During development, the lung is formed through an elaborated epithelial–mesenchymal crosstalk that drives lung specification, budding, and branching. Signaling molecules like fibroblast growth factors (Fgf), Wnt (wingless and int), Sonic hedgehog (Shh), and bone morphogenetic proteins (Bmp) are key ligands initiating the pulmonary cell fate and specifying the early lung domain at the ventral foregut endoderm (21). So far, the most convincing evidence for epithelial–mesenchymal interactions during lung development came from recombination studies where distal lung mesenchyme, grafted on the tracheal epithelium led to ectopic budding accompanied by expression of surfactant protein C as a distal epithelial marker (22–24).

The mammalian Fgf family consists of 22 members subdivided in 7 subfamilies, based on phylogenetic as well as gene loci analyses (25). Fgfs acts in a paracrine, endocrine, or intracrine fashion and have diverse biological activities during embryonic organogenesis. These growth factors act via seven main receptors (Fgfrs 1b, 1c, 2b, 2c, 3b, 3c, and 4), exhibiting different ligand-binding specificity. The Fgf receptors are encoded by four Fgfr genes (Fgfr1–Fgfr4), which undergo alternative splicing to produce the different isoforms. Each receptor comprises an extracellular ligand-binding domain with three immunoglobulin-like loops (D I, D II, D III), a transmembrane domain and an intracellular tyrosine kinase domain. Human diseases involving gain or loss of function mutations have been described. For example, loss of function of FGF3 causes deafness, heterozygous loss of function of FGF10 results in lacrimo-auriculo-dento-digital syndrome (LADD syndrome), FGF10 haploinsufficiency is also associated with chronic obstructive pulmonary disease and FGF23 gain of function leads to autosomal dominant hypophosphataemic rickets (26–29). During early (E12.5) embryonic mouse lung development, Fgf9 and Fgf10 have been shown to play an important role in branching morphogenesis and the associated differentiation of the epithelium and mesenchyme. Fgf9 is expressed in the mesothelium and the epithelium and acts through Fgfr2c- and Fgfr1c-expressing cells in the mesenchyme to maintain Fgf10 expression as well as mesenchymal progenitors proliferative and undifferentiated (30). It also can signal directly to the epithelium to promote epithelial branching by induction of Dkk1 expression and inhibition of Wnt signaling (31). Fgf10 is a diffusible key molecule orchestrating branching morphogenesis during early lung development in mice (32, 33) but the exact mechanism of action remains unknown. During the early pseudoglandular stage, Fgf10 is secreted by cells located adjacent to the mesothelium in the distal mesenchyme and signals in a paracrine manner mainly through fibroblast growth factor receptor 2-IIIb (Fgfr2b) expressed on epithelial cells. Fgf10 has a high affinity for heparan sulfate and is therefore unlikely to diffuse over a long distance. Instead, Fgf10 promotes outgrowth of the distal epithelium via a chemotactic mechanism. Several studies using transgenic mouse lines that display abnormal Fgf10/Fgfr2b signaling confirmed the importance of this pathway (Table 1). Fgf10 and Fgfr2b knockout pups display similar phenotypes. The mutant pups die shortly after birth due to lung agenesis and multiple organ agenesis/defects (salivary gland, limb, inner ear, teeth, skin, pancreas, kidney, thyroid, pituitary gland, mammary gland) (34–38).

In order to identify epithelial-specific gene expressions mediated by recombinant human FGF10 during bud morphogenesis, Lu and colleagues (39) used mesenchyme-free epithelium in culture. By using microarray analysis, they identified a panel of transcriptional Fgf10 targets, which are associated with cell rearrangement, migration, inflammatory processes, lipid metabolism, cell cycle, and tumor invasion. Interestingly, the authors did not observe a remarkable induction of genes responsible for proliferation. Moreover, Fgf10 is proposed to control the angle of the mitotic spindle in distal epithelial cells during development. Thus, Tang et al. argued that Fgf10 signals via a Ras-regulated Erk1/2 signaling pathway to shape the lung tube (40). Fgf10 is also critical for the amplification of distal epithelial cell progenitors and for the formation of multiple mesenchymal lineages during lung development. Hypomorphic Fgf10^lacZ/− pups expressing ~20% Fgf10 compared to wild type (WT) died within 24–48 h after birth due to lung defects, which included decreased branching, thickened primary septae, and vascular abnormalities with intrapulmonary hemorrhages. At the cellular level, Fgf10 deficiency led to decrease in Nkx2.1 and Sftpb-expressing cells, suggesting that adequate Fgf10 expression level is critical for the amplification of epithelial progenitors. Apart from the epithelium, constitutive decrease in Fgf10 expression also affects mesenchymal cell lineages as Pecam and αSma-positive cells are also diminished (41). Interestingly, recent experiments conducted in our lab to investigate the impact of Fgf10 levels on lung function demonstrate that even a 50% decrease in Fgf10 expression (Fgf10 heterozygous pups) leads to changes in the expression of genes relevant for lung development such as Epcam, Sftpc, Fgfr2b, Tgf-β, and Collagen. Additionally, Fgf10 heterozygous neonatal mice survive and do not display any obvious phenotypic differences compared to WT mice. However, when exposed to hyperoxia between PN0 and PN8 to trigger lung injury and mimic some of the clinical manifestations of BPD (impaired alveologenesis and inflammation), Fgf10-deficient pups display drastically increased mortality compared to WT controls. Further analysis indicates that under physiological conditions, Fgf10-deficient mice already show structural abnormalities during embryonic lung development supporting that Fgf10-deficient pups carry congenital defects. These findings suggest that Fgf10-deficient lung epithelium is more susceptible to oxygen toxicity and does not undergo normal repair after injury (Chao and Bellusci, in preparation). Additionally, recent studies suggest that Fgf10 may control basal cell density in the tracheal epithelium (42–45). This is not surprising as it has already been previously shown, by our group and others, that Fgf10 is part of the stem cell niche in the lung (20, 46, 47).

Wnt (Wnt2, Wnt2b) ligands are expressed in the mesenchyme and are important for lung domain specification of the foregut endoderm from E9.0 to E10.5. Wnt signaling is also essential for the proximo-distal patterning of the epithelium during embryonic lung development. Genetic deletion of Wnt2/2b or β-catenin leads to lung agenesis due to loss of Nkx2.1 (48, 49). Wnt2 null mice display lung hypoplasia and abnormal development of ASMCs (50). Furthermore, Mucenski and colleagues demonstrated, by using Spc-rtTA;tet(O)Cre double transgenic mice, that loss of function of β-catenin in the distal lung epithelium leads to the inhibition of distal airway formation (51). The authors showed an opposite phenotype by inducing gain of function of β-catenin signaling (52). The absence of Wnt7b results in a phenotype similar to Wnt2 null mice (53) but a combination of Wnt7b and Wnt2 loss of function leads to a more severe phenotype with decreased branching and abnormal distal endoderm patterning (54). The constitutive deletion of Wnt5a – a non-canonical Wnt ligand expressed in the mesenchyme and the epithelium – leads to increased proliferation of the mesenchyme and the distal epithelium as well as disrupted lung maturation (55). β-catenin inactivation in the mesenchyme (Dermo1-Cre line) leads to abnormal mesenchyme development with disrupted amplification of ASMC progenitors and defects in angioblast differentiation (56). Kumar and colleagues demonstrated by using a clonal cell labeling approach that ASMC progenitors are located exclusively at the tip mesenchyme and that mesenchymal Wnt signaling is able to prime the stalk mesenchyme to form an ASMC progenitor pool at the tip (57).

Bmp4 is dynamically expressed in the endoderm and in the mesenchyme during early embryonic lung development (E11.5). It is also expressed at the distal epithelial buds and has been shown to be an inhibitor of Fgf10-induced chemotaxis in the epithelium. Bmp4 controls intraepithelial crosstalk to form ASMCs. It has been shown that Fgf10 is able to upregulate Bmp4 mRNA expression. In vitro experiments demonstrated that exogenous recombinant human BMP4 inhibits Fgf10-induced bud outgrowth, providing evidence that Bmp4 is acting downstream of Fgf10 to inhibit its signaling cascade (58–61).

Other Fgf10 inhibitors are Sonic hedgehog (Shh) and Sprouty homolog 2 (Spry2), both expressed in the epithelium of the outgrowing buds. Shh is a secreted growth factor that acts through its mesenchymal receptor Patched (Ptc) to induce mesenchymal cell proliferation and differentiation. In E11.5 lung explants, exogenous recombinant SHH is able to induce expression of mesenchymal markers (Noggin, Acta2, Myosin) (19, 62). Spry2 is an intracellular inhibitor of receptor tyrosine kinase signaling (63, 64); (32). Using in vitro approaches, it has been shown that Spry2 reduction leads to increased epithelial branching and vice versa.

Apart from its important role in development, the mesenchyme is crucial in disease pathogenesis. Indeed, it has been reproducibly shown that the lung mesenchyme in preterm infants dying from BPD includes interstitial fibrosis and thickening with increased total collagen content (65–67). Similar findings were obtained in diverse animal models (rat, mice, baboon) recapitulating the conditions of preterm infants after birth (mechanical ventilation, oxygen supplementation, exogenous surfactant) leading to a human BPD-like phenotype (68–72). These pathological changes in the lung mesenchyme in BPD strengthen the concept that alveolarization depends on an intact and normally developed mesenchyme. Several studies using animal models of BPD to identify molecules located in the altered lung mesenchyme contributed to our understanding of disease pathogenesis. Some of them will be reviewed in the following section.

One of the major causes of BPD is believed to be inflammation. Inflammation is caused prenatally by chorioamnionitis and postnatally by mechanical stretch (ventilation), oxygen toxicity, as well as infection. Emerging evidence gained from in vitro and in vivo studies support this hypothesis (73–77). For example, it has been shown that lipopolysaccharides (LPS from Escherichia coli) inhibit branching morphogenesis in vitro (73). Blackwell et al. published similar results using activated resident macrophages to inhibit epithelial branching. The proposed mechanism is that LPS activates nuclear factor kappa beta (NF-kappa B), which is then accompanied by increased expression of interleukin-1beta (IL-1β) and tumor necrosis factor-a (TNF-a) in resident macrophages (74). This branching inhibitory effect caused by macrophage-mediated inflammation has been confirmed by a macrophage-depletion study in the lung. Benjamin et al. explained this inhibitory effect by linking Fgf10 signaling with inflammatory signals. Using in vitro experiments, they demonstrated that NF-kappa B, IL-1β, and TNF-a are capable of reducing Fgf10 expression in LPS-treated primary mesenchymal cells. The mechanism involved activation of toll-like receptors 2 and 4 (TLR2/4). The authors showed that FGF10-positive cells were decreased in lung samples of premature infants who died from BPD (75).

Tgf-β1 has been demonstrated to induce epithelial–mesenchymal transition (EMT) of AEC to MYF-like cells leading to extracellular matrix (ECM) deposition and thereby contributing to fibrosis and destruction of alveolar structure (78–80) (see also Table 1). Endogenous nitric oxide is proposed to attenuate EMT in AECs in an in vitro approach using primary culture of AEC II (81).

As previously mentioned, the alveologenesis phase leads to a dramatic increase in alveolar surface, which is essential for gas exchange. The current consensus is that this process is interrupted by exogenous deleterious factors leading to simplification of alveoli in BPD. Many studies confirmed that the alveolar MYF, located in the mesenchyme, is the unique cell type responsible for secondary septae formation. During alveologenesis, the alveolar myofibroblast is characterized by expression of alpha-smooth-muscle-actin (αSMA or Acta2) compared to other mesenchymal fibroblast population. By deposition of elastin and collagen, the alveolar myofibroblast initiates the process of secondary septation (82, 83). Both elastin and alveolar myofibroblast have been shown to be critical for secondary septae formation (83, 84). Expression of tropoelastin starts in the pseudoglandular stage of lung development and reaches the highest level during the alveolar stage (85, 86). The strongest evidence so far showing the importance of elastin for secondary septae formation came from the Elastin-knock-out mice that reveal a complete failure of alveologenesis leading to an emphysematous-like phenotype (87, 88) (Table 1). Interestingly, both hyperoxia and mechanical ventilation lead to increased expression of Elastin (89–91). Fgfr3 and Fgfr4 have been shown to direct alveologenesis in the murine lung by controlling elastogenesis (92). By using mice homozygous for Pdgfa-null allele, Boström and colleagues demonstrated failed alveolar formation due to loss of alveolar myofibroblasts and consequent loss of Elastin fibers (93, 94). Likewise, blocking antibody against Pdgfrα in newborn mice (PN1–PN7) led to aberrant Elastin fiber deposition and impaired alveolar septation, resulting in long-term failure in alveologenesis that lasted into adulthood. Pdgfa is expressed in the epithelium and targets its receptor (Pdgfrα) on mesenchymal cells such as alveolar myofibroblast and LIF (Table 1). Given the many mesenchymal targets of Pdgfa, it is not clear whether the impact of Pdgfa or Pdgfrα deletion on myofibroblast formation is via a direct effect of Pdgfa on alveolar myofibroblasts (and or alveolar myofibroblast progenitors) or indirectly via Pdgfa action on other targets (ASMCs and LIF). Gain and loss of function for Pdgfa/Pdgfrα signaling using cell autonomous-based approaches in specific lineages should be carried out in the future to sort out these issues.

Interestingly, increased levels of Fgf signaling in the mesenchyme also leads to arrested development of terminal airways accompanied by reduced Elastin deposition (95). The authors achieved this condition by taking advantage of Fgfr2c^+/Δ mice that develop an autocrine Fgf10–Fgfr2b signaling loop in the mesenchyme due to a splicing switch, resulting in the ectopic expression of Fgfr2b instead of Fgfr2c. The proposed mechanism of action is that mesenchymal Fgf signaling suppresses the differentiation of alveolar myofibroblast progenitors. Furthermore, the blockade of Fgfr2b ligands in the lung from E14.5 to E18.5 by overexpression of a soluble dominant negative receptor of Fgfr2b (Sftpc-rtTA/+;tetOsolFgfr2b/+) blocking all Fgfr2b ligands also leads to arrest in secondary septae formation and alveolar simplification (96) suggesting that Fgfr2b ligands, during this time period, are also important for the formation of alveolar myofibroblasts. Subsequent treatment with retinoic acid (RA, biologically active derivative of vitamin A) induced re-alveolarization and was accompanied by increased Pdgfra-positive cells and decreased αSma/Acta2-positive cells. Concurrent induction of the dominant negative Fgfr2b in these experimental conditions is able to prevent the RA-mediated alveolar regeneration. These data suggest that re-alveolarization is dependent on Fgfr2b ligands. Furthermore, the authors proposed a conceptual model that alveolar myofibroblasts (αSma/Acta2-positive) arise from Pdgfra-positive LIF. Specific lineage-tracing studies targeting subsets of lung fibroblasts (e.g., Adrp for LIF) are needed to validate this model. Chen and colleagues also demonstrated that Fgfr2b ligands are necessary for alveolar myofibroblast formation during compensatory lung growth after pneumonectomy (97). However, the blockade of Fgfr2b ligands by soluble Fgfr2b (decoy receptor) postnatally during alveolarization does not impair alveologenesis in mice. Recently, it has been shown that reduced Pdgfra expression is a primary feature of human BPD. The authors showed decreased mRNA and protein expression of PDGFR-α and PDGFR-β in MSCs isolated from tracheal aspirates of premature neonates with BPD. Similarly, lungs of infants dying from BPD display less PDGFRα-positive cells in the alveolar septae. These findings were confirmed using a BPD mouse model exposed to hyperoxia (75% oxygen) for 14 days (98).

The LIF remains a poorly characterized lipid-containing interstitial cell located in the mesenchyme in close proximity to AEC II. LIFs, which accumulate lipid vacuoles (83, 99), are abundant in the early postnatal lung and regress significantly in number after alveolar septation. The presence of LIF in rodent lungs has been demonstrated extensively. However, whether LIF reside in adult human lung remains controversial (100, 101). Because of their close localization to AEC II, LIF have been proposed to interact with AEC II. Indeed, it has been shown convincingly that LIF are involved in the trafficking of lipids to the AEC II for surfactant production (102, 103). Apart from triglycerides, LIF also secrete leptin and retinoic acid, both important for surfactant production and alveolar septation (104, 105). On the other hand, AECII secrete parathyroid hormone-related protein (Pthrp) to signal through Pthrp receptor expressed on LIF to induce expression of adipose differentiation-related protein (Adrp) via peroxisome proliferator-activated receptor gamma (Pparg) pathway (Figure 3). The current consensus is that this signaling pathway is essential for the maintenance of the LIF phenotype as well as for regulation of surfactant production (104, 106–108). By performing co-culture experiments it has been proposed recently that LIF constitute a niche for AECII cells postnatally. Co-culture of LIF with AECII cells allows the formation of alveolospheres (109).

The contribution of LIF to lung regeneration and structural maintenance in later phases of life is currently unknown. Several lung injury models including cigarette-smoke exposure induce the transdifferentiation of LIF to αSma/Acta2⁺ MYF in vitro. These αSma/Acta2⁺ MYF are highly proliferative and express high levels of collagen (110). For this reason, it has been proposed that LIF are progenitors for alveolar MYF (Figure 2). However, an alternative and more plausible possibility is that LIF give rise to the activated MYF. Activated MYF, unlike the alveolar MYF, is involved in pathological situations and is responsible for fibrosis. Supporting this possibility, we have recently shown that during alveologenesis, Fgf10-positive cells give rise to LIF rather than alveolar MYF and during adult life, a subpopulation of Fgf10-expressing cells represents a pool of resident MSCs (Cd45⁻ Cd31⁻ Sca-1⁺) (20). In addition, the LIF-to-“activated MYF” transdifferentiation would translate indeed into loss of pulmonary integrity by smoke in vivo. Such transdifferentiation can be prevented and reversed in vitro using Pparg agonists such as rosiglitazone (111). However, it remains unclear whether such transdifferentiation occurs in vivo. Of note, exposure of premature neonates to hyperoxia induces arrest of alveolar septation and thickened primary septae due to MYF hyperplasia and excessive ECM production. Therefore, LIFs are unlikely progenitors for alveolar MYF. In the future, these results will have to be confirmed by lineage-tracing experiments in the context of injury using more specific knock-in lines to target the LIF and determine their fate.

In parallel to branching morphogenesis during early embryonic lung development the lung vasculature begins to form in the mesenchyme at around E10.0 (112). This process involves angiogenesis and vasculogenesis. Angiogenesis occurs when pre-existing endothelial cells sprout to form capillaries. In comparison, vasculogenesis is characterized by migration and differentiation of endothelial progenitor cells (or hemangioblasts) in the distal mesenchyme to form new blood vessels. DeMello and colleagues investigated the early fetal development of lung vasculature by employing light and transmission electron microscopy as well as vascular casts and scanning electron microscopy. They demonstrated three features of the lung vasculature occurring between E9.0 and E20.0 in mice: (1) angiogenesis occurs in the proximal (central) lung vasculature, (2) peripheral lung vessels are established by vasculogenesis, and (3) at E13.0/E14.0 the central and peripheral parts of lung vasculature begin to connect to each other via a lytic process (113, 114). Finally, the main event of microvascular maturation takes place during the alveolar stage of lung development where the transition from a double capillary network to a single capillary system within alveolar walls occurs.

Although BPD has long been regarded as an epithelial disease due to its emphysematous aspect, much emphasis has now been placed also on the role of the lung vasculature in this disease. The temporal–spatial proximity of lung vasculature development and branching morphogenesis suggests a close interaction between these two important structures via endothelial–epithelial tissue crosstalk. The better understanding of this crosstalk in development and disease condition might be highly relevant for future therapies. Vascular endothelial growth factor receptor 2 (Vegfr2 or Flk-1) is an early marker for endothelial progenitors located in the SEM (115, 116). However, it is not yet clear whether these progenitors arise from the mesothelium or the mesenchyme (117, 118). Progenitors for VSMCs are believed to arise from Fgf10⁺ cells (20), Wnt2⁺, Gli1⁺ and Isl1⁺ cells (coming from the second heart field) (18), Pdgfrb⁺ cells (119), and mesothelial cells (117, 118).

During murine embryonic development, Vegfr2-positive cells receive the Vegfa signal from epithelial and mesenchymal cells until E14.5, after which Vegfa expression becomes restricted to the epithelium (120) (Table 1). Furthermore, it has been shown that Shh and Fgf9, secreted by the epithelium, are able to induce expression of Vegfa in the mesenchyme (121). Reciprocally, mesenchymally secreted Fgf10 leads to the upregulation of Vegf in the distal epithelium (122). In our previous work, we showed that treatment of embryonic lung explants with recombinant Vegfa not only upregulates Vegfr2 in the mesenchyme but also induces branching of the epithelium (123). However, it is unclear whether the effect of Vegfa on epithelial branching is direct or indirect. This endothelial–epithelial tissue crosstalk has been extensively examined by using in vitro recombination studies (co-culture of epithelium and mesenchyme respectively and mesenchyme alone) as well as in in vivo lung agenesis model (β-catenin knockout) (112). Using an in vivo inducible decoy receptor of Vegfr1 (solubleVegfr1), Lazarus and colleagues demonstrated that Spry2 is upregulated in the epithelium upon inhibition of Vegfr1-mediated signaling, suggesting an inhibition of Fgf signaling (as mentioned before Spry2 is an inhibitor of Fgf10), which is essential for branching morphogenesis (124). Another link in the endothelial–epithelial crosstalk came from the Pecam1-deficient mice that display a failure in endothelial cell formation accompanied by simplified alveolarization (125).

During a pathological process, Vegfa has been found down-regulated in preterm infants with BPD (8, 126, 127). Furthermore, Thebaud and colleagues demonstrated that Vegf and Vegfr2 are decreased in the hyperoxia model of BPD in newborn rats and that adenoviral administration of VEGF improved alveolar architecture and promoted capillary formation (128, 129). Although the trophic and angiogenic potential of VEGF on the lung vasculature is known, the aforementioned study, and the studies from other groups, suggest that vascular growth serves as a driving force for alveolar growth and maturation, leading to improvement of lung structure, and promoting secondary septae formation. A recent report revealed the association of a VEGF polymorphism with BPD in Japanese preterm newborns (130).

Newborn mice that are hypomorphic for Fgf10 also display reduced expression of Vegfa and Pecam. These mice suffer from an oversimplified lung with an abnormally developed lung vasculature (41). Interestingly, Fgf10 expression is reduced in lungs from BPD patients (75). Whether the effect of mesenchyme-derived growth factors (such as Fgf10) on the lung endothelium is direct needs to be demonstrated. Another animal injury model demonstrating the importance of endothelial–epithelial interactions is the pneumonectomy model in mice. An inducible endothelium-specific deletion of Vegfr2 and Fgfr1 leads to reduction of Mmp14 secretion. Mmp14 is critical for expansion of epithelial progenitor cells during compensatory lung growth by unmasking Egfr ecto-domains. This was confirmed convincingly by rescue-experiments where EGF/MMP14 administration resulted in restored alveologenesis (131).