ABSTRACT
Adult mammalian lungs exhibit a fractal pattern, as each successive generation of airways is a fraction of the size of the parental branch. Achieving this structure likely requires precise control of airway length and diameter, as the embryonic airways initially lack the fractal scaling observed in the adult. In monolayers and tubes, directional growth can be regulated by the planar cell polarity (PCP) complex. Here, we characterized the roles of PCP complex components in airway initiation, elongation and widening during branching morphogenesis of the lung. Using tissue-specific knockout mice, we surprisingly found that branching morphogenesis proceeds independently of PCP complex function in the lung epithelium. Instead, we found a previously unreported Celsr1-independent role for the PCP complex components Vangl1 and Vangl2 in the pulmonary mesenchyme, where they are required for branch initiation, elongation and widening. Our data thus reveal an explicit function for Vangl1 and Vangl2 that is independent of the core PCP complex, suggesting a functional diversification of PCP complex components in vertebrate development. These data also reveal an essential role for the embryonic mesenchyme in generating the fractal structure of airways in the mature lung.
INTRODUCTION
The airway epithelium of the mammalian lung contains thousands of terminal ends that are generated via recursive rounds of stereotyped branching (Metzger et al., 2008). The epithelium can branch by initiating a bifurcation, wherein the tip of an existing branch folds in half, or by initiating a domain branch, wherein a daughter branch erupts from the side of an existing airway. These processes are influenced by the patterned differentiation of smooth muscle cells, which are derived from the surrounding mesenchyme and provide mechanical and chemical signals to the developing epithelium (Kim et al., 2015; Goodwin et al., 2019, 2022, 2023). After airway branches are established, the resulting tubes increase in length and diameter as the lung develops, culminating in a fractal pattern in which the proximal airways are longer and wider than the distal airways (Tanabe et al., 2020; Nelson et al., 1990). This coordinated increase in airway size is essential for proper flow of air after birth.
Although branch initiation has been investigated extensively, the mechanisms that drive epithelial lengthening and widening to generate the fractal pattern of airways remain poorly understood. In the kidney, tubule elongation is promoted by epithelial-intrinsic mechanisms, such as oriented cell divisions and convergent extension. In the branching lung, the orientation of cell division has been investigated in both established airways and in actively branching tip regions. At bifurcating tips, epithelial cells in nascent branches divide perpendicular to the basement membrane, whereas those within the newly forming cleft region divide parallel to the basement membrane, increasing the distance between adjacent branches (Schnatwinkel and Niswander, 2013; El-Hashash and Warburton, 2011). Within established airways, cells undergo oriented cell divisions predominately in the axial direction, promoting airway lengthening (Tang et al., 2018, 2011). Given that airways also widen in diameter throughout development, epithelial remodeling in the developing lung is unlikely to result solely from axially oriented cell divisions. To achieve its final fractal architecture, the embryonic airway must employ additional mechanisms to tune its local length and diameter.
Planar cell polarity (PCP) refers to the long-range, collective polarization of cells along a tissue plane and is controlled by a highly conserved molecular module known as the core PCP complex. First characterized in epithelia, the core PCP complex consists of a set of transmembrane and cytoplasmic proteins that localize asymmetrically at intercellular junctions (Devenport, 2014). The transmembrane Vangl proteins complex with the cytoplasmic Prickle proteins and localize to the opposite side of the cell as the complex formed by transmembrane Frizzled (Fzd) proteins and the cytoplasmic Dishevelled proteins. Celsr proteins, which are atypical cadherins, localize to both sides of intercellular junctions and connect Fzd and Vangl complexes between neighboring cells through homotypic adhesion (Devenport, 2014). Of the Fzd receptors in mice, Fzd3 and Fzd6 function in the PCP complex (Wang et al., 2006). Broadly, the PCP complex serves as a compass, directing individual cell behaviors through cytoskeletal regulation. The PCP complex is upstream of important morphogenetic behaviors, such as oriented cell divisions, and collective cell behaviors, such as convergent extension (Tada and Smith, 2000; Heisenberg et al., 2000; Wallingford et al., 2000). Loss of PCP complex function leads to severe developmental defects in many organ systems, including a failure to close the neural tube (Curtin et al., 2003; Torban et al., 2004; Wang et al., 2006). Thus, the PCP complex is essential for regulating the shape of tissues and organs during development.
Although best understood in simple cellular sheets, the core PCP complex has also been implicated in the development of branched organs, including the kidney and mammary gland. During kidney morphogenesis, the core PCP complex drives cell intercalation through the formation and resolution of multicellular rosettes, leading to convergent extension and thus elongation of kidney tubules; when the core PCP complex is lost, kidney tubules remain short and wide (Lienkamp et al., 2012; Kunimoto et al., 2017; Brzoska et al., 2016). During mammary gland morphogenesis, loss of the core PCP protein Vangl2 results in dysregulation of tube size and disorganization of the lumenal epithelium; however, the cellular basis of the phenotype is unknown (Smith et al., 2019). Two core transmembrane PCP components, Vangl2 and Celsr1, are essential for aspects of lung development and homeostasis. In the upper airways, the core PCP complex is asymmetrically localized at epithelial junctions and orients the coordinated beating of ciliated cells that line the trachea (Vladar et al., 2012; Kunimoto et al., 2022). Cilia are misoriented in PCP mutants and fail to polarize along the proximal-distal axis of the lung epithelium (Vladar et al., 2012). The core PCP complex has also been implicated in branching morphogenesis, as PCP mutants display alterations in the number (Yates et al., 2010) and positioning (Zhang et al., 2022) of branches compared with controls. Furthermore, loss of PCP has been demonstrated to disrupt the cytoskeletal organization of pulmonary cells during branching morphogenesis (Yates et al., 2010). However, the cellular mechanisms by which loss of core PCP components affects morphogenesis of the embryonic lung have not been reported. Given that core PCP proteins are expressed in the lung from the onset of branching (Vladar et al., 2012), it has been hypothesized that the PCP complex influences the fractal shape of the airways by regulating polarized cellular behaviors throughout branch initiation, elongation and widening (Yates and Dean, 2011).
Here, we investigated the links between the PCP complex and formation of the fractal tree of airways during lung development. We reveal that Celsr1 is expressed predominantly in the lung epithelium and mesothelium, whereas Vangl2 is expressed ubiquitously throughout the tissues of the lung, including in the pulmonary mesenchyme. We found that Vangl2-mutant lungs, but not Celsr1-mutant lungs, exhibited a decrease in branch number. However, Vangl2-mutant lungs still initiated branches in the expected stereotyped pattern. To define the tissue-specific role of Vangl in airway branching morphogenesis, we took advantage of the Cre-Lox system to specifically delete Vangl1 and Vangl2 (hereafter Vangl1/2) from the airway epithelium or the pulmonary mesenchyme. We found that loss of epithelial Vangl1/2, and thus loss of epithelial PCP, did not affect branch initiation, elongation or widening. Surprisingly, we found a significant role for Vangl1/2 in the pulmonary mesenchyme: loss of mesenchymal Vangl1/2 led to a severe reduction in branch initiation and striking defects in airway elongation and widening. Our data reveal that the pulmonary mesenchyme plays a key role in shaping the morphology of the branching airway epithelium in a manner dependent on Vangl1/2, but independent of the core PCP complex.
RESULTS
Vangl2-mutant lungs have a reduction in airway epithelial volume and branch number
To examine core PCP component expression (Fig. 1A) during lung development, we harvested lungs from embryonic stages spanning the days of epithelial branching morphogenesis [embryonic day (E) 11, E13 and E15] and used immunofluorescence analysis to reveal PCP protein localization. We found that throughout the branching process, Vangl2 was membrane localized in epithelial and mesenchymal cells (Fig. 1B; Fig. S1A), whereas Celsr1 was localized at cell-cell junctions only in epithelial and mesothelial cells (Fig. 1C; Fig. S1B). Fzd6 was localized to epithelial and endothelial cell membranes (Fig. 1D; Fig. S1D). Consistently, single-cell RNA-sequencing (scRNA-seq) analysis of E11.5 lungs (Goodwin et al., 2022) revealed that although Vangl1 and Vangl2 were expressed predominantly in the airway epithelium, mesenchyme and mesothelium (Fig. 1E; Fig. S1D,E), Celsr1 was only expressed in the epithelium and mesothelium (Fig. 1F). In contrast, Celsr2 was only expressed in the epithelium, whereas the Celsr3 transcript was not detected (Fig. S1F). These scRNA-seq results were confirmed using fluorescence in situ hybridization analysis (RNAscope). Specifically, the developing brain and pulmonary epithelium expressed high levels of Celsr1 and Celsr2 transcripts, which were largely absent from pulmonary mesenchymal cells (Fig. S1G,H). Furthermore, the Celsr3 transcript was abundant in the developing brain but absent from the lung (Fig. S1I). Finally, Fzd3 and Fzd6 expression was relatively low in our dataset, likely due to underreading of transcripts (Fig. 1G; Fig. S1).
Airway branching morphogenesis is slightly delayed in Vangl2Lp/Lp mutants. (A) Schematic illustrating the core planar cell polarity complex. (B-D) Sections (10-µm-thick) from embryonic day (E) 13 lungs immunostained for Vangl2, Celsr1 or Fzd6 (magenta) and E-cadherin (Ecad; green). Nuclei were counterstained with Hoechst (blue). Images are representative of three embryos. Scale bars: 25 µm. (E-G) Violin plots showing transcript levels of Vangl2, Celsr1 and Fzd6 from single-cell RNA sequencing of E11.5-E12 lungs. (H) Z-projections of confocal slices acquired from cleared, whole-mount lungs from E12 and E13 embryos immunostained for Ecad. Scale bars: 500 µm. (I) Normalized number of terminal branches in control and Celsr1Crsh/Crsh lungs at E12 (n=5 controls and 3 mutants, P=0.27 via unpaired two-tailed Student’s t-test) and E13 (n=10 controls and 4 mutants, P=0.05 via unpaired two-tailed Student’s t-test). (J) Normalized epithelial volume in control and Celsr1Crsh/Crsh lungs at E12 (n=5 controls and 3 mutants, P=0.99 via unpaired two-tailed Student’s t-test) and E13 (n=9 controls and 4 mutants, P=0.23 via unpaired two-tailed Student’s t-test). (K) Normalized number of terminal branches in control and Vangl2Lp/Lp lungs at E12 (n=3 controls and 3 mutants, P=0.59 via unpaired two-tailed Student’s t-test) and E13 (n=6 controls and 4 mutants, P=0.04 via unpaired two-tailed Student’s t-test). (L) Normalized epithelial volume in control and Vangl2Lp/Lp lungs at E12 (n=3 controls and 3 mutants, P=0.33 via unpaired two-tailed Student’s t-test) and E13 (n=6 controls and 4 mutants, P<0.001 via unpaired two-tailed Student’s t-test). In I-L, E12 and E13 mutant lung measurements were normalized to those of E12 and E13 littermate controls, respectively. Control embryos refer to wildtype or heterozygous littermate controls. In graphs, different shapes represent distinct experimental replicates; shown are mean±s.d. ns, not significant; *P<0.05; ***P<0.001.
Airway branching morphogenesis is slightly delayed in Vangl2Lp/Lp mutants. (A) Schematic illustrating the core planar cell polarity complex. (B-D) Sections (10-µm-thick) from embryonic day (E) 13 lungs immunostained for Vangl2, Celsr1 or Fzd6 (magenta) and E-cadherin (Ecad; green). Nuclei were counterstained with Hoechst (blue). Images are representative of three embryos. Scale bars: 25 µm. (E-G) Violin plots showing transcript levels of Vangl2, Celsr1 and Fzd6 from single-cell RNA sequencing of E11.5-E12 lungs. (H) Z-projections of confocal slices acquired from cleared, whole-mount lungs from E12 and E13 embryos immunostained for Ecad. Scale bars: 500 µm. (I) Normalized number of terminal branches in control and Celsr1Crsh/Crsh lungs at E12 (n=5 controls and 3 mutants, P=0.27 via unpaired two-tailed Student’s t-test) and E13 (n=10 controls and 4 mutants, P=0.05 via unpaired two-tailed Student’s t-test). (J) Normalized epithelial volume in control and Celsr1Crsh/Crsh lungs at E12 (n=5 controls and 3 mutants, P=0.99 via unpaired two-tailed Student’s t-test) and E13 (n=9 controls and 4 mutants, P=0.23 via unpaired two-tailed Student’s t-test). (K) Normalized number of terminal branches in control and Vangl2Lp/Lp lungs at E12 (n=3 controls and 3 mutants, P=0.59 via unpaired two-tailed Student’s t-test) and E13 (n=6 controls and 4 mutants, P=0.04 via unpaired two-tailed Student’s t-test). (L) Normalized epithelial volume in control and Vangl2Lp/Lp lungs at E12 (n=3 controls and 3 mutants, P=0.33 via unpaired two-tailed Student’s t-test) and E13 (n=6 controls and 4 mutants, P<0.001 via unpaired two-tailed Student’s t-test). In I-L, E12 and E13 mutant lung measurements were normalized to those of E12 and E13 littermate controls, respectively. Control embryos refer to wildtype or heterozygous littermate controls. In graphs, different shapes represent distinct experimental replicates; shown are mean±s.d. ns, not significant; *P<0.05; ***P<0.001.
To define the role of PCP in lung development, we harvested lungs from Celsr1Crsh/Crsh and Vangl2Lp/Lp embryos, both of which harbor dominant-negative mutations that disrupt the asymmetric localization of core PCP components as well as PCP function (Devenport and Fuchs, 2008; Curtin et al., 2003; Stahley et al., 2021; Yin et al., 2012; Vladar et al., 2012). Specifically, the Celsr1Crsh allele harbors a single amino acid substitution between two cadherin repeats, thus affecting cis-interactions and assembly of PCP complexes (Stahley et al., 2021). Similarly, the Vangl2Lp allele blocks PCP complex assembly due to a point mutation in its endoplasmic reticulum export sequence that disrupts trafficking of both Vangl2 and Vangl1 to the cell surface (Yin et al., 2012). At E12 and E13, Celsr1Crsh/Crsh lungs formed the expected pattern of branches with no significant differences from those in littermate controls (Fig. 1H-J; Fig. S2A,B). Consistently, we found that when cultured for 2 days, Celsr1Crsh/Crsh lung explants branched at the same rate as controls, yielding a similar number of terminal branches with a comparable fold change in the number of terminal branches (Fig. S2C-E). Celsr1 is, therefore, surprisingly dispensable for branch initiation in the lung.
In contrast, E13 Vangl2Lp/Lp lungs were significantly smaller and displayed a reduction in the number of terminal branches and overall epithelial volume compared with the lungs of littermate controls (Fig. 1H,K,L; Fig. S2F,G). The striking difference in airway architecture between Celsr1-mutant and Vangl2-mutant lungs led us to hypothesize that Vangl2 may function independently of Celsr1 in morphogenesis of the airway epithelium.
Removal of Vangl1/2 from the lung epithelium does not affect branch initiation
Because Vangl1/2 are expressed ubiquitously in the lung, we generated tissue-specific conditional Vangl1/2 knockout mice to separately assess Vangl1/2 function in the pulmonary epithelium and mesenchyme. We first generated ShhCre; Vangl1fl/fl; Vangl2fl/fl (epithelial-conditional knockout or epiCKO) animals. In contrast to the conclusions of previous work (Zhang et al., 2022), we found that the ShhCre driver efficiently deletes Vangl1/2, as we observed loss of detectable Vangl2 protein in the epithelium, but not the mesenchyme, of epiCKO lungs (Fig. S3A). Furthermore, we previously found that removal of Vangl1/2 disrupts Celsr1 polarization in the trachea of epiCKO animals (Paramore et al., 2024); thus, using the ShhCre driver to delete Vangl1/2 causes loss of a functional PCP complex in the embryonic lung epithelium.
Whereas epiCKO lungs were indistinguishable from control lungs at E12, mutant lungs showed a significant reduction in the number of terminal branches by E13 (Fig. 2A,B; Fig. S3B). However, we also found that epiCKO embryos were significantly smaller than littermate controls and appeared developmentally delayed, likely due to loss of Vangl1/2 in Shh-expressing tissues outside of the lung (Fig. 2C,D). It was therefore possible that the decrease in branch number observed in epiCKO lungs was due to the smaller size and developmental delay of the embryo itself, rather than due to an intrinsic requirement for Vangl1/2 in the lung epithelium.
Epithelial Vangl1/2 are not required for branch initiation. (A) Z-projection of confocal slices acquired from cleared, whole-mount E12-E14 lungs from control and Vangl1/2 epithelial-conditional knockout (epiCKO) embryos immunostained for Ecad. Scale bars: 100 μm (E12); 500 μm (E13); 250 μm (E14). (B) Normalized number of terminal branches in control and epiCKO lungs at E12 (n=4 controls and 3 mutants, P=0.39 via unpaired two-tailed Student’s t-test), E13 (n=10 controls and 11 mutants, P=0.01 via unpaired two-tailed Student’s t-test) and E14 (n=3 controls and 6 mutants, P=0.002 via unpaired two-tailed Student’s t-test). (C) Representative images of E13 control and epiCKO embryos. Scale bars: 1 mm. (D) Quantification of normalized projected areas of E13 control and epiCKO embryos (n=11 controls and 10 mutants, P<0.001 via unpaired two-tailed Student’s t-test). (E) Representative time-lapse bright-field (BF) and fluorescence images of control and epiCKO lungs dissected at E12 and cultured for 48 h. Scale bars: 250 µm. (F) Number of terminal branches at 0 and 48 h of culture for control and epiCKO lungs (n=13 controls and 7 mutants, P=0.23 at 0 h, P=0.29 at 48 h via unpaired two-tailed Student’s t-test). (G) Fold change in the number of terminal branches after 48 h of culture for control and epiCKO lungs (n=13 controls and 7 mutants, P=0.40 via unpaired two-tailed Student’s t-test). In B,D, mutant measurements were normalized to those of littermate controls. Control embryos refer to Vangl1/2fl/fl littermate controls lacking the ShhCre allele. Shown are mean±s.d. ns, not significant; **P<0.01; ****P<0.0001. In all graphs, different shapes represent distinct experimental replicates.
Epithelial Vangl1/2 are not required for branch initiation. (A) Z-projection of confocal slices acquired from cleared, whole-mount E12-E14 lungs from control and Vangl1/2 epithelial-conditional knockout (epiCKO) embryos immunostained for Ecad. Scale bars: 100 μm (E12); 500 μm (E13); 250 μm (E14). (B) Normalized number of terminal branches in control and epiCKO lungs at E12 (n=4 controls and 3 mutants, P=0.39 via unpaired two-tailed Student’s t-test), E13 (n=10 controls and 11 mutants, P=0.01 via unpaired two-tailed Student’s t-test) and E14 (n=3 controls and 6 mutants, P=0.002 via unpaired two-tailed Student’s t-test). (C) Representative images of E13 control and epiCKO embryos. Scale bars: 1 mm. (D) Quantification of normalized projected areas of E13 control and epiCKO embryos (n=11 controls and 10 mutants, P<0.001 via unpaired two-tailed Student’s t-test). (E) Representative time-lapse bright-field (BF) and fluorescence images of control and epiCKO lungs dissected at E12 and cultured for 48 h. Scale bars: 250 µm. (F) Number of terminal branches at 0 and 48 h of culture for control and epiCKO lungs (n=13 controls and 7 mutants, P=0.23 at 0 h, P=0.29 at 48 h via unpaired two-tailed Student’s t-test). (G) Fold change in the number of terminal branches after 48 h of culture for control and epiCKO lungs (n=13 controls and 7 mutants, P=0.40 via unpaired two-tailed Student’s t-test). In B,D, mutant measurements were normalized to those of littermate controls. Control embryos refer to Vangl1/2fl/fl littermate controls lacking the ShhCre allele. Shown are mean±s.d. ns, not significant; **P<0.01; ****P<0.0001. In all graphs, different shapes represent distinct experimental replicates.
To distinguish between these two possibilities, we tested whether epiCKO lungs retained the same capacity for branching as control lungs when removed from the constraints of the embryonic chest cavity. We dissected lungs from E12 embryos, a stage at which we did not observe a difference between mutants and controls, and monitored their rate of epithelial branching for 48 h in culture. We found that cultured epiCKO lungs branched at the same rate as littermate controls (Fig. 2E-G; Fig. S3C). It is therefore likely that any delay in branching observed in epiCKO lungs results from the developmental delay of the embryo, rather than from loss of epithelial PCP per se. Because branch number is unaffected by loss of either Celsr1 or epithelial Vangl1/2, we conclude that branch initiation occurs independently of the core PCP complex in the murine lung.
Removal of Vangl1/2 from the lung mesenchyme leads to defects in branch initiation
In addition to being expressed in the lung epithelium, Vangl2 is also expressed at high levels in the lung mesenchyme (Fig. 1B,E). Given that epiCKO lungs initiate branches normally, we hypothesized that the branching phenotype observed in Vangl2Lp/Lp lungs indicates a role for Vangl in the pulmonary mesenchyme. To test this hypothesis, we began by breeding a mesenchymal Vangl1/2 knockout: a Dermo1Cre; Vangl1fl/fl; Vangl2fl/fl mouse line, in which Cre is expressed broadly in the splanchnic mesoderm, including in the pulmonary mesenchyme, but is absent from the airway epithelium (Chen et al., 2008). Notably, the Dermo1Cre allele and Vangl2 gene are on the same chromosome; thus, generating homozygous embryos necessitates a meiotic crossover event. Surprisingly, lungs from E13 mutant embryos appeared normal and did not recapitulate the branching defects observed in Vangl2Lp/Lp lungs (Fig. S4A,B). However, immunostaining for Vangl2 in control and Dermo1Cre; Vangl1fl/fl; Vangl2fl/fl lungs revealed significant levels of Vangl2 protein remaining in the mesenchyme at E13 (Fig. S4C). Thus, we conclude that either Dermo1Cre does not efficiently delete Vangl2 during these stages of development, or that the Vangl2 protein is stable and persists in the mesenchyme long after Cre-mediated recombination.
To circumvent this issue, we generated a Tbx4-rtTA; Tet-O-Cre; Vangl1fl/fl; Vangl2fl/fl (inducible mesenchymal conditional knockout or inducible-mesCKO) mouse line, in which expression of Cre is restricted specifically to the pulmonary mesenchyme (Zhang et al., 2013). In contrast to the Dermo1Cre driver, we found that the Tbx4-rtTA; Tet-O-Cre driver resulted in significant loss of detectable Vangl2 protein in the mesenchyme by E12.5 when doxycycline-containing drinking water was introduced to the dam at E8 (Fig. S5A). To ensure full deletion of the desired alleles, we proceeded to use the Tbx4-rtTA; Tet-O-Cre system.
We analyzed lungs from inducible-mesCKO embryos at E12-E14. By E13, the lungs of inducible-mesCKO embryos were noticeably smaller than those of littermate controls (Fig. 3A,B; Fig. S5B). In contrast to epiCKO lungs, this reduction in size cannot be attributed to an embryo-wide developmental delay: inducible-mesCKO embryos were the same size as littermate control embryos (Fig. 3C,D). Consistent with their smaller size, inducible-mesCKO lungs showed a significant reduction in the number of terminal branches at E14 (Fig. 3E). The new branches that did form appeared to be correctly stereotyped and shaped, suggesting that branch initiation is delayed but morphologically normal in mutants compared with that in controls. Consistently, E12.5 inducible-mesCKO lungs formed fewer branches when cultured ex vivo compared with littermate control lungs (Fig. 3F-H). Thus, deletion of Vangl1/2 from the pulmonary mesenchyme leads to a reduction or delay in branch initiation.
Loss of mesenchymal Vangl leads to defects in branch initiation. (A) Z-projection of confocal slices acquired from cleared, whole-mount E12, E13 and E14 lungs from control and inducible mesenchymal conditional knockout (inducible-mesCKO) embryos immunostained for Ecad. Scale bars: 100 µm (E12 lungs); 250 µm (E13 and E14 lungs). (B) Z-projection of confocal slices of individual E14 control and inducible-mesCKO cranial, medial and accessory bronchi and L1 branch, immunostained for Ecad. Scale bars: 100 μm. (C) Representative images of E13 control and inducible-mesCKO embryos. Scale bars: 1 mm. (D) Quantification of normalized projected areas of E13 control and inducible-mesCKO embryos (n=4 controls and 3 mutants, P=0.21 via unpaired two-tailed Student’s t-test). (E) Normalized number of terminal branches in control and inducible-mesCKO lungs at E12 (n=8 controls and 11 mutants, P=0.16 via unpaired two-tailed Student’s t-test), E13 (n=6 controls and 6 mutants, P=0.77 via unpaired two-tailed Student’s t-test) and E14 (n=6 controls and 5 mutants, P=0.003 via unpaired two-tailed Student’s t-test). (F) Representative time-lapse bright-field (BF) and fluorescence images of control and inducible-mesCKO lungs dissected at E12 and cultured for 48 h. Scale bars: 250 µm. (G) Number of terminal branches at 0 and 48 h of culture for control and inducible-mesCKO lungs (n=3 controls and 5 mutants, P=0.91 at 0 h, P=0.03 at 48 h via unpaired two-tailed Student’s t-test). (H) Fold change in the number of terminal branches after 48 h of culture for control and inducible-mesCKO lungs (n=3 controls and 5 mutants, P=0.02 via unpaired two-tailed Student’s t-test). In D,E, mutant measurements were normalized to those of littermate controls. Control embryos refer to Vangl1/2fl/fl littermates lacking the Tbx4-rtTA allele or the Tet-O-Cre allele, thus lacking Cre expression. Shown are mean±s.d. ns, not significant; *P<0.05; **P<0.01. In all graphs, different shapes represent distinct experimental replicates.
Loss of mesenchymal Vangl leads to defects in branch initiation. (A) Z-projection of confocal slices acquired from cleared, whole-mount E12, E13 and E14 lungs from control and inducible mesenchymal conditional knockout (inducible-mesCKO) embryos immunostained for Ecad. Scale bars: 100 µm (E12 lungs); 250 µm (E13 and E14 lungs). (B) Z-projection of confocal slices of individual E14 control and inducible-mesCKO cranial, medial and accessory bronchi and L1 branch, immunostained for Ecad. Scale bars: 100 μm. (C) Representative images of E13 control and inducible-mesCKO embryos. Scale bars: 1 mm. (D) Quantification of normalized projected areas of E13 control and inducible-mesCKO embryos (n=4 controls and 3 mutants, P=0.21 via unpaired two-tailed Student’s t-test). (E) Normalized number of terminal branches in control and inducible-mesCKO lungs at E12 (n=8 controls and 11 mutants, P=0.16 via unpaired two-tailed Student’s t-test), E13 (n=6 controls and 6 mutants, P=0.77 via unpaired two-tailed Student’s t-test) and E14 (n=6 controls and 5 mutants, P=0.003 via unpaired two-tailed Student’s t-test). (F) Representative time-lapse bright-field (BF) and fluorescence images of control and inducible-mesCKO lungs dissected at E12 and cultured for 48 h. Scale bars: 250 µm. (G) Number of terminal branches at 0 and 48 h of culture for control and inducible-mesCKO lungs (n=3 controls and 5 mutants, P=0.91 at 0 h, P=0.03 at 48 h via unpaired two-tailed Student’s t-test). (H) Fold change in the number of terminal branches after 48 h of culture for control and inducible-mesCKO lungs (n=3 controls and 5 mutants, P=0.02 via unpaired two-tailed Student’s t-test). In D,E, mutant measurements were normalized to those of littermate controls. Control embryos refer to Vangl1/2fl/fl littermates lacking the Tbx4-rtTA allele or the Tet-O-Cre allele, thus lacking Cre expression. Shown are mean±s.d. ns, not significant; *P<0.05; **P<0.01. In all graphs, different shapes represent distinct experimental replicates.
Stereotyped scaling of branch length and diameter requires Vangl1/2 expression in the mesenchyme but not the epithelium
Although the stereotyped pattern of branching has been described in detail (Metzger et al., 2008), less is known about the dynamics with which airways elongate and widen to generate the fractal tree observed in adults. Specifically, it is unclear how airway length and width scale as the lung grows. To assess changes in airway geometry over developmental time, we measured the lengths and diameters of four major airways: the bronchi of the cranial, medial and accessory lobes, and the first branch on the left lobe (L1) (Fig. 4A-D). We selected these airways because they form during the earliest stages of branching, have clear morphological landmarks and can be measured unambiguously in each lung. We plotted these measurements as a function of the number of terminal branches as a proxy for developmental time. Surprisingly, we found that the four bronchi elongated at different rates. For example, the cranial bronchus increased in length by ∼3-fold from E11 to E14.5, whereas the accessory bronchus increased in length by ∼6-fold (Fig. 4C). In contrast, all four bronchi widened at similar rates, increasing ∼4-fold in diameter from E11 to E14.5 (Fig. 4D). These observations suggest that the relative growth of the airways, which generates the fractal structure of the epithelial tree, is spatially regulated during development.
Branch elongation and widening both scale with branch complexity and require Vangl1/2 in the mesenchyme but not in the epithelium. (A,B) Schematics of lung with landmarks used for quantification of bronchus length and diameter over developmental time. (C,D) Quantification of bronchus length or diameter versus number of terminal branches in E12-E14 cranial, medial and accessory bronchi and L1 branch from wildtype lungs from CD1, C57BL/6 and C3H/HeJ backgrounds. Shown are best-fit curves. (E,F) Schematics of lung with landmarks used for quantification of bronchus length and diameter and key for epiCKO lungs. (G,H) Quantification of bronchus length or diameter versus number of terminal branches in E12-E14 cranial, medial and accessory bronchi and L1 branch from epiCKO and littermate control lungs. Shown are best-fit curves and 95% confidence intervals. (I,J) Schematics of lung with landmarks used for quantification of bronchus length and diameter and key for inducible-mesCKO lungs. (K,L) Quantification of bronchus length or diameter versus number of terminal branches in E12-E14 cranial, medial and accessory bronchi and L1 branch from inducible-mesCKO and littermate control lungs. Control embryos refer to Vangl1/2fl/fl littermates lacking the ShhCre allele (G,H) or lacking the Tbx4rtTA/Tet-O-Cre allele(s) (K,L). In all graphs, each dot represents one lung.
Branch elongation and widening both scale with branch complexity and require Vangl1/2 in the mesenchyme but not in the epithelium. (A,B) Schematics of lung with landmarks used for quantification of bronchus length and diameter over developmental time. (C,D) Quantification of bronchus length or diameter versus number of terminal branches in E12-E14 cranial, medial and accessory bronchi and L1 branch from wildtype lungs from CD1, C57BL/6 and C3H/HeJ backgrounds. Shown are best-fit curves. (E,F) Schematics of lung with landmarks used for quantification of bronchus length and diameter and key for epiCKO lungs. (G,H) Quantification of bronchus length or diameter versus number of terminal branches in E12-E14 cranial, medial and accessory bronchi and L1 branch from epiCKO and littermate control lungs. Shown are best-fit curves and 95% confidence intervals. (I,J) Schematics of lung with landmarks used for quantification of bronchus length and diameter and key for inducible-mesCKO lungs. (K,L) Quantification of bronchus length or diameter versus number of terminal branches in E12-E14 cranial, medial and accessory bronchi and L1 branch from inducible-mesCKO and littermate control lungs. Control embryos refer to Vangl1/2fl/fl littermates lacking the ShhCre allele (G,H) or lacking the Tbx4rtTA/Tet-O-Cre allele(s) (K,L). In all graphs, each dot represents one lung.
The processes of airway lengthening and widening have been attributed primarily to behaviors by the epithelium itself. Specifically, airway epithelial cells have been reported to divide predominantly along the proximal-distal axis of the airway, leading to lengthening (Tang et al., 2018). In organs such as the kidney, the core PCP complex directs oriented cell divisions and other cellular behaviors that promote lengthening, such as convergent extension (Kunimoto et al., 2017). Thus, although we found that the core PCP complex was dispensable for branch initiation, we hypothesized that epithelial PCP may regulate airway lengthening and/or widening. To determine whether the core PCP complex is required in the epithelium for airway remodeling, we compared the lengths and diameters of control and epiCKO bronchi from E12 to E14.5. After generating best-fit curves for both the control and mutant datasets, we found that the 95% confidence intervals of our control and mutant data overlapped for each bronchus analyzed (Fig. 4E-H). Although the epiCKO embryos experienced a developmental delay, the length and diameter of their bronchi still scaled with the number of terminal branches in a manner comparable with that in controls. We therefore conclude that function of the PCP complex in the epithelium is dispensable for elongation and widening of the airways during the branching morphogenesis stages of lung development.
Because inducible-mesCKO lungs appeared significantly smaller than controls, we also assessed the dynamics of branch lengthening and widening in these mutants. Strikingly, our data revealed that the rates of both were significantly decreased in inducible-mesCKO lungs: bronchi from inducible-mesCKO lungs were shorter and narrower than those from controls at the same stage of branching (Fig. 4I-L). Bronchi from inducible-mesCKO lungs were also physically collapsed, lacking the clear lumenal space observed in control lungs (Fig. S5C). These decreases in branch initiation, lengthening and widening did not appear to result from changes in cell proliferation or death: we observed no significant differences in the percentage of phospho-histone-3-positive cells in the epithelium or mesenchyme of control and inducible-mesCKO lungs at E13-E14, and the levels of cleaved caspase-3 signal in control and inducible-mesCKO lungs at E13-E14 were negligible (Fig. S5D-G). These data demonstrate that the initiation, elongation and widening of airway epithelial branches require expression of Vangl1/2 in the pulmonary mesenchyme. The pulmonary mesenchyme is therefore essential for shaping the fractal structure of the airway epithelial tree.
Loss of mesenchymal Wnt5a does not recapitulate loss of mesenchymal Vangl1/2
Both the core PCP complex generally and Vangl specifically have been hypothesized to function downstream of non-canonical Wnt signaling, although direct evidence linking Wnt ligand binding to changes in PCP complex function is sparse. However, Wnt5a has been shown to regulate phosphorylation of Vangl2 in the chondrocytes of the developing limb, leading to changes in cell shape and tissue elongation (Gao et al., 2011). Furthermore, loss of Wnt5a and Vangl2 causes similar phenotypes during late lung development, leading to speculation that a Wnt5a–Vangl signaling pathway regulates alveologenesis (Zhang et al., 2020). The Wnt5a transcript was expressed in the mesenchymal compartment at E13.5 (Fig. S6A). To determine whether Wnt5a functions upstream of mesenchymal Vangl2 in branch initiation, elongation and widening, we generated mesenchymal knockouts of Wnt5a. We began by establishing a Dermo1Cre; Wnt5afl/fl mouse line and assessed branching in control and mutant lungs at E12-E14 (Fig. S6B). Broadly, this analysis revealed that mutant lungs were misshapen and showed alterations in the angles of their airways. Surprisingly, however, we observed no decrease in the number of branches in mutant lungs compared with controls (Fig. S6C). Further, Dermo1Cre; Wnt5afl/fl embryos appeared to have axis-elongation defects and exhibited broad developmental abnormalities, likely due to loss of Wnt5a in Dermo1 (also known as Twist2)-expressing tissues outside of the lung. Thus, the Dermo1Cre driver cannot be used to uncouple the lung phenotypes caused by deleting Wnt5a from the pulmonary mesenchyme from those caused by deleting Wnt5a from other tissues.
To specifically uncover the role of Wnt5a in the lung mesenchyme, we generated a second mesenchymal Wnt5a knockout driven by Tbx4-rtTA; Tet-O-Cre. In contrast to Dermo1Cre; Wnt5afl/fl lungs, Tbx4-rtTA; Tet-O-Cre; Wnt5afl/fl lungs were indistinguishable from control lungs at E12 (Fig. S6D). Therefore, the changes in branch angles observed in Dermo1Cre; Wnt5afl/fl lungs are likely due to loss of Wnt5a from other tissues, rather than due to a role for Wnt5a in the pulmonary mesenchyme. Strangely, Tbx4-rtTA; Tet-O-Cre; Wnt5afl/fl knockout lungs showed a slight increase in the number of branches at E13, but a slight decrease in the number of branches at E14 (Fig. S6E). However, morphometric analysis revealed no significant defects in elongation or widening in the mutant lungs (Fig. S6F-I). Thus, removing Wnt5a from the pulmonary mesenchyme failed to recapitulate the branching defects observed in Vangl1/2 inducible-mesCKO lungs. Therefore, although Wnt5a may still regulate Vangl in this developmental context, this ligand is unlikely to be the sole upstream regulator. In this regard, it is possible that the expression of another non-canonical Wnt gene, such as Wnt11, compensates for loss of Wnt5a.
Loss of mesenchymal Vangl1/2 affects assembly of the airway smooth muscle layer
To understand how loss of mesenchymal Vangl1/2 leads to defects in branching morphogenesis of the adjacent epithelium, we characterized the cellular and non-cellular components of the mesenchyme in control and inducible-mesCKO lungs. Immunofluorescence analysis revealed no significant differences in the expression or localization of fibronectin (Fn1) or collagen IV (Col4a1), indicating that the changes observed in inducible-mesCKO lungs are unlikely to be due to alterations in the deposition of interstitial matrix or basement membrane (Fig. S5H,I). Similarly, we observed no differences in mesenchymal expression of Fgf10 (Fig. S5J), a key regulator of branching in the mouse lung (Yuan et al., 2018).
The embryonic pulmonary mesenchyme is compartmentalized into spatially distinct subpopulations of cells that express different markers and differentially affect epithelial morphogenesis; these compartments include the sub-epithelial mesenchyme, sub-mesothelial mesenchyme and airway smooth muscle (ASM) (Goodwin et al., 2022). Immunofluorescence analysis of E13 inducible-mesCKO lungs revealed no changes in the distribution of sub-epithelial or sub-mesothelial mesenchyme, as assessed by the expression of Lef1 and Foxp1, respectively (Fig. 5A-D). These data suggest that Vangl is not required for expression of mesenchymal markers or morphogens, nor for the patterning of mesenchymal subpopulations, consistent with the primarily structural role of this core PCP component in other tissue contexts (Boutin et al., 2014; Cetera et al., 2017; Devenport and Fuchs, 2008; Gao et al., 2011).
Loss of mesenchymal Vangl1/2 affects assembly of the airway smooth muscle layer. (A,C) Representative images of sections (10-µm thick) from E13 inducible-mesCKO and littermate control lungs immunostained for Lef1 (sub-epithelial mesenchyme) or Foxp1 (sub-mesothelial mesenchyme) and Ecad. Scale bars: 50 µm. (B,D) Quantifications of Lef1 or Foxp1 intensity profiles emanating from the epithelium or mesothelium, respectively. Shown are mean±s.d. (n=7 controls and 4 mutants). (E) Representative images of optical cross-sections through the medial bronchi of E12-E14 inducible-mesCKO and littermate control lungs immunostained for Ecad and α-smooth muscle actin (αSMA). Scale bars: 10 µm. (F) Quantification of the thickness of the airway smooth muscle (ASM) layer around major bronchi in inducible-mesCKO and littermate control lungs at E12 (n=8 control and 9 mutant bronchi, P=0.21 via unpaired two-tailed Student’s t-test), E13 (n=12 control and 12 mutant bronchi, P=0.002 via F test to compare variances) and E14 (n=12 control and 9 mutant bronchi, P<0.001 via unpaired two-tailed Student’s t-test). Shown are mean±s.d. ##variance<0.01; ***P<0.001. (G) Quantification of the percentage of phospho-histone-3+ (PH3+) ASM cells in inducible-mesCKO and littermate control lungs at E13 (n=3 controls and 3 mutants, P=0.80 via unpaired two-tailed Student’s t-test) and E14 (n=3 controls and 5 mutants, P=0.93 via unpaired two-tailed Student’s t-test) lungs. Control embryos refer to Vangl1/2fl/fl littermates lacking the Tbx4rtTA allele or the Tet-O-Cre allele, thus lacking Cre expression. Shown are mean±s.d. ns, not significant. In graphs, different shapes represent different experimental replicates.
Loss of mesenchymal Vangl1/2 affects assembly of the airway smooth muscle layer. (A,C) Representative images of sections (10-µm thick) from E13 inducible-mesCKO and littermate control lungs immunostained for Lef1 (sub-epithelial mesenchyme) or Foxp1 (sub-mesothelial mesenchyme) and Ecad. Scale bars: 50 µm. (B,D) Quantifications of Lef1 or Foxp1 intensity profiles emanating from the epithelium or mesothelium, respectively. Shown are mean±s.d. (n=7 controls and 4 mutants). (E) Representative images of optical cross-sections through the medial bronchi of E12-E14 inducible-mesCKO and littermate control lungs immunostained for Ecad and α-smooth muscle actin (αSMA). Scale bars: 10 µm. (F) Quantification of the thickness of the airway smooth muscle (ASM) layer around major bronchi in inducible-mesCKO and littermate control lungs at E12 (n=8 control and 9 mutant bronchi, P=0.21 via unpaired two-tailed Student’s t-test), E13 (n=12 control and 12 mutant bronchi, P=0.002 via F test to compare variances) and E14 (n=12 control and 9 mutant bronchi, P<0.001 via unpaired two-tailed Student’s t-test). Shown are mean±s.d. ##variance<0.01; ***P<0.001. (G) Quantification of the percentage of phospho-histone-3+ (PH3+) ASM cells in inducible-mesCKO and littermate control lungs at E13 (n=3 controls and 3 mutants, P=0.80 via unpaired two-tailed Student’s t-test) and E14 (n=3 controls and 5 mutants, P=0.93 via unpaired two-tailed Student’s t-test) lungs. Control embryos refer to Vangl1/2fl/fl littermates lacking the Tbx4rtTA allele or the Tet-O-Cre allele, thus lacking Cre expression. Shown are mean±s.d. ns, not significant. In graphs, different shapes represent different experimental replicates.
In the developing mouse lung, the sub-epithelial mesenchyme differentiates into ASM in a pattern that sculpts the epithelium into branches (Goodwin et al., 2019, 2022, 2023; Kim et al., 2015). To assess whether loss of Vangl leads to defects in ASM, we examined this mesenchymal layer using immunofluorescence analysis for α-smooth muscle actin (αSMA, encoded by Acta2). At E12, we observed no differences in the thickness of the ASM surrounding the airways between control and inducible-mesCKO lungs (Fig. 5E,F). At E13, however, we found that control ASM existed as a uniform layer approximately 5-µm thick, whereas the mutant ASM was highly variable, ranging from 2.5 to 8 µm in thickness (Fig. 5E,F). At E14, the Vangl1/2-mutant ASM had resolved into a layer that was significantly thinner than that of controls (Fig. 5E,F). This thinner layer of smooth muscle did not result from a reduction in cell proliferation (Fig. 5G). Taken together, these data reveal that although mesenchymal cell identity and compartmentalization do not appear to be altered in mesenchymal Vangl1/2 mutants, the absence of Vangl proteins results in defects in the organization of cells within the ASM tissue layer.
Loss of mesenchymal Vangl alters the morphology and polarity of airway smooth muscle cells
In most other developmental and cellular contexts, Vangl predominately regulates cellular behaviors such as motility, shape and polarization, rather than transcription. Given the differences in thickness of the ASM layer, we hypothesized that mesenchymal Vangl may affect epithelial morphogenesis by regulating ASM cell morphology. We generated a Tbx4-rtTA; Tet-O-Cre; Vangl1fl/fl; Vangl2fl/fl; Confetti mouse line, which enabled us to analyze the shapes of sparsely labeled Vangl1/2-mutant ASM cells. In control (Tbx4-rtTA; Tet-O-Cre; Confetti) lungs, αSMA+ cells were highly elongated and wrapped circumferentially around the airways (Fig. 6A). However, in inducible-mesCKO lungs, αSMA+ cells appeared rounder and less elongated (Fig. 6B). Cell-shape analysis revealed that Vangl1/2-mutant ASM cells had a significantly reduced aspect ratio and increased circularity (Fig. 6C,D). Furthermore, Vangl1/2-mutant ASM cells exhibited a reduction in circumferential polarization around airways (Fig. 6E-G). Therefore, Vangl1/2 are required for the normal elongation and circumferential wrapping of ASM cells in the developing lung. These changes in smooth muscle morphology coincide with delays in branch initiation, elongation and widening in the airways of the Vangl1/2-mutant embryos.
Loss of mesenchymal Vangl1/2 alters the morphology of airway smooth muscle cells. (A,B) Examples of airway smooth muscle (ASM) cells [immunostained for αSMA (magenta) with signal from RFP fluorescence (green)] in Tbx4-rtTA; Tet-O-Cre; Confetti or Tbx4-rtTA; Tet-O-Cre; Vangl1/2fl/fl; Confetti lungs. Scale bars: 10 µm. (C,D) Histograms of the aspect ratio (****P<0.0001, Mann–Whitney test) and circularity (****P<0.0001, Mann–Whitney test) of ASM cells from E12.5 Tbx4-rtTA; Tet-O-Cre; Confetti (n=294 cells from 3 lungs) and Tbx4-rtTA; Tet-O-Cre; Vangl1fl/fl; Vangl2fl/fl; Confetti lungs (n=205 cells from 4 lungs). (E) Schematic illustrating how ASM cell orientations were measured relative to the long axis of the airway. (F) Rose plot of ASM cell orientations from E12.5 Tbx4-rtTA; Tet-O-Cre; Confetti lungs (n=296 cells from 3 lungs). (G) Rose plot of ASM cell orientations from E12.5 Tbx4-rtTA; Tet-O-Cre; Vangl1/2fl/fl; Confetti lungs (n=224 cells from 4 lungs). (H) Schematic of proposed ‘dilational extension’ model for ASM-guided airway stabilization, lengthening and widening.
Loss of mesenchymal Vangl1/2 alters the morphology of airway smooth muscle cells. (A,B) Examples of airway smooth muscle (ASM) cells [immunostained for αSMA (magenta) with signal from RFP fluorescence (green)] in Tbx4-rtTA; Tet-O-Cre; Confetti or Tbx4-rtTA; Tet-O-Cre; Vangl1/2fl/fl; Confetti lungs. Scale bars: 10 µm. (C,D) Histograms of the aspect ratio (****P<0.0001, Mann–Whitney test) and circularity (****P<0.0001, Mann–Whitney test) of ASM cells from E12.5 Tbx4-rtTA; Tet-O-Cre; Confetti (n=294 cells from 3 lungs) and Tbx4-rtTA; Tet-O-Cre; Vangl1fl/fl; Vangl2fl/fl; Confetti lungs (n=205 cells from 4 lungs). (E) Schematic illustrating how ASM cell orientations were measured relative to the long axis of the airway. (F) Rose plot of ASM cell orientations from E12.5 Tbx4-rtTA; Tet-O-Cre; Confetti lungs (n=296 cells from 3 lungs). (G) Rose plot of ASM cell orientations from E12.5 Tbx4-rtTA; Tet-O-Cre; Vangl1/2fl/fl; Confetti lungs (n=224 cells from 4 lungs). (H) Schematic of proposed ‘dilational extension’ model for ASM-guided airway stabilization, lengthening and widening.
Vangl2 is broadly expressed throughout the mesenchyme during organogenesis
Our data show that Vangl1/2 function outside of the core PCP complex in an intricate, nonplanar, three-dimensional tissue to facilitate mesenchymal organization and epithelial morphogenesis. We hypothesize that Vangl exerts control over the cytoskeleton to correctly structure tissue layers within the mesenchyme. We were curious whether this function for Vangl might be conserved in the mesenchyme of other organs.
To assess whether there may be a broader role for Celsr1-independent Vangl2 in organogenesis, we took advantage of an endogenously tagged tdTomato-Vangl2 mouse line to examine the expression pattern of Vangl2 across various organs (Basta et al., 2021). Our analysis revealed that Vangl2 was localized to the membrane in mesenchymal cells of the developing kidney and intestine (Fig. 7A). Consistent with our data for the lung mesenchyme, scRNA-seq analysis showed that Vangl2 was expressed ubiquitously, whereas Celsr1 was excluded from the mesenchyme in these organs (Fig. 7B,C; Fig. S7A,B) (Naganuma et al., 2021; Dong et al., 2018). Thus, Vangl2 is expressed in the mesenchyme in the absence of the key core PCP component Celsr1 in several embryonic mouse organs.
Vangl2 is widely expressed in embryonic mesenchymal tissues. (A) Representative images of sections (10-µm thick) from E16.5 Vangl2-tdTomato kidney and intestine immunostained for Ecad (green) or tdTomato (magenta), with nuclei counterstained with Hoechst (blue). n=3 embryos. Scale bars: 50 µm. (B,C) Single-cell RNA-sequencing analysis of the developing kidney or intestine revealing expression of Vangl2 or Celsr1 in tissue compartments (Naganuma et al., 2021; Dong et al., 2018).
Vangl2 is widely expressed in embryonic mesenchymal tissues. (A) Representative images of sections (10-µm thick) from E16.5 Vangl2-tdTomato kidney and intestine immunostained for Ecad (green) or tdTomato (magenta), with nuclei counterstained with Hoechst (blue). n=3 embryos. Scale bars: 50 µm. (B,C) Single-cell RNA-sequencing analysis of the developing kidney or intestine revealing expression of Vangl2 or Celsr1 in tissue compartments (Naganuma et al., 2021; Dong et al., 2018).
DISCUSSION
The fractal pattern of the mature airway epithelial tree requires that lung branches differentially elongate and widen after they have formed. These processes have been attributed to the PCP complex in the branched epithelia of other organs. In these epithelia, core PCP function depends on the asymmetric localization of Celsr/Vangl and Celsr/Fzd complexes, which form intercellular junctions bridged by Celsr/Celsr homotypic binding (Devenport, 2014) (Fig. 1A). Our data reveal that neither Vangl1/2 nor Celsr1, key members of the core PCP complex, are necessary in the lung epithelium for branch initiation, elongation or widening during the pseudoglandular stage. Instead, we found a requirement for Vangl1/2 in the pulmonary mesenchyme, despite the fact that this tissue compartment lacks detectable expression of Celsr genes. Lungs lacking mesenchymal Vangl1/2 showed defects in airway elongation and widening by E13.5 and exhibited a severe reduction in branch initiation by E14.5. As reduction in branch initiation in vivo was observed after quantifiable differences in branch length and width, we hypothesize that the reduction in branching is secondary to defects in branch elongation and widening. Our study thus reveals the exciting result that Vangl1/2 expressed specifically in mesenchymal cells promotes airway initiation, elongation and widening. Taken together, our findings suggest a previously unreported, PCP complex-independent function for Vangl1/2, as the core PCP complex cannot exist in a tissue that lacks Celsr.
Our findings also suggest that the core PCP complex plays a distinct role in the lung compared with other branched organs, such as the kidney and mammary gland. Defects in branch elongation have been observed in epithelial-specific PCP complex-mutant kidneys and mammary glands (Lienkamp et al., 2012; Smith et al., 2019; Kunimoto et al., 2017), but these defects are not observed when the core PCP complex is depleted from the lung epithelium. Rather, the key role for the core PCP complex in the airway epithelium appears to be in the orientation and maintenance of cilia polarity (Vladar et al., 2012, 2016; Kunimoto et al., 2022). Our data thus indicate that the core PCP complex is not universally required in epithelia for the polarized behaviors that contribute to branching morphogenesis.
Notably, our analysis contradicts the conclusions made in previous work examining core PCP protein localization and how loss of PCP affects branching morphogenesis. Specifically, previous work concluded that Celsr1 is localized to the basal surface of epithelial cells and expressed in the pulmonary mesenchyme (Yates et al., 2010; Yates and Dean, 2011). In contrast, our immunofluorescence staining indicates that Celsr1 localizes exclusively to cell-cell junctions in the lung epithelium and mesothelium, but it is not present in the mesenchyme, consistent with our scRNA-seq and fluorescence in situ hybridization analyses. This discrepancy in Celsr1 localization may be due to differences in antigen-binding sites between the antibodies used in each study.
Furthermore, prior work inferred the extent of branching by quantifying tissue sections at E14.5 and concluded that the number of branches is reduced in Vangl2Lp/Lp and Celsr1Crsh/Crsh lungs (Yates et al., 2010). However, we observed no reduction in branching in E13.5 Celsr1Crsh/Crsh lungs in vivo or in cultured lung explants, which we assessed through whole-mount and time-lapse imaging analyses. We believe that the discrepancy between our conclusions and those of prior work using the Celsr1Crsh/Crsh strain are linked to the developmental delays observed in PCP complex-mutant embryos, which would logically lead to fewer branches by E14.5 compared with controls. Specifically, we posit that the decrease in branch number results from the gross embryonic abnormalities and reduced embryo size observed in Celsr1Crsh/Crsh embryos, a point well demonstrated by our experiments with the ShhCre; Vangl1fl/fl; Vangl2fl/fl strain, in which lungs exhibit normal branching when cultured ex vivo but appear stunted when growing within the developmentally delayed embryo. Regardless, our study demonstrates that the branching reduction observed in Vangl2Lp/Lp lungs can be attributed entirely to loss of Vangl1/2 from the mesenchymal compartment, whereas epithelial Vangl is dispensable for branch initiation, elongation and widening.
Our work also contradicts conclusions from a recent study, which reported alterations in the locations and angles of epithelial branches when Vangl1/2 are removed globally at early stages of mouse embryonic development (using Sox2Cre; Vangl1gt/gt; Vangl2fl/fl animals) (Zhang et al., 2022). We observed no obvious changes in airway location or angle in either Vangl2Lp/Lp or Vangl1/2 inducible-mesCKO lungs. Indeed, we observed caudal tilting of the cranial, medial and L1 bronchi in these mutant lungs by E13, comparable with those in littermate controls. In the same study, the authors assessed tissue-specific functions of Vangl1/2 by using the Sox9Cre and Dermo1Cre drivers to delete Vangl1/2 from the epithelium and mesenchyme, respectively. However, conclusions about tissue specificity derived from experiments with these drivers are complicated by two issues. First, in the lung, Sox9 is expressed in both the distal epithelium as well as in chondrocytes and the mesenchyme adjacent to the proximal epithelium (Turcatel et al., 2013). Second, the Dermo1Cre driver fails to deplete Vangl2 from the pulmonary mesenchyme at this stage of development (Fig. S4C). In agreement with our experiments, airway branching morphogenesis was unaffected by loss of Vangl1/2 from the lung epithelium (Sox9Cre) or mosaic depletion of Vangl1/2 from the pulmonary mesenchyme (Sox9Cre or Dermo1Cre) (Zhang et al., 2022). Although the authors interpreted this absence of phenotype to mean that Vangl1/2 are required in both tissue compartments, we found that fully removing Vangl1/2 solely from the mesenchyme via the Tbx4-rtTA; Tet-O-Cre driver resulted in defects in branching morphogenesis. These data are consistent with our conclusion that epithelial Vangl1/2 and the core PCP complex are unnecessary for branch initiation, orientation, elongation and widening in the developing mouse lung.
Wnt5a has been hypothesized to function upstream of both the core PCP complex generally and Vangl2 specifically in many tissues. Our data show that specific loss of Wnt5a from the pulmonary mesenchyme does not affect elongation or widening, although we did observe a slight decrease in branch initiation by E14. In fact, mesenchymal Wnt5a-knockout lungs appeared to be phenotypically normal at E12, in contrast to recent work concluding that loss of mesenchymal Wnt5a leads to an alteration in the position and angle of branches (Zhang et al., 2022). The different conclusions can again be attributed to the Cre drivers used in each study: both our work and the previous study demonstrate that broad deletion of Wnt5a from the splanchnic mesoderm (via Dermo1Cre) causes gross defects in the morphology of the lung, including changes in the angle of branches (Fig. S6B). However, we observed no such defects when we performed similar experiments using a highly efficient Cre driver that is restricted to the pulmonary mesenchyme (Tbx4-rtTA; Tet-O-Cre) (Fig. S6D). This discrepancy in phenotype can be explained by the expression pattern of these Cre lines: Dermo1Cre is expressed in tissues outside of the lung, whereas Tbx4-rtTA; Tet-O-Cre expression is restricted to the pulmonary mesenchyme (Zhang et al., 2013). Given that branching patterns are altered when the lung is constrained (Nelson et al., 2017; Gilbert et al., 2021), we speculate that loss of Wnt5a in Dermo1Cre-expressing non-pulmonary tissues leads to changes in the shape of the chest cavity, thereby stunting lung morphogenesis.
Our morphometric analysis revealed that the defects in airway lengthening observed in Vangl1/2 inducible-mesCKO lungs occurred concomitantly with defects in airway widening. Specifically, airways from inducible-mesCKO lungs were both shorter and narrower than their wildtype counterparts. We were initially surprised by this phenotype, as defects in tissue elongation are often coupled with abnormal widening along the orthogonal axis; for example, embryos that fail to undergo convergent extension are short and wide (Goto and Keller, 2002; Paudyal et al., 2010; Sutherland et al., 2020). However, this paradoxical phenotype has been observed previously in other mutations that affect smooth muscle. For example, loss of the potassium channel Kcnj13 leads to disorganized alignment of smooth muscle around the trachea and both shorter and narrower tracheas in mice (Yin et al., 2018). Similarly, in Ror2- and Wnt5a-mutant lungs, tracheal smooth muscle cells exhibit defects in polarization and radial intercalation, resulting in short and narrow tracheas (Kishimoto et al., 2018).
At the cellular level, we found that loss of Vangl1/2 led to alterations in ASM cell morphology. In control lungs, ASM cells extended and wrapped circumferentially around airways. When Vangl1/2 were absent, ASM cells failed to extend and instead exhibited a cobblestone-like morphology; subsequently, airways collapsed and failed to elongate or widen. Based on these data, we propose a conceptual model wherein normal wrapping and intercalation of ASM cells function to both stabilize and expand airways during branching morphogenesis. Our model first posits that the ASM layer functions as a brace, preventing inward collapse of the airway through the adhesion of the epithelium to their shared basement membrane (Fig. 6H). In support of this model, ASM is a stiffer tissue than both the airway epithelium and the surrounding undifferentiated mesenchyme (Goodwin et al., 2023, 2022), and prematurely arresting the ASM differentiation program leads to airway collapse (Young et al., 2020; Goodwin et al., 2022). We hypothesize that Vangl controls ASM cell shape and stiffness through effects on the cytoskeleton; given the change in cell shape and tissue morphology, loss of Vangl likely results in a softer smooth muscle layer, leading to the airway collapse observed in Vangl1/2-mutant lungs. The second supposition of this model is that continual axial and circumferential intercalation of nascent ASM cells along an established airway ratchets the underlying epithelial tube both longer and wider (Fig. 6H). We hypothesize that Vangl regulates the polarization of ASM cells, and thus their ability to intercalate and wrap properly around the airways. Without the requisite continual intercalation of new ASM cells, we speculate that the resulting disorganized ASM tissue layer cannot exert the forces required to promote elongation and widening of the airways. We have termed this two-part mechanism ‘dilational extension’, in contrast to the convergent extension mechanism that lengthens epithelia in other branched organs such as the kidney (Kunimoto et al., 2017; Lienkamp et al., 2012). A similar set of mesenchymal rearrangements is thought to play a role in the developing mouse trachea: the polarization and intercalation of tracheal smooth muscle cells leads to both tracheal elongation and widening (Kishimoto et al., 2018).
In previous work, we found that Vangl1/2 are required during lung sacculation, the developmental period following branching wherein the pulmonary mesenchyme thins while the epithelium expands in preparation for gas exchange after birth (Paramore et al., 2024). Specifically, we found that the pulmonary mesenchyme must be highly fluid at this later stage of development to allow sacculation to occur, and that this fluidity appears to be dependent on mesenchymally expressed Vangl1/2. In contrast, here, we demonstrate a role for Vangl in the morphology and assembly of the ASM layer. Although we hypothesize that Vangl plays a role in cytoskeletal organization in each case, it appears likely that there are temporally divergent functions for Vangl in the pulmonary mesenchyme. For example, we found that loss of mesenchymal Wnt5a phenocopies the sacculation defects observed in mesenchymal Vangl1/2-knockout lungs (Paramore et al., 2024), but not the defects in branch initiation, elongation or widening observed here. It is possible that other non-canonical Wnt ligands, such as Wnt11, may compensate for Wnt5a during branching morphogenesis but not during sacculation. Furthermore, although we propose a role for Vangl1/2 in regulating the polarity of ASM cells in this study, during sacculation, we posit that Vangl1/2 functions in a different cell population, namely, saccule-adjacent mesenchymal cells, to promote tissue fluidity. We find it unsurprising that these two different functions of Vangl2 may be downstream of two separate signaling molecules.
It is likely that Vangl2 has PCP-independent functions in organs beyond the lung. We found that Vangl2 is expressed in the absence of Celsr1 in the mesenchyme of both the kidney and intestine, suggesting a role outside of the core PCP complex in those tissues. A related polarity pathway, the Fat-Dchs pathway, has been shown to interact genetically with Vangl2 to regulate the migration and clustering of mesenchymal cells in the developing intestine (Rao-Bhatia et al., 2020). Collectively, these data suggest a role for polarity signaling in diverse, three-dimensional mesenchymal tissues. We propose that, in the context of vertebrate development, Vangl has evolved functions distinct from the core PCP complex to regulate mesenchymal cell dynamics during organogenesis. Future work will explicitly uncover the pathway via which mesenchymal Vangl2 functions to promote epithelial remodeling to form the fractal tree of airways.
MATERIALS AND METHODS
Mouse lines and breeding
All procedures involving animals were approved by the Institutional Animal Care and Use Committee (IACUC) at Princeton University. Mice were housed in an American Association for Accreditation of Laboratory Animal Care (AAALAC)-accredited facility in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. This study was compliant with all relevant ethical regulations regarding animal research. CD1 mouse embryos were used for scRNA-seq (Goodwin et al., 2022), fluorescence in situ hybridization (RNAscope), and immunostaining experiments to determine expression and localization of core PCP components. Vangl2Lp/Lp embryos (Kibar et al., 2001) and Celsr1Crsh/Crsh embryos (Curtin et al., 2003) were used to determine how the loss of core PCP function affects airway epithelial branching. The Vangl2Lp strain was maintained in a mixed C57BL/6 and C3H/HeJ background, as crossing into the C3H/HeJ background reduced the penetrance of hermaphroditism. The Celsr1Crsh strain was maintained on a mixed BalbC/CD1 background. ShhCreGFP; Vangl1fl/fl; Vangl2fl/fl; Rosa26mTmG/Rosa26mTmG embryos were used to conditionally delete Vangl1/2 from the lung epithelium (Harfe et al., 2004; Harris et al., 2006; Copley et al., 2013; Chang et al., 2016; Muzumdar et al., 2007) and the strain was maintained on a C57BL/6 background. Dermo1Cre; Vangl1fl/fl; Vangl2fl/fl; Rosa26mTmG/Rosa26mTmG embryos were used to conditionally delete Vangl1/2 from the pulmonary mesenchyme (Yu et al., 2003; Yin et al., 2008; Sosic et al., 2003) and the strain was maintained on a C57BL/6 background. Tbx4-rtTA; Tet-O-Cre; Vangl1fl/fl; Vangl2fl/fl; Rosa26mTmG/Rosa26mTmG embryos were used to conditionally delete Vangl1/2 from the pulmonary mesenchyme in an inducible manner (Zhang et al., 2013; Perl et al., 2002) and the strain was maintained on a C57BL/6 background. To induce deletion of Vangl1/2, doxycycline-medicated water (0.5 mg/ml) was administered to pregnant dams at E8. Dermo1Cre; Wnt5afl/fl embryos were used to conditionally delete Wnt5a from the pulmonary mesenchyme (Ryu et al., 2013) and the strain was maintained on a C57BL/6 background. Tbx4-rtTA; Tet-O-Cre; Wnt5afl/fl embryos were used to conditionally delete Wnt5a from the pulmonary mesenchyme in an inducible manner and the strain was maintained on a C57BL/6 background; doxycycline was administered in drinking water at E8 as described above. Tbx4-rtTA; Tet-O-Cre; Vangl1fl/fl; Vangl2fl/fl; R26R-Confetti (Livet et al., 2007; Snippert et al., 2010) embryos were used for cell shape analysis and the strain was maintained on a C57BL/6 background. tdTomato-Vangl2 and Celsr1-3xGFP embryos were used to determine the localization of Vangl2 and Celsr1 (Basta et al., 2021) and were maintained on a C57BL/6 background. For all experiments, embryos of both sexes were used. Mouse strains and genotyping primers are listed in Table S1.
Immunofluorescence analysis
Lungs were dissected from embryos in PBS and fixed in 4% paraformaldehyde (PFA). Whole-mount E11.5-E13.5 lungs were fixed for 30 min at 4°C, whereas whole-mount E14.5 lungs were fixed for 1 h at 4°C. Lungs were then washed in PBS and blocked overnight in blocking buffer composed of 4% normal donkey serum, 1% bovine serum albumin (BSA) and 1% fish gelatin in PBS with 0.2% Triton X-100. Lungs were incubated with primary antibody in blocking buffer overnight (two nights for E14.5 lungs). Lungs were washed with PBS, incubated with secondary antibody overnight (two nights for E14.5 lungs), and then washed in PBS, dehydrated through a methanol series, and cleared for imaging in 1:1 benzyl alcohol:benzyl benzoate. After fixation, lungs for sectioning were washed in PBS and taken through a sucrose gradient before embedding in OCT (Tissue Tek). 10-µm-thick or 200-µm-thick frozen sections were obtained from samples using a Leica CM3050S cryostat. 10-µm-thick sections were washed in PBS with 0.3% Triton X-100, then washed in PBS, incubated in blocking buffer for 1 h at room temperature, and then incubated in blocking buffer with primary antibody overnight. Slides were then washed in PBS, incubated with secondary antibody for 3 h at room temperature, washed, and mounted in Prolong Gold. 200-µm-thick floating sections were washed in PBS with 0.3% Triton X-100, then washed in PBS, incubated in blocking buffer overnight at 4°C, and then incubated in blocking buffer with primary antibody overnight. Sections were then washed in PBS, incubated with secondary antibody overnight at 4°C, washed in PBS and mounted in Prolong Gold. All reagents and antibodies used for staining are described in Table S1. The surfaces tool in Imaris software (Oxford Instruments) was used to calculate epithelial volume using fluorescence intensity from E-cadherin (Cdh1) immunostaining. The number of terminal branches was quantified manually in FIJI using the Cell Counter plugin; quantifications were performed on confocal z-stacks of whole-mount lungs immunostained for E-cadherin, with a z-step of 4 µm.
scRNA-seq analysis
Data for scRNA-seq analysis were generated in a previous study (GSE153069) (Goodwin et al., 2022). Briefly, the left lobes of lungs were dissected in ice-cold PBS from CD1 mouse embryos collected at E11.5, then placed in dispase and mechanically dissociated using tungsten needles. After 10 min in dispase at room temperature, DMEM (Fisher, SH3027202) without HEPES and supplemented with 5% fetal bovine serum (FBS, Atlanta Biologicals) was added, and cell suspensions were passed through a filter with 40-µm-diameter pores. The resultant cell suspensions were then processed by the Princeton Genomics Core Facility using the Chromium Single Cell 3′ Library and Gel Bead Kit v2 (10× Genomics), and then loaded onto the Chromium Controller (10× Genomics). Illumina sequencing libraries were prepared from the amplified cDNA from each sample group using the Nextra DNA library preparation kit (Illumina) and then sequenced on Illumina HiSeq 2500 Rapid flowcells (Illumina) as paired-end 26+50 nucleotide reads. Reads were processed using the Illumina sequencer control software to retain only pass-filtered reads for downstream analysis. The 10× CellRanger software v2.0.1 was used to generate gene-barcode matrices using the Mus musculus reference genome mm10-1.2.0.
Data were then processed using the Seurat package in R following standard procedures (Butler et al., 2018). The data were normalized, integrated based on 2000 variable genes, then scaled and analyzed to find neighbors and clusters. Clusters were identified based on known cell type markers, and gene expression in each cell type was visualized using violin plots.
We analyzed two additional published scRNA-seq datasets generated from embryonic mouse tissues: E15.5 kidney (GSE149134) (Naganuma et al., 2021) and E9.5-E11.5 intestine (GSE87038) (Dong et al., 2018). All analyses were carried out using the Seurat package (Butler et al., 2018). The intestinal scRNA-seq data consist of four datasets from embryos at different stages of development (two at E9.5, one at E10.5 and one at E11.5), which we merged prior to analysis. The data were first filtered to exclude cells with fewer than 500 genes, more than 30,000 unique molecular identifiers (possible multiplets), or greater than 10% mitochondrial DNA (dying cells). Following the Seurat pipeline, we then normalized the data, identified variable features, scaled gene expression for each cell, ran principal component analysis, and identified neighbors and clusters. We then generated uniform manifold approximation and projection (UMAP) plots and extracted cluster markers to identify cell types. Clusters representing distinct cell types were identified by color coding the UMAP based on known markers of epithelial and mesenchymal cells in the kidney (Cdh1 and Cdh2) and the intestine (Cdh1 and Foxf1). We then examined the expression patterns of Celsr1 and Vangl2 by color coding the UMAP and comparing cell-level expression of these two genes with that of the cell type-specific markers listed above.
Organ explant culture
E12.5 lungs were dissected in sterile PBS and cultured at an air-liquid interface on porous membranes (nucleopore polycarbonate track-etch membrane, 8-μm pore size, 25-mm diameter; Whatman) in DMEM/F12 medium (Fisher, 11965092; without HEPES) supplemented with 5% FBS and antibiotics (50 U/ml of penicillin and streptomycin). Explants were fixed for 30 min in 4% PFA at room temperature and stained and imaged as described above. The number of terminal branches was quantified manually in FIJI using the Cell Counter plugin; quantifications were performed on confocal z-stacks of whole-mount lungs immunostained for E-cadherin, with a z-step of 4 µm (representative examples of quantification are in Movie 1).
Fluorescence in situ hybridization (RNAscope) analysis
Lungs were isolated at E12.5-E13.5 and immediately fixed in 4% PFA in diethyl pyrocarbonate (DEPC)-treated PBS for 24 h at 4°C. Lungs were taken through a sucrose gradient (20% sucrose in DEPC-treated PBS for 1 h at 4°C, then 30% sucrose in DEPC-treated PBS overnight at 4°C), embedded in OCT, and frozen on dry ice. 10-µm-thick frozen sections were obtained from each sample using a Leica CM3050S cryostat. Fluorescence in situ hybridization was performed using the standard RNAscope Multiplex Fluorescent V2 Assay (Advanced Cell Diagnostics) protocol for fixed-frozen samples. The probes were for M. musculus Celsr1 (channel 1, 317931), Celsr2 (channel 1, 317941), Celsr3 (channel 1, 319241), Fgf10 (channel 2, 446371) and Wnt5a (channel 2, 316791), and the fluorophores were Opal Polaris 520 and Opal 620 (Akoya Biosciences, FP1487001KT and FP1495001KT). Sections were imaged on a Nikon A1RSi confocal microscope with a 40× objective.
Airway smooth muscle cell-shape analysis
Images of RFP-labeled control or Vangl1/2-mutant ASM cells were obtained by imaging lungs from Tbx4-rtTA; Tet-O-Cre; R26R-Confetti and Tbx4-rtTA; Tet-O-Cre; Vangl1fl/fl; Vangl2fl/fl; R26R-Confetti E12 embryos. ASM cells were identified by their proximity to airways and by their expression of αSMA. ASM cells were segmented and binarized using the ilastik 2-stage autocontext workflow (Berg et al., 2019), using the signal from cytoplasmic RFP+ mesenchymal cells. Segmented images were then analyzed in MATLAB by labeling each connected component using the bwlabel function. The segmented cell metrics of circularity, major and minor axis lengths were then calculated using the regionprops function. Aspect ratio was calculated as the ratio of the major and minor axis lengths. Histograms were then generated using R and the ggplot2 package (https://www.r-project.org/; Villanueva and Chen, 2019).
Acknowledgements
We thank Dr Wei Shi (Keck School of Medicine, University of Southern California) for generously providing the Tbx4-rtTA; Tet-O-Cre mice. S.V.P. is very grateful to Katie Little for her work on maintaining mice, genotyping and assisting with revisions. We thank Sha Wang and the Microscopy Core Facility at Princeton University, a Nikon Center for Excellence, for their assistance with imaging.
Footnotes
Author contributions
Conceptualization: S.V.P., D.D., C.M.N.; Methodology: S.V.P.; Validation: S.V.P.; Formal analysis: S.V.P., K.G.; Investigation: S.V.P., E.W.F.; Writing - original draft: S.V.P.; Writing - review & editing: D.D., C.M.N.; Supervision: D.D., C.M.N.; Project administration: D.D., C.M.N.; Funding acquisition: D.D., C.M.N.
Funding
This work was supported in part by grants from the National Institutes of Health (HL110335, HL118532, HL120142, HL164861, HD099030, HD105009, HD111539, AR066070, AR068320), the Genetics and Molecular Biology Training Grant of the Molecular Biology Department at Princeton University (T32 GM007388), and a Faculty Scholars Award from the Howard Hughes Medical Institute. S.V.P. was supported in part by the National Cancer Institute Ruth L. Kirschstein (F31) Fellowship. K.G. was supported in part by the postgraduate scholarship-doctoral (PGS-D) program from the Natural Sciences and Engineering Research Council of Canada, the Dr Margaret McWilliams Predoctoral Fellowship from the Canadian Federation of University Women, a predoctoral fellowship from the American Heart Association, and the Princeton University Procter Fellowship. Open access funding provided by Princeton University. Deposited in PMC for immediate release.
Data availability
All relevant data can be found within the article and its supplementary information.
Peer review history
The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202692.reviewer-comments.pdf
References
Competing interests
The authors declare no competing or financial interests.