epithelial cell proliferation (Wang et al., 2018). Ectopic activation of HH signaling in the distal mesenchyme leads to emphysema-like phenotypes (Wang et al., 2018). These results highlight a central function for HH signaling in adult lung homeostasis. In parallel, several genome-wide association studies (GWAS) have identified polymorphisms in genes encoding HH signaling components in cohorts of individuals with asthma and COPD, implicating HH signaling in chronic lung disease initiation or progression (Kugler et al., 2015; Wang et al., 2019). Overall, genetic studies of HH signaling in the mouse embryonic and adult lungs have focused on the paracrine role of epithelial SHH in controlling the patterning of the mesenchyme. Mesenchymal cells respond to SHH by the restricted activation of distinct programs that, in turn, guide epithelial morphogenesis and tissue repair responses in the adult. Although the paracrine functions of SHH are instrumental in the lung, its potential autocrine function in respiratory tract development has not been formally investigated. Recent pharmacological evidence has suggested a role for SHH in the proliferation and differentiation of cultured human primary nasal epithelial (HNE) cells (Belgacemi et al., 2020), but the role of epithelial HH signaling in the lung remains elusive.

The tracheal tube consists of endoderm-derived epithelium surrounded by mesoderm-derived cartilage, connective tissue and smooth muscle (Brand-Saberi et al., 2014). The tracheal epithelium is composed of several cell types, including basal progenitor cells, ciliated cells, club cells, goblet cells, neuroendocrine cells, tuft cells and ionocytes (Montoro et al., 2018; Plasschaert et al., 2018). Like the lung airways, tracheal tube formation depends on complex epithelial-mesenchymal interactions (Sala et al., 2011; Hines et al., 2013; Snowball et al., 2015; Gerhardt et al., 2018) that orchestrate cellular proliferation (Snowball et al., 2015) and differentiation (Gerhardt et al., 2018). During embryonic development, inactivation of HH signaling impairs tracheal cartilage formation by reducing SOX9⁺ chondrocyte proliferation (Park et al., 2010), differentiation (Park et al., 2010) and condensation (Yin et al., 2018). This process involves the interplay of SHH emanating from the tracheal epithelium and FGF10 from the mesenchyme (Sala et al., 2011). A recent genetic study in children associated HH signaling with a tracheal ring deformity that was characterized by complete cartilaginous rings surrounding the trachea and tracheal stenosis (Sinner et al., 2019).

Here, we investigate the potential role of HH signaling in tracheal epithelial cells. We first show that HH signaling components are dynamically expressed in the tracheal epithelium and mesenchyme. Conditional Shh inactivation in Nkx2.1-expressing epithelial cells led to early postnatal lethality accompanied by respiratory defects. The tracheal cartilage rings were discontinuous and the tracheal tube collapsed. Epithelial cell proliferation and the number of differentiated secretory and multiciliated cells were reduced in the Shh mutant trachea. Inactivation of the transmembrane protein Smo or expression of a dominant active form of Smo in the lung endoderm caused opposite defects in tracheal epithelial cell proliferation and differentiation, arguing for an autocrine role of HH signaling in directly promoting proliferation and differentiation in the tracheal epithelium. We used an in vitro air-liquid interface (ALI) differentiation model of primary human bronchial cells to test whether the autocrine role of HH signaling functions in isolated human epithelial cells. Chemical inhibition of SMO or of GLI led to defects in epithelial cell proliferation and differentiation, arguing for an epithelial cell-autonomous function of HH signaling in human bronchial cells as well. Our results reveal a new autocrine function of HH signaling in addition to its previously characterized paracrine roles in respiratory organ development, and prompt further investigations of epithelial SMO-mediated HH signaling in lung epithelial cells during embryonic development and lung disease progression.

To examine the spatiotemporal mRNA expression of HH signaling components in the developing trachea, we performed multiplex fluorescence in situ hybridizations using SCRINSHOT (single cell resolution in situ hybridization on tissues) during embryonic stages (Sountoulidis et al., 2020). We examined the cellular colocalization of Shh, Smo, patched 1 (Ptch1), Ptch2, hedgehog interacting protein (Hhip), Gli1, Gli2 and Gli3 mRNAs in epithelial and mesenchymal cells from E13.5 to E15.5 (Fig. 1A,B). To mark epithelial cells, we used probes against the Cdh1 and Trp63 mRNAs. Shh was expressed in the epithelium and Hhip was expressed in the mesenchyme, and both displayed reduction in expression levels from E13.5 to E15.5 (Fig. 1A,B). The HH signaling transducer gene Smo was expressed in both epithelial and mesenchymal cells, including putative chondroblasts (Fig. 1A,B and Fig. S1). Similarly, the HH signaling transcription factor genes (Gli1, Gli2 and Gli3) and transmembrane receptor genes (Ptch1 and Ptch2) were detected in both epithelial and mesenchymal cells, and displayed higher expression levels at E13.5 compared with E15.5 (Fig. 1A,B). We also examined Shh, Smo, Gli1 and Gli2 mRNA levels in whole tracheal extracts by quantitative reverse transcription PCR (RT-qPCR). Shh, Smo, Gli1 and Gli2 mRNAs were detectable as early as E12.5, and became gradually reduced until E18.5 (Fig. S2A-D). Smo mRNA levels did not dramatically change, possibly owing to mesenchymal expression. These data show dynamic expression of HH signaling components in the developing trachea and suggest that HH signaling may operate in both epithelial and mesenchymal tracheal cell development.

To identify single embryonic cells that express HH signaling components and relate their expression to known cell types of the trachea, we used publicly available single cell transcriptomic data, from E12.5 to E18.5 stage wild-type tracheal epithelial cells (Kiyokawa et al., 2021). We projected the expression of a panel of HH signaling components on these data (Fig. S3) and kept the cell type classification of the published study. We found Shh and Smo to be highly expressed in embryonic tracheal epithelial cells, and the majority of the embryonic cells were positive for Shh or Smo. There were also a few embryonic cells that co-expressed both the ligand and the transducer. At later stages (E16.5 and E18.5), when known cell type markers initiate their expression, we detected the expression of Shh or Smo in a few ciliated, club and basal cells (Fig. 1C). Moreover, co-expression of Shh and Smo was detected only in rare club or basal cells but not in ciliated cells. The number of epithelial cells expressing Shh signaling components increased up to E15.5 and was reduced in later stages when cell type-specific markers were expressed (Fig. 1D). Recently, Ihh was also reported to be expressed by alveolar epithelial cells (Nikolić et al., 2017), but at comparatively very low levels and in only a few embryonic tracheal epithelial cells (Fig. S3D). These data suggest an autocrine function of HH signaling on epithelial cells in early stages of trachea development (up to stage E16). We conclude that epithelial HH signaling is activated in the trachea and hypothesize that the autocrine function of HH signaling on the tracheal epithelium could occur in two ways, either in a cell-autonomous autocrine fashion or in a paracrine fashion between epithelial cells of different types.

Shh mutants exhibit defects in tracheal cartilage formation during embryonic stages (Park et al., 2010; Yin et al., 2018). We thus hypothesized that epithelial deletion of Shh might cause tracheomalacia with respiratory distress. To inactivate Shh expression specifically in the tracheal epithelium, we used a BAC transgenic mouse line where Cre recombinase is expressed under the control of Nkx2.1 control elements (Nkx2.1^Cre) (Xu et al., 2008). The recombination efficiency of Nkx2.1^Cre in the trachea and lungs has been previously characterized in Nkx2.1^Cre;ROSA26R-LacZ embryos (Sala et al., 2011; Tiozzo et al., 2009). In these experiments, LacZ activity was detected in the lung epithelium as early as E10.5 and the pattern of LacZ activity was nearly homogeneous throughout the tracheal epithelium at E13.5 (Sala et al., 2011; Tiozzo et al., 2009). Thus, the Nkx2.1^Cre strain induces recombination in tracheal epithelial cells with high efficiency. We further evaluated the recombination efficiency of the Shhflox allele, by RT-qPCR for Shh mRNA expression in wild-type embryos and in Nkx2.1^Cre;Shh^flox/flox (ShhCKO) mutants. We observed a strong reduction of Shh mRNA levels in both E11.5 and E18.5 stages in ShhCKO tracheae compared with controls (Fig. S4). ShhCKO mice displayed cyanosis (Fig. 1E) and neonatal respiratory distress (Fig. 1F and Movie 1), and died within 24 h of birth. These mutants were born in the expected Mendelian ratio with a collapsed tracheal lumen (Fig. 1G,H) and fractured cartilage rings instead of the intact ventrolateral cartilage rings seen in control siblings (Fig. 1I,J), indicating that epithelial inactivation of Shh does not cause embryonic lethality. ShhCKO mice exhibited no trachea-esophageal fistula at E17.5. Although Shh mRNA levels in E11.5 ShhCKO lungs were already severely reduced, the lack of fistula may suggest that Shh inactivation in this strain occurs after the separation of the trachea from the esophagus (Fig. S5A,B). To test for a role of Shh in lung development, we analyzed the embryonic lungs at E17.5. As expected, ShhCKO animals also exhibited pulmonary hemorrhage (Fig. 1K,L) and cysts (Fig. 1K,M), mainly in distal regions compared with controls (Miller et al., 2004). These data indicate that the mutants die from compromised respiratory function due to defective formation of the tracheal tube with high airflow resistance limiting breathing frequency and respiratory air sacs additionally impacting gas exchange.

Sonic hedgehog signaling controls epithelial differentiation in endodermal organs, including the esophagus and thymus (Freestone et al., 2003; van Dop et al., 2013; Saldaña et al., 2016). To test for a possible role of Shh signaling in tracheal epithelial cell differentiation, we performed immunostaining for characteristic markers of major tracheal epithelial cell types. FOXJ1, a ciliated cell marker, is expressed in the trachea around E15.5 (Stauber et al., 2017). We observed several FOXJ1⁺ but no acetylated α-tubulin⁺ cells in the wild-type tracheal epithelium at E15.5 (Fig. S6A,B). Acetylated α-tubulin⁺ ciliated cells were detected at E16.5 (Fig. S6B), indicating that mature ciliated cells first appear in the trachea between E15.5 and E16.5. From E18.5, an even distribution of acetylated α-tubulin⁺ ciliated cells was observed in the tracheal epithelium (Fig. S6B). Interestingly, ShhCKO animals at E18.5 exhibited reduced FOXJ1⁺ ciliated cells (Fig. 2A,B) with decreased Foxj1 mRNA levels compared with controls (Fig. 2C). Consistently, the relative number of mature ciliated cells expressing acetylated α-tubulin (Fig. 2D,E), and the overall Tubb4b mRNA levels (Fig. 2C), were also reduced in ShhCKO animals. Next, we examined markers of club cell maturation in the mutant epithelium. SCGB1A1, a mature club cell marker, was readily detected in the wild-type tracheal epithelium at E18.5 (Fig. S6C). These SCGB1A1⁺ cells were distributed evenly in the apical epithelial surface and expressed variable SCGB1A1 levels at this stage of development (Fig. S6C). ShhCKO tracheal sections from E18.5 embryos exhibited fewer SCGB1A1⁺ club cells (Fig. 2F,G) and decreased Scgb1a1 mRNA levels compared with controls (Fig. 2C). Intriguingly, the earlier secretory club cell fate markers SCGB3A2 and SCGB3A1 (Guha et al., 2012; Reynolds et al., 2002), which were already detected in the wild-type trachea at E17.5 (Fig. S6D,E), exhibited a similar distribution in ShhCKO and wild-type tracheae (Fig. S7A-D). The overall levels of Scgb3a2 and Scgb3a1 mRNA in the mutant trachea were also similar to controls (Fig. S7E). These results indicate that HH signaling is required for the progression but not for the initiation of the secretory club cell differentiation program. To further examine epithelial differentiation defects in the trachea at a postnatal stage, we performed immunostaining in tissue sections at P0. ShhCKO tracheae exhibited persistently reduced numbers of acetylated α-tubulin⁺ ciliated cells (Fig. S8A,B) and of SCGB1A1⁺ club cells (Fig. S8A,C) compared with controls, arguing that these defects are not due to a general delay of differentiation. As the transcription factors FOXP1 and FOXP4 cooperatively modulate club cell differentiation in the airways (Li et al., 2012), we looked for changes in Foxp1 and Foxp4 expression. ShhCKO tracheae exhibited reduced Foxp1 expression (Fig. 2C), suggesting that Foxp1 is a potential target of HH signaling during club cell development.

HH signaling has been reported to modulate cell proliferation in several contexts (Berman et al., 2003; Thayer et al., 2003; van den Brink et al., 2004; Merchant et al., 2010; Plaisant et al., 2011; Gagné-Sansfaçon et al., 2014; Raleigh et al., 2018). We thus tested whether epithelial deletion of Shh might lead to tracheal epithelial cell proliferation defects. We examined epithelial cell proliferation first by immunostaining for Ki67, a cell cycle marker, and found that ShhCKO tracheae exhibited a reduced number of Ki67⁺ cells in the CDH1 marked epithelium compared with controls (Fig. 2H,I). Next, we performed an EdU incorporation assay to assess cells that had entered S-phase during the labeling period. After an ex vivo, 21 h long EdU treatment, ShhCKO tracheae displayed fewer EdU⁺ cells in their epithelium (Fig. 2J,K), indicating that Hedgehog inactivation compromises cell cycle entry. In a more-detailed analysis, we observed a reduced number of Ki67⁺SCGB1A1⁺ cells (Fig. 2L,M), but a similar number of PDPN⁺ (Fig. S9A,B) and Ki67⁺PDPN⁺ basal cells (Fig. S9A,C) in ShhCKO tracheae compared with controls. Interestingly, ShhCKO tracheae exhibited increased ratios of KRT5⁺ basal cells to acetylated α-tubulin⁺ ciliated cells and to SCGB1A1⁺ club cells compared with controls (Fig. S10A-D). We did not find obvious alterations in caspase3 staining between ShhCKO tracheal epithelium and controls at E18.5, suggesting that apoptosis is not affected in the mutants (Fig. S9D). To examine for changes in the surrounding mesenchyme, we analyzed SOX9⁺ chondroblasts, α-SMA⁺ smooth muscle cells and mesenchymal cell proliferation. ShhCKO tracheae exhibited SOX9⁺ chondroblast condensation defects (Fig. S11A), but no obvious SOX9⁺ chondroblast differentiation defects (Fig. S11B-D). We also detected reduced numbers of α-SMA⁺ smooth muscle cells and increased numbers of CDH1⁻Ki67⁺ mesenchymal cells compared with controls (Fig. S11E-G). As HH signaling modulates Fgf10 expression in the lung mesenchyme (Volckaert et al., 2013; Balasooriya et al., 2017), we examined Fgf10 mRNA expression levels in ShhCKO tracheae. We detected a 1.5-fold increase in Fgf10 mRNA in ShhCKO mutants (Fig. S11H) but this was not sufficient to increase basal cell number, as might have been expected from previous studies. Altogether, these results indicate that loss of Shh function reduces club cell proliferation and interferes with ciliated cell differentiation and secretory cell maturation in the tracheal epithelium. In the tracheal mesenchyme, Shh promotes chondrocyte and smooth muscle morphogenesis.

To investigate whether epithelial inactivation of HH signaling causes phenotypes similar to those observed in the ShhCKO tracheal epithelium, we selectively deleted the Shh effector Smo in epithelial cells by using Nkx2.1^Cre;Smo^flox/flox (SmoCKO) mice. The recombination efficiency of the Shhflox and Smoflox alleles has been investigated with a Nestin-driven Cre transgene (Nestin^Cre) (Machold et al., 2003). In the neural progenitor cells, the phenotypes of Shh^null/Shh^flox and Smo^null/Smo^flox mutant mice were indistinguishable, suggesting that the Shhflox and the Smoflox are recombined with similarly high efficiency (Machold et al., 2003). Additionally, Act2^Cre;Smo^flox/Smo^flox embryos display similar phenotypes to the Smo^null/Smo^null mutants, indicating that Smo^flox recombines with high efficiency. ShhCKO embryos display a strong reduction of Shh mRNA in their lungs, suggesting that SmoCKO and ShhCKO represent strong loss-of-function mutants (Fig. S4). We detected decreased numbers of FOXJ1⁺ ciliated cells in SmoCKO mice (Fig. 3A,B), and an overall reduction of Foxj1 mRNA levels (Fig. 3C) in mutant trachea compared with controls at E18.5. Similarly, we detected fewer acetylated α-tubulin⁺ mature ciliated cells (Fig. 3D,E) and reduced Tubb4b mRNA levels (Fig. 3C) in SmoCKO animals. SmoCKO tracheae also exhibited decreased numbers of SCGB1A1⁺ club cells (Fig. 3F,G) and lower Scgb1a1 mRNA levels throughout the tissue (Fig. 3C). Interestingly, SmoCKO tracheae exhibited increased ratios of KRT5⁺ basal cells to acetylated α-tubulin⁺ ciliated cells and to SCGB1A1⁺ club cells compared with controls (Fig. S12A-D). As in the ShhCKO mutants, the SmoCKO tracheae did not exhibit significant defects in the numbers of SCGB3A2⁺ (Fig. S13A,B) or SCGB3A1⁺ (Fig. S13C,D) cells, or in the overall expression levels of Scgb3a2 or Scgb3a1 (Fig. S13E) compared with controls. SmoCKO tracheae also displayed reduced EdU⁺ incorporation in CDH1⁺ epithelial cells (Fig. 3H,I) compared with controls, suggesting that epithelial Smo controls cell proliferation. However, we did not detect any defects in cartilage ring formation or in chondroblasts and smooth muscle cells (Fig. S14A-E) in SmoCKO tracheae. Notably, the observed phenotypes in SmoCKO mutants are not as severe as those in ShhCKO animals, suggesting that the increased severity in the ShhCKO tracheae may be due to the reciprocal signaling from loss of Shh signaling in the surrounding mesenchyme. SHH activation in mesenchymal cells may also indirectly contribute to tracheal epithelial development.

To investigate whether epithelial activation of HH signaling is sufficient to drive tracheal epithelial cell differentiation, we overexpressed a constitutively active form of Smo, SmoM2 (Jeong et al., 2004), in epithelial cells using Nkx2.1^Cre;R26^SmoM2 mice. Nkx2.1^Cre;R26^SmoM2 tracheae displayed increased numbers of acetylated α-tubulin⁺ ciliated cells (Fig. 3J,K) and SCGB1A1⁺ club cells (Fig. 3L,M) compared with controls. Overactivation of Smo also resulted in a weak increase in the proportion of TRP63⁺KRT5⁺ basal cells in the epithelium but this difference was not statistically significant compared with controls (Fig. S15). Collectively, these data indicate that activation of epithelial HH signaling is required for the proliferation of a subset of tracheal epithelial cells and for the differentiation of secretory and ciliated cell types.

The epithelial structures of the mouse trachea share similarities with human bronchioles, including the cartilaginous rings overlaid by basal progenitor cells and apical secretory and ciliated cells (Rock et al., 2010; Danopoulos et al., 2019). To examine whether HH signaling is required for human bronchial epithelial (HBE) cell differentiation, we used primary bronchial epithelial cells cultured in vitro at an air-liquid interface (ALI) as a model of human airway epithelial development (Schmid et al., 2017). We seeded undifferentiated HBE cells onto transwell filters and removed the medium from the upper chamber to initiate cell differentiation (defined as day 0). We then examined the expression levels of sonic hedgehog signaling components in HBE cells and found that SHH, SMO, GLI1 and GLI2 mRNA were dynamically expressed from day 0 to day 14 (Fig. 4A-D).

HH signaling inhibition by the SMO inhibitor cyclopamine (Chen et al., 2002) reduces epithelial cell differentiation in the colon (van den Brink et al., 2004), ureter (Bohnenpoll et al., 2017) and nose (Belgacemi et al., 2020). A 9-day cyclopamine treatment led to decreased differentiation of FOXJ1⁺ and RFX3⁺ ciliated cells, and of MUC5AC⁺ secretory cells (Fig. 4E-G and Fig. S16A-D), but unaltered TP63⁺ basal cell numbers (Fig. S16A,C,E). Similarly, we detected reduced mRNA levels of FOXJ1, TP73, MYB, MUC5AC and HHIP mRNA levels, but unchanged TP63 mRNA levels compared with controls (Fig. 4H, Fig. S17A-D). Cyclopamine treatment also caused decreased differentiation of acetylated α-tubulin⁺ ciliated cells and MUC5B⁺ secretory cells (Fig. 4I-K), and an overall reduction of MUC5B mRNA levels (Fig. 4H). Next, we examined the expression of SPDEF, a transcription factor gene involved in goblet cell differentiation (Park et al., 2007; Chen et al., 2009), and observed significantly reduced SPDEF mRNA levels after cyclopamine treatment compared with controls (Fig. 4H). To test whether the epithelial differentiation defects due to HH signaling inhibition correlate with alterations in the Notch pathway, we examined the levels of Notch signaling components. We did not detect significant changes in JAG1, JAG2, NOTCH2, HES1 or HEY1 mRNA levels upon cyclopamine treatment (Fig. S17E-I). Moreover, we did not observe significant changes in the number of NGFR⁺ basal cells (Fig. S18A,B). A 9-day treatment with 10 μM GANT 58, a GLI1 inhibitor (Lauth et al., 2007; Beauchamp et al., 2009), also decreased the relative numbers of FOXJ1⁺ ciliated cells (Fig. S19A,B) and MUC5AC⁺ secretory cells (Fig. S19A,C), as well as HHIP mRNA levels (Fig. S19D), suggesting that HH signaling regulates HBE cell differentiation in a GLI1-dependent manner. Next, we asked whether SHH could induce HBE cell differentiation. A 9 day-long treatment with recombinant human SHH (H-SHH) increased the relative number of FOXJ1⁺ ciliated cells (Fig. 4L,M) and MUC5AC⁺ secretory cells (Fig. 4L,N), accompanied by upregulated FOXJ1, MUC5AC and HHIP mRNA levels (Fig. 4O and Fig. S19E), at day 14 compared with controls. These data indicate that canonical sonic hedgehog signaling drives HBE cell differentiation.

To investigate HH signaling in HBE cell proliferation, we analyzed the number of HBE cells in S-phase in ALI cultures with or without cyclopamine treatment. We conducted a dual pulse labeling using EdU and BrdU incorporation during HBE cell proliferation from day 0 to day 7 to follow proliferation at different stages of culture and differentiation (Fig. 5A). High Alexa Fluor 555-derived fluorescence intensity was observed in most HBE cells after a 12 h EdU treatment at day 0 (Fig. S20A,B), indicating that undifferentiated HBE cells undergo active proliferation. To examine proliferation characteristics of these EdU⁺ cells, we traced cell proliferation at later stages by adding BrdU into the culture medium from day 2.5 to day 6.5, and performing staining and analysis 12 h after each pulse labeling (Fig. 5A). At day 3, both high and low Alexa Fluor 555-derived fluorescence intensities were observed in HBE cells (Fig. 5B), indicating different proliferative activities among HBE cells from day 0 to day 3. High Alexa Fluor 488-derived fluorescence intensity reflecting high BrdU incorporation was observed only in some of the cells with low Alexa Fluor 555-derived fluorescence intensity at day 3 (Fig. 5B), indicating further proliferation of some dividing cells from day 0. Ki67 was detected mainly in BrdU⁺ cells (Fig. 5B) and a subset of cells with low Alexa Fluor 555-derived fluorescence intensity (Fig. 5B). At day 5, cells with lower Alexa Fluor 555-derived fluorescence intensity were observed compared with day 3 (Fig. 5B,C), and BrdU⁺ or Ki67⁺ signals were detected in only a subset of these cells (Fig. 5C), indicating that few of the EdU⁺ cells divided from day 3 to day 5. At day 7, even fewer BrdU⁺ or Ki67⁺ cells exhibited low or undetectable Alexa Fluor 555-derived fluorescence intensity compared with day 5 (Fig. 5D), suggesting that only a few of the EdU⁺ labeled cells at day 0 continue to divide from day 5 to day 7, or that the label became diluted after serial rounds of proliferation. Cyclopamine treatment led to a general decrease in BrdU incorporation in cells with Alexa Fluor 555-derived fluorescence intensity as well as reduced Ki67 immunostaining at day 3 (Fig. 5E,F), day 5 (Fig. 5G,H) and day 7 (Fig. 5I,J), indicating that HH signaling is required continuously for DNA synthesis in proliferating HBE cells. To test whether HH signaling is required for cell proliferation during differentiation of HBE cells, we treated HBE cells in ALI with 10 μM cyclopamine from day 5 to day 14, when ciliated cells and secretory cells begin to differentiate. Again, after a 9-day cyclopamine treatment, we observed an overall decrease in proliferation of epithelial cells (Fig. S21A,B). Altogether, these results suggest that HH signaling is required for HBE cell proliferation at both the cell expansion and differentiation stages.

Genetic studies in mice have linked HH signaling to several human congenital malformations of the trachea and bronchi, including tracheoesophageal fistula and possible tracheomalacia (Sala et al., 2011; Miller et al., 2004; Motoyama et al., 1998; Park et al., 2010). Recently, whole-exome sequencing of pediatric patients and their parents revealed a de novo mutation in SHH (p.Asp279Tyr) in children with complete tracheal ring deformity (CTRD) (Sinner et al., 2019), a condition characterized by circumferentially continuous or nearly continuous cartilaginous tracheal rings (Sahoo et al., 2009; Faust et al., 1998), variable degrees of tracheal stenosis and/or shortening, and/or pulmonary arterial sling anomaly (Berdon et al., 1984; Lee et al., 1996). Given the importance of SHH in tracheal development and congenital disorders, it is important to understand the specific contributions of paracrine and autocrine signaling activation. Our analysis of the epithelial ShhCKO phenotypes suggests that epithelial-derived SHH mediate tracheal cartilage formation, possibly by activating HH signaling in mesenchyme. Sox9 and Col2a1 mRNA levels are both reduced in the tracheal mesenchyme of Shh null mice (Park et al., 2010), and presumably other targets involved in tracheal cartilage formation are also affected. Previous work has also shown that Shh and Fgf10 exhibit genetic interactions during tracheal cartilage ring formation, and suggest that mesenchymal Fgf10 control epithelial growth, cartilage patterning and epithelial Shh expression (Sala et al., 2011). Nevertheless, our epithelial cell specific deletion of Shh leads to more severe tracheal cartilage defects compared with the Fgf10 or Fgfr2b null allele (Sala et al., 2011), suggesting that Shh acts via additional target(s) in the mesenchyme to regulate tracheal patterning. In bladder, epithelial cell-derived SHH increases stromal expression of Wnt2 and Wnt4, which in turn stimulates the proliferation of both epithelial and stromal cells on injury (Shin et al., 2011). Interestingly, both global knockout of Wnt4 and epithelial cell specific deletion of Wntless, a cargo receptor that facilitates Wnt ligand secretion (Bänziger et al., 2006), causes defects in tracheal cartilage ring formation (Snowball et al., 2015; Caprioli et al., 2015). Wnt signaling is also essential for human airway epithelial cell differentiation and proliferation (Schmid et al., 2017). However, it is unknown whether Wnt signaling is required for tracheal epithelial cell differentiation or proliferation in vivo. It would be interesting to examine epithelial cell differentiation in Wnt signaling inactivation mutants, and test for the possible crosstalk between HH and Wnt signaling during tracheal formation.

The SmoCKO analysis also uncovers a new epithelial function for Shh, where Smo controls cell-autonomously the proliferation of epithelial cells and their differentiation into SCGB1A1⁺ club cells and ciliated cells. The decreased number of ciliated and SCGB1A1⁺ cells in the SmoCKO trachea could be due to the proliferation defect of a common progenitor or due to a continuous requirement of epithelial Smo for cell differentiation of the progenitors towards the ciliated and SCGB1A1⁺ lineages. This autocrine function of HH signaling is also retained in human bronchial cells, where both SMO and GLI inhibition reduce epithelial proliferation and differentiation towards the secretory and ciliated lineages. The overexpression analysis of the activated SmoM2 allele in the mouse and the effects of SHH addition to the human bronchial epithelial cell further argue for a direct role of HH signaling in epithelial cell differentiation. Our work suggests a functional separation of the tracheal defects observed in mouse mutants and in individuals with mutations in HH pathway genes. Cartilage abnormalities and other mesenchymal defects are due to the paracrine function of Shh, whereas tracheal lumen stenosis and respiratory dysfunctions maybe due to the combined functions of HH and its downstream effectors in both the epithelium and mesenchyme.

All mouse experiments were performed under standard conditions in accordance with German federal ethical and animal welfare guidelines (Ethical Permit number GI 20/10, Nr. G 29/2019) and Swedish ethical regulations approved by the Northern Stockholm Animal Ethics Committee (Ethical Permit numbers N254/2014 and 15196/2018). All breeding colonies were maintained under cycles of 12 h light and 12 h dark. Nkx2.1^Cre (Xu et al., 2008), Shh^flox (Sala et al., 2011), Smo^flox (Long et al., 2001), R26Smo^M2 (Jeong et al., 2004) alleles have been previously described.

E13.5 and E15.5 wild-type tracheae were fixed in 4% PFA for 30-45 min at 4°C. Longitudinal cryosections of the trachea (10 μm) were prepared and the SCRINSHOT protocol (Sountoulidis et al., 2020) was used for probe design and in situ detection of mRNA transcripts. Sequences of probes are listed in the Table S1.

Tracheal cryosections (10 µm) were fixed in 4% paraformaldehyde for 20 min, treated with 3% acetic acid solution for 3 min, stained in 0.05% Alcian Blue for 10 min and counterstained with 0.1% nuclear Fast Red solution for 5 min. For whole-mount staining of tracheal cartilage, dissected tracheae were fixed in 95% ethanol for 12 h followed by overnight staining with 0.03% Alcian Blue dissolved in 80% ethanol and 20% acetic acid. Samples were cleared in 2% KOH.

Tracheae were dissected in PBS, fixed in 4% paraformaldehyde overnight at 4°C, incubated in 10% sucrose and 30% sucrose for 24 h each at 4°C, mounted in OCT embedding compound and sectioned at 10 µm. To perform immunostaining, sections were fixed in 4% paraformaldehyde for 10 min at 4°C, followed by incubation in permeabilization solution (0.3% Triton X-100/PBS) for 15 min at room temperature. They were then incubated in blocking solution (5% FBS/PBS/3% BSA) for 1 h at room temperature, incubated in primary antibodies overnight at 4°C, washed, incubated in secondary antibodies for 2 h at room temperature, washed and then mounted for imaging.

Tracheae and lungs were isolated from E18.5 embryos and cultured using an established protocol (Del Moral and Warburton, 2010). For EdU incorporation, isolated tracheae and lungs were cultured in DMEM/F-12 medium containing 10 µM EdU at 37°C in a 5% CO₂ incubator for 20 h. 0.1% DMSO was used as a control. Tracheae were fixed in 4% paraformaldehyde overnight at 4°C, incubated in 10% sucrose and 30% sucrose for 24 h each at 4°C, mounted in OCT embedding compound and sectioned at 10 µm. To perform EdU detection and CDH1 immunostaining, sections were fixed in 4% paraformaldehyde for 10 min at 4°C followed by incubation in permeabilization solution (0.3% Triton X-100/PBS) for 15 min and in blocking solution (5% FBS/PBS/3% BSA) for 1 h at room temperature. Samples were incubated in EdU reaction cocktail prepared according to the manufacturer's instruction (Click-iT EdU Imaging Kit, Thermo Fisher Scientific, C10338) for 30 min, washed, incubated in the CDH1 primary antibody for 4 h at room temperature, washed, incubated in the secondary antibody for 2 h at room temperature, washed and then mounted for imaging.

Primary human bronchial epithelial (HBE) cells (material obtained from the UGMLC/DZL-Biobank, Dr C. Ruppert and Prof. A. Günther, and approved by the local ethics committee, AZ 85/15) were isolated and cultivated under air-liquid interface conditions to form well-differentiated, pseudostratified cultures as described previously (Bals et al., 2004). Isolated HBE cells were maintained and expanded (one passage) in T75 flasks in hormone- and growth factor-supplemented airway epithelial cell growth medium (AEGM, ready-to-use; PromoCell) at 37°C in a 5% CO₂ incubator. At 80% confluence, cells were detached with 0.05% trypsin-EDTA (Gibco) and seeded on membrane supports (12 mm Transwell culture inserts, 0.4 µm pore size, Costar) coated with 0.05 mg collagen from calf skin (Sigma-Aldrich) in ready-to-use AEGM supplemented with 1% penicillin/streptomycin. HBE cells were cultured for 2 days until they reached complete confluence. The apical medium was then removed and the basal medium was replaced by a 1:1 mixture of DMEM (Sigma) and ready-to-use AEGM supplemented with 60 ng/ml retinoic acid (Sigma). Cultures were maintained under air-liquid interface conditions by changing the medium in the basal filter chamber three times a week. For cyclopamine (1623, TOCRIS) treatment, a 5 mM stock solution was diluted to 10 µM. For GANT 58 (3889, Tocris) treatment, a 10 mM stock solution was diluted to 10 µM. For recombinant human SHH (H-SHH) (R&D Systems, 1845-SH-025) treatment, a 200 µg/µl stock solution was diluted to 100 ng/ml. Epithelial cells were cultured in differentiation medium containing the above chemicals or H-SHH at 37°C in a 5% CO₂ incubator from day 5 to day 14. The medium was replaced every 24 h before collection for analysis.

For EdU incorporation in HBE cells, a 10 mM EdU stock solution was diluted to 10 µM. For BrdU incorporation, a 10 mM BrdU stock solution was diluted to 10 µM. HBE cells were cultured in differentiation medium with EdU or BrdU at 37°C in a 5% CO₂ incubator for 12 h at the indicated time point. For EdU and BrdU detection, ALI cultures were fixed in 4% paraformaldehyde 20 min at room temperature, followed by incubation in permeabilization solution (0.3% Triton X-100/PBS) for 15 min and in blocking solution (5% FBS/PBS/3% BSA) for 1 h at room temperature. Samples were incubated in EdU reaction cocktail prepared according to the manufacturer's instruction (Click-iT EdU Imaging Kit, Thermo Fisher Scientific, C10338) for 30 min, washed, incubated in the BrdU antibody for 4 h at room temperature, washed, incubated in the secondary antibody for 2 h at room temperature, washed and then mounted for imaging.

Total RNA extraction was conducted using a miRNeasy Mini Kit (Qiagen, 217004). cDNA was synthesized using the Maxima First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, K1641), according to manufacturer's instructions. Quantitative real-time PCR was performed using LightCycler 480 II (Roche) and iTaq Universal SYBR Green Supermix (Bio-Rad, 1725122). The following primers were used: mActb forward, 5′-CGGCCAGGTCATCACTATTGGCAAC-3′ and mActb reverse, 5′-GCCACAGGATTCCATACCCAAGAAG-3′; mShh forward, 5′-GGCTGATGACTCAGAGGTGCAAAG-3′ and mShh reverse, 5′-GCTCGACCCTCATAGTGTAGAGAC-3′; mGli1 forward, 5′-GCCTGGAGAACCTTAGGCTGGA-3′ and mGli1 reverse, 5′-ACAGGTGCGCCAGCGTG-3′; mGli2 forward, 5′-GCCCTGGAGAGTCACCCTT-3′ and mGli2 reverse, 5′-TGCACAGACCGGAGGTAGT-3′; mSmo forward, 5′-GAGCGTAGCTTCCGGGACTA-3′ and mSmo reverse, 5′-CTGGGCCGATTCTTGATCTCA-3′; mFoxj1 forward, 5′-GAGTGAGGGCAAGAGACTGG-3′ and mFoxj1 reverse, 5′-TCAAGTCAGGCTGGAAGGTT-3′; mTubb4b forward, 5′-AACCCGGCACCATGGACTCTGT-3′ and mTubb4b reverse, 5′-TGCCTGCTCCGGATTGACCAAATA-3′; mScgb1a1 forward, 5′-ATGAAGATCGCCATCACAATCAC-3′ and mScgb1a1 reverse, 5′-GGATGCCACATAACCAGACTCT-3′; mScgb3a1 forward, 5′-ACCACCACCTTTCTAGTGCTC-3′ and mScgb3a1 reverse, 5′-GGCTTAATGGTAGGCTAGGCA-3′; mScgb3a2 forward, 5′-GCTGGTATCTATCTTTCTGCTGGTG-3′ and mScgb3a2 reverse, 5′-ACAACAGGGAGACGGTTGATGAGA-3′; mFoxp1 forward, 5′-CGAATGTTTGCTTACTTCCGACGC-3′ and mFoxp1 reverse, 5′-ACTTCATCCACTGTCCATACTGCC-3′; hRPLRO forward, 5′-CCAGCAGGTGTTCGACAAT-3′ and hRPLRO reverse, 5′-CAGGAAGCGAGAATGCAGA-3′; hSHH forward, 5′-AAGGACAAGTTGAACGCTTTGG-3′ and hSHH reverse, 5′-TCGGTCACCCGCAGTTTC-3′; hGLI1 forward, 5′-TCTCAAAGTGGGAGGCACAA-3′ and hGLI1 reverse, 5′-CCCTTAGGAAATGCGATCTG-3′; hGLI2 forward, 5′-TTATGGGCATCCTCTCTGGT-3′ and hGLI2 reverse, 5′-CGGAGCAGAGTATCCAGTAT-3′; hSMO forward, 5′-GAAGTGCCCTTGGTTCGGA-3′ and hSMO reverse, 5′-GCAGGGTAGCGATTCGAGTT-3′; hFOXJ1 forward, 5′-GCCTCCCTACTCGTATGCCA-3′ and hFOXJ1 reverse, 5′-GCCGACAGGGTGATCTTGG-3′; hMUC5AC forward, 5′-GCTCAGCTGTTCTCTGGATGAG-3′ and hMUC5AC reverse, 5′-TTACTGGAAAGGCCCAAGCA-3′; hMUC5B forward, 5′-ACATGTGTACCTGCCTCTCTGG-3′ and hMUC5B reverse, 5′-TCTGCTGAGTACTTGGACGCTC-3′; hNOTCH2 forward, 5′-TGGTGGCAGAACTGATCAAC-3′ and hNOTCH2 reverse, 5′-CTGCCCAGTGAAGAGCAGAT-3′; hJAG1 forward, 5′-GAATGGCAACAAAACTTGCAT-3′ and hJAG1 reverse, 5′-AGCCTTGTCGGCAAATAGC-3′; hJAG2 forward, 5′-GAGCTCTGCGACACCAATC-3′ and hJAG2 reverse, 5′-TCATTGACCAGGTCGTAGCA-3′; hHES1 forward, 5′-TTACGGCGGACTCCATGT-3′ and hHES1 reverse, 5′-AGAGGTGGGTTGGGGAGT-3′; hHEY1 forward, 5′-GATGATCAGCTTTATCCAAGAAAGA-3′ and hHEY1 reverse, 5′-CAGTTTGTACATTCACCTTTCTGC-3′; hMYB forward, 5′-CCGGGAAGAGGATGAAAAAC-3′ and hMYB reverse, 5′-TTTCCAGTCATCTGTTCCATTC-3′; hTP73 forward, 5′-CCACTGGTGGACTCCTATCG-3′ and hTP73 reverse, 5′-CTGTAGGTGACTCGGCCTCT-3′; hTP63 forward, 5′-GAAGATCAAAGAGTCCCTGGAA-3′ and hTP63 reverse, 5′-GCTGTTGCCTGTACGTTTCA-3′.

The following antibodies were used: rat anti-CDH1 (1:200, Santa Cruz, sc-59778); mouse anti-CDH1 (1:100, BD Biosciences, 560062); rabbit anti cleaved caspase-3 (1:600, Cell Signaling Technologies, #9661); goat anti-SCGB1A1 (1:200, Santa Cruz, T-18); rabbit anti-NKX2.1 (1:400, Santa Cruz, H-190); mouse anti-acetyl-α tubulin (1:2000, Sigma, MABT868); mouse anti-FOXJ1 (1:400, Thermo Fisher Scientific, 14-9965-82); rabbit anti-MUC5AC (1:400, Santa Cruz, H-160); rabbit anti-MUC5B (1:400, Novus Biologicals, NBP1-92151); mouse anti-BrdU (1:400, Thermo Fisher Scientific, B35141); hamster anti-PDPN (1:20, DSHB, 8.1.1); rat anti-SCGB3A1 (1:200, R&D Systems, MAB2954); rat anti-SCGB3A2 (1:200, R&D Systems, MAB3465); rabbit anti-Ki67 (1:400, Thermo Fisher Scientific, PA5-19462); goat anti-NGFR (1:200, Santa Cruz, C-20); mouse anti-α-SMA-Cy3 (1:1000, Sigma-Aldrich, C6198); rabbit anti-SOX9 (1:400, Millipore, AB5535MA); goat anti-SOX9 (1:500, R&D Systems, AF3075); rat anti-Cd140a (1:100, Biolegend, 135901); rabbit anti-RFX3 (1:500, Sigma, HPA035689); mouse anti-TRP63 (1:500, Abcam, ab735); rabbit anti-TRP63 (1:100, Cell Signaling Technology, 13109S) and chicken anti-KRT5 (1:400, Biosite, 905901).

Imaging of whole-mount tracheae, trachea sections and ALI cultures was performed using a Nikon SMZ25 or a Zeiss 880 upright laser scanning confocal microscope, a Zeiss 780 laser scanning confocal microscope or a Zeiss Axio Observer Z.2 fluorescent microscope. SCRINSHOT imaging was performed using a Zeiss Axio Observer Z.2 fluorescent microscope with a Colibri led light source, equipped with a Zeiss AxioCam 506 Mono digital camera and an automated stage. Quantification of cell number was performed using ImageJ (http://rsbweb.nih.gov/ij/).

Statistical analyses were performed using GraphPad software. P values were calculated by Student's t-test P values were calculated by Student's t-test. P<0.05 was considered significant.

We thank Radhan Ramadass, Anoop Cherian and Yu Hsuan Carol Yang for imaging assistance, Ramesh-Kumar Krishnan, Thomas Sontag and Sigrid Einemann for discussions and/or assistance.