Hippo signaling promotes lung epithelial lineage commitment by curbing Fgf10 and β-catenin signaling

The mammalian lung has a well-designed anatomical structure that allows for efficient gas exchange: an arborized network of airways, each ending distally in a vast number of alveoli that form an extensive surface area within a limited chest volume. The framework of the respiratory tree is sculpted in the branching morphogenesis stage of lung development, which is completed at ∼E16 in the mouse and followed by an alveolar epithelial differentiation program (Alanis et al., 2014; Chang et al., 2013). During branching morphogenesis, distal tip epithelial progenitors are maintained and undergo a branching program driven by the transcription factor Sox9, the expression of which is induced by Fgf10. Fgf10, expressed and released by the adjacent lung mesenchyme, binds and activates Fgfr2 on distal tip epithelial progenitors, which results in the induction of KRAS and β-catenin signaling and downstream Sox9 expression (Abler et al., 2009; Bellusci et al., 1997; Mailleux et al., 2005; Ramasamy et al., 2007; Ustiyan et al., 2016; Volckaert et al., 2013; Volckaert and De Langhe, 2015). As progeny of distal tip epithelial progenitors becomes displaced from the source of Fgf10, Sox9 expression steadily decreases, whereas Sox2 expression is induced allowing this progeny to differentiate into bronchial epithelial cells (Abler et al., 2009; Bellusci et al., 1997; De Langhe et al., 2008, 2006; Nyeng et al., 2008; Ramasamy et al., 2007; Volckaert et al., 2013). At E16 in the mouse, a fate switch occurs during which Sox9^pos epithelial progenitors no longer give rise to Sox2^pos bronchial epithelial cells but instead give rise to Sox2^neg bipotential alveolar epithelial progenitors, which differentiate into alveolar type 1 (AT1) and 2 (AT2) epithelial cells (Desai et al., 2014). Although it is unclear how reduced Fgf10 signaling proximally orchestrates the differentiation of the progeny of Sox9^pos distal tip epithelial progenitors, overexpression of Fgf10/KRAS from E15.5 until E18.5 was shown to maintain Sox9 expression and to block alveolar epithelial differentiation (Alanis et al., 2014; Chang et al., 2013; Rockich et al., 2013; Volckaert et al., 2013), indicating that Fgf10 expression and signaling is delicately controlled.

Cues provided by the extracellular matrix (ECM) are key regulators of cell proliferation, death and differentiation. However, whether and how ECM signals impinge on the epithelial-mesenchymal growth factor signaling crosstalk during lung development is not known. ECM-derived signaling is mediated by integrins (Sun et al., 2016), which link matrix proteins through dynamically regulated signaling hubs (also called integrin adhesome) to the actomyosin cytoskeleton (Winograd-Katz et al., 2014). Integrins are indispensable for branching morphogenesis and cellular differentiation in several branching organs, and the lung is no exception (Chen and Krasnow, 2012; Mathew et al., 2017; Plosa et al., 2014; Shih et al., 2016; Zhang et al., 2009). ECM-integrin signals have been shown to regulate the nuclear translocation of the transcriptional activators Yap and Taz, which are downstream effectors of the Hippo pathway: when active, Hippo signaling activates a kinase cascade, in which Mst1 and Mst2 phosphorylate and activate Lats1 and Lats2, which in turn phosphorylate Yap and Taz, and thereby promote their cytoplasmic retention and/or degradation (Panciera et al., 2017). However, if Hippo signaling is curbed, Yap and Taz translocate into the nucleus, associate with TEAD and TEF transcription factors, and activate gene transcription required for cell proliferation and survival.

Genetic studies in mice have shown that Yap and Taz play important roles during lung development. Yap is nuclear in the distal Sox9^pos epithelial progenitors during lung development and becomes phosphorylated and therefore cytoplasmic as progeny acquire the Sox2^pos bronchial epithelial fate (Mahoney et al., 2014). Yet, counterintuitively, inactivation of Yap expands the Sox9^pos domain (Mahoney et al., 2014), which is indicative of a failure in lineage commitment of Sox9^pos distal tip epithelial progenitors and suggests an important role for cytoplasmic Yap. On the other hand, a permanent nuclear retention of Yap and Taz as a result of the epithelial-specific inactivation of the Mst1 (Stk4) and Mst2 (Stk3) genes in the developing lung, appears to maintain the branching program and impairs alveolar epithelial differentiation (Chung et al., 2013), similar to lungs overexpressing Fgf10 or Kras (Alanis et al., 2014; Chang et al., 2013; Rockich et al., 2013; Volckaert et al., 2013). A reason for these defects could be the ability of cytoplasmic Yap to orchestrate the degradation of β-catenin (Azzolin et al., 2014), the nuclear effector of the canonical Wnt pathway and a known inducer of Sox9 expression in the distal progenitors (Hashimoto et al., 2012; Ostrin et al., 2018; Volckaert et al., 2013). Although this possibility has never been directly investigated in the lung, it has been shown that β-catenin is only stabilized and present in the nucleus downstream of Wnt signaling, when Yap and Taz are also in the nucleus or when Yap and Taz expression is abrogated (Azzolin et al., 2014, 2012). Conversely, phospho- or cytoplasmic β-catenin is essential for Taz degradation (Azzolin et al., 2014, 2012) and ablation of Taz leads to defective alveologenesis (Makita et al., 2008; Park et al., 2004).

Despite the prominent role of Hippo signaling in the developing lung, it remains unclear how the Hippo signaling cascade is turned on and off or how its signaling strength is modulated. There is evidence that growth factors such as Tgfβ, Wnt and Notch (Azzolin et al., 2014; Esteves de Lima et al., 2016; Kim et al., 2017; Saito and Nagase, 2015), and adhesion molecules such as integrins influence the Hippo signaling pathway by regulating cell size, polarity and contractility (Dupont, 2016; Lee and Streuli, 2014; Panciera et al., 2017; Szymaniak et al., 2015). How growth factors and integrins engage the Hippo signaling pathway remains still elusive. A central and essential adaptor protein in integrin-containing adhesion sites [also called focal adhesion (FA)] is the integrin-linked pseudo-kinase (Ilk), which associates, via its pseudo-kinase domain, with integrins and, via its binding partners PINCH and parvin, with the actomyosin cytoskeleton to regulate cell migration, polarity, matrix remodeling and proliferation (Ghatak et al., 2013; Widmaier et al., 2012). We have shown that Ilk inactivation in the airway epithelium of adult lungs inhibits Hippo signaling by destabilizing Merlin, a well-known Hippo activator (Volckaert et al., 2017). The resulting nuclear localization of Yap and Taz drives epithelial Wnt7b secretion, which acts on adjacent airway smooth muscle cells to release Fgf10. This epithelial-mesenchymal cascade is activated in injured airways to initiate regeneration (Volckaert et al., 2011). However, if this crosstalk is sustained after regeneration is completed, epithelial Fgf10 signaling becomes chronic, resulting in airway remodeling and fibrosis. We also reported that changes in the biophysical quality of the injured lung tissue are sensed by lung cells, resulting in a series of molecular events between epithelium and mesenchyme that re-direct epithelial fate (Volckaert et al., 2017).

In the present paper, we show that Ilk during lung development activates Hippo signaling to promote epithelial lineage commitment by inhibiting β-catenin and Fgf10 signaling. At the same time, we show that ablation of Yap results in increased β-catenin and Fgf10 signaling, and that simultaneous genetic ablation of the Ctnnb1 gene rescues the cystic and hypoplastic phenotype of lungs lacking Yap in epithelial cells. We further demonstrate redundant and non-redundant functions for the two nuclear effectors of the Hippo pathway during lung development in which they regulate each other's stability. Together, our data point to a role for cytoplasmic Yap in promoting maturation of the developing lung epithelium by curbing β-catenin and Fgf10 signaling.

In the adult lung, Ilk positively regulates the Hippo pathway to inhibit epithelial Wnt7b and downstream mesenchymal Fgf10 expression (Volckaert et al., 2017). Tightly regulated Fgf10 expression is essential to determine the fate of distal Sox9^pos epithelial progenitors during lung development (Volckaert et al., 2013). To investigate whether Ilk also engages the Hippo pathway to tune mesenchymal Fgf10 expression by activating Hippo signaling during lung development, we inactivated Ilk specifically in the lung epithelium using Shh-Cre;Ilk^f/f;Fgf10^LacZ mice. Shh-Cre;Ilk^f/f;Fgf10^LacZ lungs showed an expansion of distal Sox9^pos epithelial progenitor cells featuring strong nuclear localization of Yap, increased expression of Wnt7b and downstream mesenchymal Fgf10 expression (Fig. 1A-C,F and Fig. S1A-C) (De Langhe et al., 2008; Rajagopal et al., 2008; Volckaert et al., 2013) as well as severe lung branching defects, with 68±11% (n≥4; P=0.009) fewer and dilated distal tips. These changes were accompanied by the increased expression of the Fgf10 target genes Spry2 and Bmp4 (Fig. 1B,C) (Bellusci et al., 1997; Mailleux et al., 2005, 2001). Sox9 and Wnt7b are β-catenin target genes (Hashimoto et al., 2012; Ostrin et al., 2018), suggesting that Ilk could curb β-catenin signaling by preventing nuclear Yap translocation, which has been shown to promote β-catenin degradation (Azzolin et al., 2014, 2012). To investigate whether inactivation of Ilk resulted in increased canonical epithelial Wnt signaling, we generated Shh-Cre;Ilk^f/f;Topgal mice and found an increase in epithelial Wnt signaling, which was confirmed by increased levels of Axin2 expression (Fig. S1A and Fig. 1C).

To investigate whether inactivation of Yap also results in increased β-catenin signaling and downstream Wnt7b and Fgf10 expression, we inactivated Yap specifically in the lung epithelium using Shh-Cre;Yap^f/f mice. Our data indicate increased nuclear β-catenin, as well as increased expression of the β-catenin target genes Wnt7b and Axin2, and downstream Fgf10 expression (Fig. 1D,E). Interestingly, Shh-Cre;Yap^f/f lungs also showed a marked reduction in Shh expression, another negative regulator of Fgf10. Immunostaining further revealed that loss of Yap resulted in an expansion of the Sox9-expressing domain (Fig. 1F,G and Fig. S2A,B) in E14.5 Shh-Cre;Yap^f/f lungs, similar to what was previously reported (Mahoney et al., 2014). However, surprisingly we also found a simultaneous expansion of the Sox2-expressing domain all the way down to the distal tip (Fig. 1F,G and Fig. S2A,B), suggesting that both nuclear Yap and β-catenin are required to prevent the differentiation of distal tip epithelial progenitors and that nuclear Yap may inhibit Sox2 expression, whereas nuclear β-catenin promotes Sox9 expression. Interestingly, we also identified a striking lung defect upon deletion of one Yap allele in Shh-Cre;Yap^f/+ lungs, which have dilated distal tips and an expanded Sox9- but not Sox2-expressing domain. The finding that both increased nuclear Yap and loss of Yap result in increased epithelial β-catenin signaling, Wnt7b and Fgf10 expression, and an expansion of the Sox9^pos domain of distal tip epithelial progenitors suggests a role for cytoplasmic Yap in regulating epithelial lineage commitment in the developing lung likely in part by orchestrating the degradation of β-catenin (Azzolin et al., 2014; Hashimoto et al., 2012; Ostrin et al., 2018; Volckaert et al., 2013).

To dissect the role of Ilk in epithelial cell lineage commitment, we next ablated the Ilk gene using the Nkx2.1-Cre line, which is active late during lung development. In line with published observations (Volckaert et al., 2017), inactivation of Ilk in Nkx2.1-Cre;Ilk^f/f mice resulted in the destabilization of Merlin, reduced phosphorylation of Mst1, Mst2 and Yap, and increased levels of nuclear Taz (Fig. 2A-O).

Until E16.5, the mouse lung undergoes a branching program driven by the transcription factor Sox9, which is expressed in epithelial progenitors at the distal branch tips and regulated by Fgf10, which itself is expressed in the distal mesenchyme (Bellusci et al., 1997; Mailleux et al., 2005; Volckaert et al., 2013, 2011). Fgf10 maintains the distal tip epithelial progenitors by inducing Sox9 expression (Volckaert et al., 2013) and suppressing Sox2 expression, thereby preventing them from differentiating into proximal (bronchial) epithelial cells (Bellusci et al., 1997; De Langhe et al., 2008; Volckaert et al., 2013). At around E16.5 in the mouse, the lung transitions from a branching program into an alveolar epithelial differentiation program. During this stage, Sox9-expressing distal epithelial progenitors no longer give rise to Sox2-expressing bronchial epithelial cells but instead start to differentiate into bipotential alveolar epithelial progenitors that, over time, give rise to alveolar type I (AT1) and AT2 epithelial cells (Desai et al., 2014; Treutlein et al., 2014). We and others have previously shown that ubiquitous overexpression of Fgf10 from E15.5 until E18.5 blocks the alveolar epithelial differentiation program by maintaining Sox9 expression (Alanis et al., 2014; Chang et al., 2013; Rockich et al., 2013; Volckaert et al., 2013). Interestingly, we found that Nkx2.1-Cre;Ilk^f/f lungs also failed to transition from a branching program into an alveolar epithelial differentiation program, and featured an expansion of Sftpc^pos/Rage^pos bipotential progenitors at E18.5 compared with control lungs, which featured proper AT1 and AT2 cell differentiation (Fig. 3A,B,D,U, Fig. 4H,I,X,Z and Fig. S4A-C). In addition, Nkx2.1-Cre;Ilk^f/f lungs featured an increase in Cgrp-expressing neuroendocrine bodies (NEBs) (Figs 3E,F,H,U and 4T,U,X), impaired ciliated cell differentiation showing reduced Foxj1 expression (Fig. 3I,J,U) and ectopic differentiation of basal cells in the conducting airways (Figs 3L,M,U and 4P,Q,X,Y) similar to what we previously reported for lungs overexpressing Fgf10 (Fig. 4R,V,X) (Volckaert et al., 2013). Interestingly, we also found increased nuclear Yap in lungs overexpressing Fgf10 (Fig. S3D-I), suggesting that Fgf10-Yap may signal in a feedback loop.

To investigate whether the increased Fgf10 expression in Nkx2.1-Cre;Ilk^f/f could be rescued by reducing Yap and Taz levels we inactivated only one allele of Yap and Taz in Nkx2.1-Cre;Ilk^f/f;Yap^f/+;Taz^f/+ mice, which partially rescued the Nkx2.1-Cre-Ilk^f/f phenotype. Nkx2.1-Cre;Ilk^f/f;Yap^f/+;Taz^f/+ lungs displayed reduced Wnt7b, Fgf10 and Sox9 expression (Fig. 3U), and exhibited proper alveolar epithelial differentiation (Fig. 3A-D,U). Furthermore, lineage specification of conducting airway epithelium along the ciliated cell lineage was normal (Fig. 3I-K,U) without a basal cell expansion (Fig. 3L-N,U). However, we still observed increased numbers of neuroendocrine cells (Fig. 3E-H,U). Immunostaining and qPCR for Ilk were performed to demonstrate efficient Ilk deletion in both mutant and rescue lungs (Fig. 3O-U).

When both Yap and Taz alleles were completely inactivated in combination with Ilk in Nkx2.1-Cre;Ilk^f/f;Yap^f/f;Taz^f/f triple knockouts, we observed cystic lungs with severe defects in alveolar epithelial differentiation (Fig. S3A-C), similar to recently reported Sftpc-rtTA;Tet-cre;Yap^f/f;Taz^f/f lungs induced from E15.5 (Nantie et al., 2018) and Nkx2.1-Cre;Yap^f/f (Fig. 5B,F), Shh-Cre;Yap^f/f lungs (Lin et al., 2017; Mahoney et al., 2014) or lungs expressing a hyperactive β-catenin in epithelial cells (Hashimoto et al., 2012; Mucenski et al., 2005; Okubo and Hogan, 2004). Together, these findings indicate that Ilk promotes epithelial differentiation and sacculation by positively regulating Hippo signaling and point to a role for cytoplasmic Yap (active Hippo signaling) in epithelial lineage commitment.

Sox9 is expressed in distal tip epithelial progenitors, where it acts downstream of Fgf/Kras and Fgf/β-catenin to orchestrate branching and inhibit premature initiation of alveolar epithelial differentiation (Chang et al., 2013). In agreement with this observation, we also found Sox9 expression limited to a few clustered and cuboidal epithelial progenitors in the distal branches of control E18.5 lungs (Fig. 4D,H). In sharp contrast, the distal tip epithelium of E18.5 Nkx2.1-Cre;Ilk^f/f lungs consisted of many, densely packed cuboidal epithelial progenitor cells (Fig. 4E,I), which are normally only observed during the branching or pseudoglandular stage of lung development. In line with the expansion of the progenitor cell population, the expression of the distal tip epithelial progenitor markers Sox9 and Sftpc was also dramatically increased (Fig. 4D,E,H,I,X,Z), indicating sustained activity of the branching program and inhibition of alveolar epithelial differentiation in Nkx2.1-Cre;Ilk^f/f lungs. Indeed, flattened differentiated AT1 cells were absent in Ilk-deficient distal lung epithelium, whereas flattened RAGE^pos AT1 cells were clearly visible in control lungs (Fig. 4L,M,X,Z). Interestingly, arrested lung maturation and sustained activation of the branching program was also observed in lungs overexpressing Fgf10 (Fig. 4F,J,N,X,Z) (Volckaert et al., 2013) or Kras (Alanis et al., 2014; Chang et al., 2013), and in Nkx2.1-Cre;Mst1/2, Shh-Cre;Mst1/2^f/f (Chung et al., 2013; Lange et al., 2015) and Shh-Cre;Nf2^f/f lungs (Fig. S4D-L).

As distal tip epithelial progenitor cell maintenance is crucially dependent on the expression of Fgf10 by the lung mesenchyme, we hypothesized that mesenchymal Fgf10 expression is increased in Nkx2.1-Cre;Ilk^f/f mutant lungs. To test this hypothesis, we monitored Fgf10 expression in Nkx2.1-Cre;Ilk^f/f mutant lungs by crossing them with an Fgf10^LacZ line, and indeed found elevated numbers of Fgf10-expressing lipofibroblasts in the mesenchyme, which was abnormally thickened (Fig. 4A-C). These observations were further confirmed by qPCR analysis, which revealed increased Fgf10 as well as Wnt7b expression in Nkx2.1-Cre-Ilk^f/f lungs (Fig. 4X).

The increase in Fgf10 expression in Nkx2.1-Cre;Ilk^f/f lungs and the similar phenotype in Fgf10-overexpressing lungs, Nkx2.1-Cre;Mst1/2^f/f lungs (Chung et al., 2013), Shh-Cre;Mst1/2^f/f and Shh-Cre;Nf2^f/f lungs (Fig. S4D-L), suggest that elevated Fgf10 signaling is the underlying cause of epithelial maturation defects produced by inactivation of the Hippo pathway.

To test this hypothesis, we investigated whether the Nkx2.1-Cre-Ilk^f/f lung phenotype can be rescued by inhibiting Fgfr2b signaling. To achieve this, we crossed a transgenic line overexpressing sFgfr2b, a secreted, Fgf10-sequestering Fgfr2b receptor, with the Nkx2.1-Cre;Ilk^f/f strain to obtain Nkx2.1-Cre;Ilk^f/f;Rosa26-rtTa;Tet-sFgfr2b mice. Overexpressing sFgfr2b at E15.5 in the absence of Ilk partially rescued the alveolar epithelial differentiation defects in E18.5 Nkx2.1-Cre;Ilk^f/f lungs, as the majority of distal tip epithelial progenitors in Nkx2.1-Cre;Ilk^f/f;Rosa26-rtTa;Tet-sFgfr2b lungs switched from the branching to the alveolar epithelial differentiation program, demonstrated by reduced Sox9 expression and the presence of flattened RAGE-expressing AT1 cells (Fig. 4G,K,O,X,Z). We also observed ciliated cells and normal numbers of neuroendocrine cells (Fig. 4S,W-Y) indicating that the conducting airway underwent epithelial lineage specification and that epithelial differentiation defects in Nkx2.1-Cre;Ilk^f/f lungs are indeed a consequence of elevated Fgf10 expression. Together, these data indicate that Ilk promotes bronchial and alveolar epithelial lineage commitment by engaging Hippo signaling and curbing β-catenin and Fgf10 signaling.

Conditional epithelial inactivation of Yap during early lung development (Shh-Cre;Yap^f/f) results in an expansion of both the Sox9^pos (Mahoney et al., 2014) and Sox2^pos domain (Fig. 1F,G and Fig. S2A,B), pointing to a role for nuclear Yap in repressing and cytoplasmic Yap in promoting differentiation. This finding is in line with the observation that the Hippo pathway is active in differentiated epithelium (Fig. 2G,M and Fig. S2B) and inactive in distal tip epithelial progenitors (Fig. 1A,F; Figs S1B,C and S2B) during lung development (Mahoney et al., 2014; Szymaniak et al., 2015). Altogether, these findings suggest that inactivation of Hippo signaling (leading to nuclear Yap) represses differentiation, whereas activation of Hippo signaling (leading to cytoplasmic accumulation of Yap) promotes differentiation. As cytoplasmic Yap was shown to destabilize β-catenin (Azzolin et al., 2014, 2012), we hypothesized that a similar crosstalk between the Hippo and β-catenin signaling pathways ensures normal lung maturation during development. To investigate whether similar interactions between Yap and β-catenin may also occur in the lung, we performed immunostaining for Yap and β-catenin on NEBs in the developing lung, which are known to feature strong nuclear β-catenin (Li et al., 2009, 2013). Interestingly, we found that NEBs did not express Yap but showed strong nuclear β-catenin, suggesting that Yap may also block β-catenin in the lung (Fig. S5).

To further investigate this, we generated Nkx2.1-Cre;Yap^f/f and Nkx2.1-Cre;Yap^f/f;βcat^f/f compound mutant mice. Although it would theoretically be interesting to also generate Shh-Cre;Yap^f/f;βcat^f/f mice, Shh-Cre;βcat^f/f mice have previously been shown not to develop lungs (Goss et al., 2009; Harris-Johnson et al., 2009), therefore making this experiment impossible. Interestingly, we found that Nkx2.1-Cre;Yap^f/f lungs featured large dilated epithelial sacs (Fig. 5A,B,D,E). The proximal epithelium in Nkx2.1-Cre;Yap^f/f lung was correctly specified, containing club and ciliated cells (Fig. 5Q), but demonstrated an increase in Cgrp (Calca) expression indicative of more and/or larger NEBs in the proximal developing conducting airways of Nkx2.1-Cre;Yap^f/f mice (Fig. 5Q and Fig. S3J-N). The increase in NEBs was also observed in lungs expressing an activated version of β-catenin or lacking Apc in the lung epithelium (Li et al., 2009, 2013), whereas fewer NEBs exist in developing airways that overexpress the Wnt antagonist Dkk1, which inactivates the Wnt signaling pathway (Volckaert et al., 2013). Interestingly, we found Taz, a binding partner of β-catenin in the nucleus, to be upregulated in Nkx2.1-Cre;Yap^f/f lungs, indicating that Yap also negatively regulates Taz (Fig. 5G-J).

In contrast to Nkx2.1-Cre;Yap^f/f mice, Nkx2.1-Cre;Yap^f/f;βcat^f/f lungs showed normal sacculation and epithelial differentiation but reduced Taz expression (Fig. 5B,C,E,F,H-J,Q), suggesting that Yap, Taz and β-catenin signaling are functionally linked to orchestrate the late stages of lung development. Surprisingly, rescued Nkx2.1-Cre;Yap^f/f;βcat^f/f lungs showed an increase in basal cells (Fig. 5N-Q), which was unexpected as Yap is thought to be required for basal cell maintenance (Volckaert et al., 2017; Zhao et al., 2014), suggesting that increased β-catenin signaling downstream of Yap ablation may cause the loss of basal cells. To verify β-catenin and Yap deletion, we performed immunostaining and qPCR for Yap and β-catenin on control, Nkx2.1-Cre;Yap^f/f and Nkx2.1-Cre;Yap^f/f;βcat^f/f lungs (Fig. 5K-M,Q). To verify the specificity of the Taz antibody, we also performed immunostaining for Taz on Shh-Cre;Taz^f/f lungs (Fig. S6I-K). Interestingly, we also found a reduction in AT1 cell differentiation in Shh-Cre-Cre;Taz^f/f lungs at E18.5 compared with control lungs (Fig. S6A,B,E,F) and a cumulative even stronger phenotype resembling Nkx2.1-Cre;Yap^f/f knockouts in lungs in Shh-Cre-Cre;Taz^f/f;Yap^f/+ lungs but not Shh-Cre-Cre;Taz^f/+;Yap^f/+ lungs (Fig. S6A-H). Together, our data suggest redundant and non-redundant roles for the nuclear effectors of the Hippo pathway, whereby each transcription factor may regulate the other as well as the nuclear activity of β-catenin.

Taz, the other nuclear effector of the Hippo pathway, is known to be required for alveologenesis (Makita et al., 2008). Our data using Shh-Cre-Cre;Taz^f/f lungs suggest that epithelial Taz executes these effects. To further investigate the role of Taz in alveologenesis, we inactivated Taz using Nkx2.1-Cre;Taz^f/f mice. At P21, Nkx2.1-Cre;Taz^f/f lungs displayed defective alveologenesis and reduced levels of the AT1 cell marker RAGE (Fig. 6A,B,P). As Yap and Taz have redundant and non-redundant, gene dose-dependent roles, we also investigated whether Yap partially compensates for Taz inactivation by simultaneously inactivating one allele of Yap. We found that Nkx2.1-Cre;Taz^f/f;Yap^f/+ lungs demonstrated further deterioration of alveolar epithelial integrity, eventually resulting in the development of lung fibrosis (Fig. 6A-P). Interestingly, the distal alveolar epithelium in Nkx2.1-Cre;Taz^f/f;Yap^f/+ lungs is lined with immature Sox9^pos alveolar epithelial progenitors (Fig. 6D-F,P). Finally, Taz immunostaining revealed that Taz is primarily expressed in AT1 cells and is efficiently deleted in both Nkx2.1-Cre;Taz^f/f and Nkx2.1-Cre;Taz^f/f;Yap^f/+ lungs (Fig. 6M-O). Together, our findings indicate that, during early lung development as well as bronchial epithelial lineage differentiation, Hippo signaling is primarily mediated by Yap, whereas both nuclear effectors of the Hippo pathway are required for proper epithelial lineage commitment and maintenance during alveologenesis, with Taz playing a major role in AT1 cell differentiation.

Signals emanating from ECM-integrin adhesion sites can be converted into biochemical responses by the transcription factors Yap and Taz (Dupont et al., 2011; Halder and Johnson, 2011). How exactly these cues are transduced to regulate Yap and Taz activity is less well understood. We have recently uncovered a role for the adaptor protein Ilk in maintaining adult airway epithelial quiescence by preventing nuclear Yap translocation and inhibiting the regenerative Wnt7b-Fgf10 signaling crosstalk (Volckaert et al., 2017). Here, we analyzed the role of Ilk during lung development and show that Ilk prevents nuclear accumulation of Yap and Taz in the developing airway epithelium, and thereby promotes epithelial lineage commitment of distal tip epithelial progenitors by curbing Fgf10 and β-catenin signaling.

We found that Nkx2.1-Cre;Ilk^f/f lungs fail to enter the alveolar epithelial differentiation program and sacculation phase, and were able to rescue this phenotype by reducing Yap and Taz expression levels or by inhibiting Fgf10 signaling. These findings are in line with previous reports describing that Fgf10/KRAS overexpression during late lung development keeps the lung in the branching program and inhibits alveolar development (Alanis et al., 2014; Chang et al., 2013). A similar lung maturation arrest and altered respiratory epithelial cell differentiation occurs in Nkx2.1-Cre;Mst1/2^f/f mice (Chung et al., 2013) and Shh-Cre;Mst1/2^f/f and Shh-Cre;Nf2^f/f lungs.

Our data uncover Ilk as an upstream negative regulator of Yap and Taz, which functionally interact with β-catenin signaling to control lineage commitment of conducting airway and alveolar epithelial cells during lung development. Yap/Taz are integral members of the β-catenin destruction complex, and their nuclear translocation helps β-catenin to escape degradation and accumulate in the nucleus (Azzolin et al., 2014). As genetic Yap deletion cripples the β-catenin destruction complex, β-catenin becomes stabilized, which explains the expansion of the distal Sox9^pos domain (Mahoney et al., 2014). Interestingly, as a corollary, phospho- or cytoplasmic β-catenin is essential for Taz degradation (Azzolin et al., 2014, 2012), which may explain the observed increase in nuclear Taz in Yap knockout lungs. Taz is thought to mediate about 50% of downstream Wnt/β-catenin-mediated transcription (Azzolin et al., 2014, 2012). However, although Yap blocks both β-catenin and Taz, Taz is not expressed during early lung development. Therefore, Yap inactivation during early lung development versus late lung development results in a different signaling outcome.

Growth factor signaling provides another layer of Yap/β-catenin-mediated control over branching and differentiation programs, as Fgf10/Kras and Fgf10/β-catenin signaling induce the expression of the branching program driver Sox9 in distal tip epithelial progenitors (Chang et al., 2013; Chatzeli et al., 2017; Seymour et al., 2012; Volckaert et al., 2013), and overexpression of Fgf10 is sufficient to drive β-catenin activation (Volckaert et al., 2013, 2011) and nuclear Yap. Our data demonstrate that mechanical cues via an ECM-integrin-ILK-Merlin-Hippo-Yap-β-catenin-Fgf10 regulatory network drive lung development. Although it is well-established that Wnt ligand-mediated β-catenin signaling plays an important role in epithelial specification and differentiation (De Langhe et al., 2005; Li et al., 2005; Volckaert et al., 2013; Volckaert and De Langhe, 2015), our data suggest that Yap functions as a rheostat of epithelial β-catenin activity during lung development (Fig. 7). It is clear that the processes of branching morphogenesis and epithelial differentiation must be tightly coordinated to generate an optimal gas-exchange surface. Our data indicate that Ilk-mediated mechanical control of Yap/Taz feed back into the same pathways that underlie tissue patterning and provide a means to coordinate branching morphogenesis with epithelial lineage commitment.

All mice were bred and maintained in a pathogen-free environment with free access to food and water. Both male and female mice were used for all experiments. Nkx2.1-Cre (JAX 008661), Shh-Cre (Harfe et al., 2004), Tet-sFgfr2b (JAX 0025672) (Hokuto et al., 2003), Ctnnb1^f/f (JAX 004152), Taz^f/f;Yap^f/f (JAX 030532), Tet-Fgf10 (Clark et al., 2001), Rosa26-rtTa (Volckaert et al., 2011), Fgf10^LacZ (Kelly et al., 2001), TOPGAL (JAX 004623), Ilk^f/f (Grashoff et al., 2003), Mst1/2^f/f (JAX 017635), Nf2^f/f (Giovannini et al., 2000) and Wnt7b^f/f (Rajagopal et al., 2008) mice have been previously described. All experiments were approved by the University of Alabama at Birmingham Institutional Animal Care and Use Committee (IACUC). For doxycycline induction: pregnant dams were placed on doxycycline-containing food (rodent diet with 625 mg/kg dox; Harlan Teklad TD.09761). The day on which a vaginal plug was observed was denoted as E0.5.

Fgf10^LacZ and TOPGAL lungs were dissected and fixed in 4% formalin in PBS at room temperature for 5 min and rinsed in PBS. For staining, X-Gal (Research Products International, B71800) was dissolved in dimethylformamide and freshly added to LacZ staining solution [5 mM potassium hexacyanoferrate(III), 5 mM potassium hexacyanoferrate(II) trihydrate, 2 mM MgCl₂, 81 mM Na₂HPO₄ and 20 mM NaH₂PO₄] at a final concentration of 1 mg/ml. Lungs were stained at 37°C in the dark until signal was strong enough. After rinsing with PBS, lungs were postfixed in 4% formalin in PBS at room temperature overnight.

All staining was carried out on paraffin sections of formalin-fixed lungs. Immunofluorescent staining was performed with the following primary antibodies: mouse anti-β-tubulin (1:500; 3F3-G2; Seven Hills Bioreagents), goat anti-Scgb1a1 (1:200; clone T-18; sc-9772; Santa Cruz Biotechnology), rabbit anti-Scgb1a1 (1:500; WRAB-CCSP; Seven Hills Bioreagents), guinea pig anti-ADRP (Adipophilin) (1:200; 20R-AP002; Fitzgerald Industries), rabbit anti-β-galactosidase (1:500; 100-4136; Rockland Immunochemicals Inc.), mouse anti-α-actin (1:500; clone 1A4; sc-32251; Santa Cruz Biotechnology), rabbit anti-keratin 5 (1:200; clone EP1601Y; MA5-14473; Thermo Fisher Scientific), rabbit anti-ILK (1:50; 3862S; Cell Signaling Technology), mouse anti-p63 (1:200; clone 4A4; CM163B; Biocare Medical), rabbit anti-phospho-Mst1/2 (1:250; 3681S; Cell Signaling), rabbit anti-Sftpc (1:200; WRAB-9337; Seven Hills Bioreagents), goat anti-Sftpc (1:500; M-20; sc-7706; Santa Cruz Biotechnology), rabbit anti-phospho Yap (1:200; S127; Cell Signaling Technology), rabbit anti-YAP (1:50; 4912S; Cell Signaling Technology), mouse anti-Hopx (1:100; E-1; sc-398703 Santa Cruz Biotechnology), rabbit anti-Merlin (Nf2) (1:250; clone A-19; sc-331; Santa Cruz Biotechnology), mouse anti-Sox2 (1:25; E-4; sc-365823; Santa Cruz Biotechnology), rabbit anti-Sox2 (1:1000; WRAB-1236; Seven Hills Bioreagents), mouse anti-E-Cadherin (1:200; 610181; BD Transduction Laboratories), goat anti-Sox9 (1:500; R&D Systems; AF3075), rabbit anti-Taz (1:1000; HPA007415; Sigma), mouse anti-β-catenin (1:200; 610154; BD Transduction Laboratories), rat anti-RAGE (1:500; MAB1179; R&D Systems), rabbit anti-CGRP (1:5000; C8198; Sigma) and chicken anti-CGRP (1:75; CH14100; Neuromics). After deparaffinization, slides were rehydrated through a series of decreasing ethanol concentrations, antigen unmasked by either microwaving in citrate-based antigen unmasking solution (Vector Labs, H-3000) or by incubating sections with proteinase K (7.5μg/ml) (Invitrogen, 25530-049) for 7 min at 37°C. Tissue sections were then washed in TBS with 0.1% Tween-20 and blocked with 3% bovine serum albumin (BSA), 0.4% Triton in TBS for 30 min at room temperature followed by overnight incubation of primary antibodies diluted in 3% BSA, 0.1% Triton in TBS. The next day, slides were washed in TBS with 0.1% Tween-20 and incubated with secondary antibodies diluted in 3% BSA and 0.1% Triton in TBS for 3 h at room temperature. All fluorescent staining was performed with appropriate secondary antibodies from Jackson ImmunoResearch. Slides were washed with TBST and incubated with DAPI (Life Technologies, D1306) in TBST for 5 min, washed in TBST and mounted using Vectashield (Vector Labs, H-1000). All antibodies are commercially available and have been rigorously tested and validated by ourselves and other groups.

The whole-mount in situ hybridization protocol used was based on a previously described method (Winnier et al., 1995). Embryonic lungs were dissected and fixed in 4% formalin in PBS overnight. The next day, lungs were rinsed in PBS and dehydrated to 100% ethanol. Lungs were then bleached in four parts ethanol, one part 30% hydrogen peroxide for 1 h at room temperature. Lungs were sequentially washed in PBS containing 0.1% Tween-20 (PBST), treated with 1 ml of 7.5 μg/ml proteinase K (Invitrogen, 25530-049) in PBST for 5 min at room temperature, washed in PBST containing 2 mg/ml glycine, washed in PBST, fixed in 4% formaldehyde/0.2% glutaraldehyde in PBS for 20 min at room temperature and washed in PBST. Lungs were then placed in hybridization solution (prewarmed to 70°C) (50% deionized formamide, 5× SSC buffer, 0.05 mg/ml tRNA, 1% SDS and 0.05 mg/ml heparin), replaced with fresh hybridization solution and prehybridized at 70°C for 1 h with gentle agitation. Lungs were then hybridized in probe/hybridization solution overnight at 70°C with gentle agitation. The next day, lungs were washed in Solution I (50% deionized formamide, 5× SSC buffer, 1% SDS) for 2×30 min at 70°C. After solution I washes, lungs were washed in 1:1 Solution I:Solution II [0.5 M NaCl, 10 mM Tris (pH 7.5), 0.1% Tween-20] for 10 min at 70°C, washed in Solution II for 3×5 min at room temperature, washed in 100 μg/ml RNAse A (Roche, 109169) in Solution II for 2×30 min at 37°C, washed in Solution III (50% formamide, 2×SSC) for 2×30 min at 65°C and washed with TBST/0.5 mg/ml levamisole for 3×10 min. Lungs were then blocked in blocking solution containing 2% blocking reagent (11096176001, Sigma-Aldrich), 10% heat-inactivated fetal bovine serum (HI-FBS) in TBST/0.5 mg/ml levamisole for 1 h at room temperature and then incubated with sheep anti-digoxigenin-AP (1:2000, 11093274910, Sigma-Aldrich) in 1% blocking reagent, 1% HI-FBS in TBST/0.5 mg/ml levamisole at 4°C overnight. Blocking reagent was prepared by dissolving in maleic acid buffer (100 mM maleic acid, 150 mM NaCl at pH 7.5). The next day, samples were washed five or six times with TBST/0.5 mg/ml levamisole and washed in the same solution overnight at 4°C. The following day, lungs were washed with NTMT/alkaline phosphatase buffer [100 mM NaCl, 100 mM Tris (pH 9.5), 50 mM MgCl, 0.1% Tween-20 and 0.5 mg/ml levamisole] 2× for 20 min at room temperature. Lungs were then incubated in BM purple solution (11442074001, Sigma-Aldrich) at 37°C in the dark until the signal appeared. Finally, lungs were washed in PBS, post-fixed in 4% formalin in PBS at room temperature. The following mouse cDNAs were used as templates for the synthesis of digoxigenin (DIG)-labeled riboprobes using DIG RNA labeling mix (Roche, 11277073910): a 584 bp fragment of mouse Fgf10, a 948 bp full-length Spry2, a 1.5 kb full-length Bmp4 and a 300 bp fragment of Wnt7b (Bellusci et al., 1997; De Langhe et al., 2006, 2005).

Tissue was imaged using a micrometer slide calibrated Zeiss Axioimager microscope using Axiovision software, a Leica MZ16FA stereomicroscope and a Zeiss LSM800 confocal microscope using ZEN imaging software. Magnification tile images (20×) were taken covering sections through the entire lung, which were stitched to create a composite image using Zen blue software. Images were processed and analyzed using ImageJ/Fiji (NIH), Adobe Photoshop Creative Suite 3 (Adobe) and ZEN software. Immunostaining intensity was quantified after thresholding of pixel intensities. Intensity measurements reflect either maximum intensity levels or sum of intensities per region (indicated in figure legends). Length of Sox9 and Sox2 domains was measured on paraffin sections through the left and lower right intralobar main airway from E14.5 lungs using ZEN software. Basal cells were quantified by counting the total number of K5/P63⁺ basal cells in cross-sections of the whole main airway. NE cells were quantified on sections by counting the number of CGRP⁺ cells (solitary and clustered NE cells) per area. Three or more clustered NE cells in a section were scored as an NEB. Image quantification and analysis was performed in a blinded fashion.

Total RNA was extracted using RNALater (Invitrogen, AM7021) and Total RNA Kit I (Omega Biotek, R6834-02) according to the manufacturer's instructions. RNA concentration was determined by spectrophotometry. cDNA was generated using a Maxima First Strand cDNA Synthesis kit (Fisher Scientific, FERK1642) according to the manufacturer's instructions. Gene expression was analyzed by quantitative RT-PCR using Taqman Gene Expression Assays (Applied Biosystems, 4369016) directed against the mouse targets β-glucuronidase (Mm00446953_m1), Foxj1 (Mm00807215_m1), keratin 5 (K5) (Mm01305291_g1), p63 (Trp63) (Mm00495788_m1), Spry2 (Mm00442344_m1), Bmp4 (Mm00432087_m1), Wnt7b (Mm00437358_m1), Scgb1a1 (Mm00442046_m1), Calca (Cgrp) (Mm00801463_g1), Sftpc (Mm00488144_m1), Ager (Mm01134790_g1), Sox9 (Mm00448840_m1), Ctnnb1 (Mm00483039_m1), Ilk (Mm01274281_g1), Yap1 (Mm01143263_m1), Taz (Mm00504978_m1), Shh (Mm00436528_m1), Fgfr2 (Mm01269938_m1), Axin2 (Mm00443610_m1, Mm01265783_m1) and Fgf10 (Mm00433275_m1). Quantitative real-time PCR was performed using a StepOne Plus system (Applied Biosystems). Data are presented as mean expression relative to the housekeeping gene β-glucuronidase±s.e.m. (s.e.m.). Each experiment was repeated with samples obtained from at least six different lung preparations.

Peripheral lung buds were counted and statistical analysis was performed using Student's t-test. All results are expressed as mean values±s.e.m. The ‘n’ represents biological replicates and can be found in the figure legends. The significance of differences between two sample means was determined by unpaired Student's t-test (assuming unequal or equal variances as determined by the F-test of equality of variances). All datasets followed a normal distribution and P<0.05 was considered statistically significant. The number of samples to be used was based on the number of experimental paradigms multiplied by the number in each group that is necessary to yield statistically significant results [based on power analysis, to reject the null hypothesis with 80% power (type I error=0.05)]. Experiments were replicated in the laboratory once by a different investigator.