WNT through the Ranks: A Mesenchymal FOXF1-WNT5A Axis Sways Hierarchical Decisions in Alveolar Epithelial Development

Lung morphogenesis is elegantly orchestrated by cell–cell communication that is mostly mediated by diffusible, paracrine-acting signaling molecules. Among the most studied developmental pathways is WNT signaling, which has been shown to control cell behavior, including dictating cell-fate decisions, in both the epithelial and mesenchymal compartments of the embryonic and postnatal lung. WNT signaling operates via two arms: a β-catenin–dependent canonical arm and a β-catenin–independent noncanonical arm. Although canonical WNT signaling has been thoroughly studied in the context of lung development and injury, noncanonical WNT is less understood, with WNT5A being regarded as the prototypical noncanonical WNT ligand (1, 2), although it has also been reported to signal in a canonical manner (3, 4). Importantly, the noncanonical arm also regulates its canonical counterpart, thus adding to the overall complexity of this critical signaling pathway.

The formation of the alveolar epithelial lineage has gained increasing interest in recent years, as the mechanisms that are in play during lung development are likely involved in alveolar repair after injury, such as in the case of acute respiratory distress syndrome. Type 2 alveolar epithelial cells (AEC2s) are important progenitor/stem cells in the distal lung, as they self-renew and give rise to type 1 alveolar epithelial cells (AEC1s) (5). A subset of WNT-responsive AEC2s is particularly important for alveolar repair after influenza virus–induced pneumonia or hyperoxic injury (3, 6). Although it is well established that β-catenin signaling favors the formation of the AEC2 lineage (3, 7), upstream factors that control this process are still elusive. In this context, AEC2 hyperplasia is a feature of lung pathologies such as idiopathic pulmonary fibrosis and alveolar capillary dysplasia with misalignment of pulmonary veins (ACDMPV). The latter is a rare congenital disorder that leads to respiratory distress and pulmonary hypertension shortly after birth and is predominantly lethal. The vast majority of patients with ACDMPV carry mutations in the transcription factor FOXF1 (forkhead box f1) gene (8). Mice that are heterozygous for Foxf1 or carry the S52F Foxf1 mutation (a mutation that renders FOXF1 transcriptionally inactive) exhibit an alveolar capillary dysplasia phenotype (9, 10).

WNT5A expression is dysregulated in many human lung diseases, such as asthma (11), chronic obstructive pulmonary disease (12), idiopathic pulmonary fibrosis (13), pulmonary arterial hypertension (14), and bronchopulmonary dysplasia (4, 15). Wnt5a expression has been shown to be induced during AEC2-to-AEC1 differentiation in vitro upon treatment with IGF-I (insulin-like growth factor 1) (16). Moreover, in vivo studies have shown that loss of function of Wnt5a impairs AEC1 differentiation (15), while gain of function of WNT5a promotes AEC2-to-AEC1 differentiation (17). The latter study also showed that WNT5A inhibits canonical WNT signaling in a subset of AEC2s defined by low expression of Sftpc (surfactant protein C) (17). Another study showed that mesenchyme-derived WNT5A negatively regulates β-catenin signaling in AECs and impairs endogenous repair in emphysema (12). Furthermore, treatment with recombinant WNT5A reduces the formation of SFTPC⁺ human lung organoids (18). Last but not least, mesenchymal WNT5A mediates hyperoxic injury in the developing lung, therefore representing a potential therapeutic target in bronchopulmonary dysplasia (4).

In this issue of the Journal, Reza and colleagues (pp. 430–443) describe their use of an elegant model of alveolar organoids to link the ACDMPV-associated S52F Foxf1 mutation in lung fibroblasts to 1) increased number and size of the organoids, 2) increased AEC2:AEC1 ratio, and 3) increased AEC2s exhibiting proliferation and active β-catenin signaling (19). These results agree with histological examination of Embryonic Day 18.5 lungs isolated from control and S52F Foxf1 embryos in which the AEC2:AEC1 ratio was skewed toward AEC2s, with more AEC2s exhibiting activated β-catenin signaling. The organoid phenotype in terms of number and size could be reproduced using Foxf1^+/− fibroblasts. Importantly, the authors also observed decreased release of WNT5A by S52F Foxf1 fibroblasts in parallel to the enhanced epithelial β-catenin activation. Further chromatin immunoprecipitation–quantitative PCR, site-directed mutagenesis, and luciferase assays demonstrated that FOXF1 binds to an evolutionally conserved intronic enhancer within the Wnt5a genomic sequence and that such binding is decreased in S52F Foxf1 fibroblasts, thus leading to decreased Wnt5a expression. Finally, the authors show that the organoid phenotype obtained using S52F Foxf1 fibroblasts could be partially rescued by treatment with exogenous WNT5A and, conversely, that such phenotype could be mimicked in wild-type organoids by treatment with blocking antibodies against WNT5A.

Although these data clearly have the potential to influence future treatment strategies against ACDMPV, such as by pharmacological activation of FOXF1 or WNT5A, many open questions remain to be answered. Although the analysis of wild-type and S52F Foxf1 fibroblasts cultured in vitro showed similar expression of typical mesenchymal markers and did not show altered proliferation, it cannot be excluded that the S52F Foxf1 mutation causes intrinsic abnormalities in mesenchymal cells leading to impaired lung development independently of WNT5A-mediated mesenchymal–epithelial cross-talk. It has been shown that WNT5A directly acts on human lung fibroblasts to increase proliferation and resistance to apoptosis (20). Conditional deletion of the WNT5A receptors Ror1 and Ror2 in GLI1⁺ (glioma-associated oncogene 1–positive) mesenchymal cells, which are progenitors for alveolar myofibroblasts, impairs alveologenesis during late lung development (15). Moreover, treatment of early postnatal lung fibroblasts with recombinant WNT5A inhibits canonical WNT signaling and proliferation (17). Furthermore, mesenchymal inactivation of Pten (phosphatase and tensin homolog) leads to an alveolar capillary dysplasia–like phenotype via increased mesenchymal proliferation and defective angioblast-to-endothelial cell differentiation, thus leading to abnormal epithelial–capillary coupling and disorganized capillary beds (21). Although the phenotype reported in the latter study was accompanied by enhanced paracrine FGF10 (fibroblast growth factor 10)/FGFR2b (fibroblast growth factor receptor 2-IIIb) signaling that activates WNT signaling in epithelial progenitors, thus enhancing AEC2 abundance and inhibiting epithelium-derived Vegfa (vascular endothelial growth factor A) expression, there was a strong mesenchymal phenotype in terms of proliferation and differentiation (21). Such studies and others suggest that autocrine and/or paracrine signaling in the lung mesenchyme per se might also contribute to the phenotype driven by Foxf1 loss of function, independently of, but in addition to, the impact on the alveolar epithelium.

In this context, although the S52F Foxf1 mutation was confined to the fibroblast compartment in the organoid setting, the link to the in vivo setting and phenotype was established using a global and constitutive approach (i.e., ubiquitous loss of function of Foxf1 in all cell types throughout lung development). Therefore, an inducible, cell type–specific approach to manipulate the FOXF1–WNT5A axis in various cell types, including distinct mesenchymal cell subsets, at different intervals would help dissect epithelial versus intrinsic mesenchymal abnormalities and overlay and/or contrast the results with the epithelial–mesenchymal cross-talk abnormality reported in this study. Such analysis would further strengthen the causative link between mesenchymal Foxf1 loss of function and ACDMPV development and provide deeper mechanistic insights into the associated pathological events.

One final aspect is whether regulation of alveolar epithelial differentiation by mesenchymal FOXF1–WNT5A occurs at the level of a bipotential progenitor (favoring bipotential progenitor–to–AEC2 versus bipotential progenitor–to–AEC1 differentiation) or directly acting on the AEC2 progenitor pool and controlling its differentiation to AEC1s. In support of the latter scenario, the authors report that there was no colocalization between AEC1 and AEC2 markers in their organoid system, and previous studies have shown that WNT5A promotes AEC2-to-AEC1 differentiation (15, 17). However, whether the S52F Foxf1 mutation preferentially affects bipotential progenitors versus AEC2s and their progenitors in vivo requires further investigation.