A WNT4- and DKK3-driven canonical to noncanonical Wnt signaling switch controls multiciliogenesis Free

The multiciliated cells lining the conducting airways are our first line of defense against inhaled pathogens, allergens, toxins and debris (Brooks and Wallingford, 2014). They each have 200–300 cilia whose concerted, directional motility sweeps contaminants out of the lungs. Diminished mucociliary clearance due to multiciliated cell loss and ciliary dysfunction is a major driver of chronic inflammatory airway diseases, including asthma, cystic fibrosis, chronic obstructive pulmonary disease and chronic rhinosinusitis (Adam et al., 2015; Gohy et al., 2019; Leung et al., 2019; Van Bruaene and Bachert, 2011). Thus, it is critical to ensure that the proper number of functional multiciliated cells are generated and maintained within the airways.

Multiciliated cells are derived from basal stem cells or secretory progenitors (Brooks and Wallingford, 2014). Upon cell fate acquisition, nascent multiciliated cells turn on a massive multiciliated cell-specific transcriptional program to express hundreds of ciliary genes. During ciliogenesis, basal bodies (the base of cilia) first assemble in the cytoplasm, then traffic to and dock with the apical plasma membrane, and finally extend the motile axoneme (the external portion of cilia).

Wnt signaling has emerged as a critical regulator of multiciliated cell formation and function (Boscke et al., 2017; Haas et al., 2019; Malleske et al., 2018; Schmid et al., 2017; Boutin et al., 2014; Park et al., 2008; Vladar et al., 2012). Wnt signaling comprises canonical and noncanonical pathways (Nusse and Clevers, 2017). Both types are initiated by the binding of a secreted WNT ligand to a Frizzled (FZD) family receptor, which leads to the activation of its intracellular binding partner Dishevelled (DVL) (Nusse and Clevers, 2017). In the canonical or β-catenin-dependent pathway (Wnt/β-cat), WNT stimulation leads to the stabilization and nuclear entry of the β-catenin effector to turn on target genes (Valenta et al., 2012). The Wnt/planar cell polarity (Wnt/PCP) program, the best-understood noncanonical pathway, regulates directional cell behaviors like polarized morphology, division and migration via the cytoskeleton, independently of β-catenin (Vladar et al., 2009).

During multiciliated epithelial differentiation, Wnt/β-cat is activated to specify multiciliated cell fate; then it is turned off. Wnt/PCP is then activated to control the apical transport of basal bodies during ciliogenesis (Park et al., 2008; Boutin et al., 2014). Subsequently, Wnt/PCP regulates the distal to proximal (lung to oral) alignment of motile cilia for directional mucociliary clearance (Vladar et al., 2012). In each multiciliated cell, individual cilia are physically oriented by microtubules towards a membrane domain occupied by the FZD–DVL complex on the proximal side apical cell–cell junctions. Once thought to be distinct, canonical and noncanonical Wnt pathways are now known to share regulatory components, and cells can switch between modes in a context-dependent manner. Proper multiciliated cell formation has been shown to require a switch from Wnt/β-cat to Wnt/PCP as prolonged Wnt/β-cat activation either by a small-molecule agonist or WNT3A canonical ligand treatment blocked Wnt/PCP activation and ciliogenesis (Boscke et al., 2017; Haas et al., 2019; Malleske et al., 2018). This led us to propose that instead of two independent Wnt signaling events, a canonical to noncanonical Wnt signaling switch regulates multiciliated cell formation. Central to this proposed mechanism is the timely inhibition of Wnt/β-cat signaling after cell fate acquisition to permit the Wnt/PCP-dependent differentiation of nascent multiciliated cells.

We hypothesized that WNT ligands and secreted canonical Wnt/β-cat regulators might play critical upstream roles in regulating ciliogenic Wnt signaling. The airway epithelium is known to express many of the 19 mammalian WNTs and about as many secreted regulators (Nusse, 1997-2023), although none has been specifically implicated in multiciliated cell formation. We previously revealed that the FZD3 and FZD6 receptors are likely to be involved in the orientation of cilia via Wnt/PCP signaling (Vladar et al., 2012), but their ligand(s) are unknown. WNT5A, the prototypical noncanonical ligand is an attractive candidate, as it regulates Wnt/PCP in the inner ear and the embryonic node (Minegishi et al., 2017; Qian et al., 2007). Wnt5a-knockout mice display airway axis elongation defects, although this was shown to be due to the loss of mesenchymal WNT5A (Kishimoto et al., 2018; Li et al., 2002). WNT4, another noncanonical ligand and known binding partner of FZD6 (Lyons et al., 2004), was demonstrated to be expressed during airway epithelial differentiation (Schmid et al., 2017) and to regulate distal lung morphogenesis (Caprioli et al., 2015). WNT3A and WNT10A have both been shown to regulate canonical Wnt/β-cat signaling in the stem cells of the developing submucosal glands of the airways (Driskell et al., 2007; Lynch et al., 2016). Secreted canonical Wnt/β-cat regulators comprise multiple families of related factors (Cruciat and Niehrs, 2013). Members of the secreted Frizzled related protein (sFRP1–sFRP5) and the Dickkopf WNT signaling pathway inhibitor (DKK1–DKK4) family are the most well characterized. Many, such as sFRP1 and DKK1, are well known secreted inhibitors of canonical Wnt/β-cat, whereas others, such as DKK3, have been shown to activate or suppress Wnt/β-cat signaling. Most have yet to be implicated in airway epithelial development, although DKK1 has been shown to be expressed during multiciliated cell differentiation (Schmid et al., 2017), and sFRP2 has been linked to promoting inflammation in chronic lung disease (Zhou et al., 2019). Identifying specific upstream factors and mechanisms in ciliogenic Wnt signaling will provide much needed insight into multiciliated cell generation and maintenance.

Here, we identify WNT4, a noncanonical ligand and DKK3, a member of secreted canonical Wnt regulator family as novel drivers of multiciliated cell formation and polarized ciliary orientation. We show both in mouse airway tissues and human primary cultures that WNT4 is expressed in the airway epithelium predominantly during developmental and regenerative differentiation and to a lesser extent during adult tissue maintenance. WNT4 is expressed by basal stem cells. Our data indicate that it signals to multiciliated target cells via the FZD6 noncanonical Wnt/PCP receptor. We demonstrate that DKK3 is contemporaneously expressed with WNT4 by the same population of basal cells, and our data suggest that it binds to the multiciliated cell surface via KREMEN1. We show that both WNT4 and DKK3 are necessary for multiciliated cell formation and act by suppressing the canonical Wnt/β-cat pathway and promoting a switch to noncanonical Wnt/PCP to regulate the formation of properly oriented motile cilia for optimal mucociliary clearance.

We sought to identify secreted Wnt regulator(s) that might control multiciliated cell formation in the airway epithelium. We assessed the expression of all 19 WNT ligands, the Porcupine (PORCN) and Wntless (WLS) WNT ligand secretion regulators, and 18 well-known secreted Wnt regulators (alternative ligands and signaling inhibitors and activators) (see http://web.stanford.edu/group/nusselab/cgi-bin/wnt/; accessed 13 July 2023) in recent, publicly available human fetal lung and adult airway single-cell RNA sequencing (scRNAseq) datasets [Deprez et al., 2020; Miller et al., 2020; accession numbers EGAS00001004082 (European Genome-Phenome Archive) and E-MTAB-8221 (ArrayExpress), respectively] (Fig. S1A,B). As expected, based on the broad and fundamental roles for the Wnt signaling pathway in lung development (Beers and Morrisey, 2011), transcriptional analysis suggested some form of Wnt pathway activation in all compartments of the lung. During fetal development, secreted Wnt regulator expression was mainly detected in mesenchymal cells and airway epithelial cells. In the adult, virtually all Wnt ligand expression and secretion (based on PORCN and WLS expression) was in airway epithelial cells, whereas some secreted Wnt regulators (FRZB, RSPO3 and SFPR4) remained only mesenchymally expressed. Given that Wnts are short-range ligands (Boutros and Niehrs, 2016), we focused on factors with strong airway epithelium-enriched expression, namely, WNT2B, WNT3A, WNT4, WNT5A, WNT5B, WNT7B, WNT9A, WNT10A, DKK1 and DKK3.

Primary human bronchial epithelial cells (HBECs) cultured at an air–liquid interface (ALI) activate both canonical Wnt signaling to determine multiciliated cell fate and noncanonical Wnt/PCP signaling to build and orient motile cilia during in vitro differentiation (Boscke et al., 2017; Qian et al., 2007; Schmid et al., 2017; Vladar et al., 2012, 2015) (Fig. 1A; Fig. S2A,B). Thus, we used qRT-PCR to test the expression of candidate secreted Wnt regulators during a time-course of HBEC culture spanning proliferation, differentiation and maturity (Fig. 1B; Fig. S1C). Airway epithelial expression was confirmed for all candidates. WNT4, WNT5A, WNT10A and DKK3 showed a peak of expression during early differentiation coincident with multiciliated cell formation, as indicated by the peak of expression for the FOXJ1 multiciliated cell specific transcription factor. WNT5B and WNT9A expression peaked in mature HBECs, which might indicate a multiciliated cell maintenance role for these ligands. We selected two noncanonical Wnt ligands for further analysis – WNT4, a FZD6-binding partner (Lyons et al., 2004), and WNT5A, a known Wnt/PCP regulator (Qian et al., 2007).

To test whether WNT4 or WNT5A are potential regulators of airway epithelial Wnt/PCP signaling, we used transmission electron microscopy (TEM) to assess polarized ciliary orientation in Wnt4 and Wnt5a germline knockout (Caprioli et al., 2015; Li et al., 2002) mouse tracheas. In the developing mouse airways, multiciliated cells first appear at embryonic day 16.5 (E16.5) and continue to form and mature until the first weeks of life (Francis et al., 2009). We previously demonstrated that, during development, the orientation of ciliary basal feet (polarized appendages on cilia that indicate the beat direction; Boisvieux-Ulrich et al., 1985) becomes increasing more refined towards the proximal (oral) direction for anatomically directed mucociliary clearance (Vladar et al., 2012, 2015) (Fig. S2C). The angle of deviation from the proximal direction for basal feet is used to derive metrics of intracellular (r_cell) and tissue-wide (r_trachea) ciliary orientation (Fig. S2D). Both Wnt4 and Wnt5a homozygous germline mutants die at birth (Caprioli et al., 2015; Li et al., 2002), so we assessed basal feet at E18.5. Wnt5a^−/− homozygous mutants (r_cell=0.90±0.04; r_trachea=0.74; mean±s.e.m.) and wild-type littermates (r_cell=0.85±0.03; r_trachea=0.79) both exhibited similarly proximally oriented cilia (Fig. 1C). In contrast, Wnt4^−/− homozygous mutants had strongly misoriented cilia at E18.5 (r_cell=0.68±0.04; r_trachea=0.47±0.06) compared to wild-type littermates (r_cell=0.88±0.03; r_trachea=0.77±0.02; Fig. 1D). Adult Wnt4^+/− heterozygous mutants also exhibited slightly misoriented cilia (r_cell=0.77±0.02; r_trachea=0.88±0.03) compared to wild-type littermates (r_cell=0.97±0.01; r_trachea=0.95±0.01, Fig. 1E). The extent of ciliary misorientation in Wnt4^−/− embryos was comparable to what we previously reported for E18.5 Vangl1CKO^−/− or Vangl2^Lp/Lp Wnt/PCP mutant embryos (Vladar et al., 2012). Lack of misorientation in Wnt5a^−/− mutants suggests that Wnt5a is dispensable for ciliary planar polarity, although we cannot rule out its contribution to other noncanonical Wnt signaling events in the airway epithelium. We conclude that Wnt4 is required for ciliary orientation potentially by regulating airway epithelial Wnt/PCP signaling.

As WNT4 is a secreted ligand, we used RNA fluorescence in situ hybridization (RNAscope) to characterize its spatiotemporal expression in the airway epithelium. First, we took advantage of the proximal-distal developmental gradient of the mouse airways to compare the more well-differentiated tracheal to the relatively undifferentiated distal airways at E18.5 (Warburton et al., 2010). Epithelial Wnt4 expression was present throughout the airway tree but was more robust in the proximal compared to the distal portions (Fig. 2A; Fig. S3A). Wnt4 was detectible prior to the appearance of mature multiciliated cells (Fig. S3A). Primary HBEC culture differentiation qRT-PCR (Fig. 1B) and RNAscope (Fig. 2B) time-course data also showed that the highest WNT4 expression was coincident with multiciliated cell formation. HBEC time-course WNT4 protein expression matched the gene expression profile (Fig. S3B). Consistent with secreted ligand function, WNT4 protein was only detected in cell culture medium harvested from differentiating, an not proliferating, HBECs (Fig. S3C).

Wnt4 and WNT4 was detected in the human and the mouse adult airway epithelium, respectively (Fig. 2C,D). In the adult mouse lung, we found uniformly abundant epithelial Wnt4 expression along the proximal-distal airway tree from the trachea to the bronchoalveolar duct junction (Fig. S3D). Consistent with a role in distal lung formation (Caprioli et al., 2015), we also detected parenchymal Wnt4 expression at E18.5, which was lower than Wnt4 expression in the airway epithelium (Fig. S3A). Adult parenchymal tissues had only weak Wnt4 signal (Fig. S3D).

We asked whether WNT4 plays a role in regeneration following airway epithelial luminal cell destruction by naphthalene injury. We detected strongly increased Wnt4 in regenerating airway epithelia 7 days post-injury as luminal cells underwent differentiation from progenitor cells (Fig. S3E). This is consistent with increased WNT4 expression during human primary HBEC culture differentiation (Figs 1B, 2B), a model of in vitro airway epithelial regeneration from adult stem cells. In sum, our data indicate that WNT4 is predominantly an airway epithelial ligand that is secreted during developmental and regenerative airway epithelial differentiation as well as adult tissue maintenance. Although WNT4 is critical for the proper orientation of cilia, the lack of graded expression along the proximal-distal axis of the lung (Fig. S3D) suggests that likely it is not a directional cue for Wnt/PCP signaling.

To determine the cellular basis of WNT4 signaling, we sought to identify the cell type(s) in the airway epithelium that express WNT4. scRNAseq data from human fetal lung and adult airways (Deprez et al., 2020; Miller et al., 2020) indicate that WNT4 is expressed predominantly by airway epithelial basal stem cells (Fig. S1A,B). WNT4 expression is enriched in the TP63 transcription factor positive basal cells (Vieira Braga et al., 2019) (TP63+ BCs, Fig. 3A), which is consistent with prior studies identifying WNT4 as a direct TP63 transcriptional target (Osada et al., 2006). WNT4 basal cell-enriched expression was confirmed using RNAscope with KRT5 co-immunolabeling in adult human bronchial tissue sections and in a wholemount labeled HBEC differentiation time-course (Fig. 2B,C). WNT4 transcripts were also weakly detectible in rare luminal cells, but not in mesenchymal cell types. Colabeling adult mouse tracheal sections with Krt5 revealed that Wnt4 is also enriched in murine basal cells (Fig. 2A). In the mouse small airways that lack basal cells, Wnt4 was detected in and might derive from SCGBA1A-expressing secretory progenitor cells (Fig. 2A).

Compellingly, scRNAseq data showed that the same TP63+ BCs that express WNT4 also express DKK3, a member of the DKK family of secreted regulators of the canonical Wnt/β-cat pathway (Schmid et al., 2017) (Fig. 3A). Our HBEC qRT-PCR time-course demonstrated that WNT4 and DKK3 are also expressed at the same time, peaking contemporaneously with early multiciliated cell formation in the airway epithelium (Fig. S1C). This suggests that WNT4 and DKK3 might co-regulate ciliogenic Wnt signaling.

To identify WNT4 target cell(s), we carried out wholemount immunolabeling of differentiating HBECs with a WNT4 antibody. Non-ciliated cells contained a very faint, uniform WNT4 signal (Fig. 3B). Multiciliated cells contained two pools of WNT4, first, an asymmetric localization overlapping the apical junctions, and second, localization to large cytoplasmic puncta (Fig. 3B). WNT4 localization to multiciliated cells was greatly enhanced in HBECs treated with recombinant human WNT4 (rhWNT4, Fig. 3B). We observed asymmetric WNT4 membrane localization in both ciliating and mature multiciliated cells [distinguished by the localization and abundance of the acetylated α-tubulin signal (Vladar and Stearns, 2007), Fig. 3B]. Ciliating cells had more cytoplasmic puncta compared to mature multiciliated cells (Fig. 3B,C). We interpret these localization patterns as WNT4 binding to the asymmetrically localized FZD3 and/or FZD6 receptors (Fig. 1A) and being internalized by ciliating cells. Lack of compatible antibodies prevented directly testing WNT4 colocalization with the FZDs in HBECs. However, WNT4 localization to multiciliated cells was eliminated in FZD6 CRISPR knockout (FZD6 KO) HBECs (Fig. 3B; Fig. S4A–D).

Antibody labeling of HBECs revealed that DKK3 also localizes to multiciliated cells (Fig. 4A), which is completely eliminated in DKK3 CRISPR KO HBECs (Fig. S4E). Unlike WNT4, which localizes asymmetrically to the apical junctions, DKK3 is uniformly distributed on the multiciliated cell apical surface, with faint signal also detectible on the ciliary axonemes (Fig. 4A). Furthermore, we show that DKK3 colocalizes with the DKK family receptor KREMEN1 (Cruciat and Niehrs, 2013; Nakamura and Hackam, 2010) on the multiciliated cell surface (Fig. 4B,C), with a Manders' coefficient of colocalization of 0.97±0.03 (mean±s.e.m., 1.00=perfect colocalization). In support of KREMEN1 acting as a multiciliated cell-resident receptor for DKK3, we show that DKK3 signal is eliminated in KREMEN1 CRISPR KO HBECs (Fig. 4D). Although colocalization and dependence for multiciliated cell localization cannot be considered as direct evidence of WNT4–FZD6 or DKK3–KREMEN1 binding, our gene expression and localization data (Figs 1B, 2C; Fig. S3B,C) strongly suggest that in the majority of the conducting airways, multiciliated cells act as signal-receiving cells for airway stem-cell derived WNT4 via FZD6 and for DKK3 via KREMEN1 during multiciliated cell formation and maintenance.

To better characterize the role of WNT4 in multiciliated cell formation in the airway epithelium, we generated two independent WNT4 CRISPR knockout (WNT4 KO) HBECs (Fig. S4A–C). Multiciliated cells were identified based on nuclear FOXJ1 signal, which marks both ciliating and mature multiciliated cells, or by labeling with γ-tubulin (marks basal bodies) and acetylated α-tubulin (marks axonemes), which can distinguish ciliating versus mature multiciliated cells (Vladar and Brody, 2013; Vladar and Stearns, 2007) (see Fig. S2A). Interestingly, we observed that WNT4 KO HBECs contained overall fewer FOXJ1+ multiciliated cells compared to control (scrambled CRISPR guide RNA, scr; Fig. 5A,B). Multiciliated cell numbers were strongly reduced during differentiation [10 days (10d) of culture after air–liquid interface (ALI) creation; denoted ALI+10d], but still remained significantly lower in mature (ALI+21d) cultures. Colabeling with γ-tubulin and acetylated α-tubulin also showed that compared to scr control, ALI+21d WNT4 KO HBECs contained fewer ciliated cells and more immature multiciliated cells based on the presence of cells with basal bodies (γ-tubulin), but no or only short and sparse axonemes (acetylated α-tubulin, Fig. 5C). In contrast, WNT4 KO HBECs had no change in the number of mucus secretory cells (Fig. S5A), indicating a specific defect in making multiciliated cells. Conversely, we demonstrated that rhWNT4 treatment during differentiation increased FOXJ1+ multiciliated cell numbers in a dose-dependent manner (Fig. 5D,E). Ectopic rhWNT4 treatment of mature HBECs was also able to slightly increase multiciliated cell number (Fig. S5B), indicating that progenitors remain responsive to WNT4.

We tested the specificity of WNT4 in regulating multiciliated cell formation by generating CRISPR KO HBECs of other Wnt family members expressed in the airway epithelium during differentiation (Fig. S1C). Consistent with a previous study where all secreted Wnts were eliminated in the mouse airways (Aros et al., 2020), we show that CRISPR KO of Wnt ligand secretion mediator (WLS) in HBECs blocks multiciliated cell formation (Fig. S5C). WNT5B, WNT9A and WNT10A CRISPR KO HBECs all had slightly lower numbers of multiciliated cells during differentiation at ALI+10d, but normal numbers at maturity at ALI+21d (Fig. S5C). This suggests that although WNT4 has a critical role in the differentiation of multiciliated cells, we cannot rule out a minor direct or indirect contribution from other WNTs. Similar to WNT4, FZD6 KO HBECs had reduced multiciliated cell numbers at both ALI+10d and ALI+21d (Figs S4D and S5C), further supporting a role for WNT4–FZD6 signaling for multiciliated cell formation.

We demonstrated that, similar to WNT4, DKK3 is also expressed during differentiation by the same subset of basal cells and is detected on the multiciliated cell surface (Fig. S1C;Figs 3A, 4A). Thus, we set out to test whether DKK3 is also required to form multiciliated cells. We found that DKK3 KO HBECs (Fig. S4E) had a profound loss of multiciliated cells as indicated by loss of both FOXJ1-positive cells (Fig. 6A,B) and γ-tubulin and acetylated α-tubulin markers in multiciliated cells (Fig. 6C; Fig. S5D). In contrast, rhDKK3 increased multiciliated cell numbers in differentiating HBECs (Fig. 6D,E).

Consistent with KREMEN1 acting as a receptor for DKK3 (Nakamura and Hackam, 2010), we showed that KREMEN1 CRISPR KO (Fig. S4F) phenocopied the multiciliated cells loss displayed by DKK3 KO HBECs (Fig. 6A–C). DKK3 KO and KREMEN1 KO HBEC epithelia were poorly organized, with irregular cell shapes and sizes (Fig. 6C; Fig. S4E,F). However, we observed no significant change in mucus secretory cell numbers (Fig. S5A), suggesting that loss of DKK3 or KREMEN1 does not simply block epithelial differentiation, but specifically disrupts multiciliated cell formation. This is further supported by the ability of DKK3 KO and KREMEN1 KO HBECs to generate nearly normal numbers of multiciliated cells in response to treatment with DAPT (Fig. 6F,G), an inhibitor of Notch signaling, which promotes multiciliated cell formation (Stubbs et al., 2012; Tsao et al., 2009).

Our screen for airway epithelial secreted Wnt regulators also identified DKK1 (Fig. S1A), another well-known member of the DKK family of canonical Wnt/β-cat regulators that was previously shown to localize to multiciliated cells (Schmid et al., 2017). However, we eliminated DKK1 as a candidate ciliogenesis regulator, as its expression was enriched during HBEC proliferation, not differentiation (Fig. S1C). Furthermore, unlike rhDKK3, rhDKK1 treatment had no effect on multiciliated cell formation in HBECs (Fig. 5D,E).

Consistent with the loss of multiciliated cells in DKK3 KO HBECs, we also observed defective multiciliated cell formation in adult Dkk3^−/− germline knockout mouse (Barrantes Idel et al., 2006) airways. Dkk3^−/− airways contained fewer ciliated cells as identified by both FOXJ1 and acetylated α-tubulin immunolabeling (Fig. S6A–C). Multiciliated cell loss was evident throughout the airway tree, including in the trachea (not shown), but was most severe in the small and terminal airways. Greatly reduced acetylated α-tubulin signal intensity and volume also suggests that Dkk3^−/− multiciliated cells might contain shorter and/or sparser motile cilia (Fig. S6A). However, similar to normal numbers of mucus secretory cells in DKK3 KO HBECs (Fig. S5A), we did not detect changes in the number of SCGB1A1-expressing secretory cells in Dkk3^−/− airways (Fig. S6A). In sum, our data suggest that DKK3 to KREMEN1 signaling is required for multiciliated cell formation.

We and others have shown that prolonged canonical Wnt/β-cat signaling blocks multiciliated cell formation and Wnt/PCP activation (Boscke et al., 2017; Haas et al., 2019; Malleske et al., 2018; Schmid et al., 2017). Given that WNT4 has been shown to block canonical Wnt/β-cat signaling (Bernard et al., 2008), and DKK3 can act as a canonical Wnt/β-cat inhibitor (Caricasole et al., 2003), we asked whether they promote multiciliated cell formation by downregulating Wnt/β-cat. To detect active Wnt/β-cat signaling, we carried out qRT-PCR for AXIN2, a well-known Wnt/β-cat target gene. We first validated this assay in HBECs by showing increased AXIN2 expression upon treatment with the Wnt/β-cat agonist CHIR-99021 (Fig. S7A). We show that in a healthy HBEC differentiation time-course, AXIN2 peaks during differentiation, then its expression plummets indicating that Wnt/β-cat is turned off as the culture reaches maturity (Fig. 7A). We found that in contrast to scrambled control KO HBECs, mature WNT4 KO and DKK3 KO HBECs had increased AXIN2 gene expression at ALI+21d (Fig. 7B), indicating a failure to downregulate the Wnt/β-cat pathway.

Next, we asked whether loss of WNT4 or DKK3 disrupts the Wnt/PCP pathway. Asymmetric localization of VANGL1 and FZD6, central regulators of Wnt/PCP, central regulators of Wnt/PCP the pathway is the gold standard readout for Wnt/PCP activity in the airway epithelium (Vladar et al., 2012, 2015). We asked whether WNT4 and/or DKK3 are required for VANGL1 and FZD6 asymmetric junctional localization by immunolabeling CRISPR KO HBECs at ALI+21d. We found that VANGL1 and FZD6 asymmetric localization was still detectible in some WNT4 KO cells; however, their levels were greatly diminished (Fig. 7C,D). This suggests that WNT4 signaling through FZD6 is required to establish or maintain robust core Wnt/PCP membrane domains. However, the reciprocal was not observed, as we detected unperturbed Wnt4 expression in Wnt/PCP mutant (Vangl1CKO^−/−) mouse tracheas (Fig. S7B). In contrast, asymmetric junctional VANGL1 and FZD6 were undetectable in DKK3 KO HBECs, indicating a defect in Wnt/PCP (Fig. 7C,D). In sum, our data are consistent with a model where multiciliated cell formation requires the contemporaneous secretion of WNT4 and DKK3 from basal cells to block canonical Wnt/β-cat signaling and turn on Wnt/PCP in nascent multiciliated cells (Fig. 7E).

Our studies demonstrate that multiciliated cell formation requires a canonical to noncanonical signaling switch controlled by basal cell derived WNT4, a noncanonical Wnt ligand, and DKK3, a member of the DKK family of secreted canonical Wnt regulators. We propose that this novel mechanism links multiciliated cell fate acquisition to ciliogenesis and ciliary orientation to ensure that the airway epithelium contains the proper number of multiciliated cells and that these mature in a timely manner for optimal mucociliary clearance. The control of multiciliated cell formation by sequentially deployed Wnt pathways has been previously established (Boscke et al., 2017; Haas et al., 2019; Malleske et al., 2018; Schmid et al., 2017; Boutin et al., 2014; Park et al., 2008; Vladar et al., 2012), but the mechanisms that control the switch from canonical to noncanonical signaling were not known. By eliminating all Wnt ligand secretion using Porcn mutant mice (Aros et al., 2020), a basal cell-derived Wnt ligand was previously shown to be important for multiciliated cell formation. Here we sought to identify the Wnt ligand(s) controlling this process.

Through a gene expression-based screen, we identified WNT4 as a novel regulator of multiciliated cell formation. WNT4 was already known to be crucial for lung development as Wnt4 germline knockout mice display hypoplastic distal lungs and defective cartilage formation (Caprioli et al., 2015). In humans, WNT4 loss of function leads to SERKAL (sex reversion, kidneys, adrenal and lung dysgenesis) syndrome (Mandel et al., 2008). Mechanisms underlying these defects are still incompletely understood. Our data show that WNT4 is expressed predominantly in the airway epithelium and to a lesser extent in parenchymal cells, suggesting that it might regulate multiple processes. We implicate airway epithelial WNT4 in the regulation of multiciliated cell formation by showing that Wnt4 germline knockout mice have misaligned motile cilia, and WNT4 CRISPR knockout airway epithelial cells make fewer multiciliated cells.

Our studies are consistent with a mechanism whereby airway epithelial WNT4 expressed by basal cells signals to multiciliated cells in a FZD6-dependent manner to block Wnt/β-cat and stimulate Wnt/PCP signaling. This model is strongly supported by WNT4 biology established in prior studies. First, WNT4 was previously shown to be expressed in the airway epithelium during differentiation (Schmid et al., 2017). Here, we show that it is specifically expressed by the TP63+ subset of basal cells. This is strengthened by the fact that WNT4 is a known direct transcriptional target of TP63 (Osada et al., 2006). However, given that TP63+ BCs are present during differentiation and homeostasis, additional factors are likely needed to control the temporal expression of WNT4. Second, WNT4 has been shown to physically interact with the extracellular domain of FZD6 in vitro and signal via FZD6 in vivo in Xenopus (Lyons et al., 2004). FZD6 is a well-established regulator of Wnt/PCP, including, as demonstrated through our work, in the orientation of airway cilia (Guo et al., 2004; Vladar et al., 2012; Wang et al., 2006). Here, we demonstrate that FZD6 is necessary for WNT4 to target to and to be internalized by multiciliated cells. Finally, supporting our result that WNT4 controls the ciliogenic Wnt signaling switch, studies have shown that WNT4 can suppress the Wnt/β-cat pathway and activate β-catenin-independent signaling (Bernard et al., 2008).

We identified several other airway epithelial WNTs that we cannot completely rule out as additional regulators of multiciliated cell formation or function. WNT5B, WNT9A and WNT10A CRISPR KO HBECs also had slightly reduced multiciliated cell numbers during differentiation, but unlike WNT4, multiciliated cells were normal at maturity. This might indicate that these proteins have a minor contribution to ciliogenesis for these other ligands or potentially broader functions in epithelial differentiation or structure that could delay multiciliated cell formation. Curiously, WNT9A is exclusively expressed by multiciliated cells, but more studies are needed to test its signaling function.

We also identify DKK3 as a novel regulator of multiciliated cell formation, and we propose that it acts by blocking Wnt/β-cat during the ciliogenic Wnt switch. These results are strengthened by the loss of multiciliated cells in both DDK3 CRISPR KO HBECs and in homozygous mutant Dkk3 germline knockout mouse airways. DKK3 is a member of the Dickkopf WNT signaling pathway inhibitor (DKK1–DKK4) family. Although other DKK family members, especially DKK1 are better characterized, DKK3 is less well understood. DKK1 was previously shown to be expressed during ciliogenesis and localize to multiciliated cells (Schmid et al., 2017); however, our qRT-PCR results indicate that DKK1 is exclusively expressed during HBEC proliferation. We also show that ectopic DKK3, but not DKK1, promotes multiciliated cell formation. Although we cannot rule out a contribution by DKK1, our data are more consistent with DKK1 regulating Wnt-dependent proliferative events instead of multiciliated cell formation.

DKK family members act by binding to the KREMEN1 receptor, which then inhibits the interaction between FZD and the LRP5 and LRP6 (LRP5/6) co-receptor necessary to transduce canonical Wnt/β-cat signals (Cruciat and Niehrs, 2013). LRP5/6 does not participate in noncanonical pathways. Therefore, DKK-dependent sequestration of LRP5/6 might not only block Wnt/β-cat, but also allow FZDs to switch to transducing noncanonical signals. DKK3 has been demonstrated to interact with KREMEN1 (Nakamura and Hackam, 2010). Our data showing that KREMEN1 and DKK3 colocalize and exhibit identical loss-of-function phenotypes in HBECs and that DKK3 can no longer be detected on the multiciliated cell surface in KREMEN1 KO HBECs are strongly supportive of this interaction. Further studies are needed to demonstrate direct DKK3–KREMEN1 binding in multiciliated cells. Although DKK1, DKK2 and DKK4 have been shown to act as inhibitors of Wnt/β-cat signaling, DKK3 is not a proven inhibitor. Studies indicate that DKK3 can potentiate as well as inhibit Wnt/β-cat signaling (Caricasole et al., 2003; Nakamura and Hackam, 2010). Our data showing increased AXIN2 Wnt/β-cat target gene expression in DKK3 KO HBECs suggests that it acts as a Wnt/β-cat inhibitor in multiciliated cells.

Our data indicate that DKK3 and WNT4 might act together on nascent multiciliated cells. They are contemporaneously expressed by the same TP63+ BCs. Compellingly, similar to WNT4, DKK3 is also a known TP63 transcriptional target (Kajiwara et al., 2018). Both WNT4 and DKK3 are detected on the multiciliated cell surface in a pattern that supports binding to their putative cognate receptors; FZD6 for WNT4, and KREMEN1 for DKK3. And finally, WNT4 and DKK3 share CRISPR knockout and ectopic expression phenotypes. Thus, we propose a model where upon multiciliated cell fate acquisition, DKK3 blocks Wnt/β-cat via interacting with KREMEN1, which allows WNT4-dependent activation of FZD6 to turn on Wnt/PCP to promote cilium biogenesis and orientation. Future studies are needed to delineate intracellular Wnt/β-cat mechanisms and Wnt/PCP-dependent cytoskeletal dynamics downstream of FZD6 activation.

Multiciliated cell formation is controlled by multiple signaling pathways, including Notch, IL-6-STAT3 and TGF-β (Brooks and Wallingford, 2014). It will be critical to understand the relationship between ciliogenic Wnt signaling and these other key regulators. We show that inhibition of Notch signaling, which is known to be required for generation of multiciliated cells (Tsao et al., 2009), was able to rescue the lack of multiciliated cell formation in DKK3 and KREMEN1 KO HBECs. This suggests that Wnt/β-cat signaling might be acting upstream of the Notch competition required to select cells for multiciliated versus secretory cell fates. Similarly, it will be critical to evaluate Wnt switch mechanisms in diseased epithelia with multiciliated cell loss and ciliary dysfunction. Canonical and noncanonical Wnt signaling are both known to be aberrantly regulated in chronic airway diseases (Baarsma and Konigshoff, 2017; Vladar and Konigshoff, 2020). Using chronic rhinosinusitis as a model, we previously showed that Wnt/PCP is blocked in these airways, and Wnt ligand expression, including WNT4, is broadly disrupted (Boscke et al., 2017). WNT4 is known to be aberrantly expressed in COPD (Durham et al., 2013), but detailed studies are needed to establish mechanisms, which might implicate a Wnt switch failure in mucociliary clearance dysfunction.

Wnt4, Wnt5a, and Dkk3 germline knockout and Vangl1CKO conditional knockout mice have been previously described (Caprioli et al., 2015; Li et al., 2002; Barrantes Idel et al., 2006; Vladar et al., 2012). E18.5 Wnt4^−/− and Wnt5a^−/− and adult Wnt4^+/− and Dkk3^−/− lungs were obtained from heterozygote crosses. Adult (8–12 weeks) Vangl1CKO^−/− and Vangl1CKO^+/− mice were obtained from heterozygote crosses of Vangl1CKO mice that were mated to the HPRT-Cre deleter line (JAX) to generate germline deletion of Vangl1 (Vladar et al., 2012). C57BL/6 wild-type mice were obtained from Jackson Laboratory. Male and female mice were used in this study. Naphthalene (NA) airway epithelial injury was carried out using 200 µl total of NA diluted in corn oil (200 µg NA per g mouse weight) or corn oil only administered intraperitoneally. NA leads to luminal cell death in the trachea, which is fully regenerated over 2 weeks (Van Winkle et al., 1995; Hong et al., 2004). All procedures involving animals were approved by the Institutional Animal Care and Use Committee of Stanford University School of Medicine and The University of Texas MD Anderson Cancer Center, and in accordance with established guidelines for animal care.

Primary human bronchial epithelial cells (HBECs) were isolated from non-transplanted healthy human lungs obtained in a deidentified manner via the National Jewish Health Human Lung Tissue Consortium (Denver, CO). The use of human tissues and cells was carried out under approval from the Institutional Review Board of the University of Colorado School of Medicine. Informed consent was obtained for all tissue donors and all investigation has been conducted according to the principles expressed in the Declaration of Helsinki. Primary air–liquid interface (ALI) culture of bronchial epithelial cells from passage 1 (P1) basal stem cells was carried out as previously described (Vladar et al., 2016; Fulcher et al., 2005). HBECs were cultured in medium made in-house as per Vladar et al. (2016). Proliferation medium is formulated using Ham's F-12 medium (Thermo Fisher Scientific, 11330-032) with 1.5 mM L-glutamine (Thermo Fisher Scientific, 25030-149), 0.03% sodium bicarbonate (Thermo Fisher Scientific, 25080-094), 100 U/ml penicillin-streptomycin (Thermo Fisher Scientific, 15140-148), 0.1% Fungizone (Thermo Fisher Scientific, 15290-018), 10 µg/ml insulin (Sigma, I1882), 25 ng/ml epidermal growth factor (BD Biosciences, 354001), 5 µg/ml apo-transferrin (Sigma, T1147), 0.1 µg/ml cholera toxin (Sigma, C8052), 0.5% bovine pituitary extract (Hammond Cell Tech, 1078-NZ), 5% fetal bovine serum (Thermo Fisher Scientific, 26140-079), and 50 nM retinoic acid (Sigma, R2625). Differentiation medium is formulated using Ham's F-12 medium with 1.5 mM L-glutamine, 0.03% sodium bicarbonate, 100 U/ml penicillin-streptomycin, 0.1% Fungizone, 2% NuSerum I (BD Biosciences, 355100) and 50 nM retinoic acid. Briefly, epithelial cells were dissociated from airway segments using overnight enzymatic digestion. Basal cells were first expanded on collagen I-coated plastic dishes in proliferation medium supplemented with Y-27632 (ROCK inhibitor, 10 µM), DMH-1 (BMP inhibitor, 1 µM), A-83-01, (TGF-β inhibitor, 1 µM) and CHIR-99021 (WNT agonist, 1 µM) (Mou et al., 2016), all from Selleckchem, then cryopreserved at P1. Freshly thawed P1 HBECs were seeded onto Transwell filters, initially cultured submerged in proliferation medium until confluency, then lifted to ALI (considered as ALI+0d of culture) by supplying differentiation medium only from the bottom compartment. HBECs are considered mature at ALI+21d of culture. HBECs from at least n=3 age-matched healthy donors were used for CRISPR knockout, qRT-PCR, and immunohistochemistry experiments. HBECs used in the study were cultured in medium made in-house as per Vladar et al. (2016). HBECs were treated with 250 or 400 ng/ml recombinant human WNT4 or 100 ng/ml DKK1 or DKK3 (R&D Systems) in the differentiation medium from ALI+7 to 10d of culture. HBECs were treated with 1 µM DAPT (Selleckchem) in the differentiation medium from ALI+14 to ALI+21d of culture. See Fig. S2B for more info.

Lentiviral CRISPR knockout vectors were generated using two independent guide RNA sequences per target gene. Guides were designed using the Benchling software, purchased as oligonucleotides and subcloned into the lentiCRISPRv2 vector (Addgene, plasmid no. 52961). Alternately, ready-made lentiviral CRISPR plasmids were ordered from VectorBuilder, Inc. For the KREMEN1 knockout, both guide RNAs were cloned tandem into a single vector (designated as KO1/2). For all other targets, each guide RNA was cloned into a separate vector (designated as KO1 and KO2). For CRISPR guide RNA sequences, see Table S1. Lentivirus was prepared in house in the 293 T/17 cell line using the psPAX2 and pMD2.G helper plasmids (Addgene) according to published methods (Vladar and Brody, 2013). 293 T/17 cells were purchased directly from ATCC, cryopreserved after two passages, and used only up to passage 10, after which new frozen aliquots wrtr used. No additional authentication is used. P1 bronchial epithelial basal cells cultured on collagen I-coated plastic dishes were infected, puromycin selected, expanded and cryopreserved. The now P2 scrambled guide RNA control (scr) and CRISPR knockout basal cells were thawed directly onto Transwell filters and ALI differentiated as above. CRISPR knockout was validated by high resolution melt curve analysis according to published methods (Everman et al., 2018). Briefly, genomic DNA was isolated from scrambled guide RNA control and CRISPR knockout HBEC cultures. The CRISPR target region was amplified using the MeltDoctor HRM Kit (Thermo Fisher Scientific) and analyzed using the high-resolution melt curve module in a QuantStudio 7 Real-Time PCR System (Thermo Fisher Scientific). For primer sequences, see Table S2.

cDNA was prepared from HBECs using standard methods. qRT-PCR was performed in triplicate with Power SYBR Green Master Mix (Thermo Fisher Scientific) in a QuantStudio 7 Real-Time PCR System (Thermo Fisher Scientific). Gene expression was evaluated using the ΔΔCt method. For primer sequences, see Table S2.

WNT4 was detected in HBEC ALI culture lysates by western blotting according to standard methods. Secreted WNT4 was detected in conditioned medium harvested from pooled HBEC preALI and ALI culture media as previously described (Rao et al., 2019). Conditioned medium harvested from 293 T/17 cells transduced by the WNT4:pLX304 overexpression construct (gift from M. Sikora, University of Colorado) was used as positive control. For antibodies, see Table S3.

For wholemount immunofluorescence, HBECs were fixed in −20°C methanol or 4% paraformaldehyde for 10 min as previously described (Vladar and Brody, 2013). Transwell filters were cut out of the plastic supports and placed in a humid chamber for staining. Samples were blocked in 10% normal horse serum and 0.1% Triton X-100 in PBS and incubated with primary antibodies (Table S3) for 1–2 h, then with Alexa Fluor dye-conjugated secondary antibodies (Thermo Fisher Scientific) for 30 min at room temperature. Filters were mounted in Mowiol mounting medium containing 2% N-propyl gallate (Sigma). RNA in situ hybridization (RNAscope, ACD Bio-Techne) in conjunction with immunolabeling was carried out according to the manufacturer's instructions using the human WNT4 (cat. no. 429441) or the mouse Wnt4 (cat. no. 401101) probes. Samples were imaged with a Leica SP8 or Zeiss LSM900 confocal microscope. Mouse immunohistochemistry and RNAscope images are representative of left and right lung sections from n>3 mice. Human immunohistochemistry and RNAscope images are representative of tissues from n=3 donors. HBEC images are representative of cultures derived from n=3 donors immunolabeled in triplicate. To test for colocalization, Manders' overlap coefficient (reported as 0–1; 1=perfect overlap) was calculated using the JACoP Plugin in ImageJ (NIH). For antibodies, see Table S3.

Mouse tracheal tissue preparation and image analysis were as previously described (Vladar et al., 2012, 2015). Briefly, excised tracheas were fixed in 2% glutaraldehyde, 4% paraformaldehyde in 0.1 M NaCacodylate buffer, pH 7.4 at 4°C overnight. Samples were osmicated, stained with uranyl acetate, then dehydrated with a graded ethanol series and infiltrated with EMbed-812 (Electron Microscopy Sciences). 80–100-nm sections were mounted onto copper grids and analyzed with a JEOL JEM-1400 microscope using a Gatan Orius Camera at the Stanford Cell Sciences Imaging Facility (CSIF). Proximal airway direction was tracked throughout the procedure. To assess basal body orientation, the angle of basal foot orientation with respect to the proximal direction was measured in TEM images using the ImageJ software (NIH). The mean vector of basal body orientation per cell (a_cell) and the length of the mean vector (r_cell=1 – circular variance), representing the dispersion of basal body directions per cell, were calculated using the Oriana v3.0 software (Kovach Computing Services). The degree of intracellular orientation (r_cell) was reported as mean±s.e.m. and compared among conditions using a two-tailed unpaired Student's t-test. To describe intercellular orientation in the trachea, the tracheal mean vector (a_trachea) was determined by averaging a_cell values per condition, and the length of the mean tracheal vector (r_trachea) was calculated to show the dispersion of directions. Note that thin TEM cross sections do not capture all the basal bodies in a cell. Data are represented on circular plots with the proximal direction set at 0°, and each arrow describes the orientation of a single multiciliated cell with the angle of the arrow showing the mean vector of basal body orientation and the length of the mean vector representing the coherence of orientation within that cell (r_cell). See Fig. S2C,D for more info.

We thank Dr Roel Nusse (Stanford) for the gift of the Wnt5a mice, Dr Matthew Sikora (University of Colorado) for help with WNT4 protein assays, Klara Fekete (Stanford University) for help with animal husbandry and John Perrino (Stanford Cell Sciences Imaging Facility) for help with electron microscopy. This work is dedicated in fond memory of Kara Patricia Geraci, a passionate and talented young scientist, who died a few months after successfully completing the human and mouse tissue expression experiments.