Our lungs are highly complex organs that are exquisitely specialized for gas exchange and host defense. They have a large surface area exposed to the external environment, which makes them vulnerable to both infection and environmental damage. The adult lung has a tremendous capacity to regenerate following an acute insult. An example of an acute lung injury is partial pneumonectomy. If the entire left lobe of a mouse lung is removed, total lung volume is restored within 3 weeks by compensatory growth of the right lung. Detailed stereological analysis has shown that this compensatory growth involves both expansion of remaining alveoli and the formation of new alveolar units (neo-alveolarization; Voswinckel et al., 2004; Fehrenbach et al., 2008). In contrast, many chronic injuries are not repairable and can lead to lung conditions, such as emphysema in which alveolar units are lost. Numerous labs are characterizing adult lung stem cells with the long-term aim of manipulating their behavior for therapeutic purposes. One successful clinical example of stem cell therapy for the lung has been reported: transplantation of a decellularized trachea reseeded with the patient's own bronchial epithelial cells (Macchiarini et al., 2008). Similar techniques have been used to obtain bioengineered rat lungs, which were able to function for up to 6 hours after transplantation (Ott et al., 2010; Petersen et al., 2010). However, the successful bioengineering of an entire human lung is vastly more complex, and these advances remain far from routine clinical use. One area of current and future research of great clinical relevance is the normal developmental biology of the mammalian lung.
Variation in adult lung function, and lung conditions, can result from changes in our lungs during embryogenesis or infancy. It has become apparent that many developmental abnormalities of the lung cannot be corrected postnatally. These can result from premature birth, genetic variation, or antenatal conditions such as intrauterine infection (Baraldi and Filippone, 2007; Shi et al., 2007). Longitudinal studies of individuals born prematurely, with or without acute lung disease at birth, show that these people generally have decreased lung function and/or more respiratory problems throughout adolescence (Vrijlandt et al., 2006; Narang et al., 2008; Wong et al., 2008). Furthermore, it is often suggested that these individuals will be at increased risk of developing diseases associated with the normal age-related decline in lung function, such as chronic obstructive pulmonary disease (COPD; Maciewicz et al., 2009). Even in normal healthy individuals there is variation in lung function. Meta-analysis of genome-wide studies (16 studies totaling >40,000 individuals) identified loci associated with natural variations in lung function of healthy volunteers (Hancock et al., 2010; Repapi et al., 2010). These loci included HHIP and PTCH1 which are both required for lung development in mice (Chuang et al., 2003). A better understanding of embryonic lung development will improve our knowledge of these developmental conditions and may eventually lead to additional therapeutic options.
The adult lung consists of a pair of primary bronchi that branch from the trachea and give rise to progressively finer branches of the conducting airways (the bronchioles) which open to the ductal alveolar region where gas exchange occurs (Fig. 1A). The different regions of the lung contain specific specialized epithelial and mesenchymal cells. In mice, the primary bronchi are lined with a columnar epithelium consisting of basal, Clara (secretory), ciliated, and a small number of neuroendocrine (NE) cells. The mesenchyme of these airways has evenly spaced C-ring cartilage on the ventral side and smooth muscle (parabronchial smooth muscle) on the dorsal side (Fig. 1B). The bronchioles are lined by a similar columnar epithelium consisting of Clara and ciliated cells and clusters of NE cells known as neuroendocrine bodies, or NEBs (Fig. 1C,D). The epithelial composition of the bronchioles, particularly the ratio of Clara:ciliated cells, changes stereotypically along the proximal–distal axis; the smaller airways having an increased proportion of Clara cells (Toskala et al., 2005). The bronchioles are completely circled by parabronchial smooth muscle and do not have cartilage (Fig. 1F). Running parallel with the branched network of conducting airways are similar networks of blood vessels (arteries and veins, composed of endothelium, vascular smooth muscle, and pericytes), nerves (sympathetic and parasymphathetic neurons innervating both the airways and vasculature), and lymphatic tissue. The terminal branch of the bronchioles opens directly into the alveolar ducts where gas exchange occurs. The alveolar epithelium is squamous and consists of surfactant-secreting type 2 cells and highly elongated type 1 cells (Fig. 1E–G). The type 1 cells line the entire alveolar epithelial surface and run adjacent to capillaries in the underlying mesenchyme, providing a minimal barrier for gas exchange. The mesenchyme of the alveoli also contains alveolar smooth muscle cells and stromal fibroblasts, including lipofibroblasts and myofibroblasts, although these cell types may be interconvertible (Kapanci et al., 1974; Lindahl et al., 1997; McGowan et al., 2008; Perl and Gale, 2009). In all regions of the lung, it is difficult to distinguish the different mesenchymal cells, either morphologically or with molecular markers, and there may be additional undescribed subtypes. The entire lung is surrounded by a layer of squamous epithelium, the mesothelium or pleura. This provides a lubricated surface for movement of the organ.
An Overview of Lung Development
When and where do all these different cell types originate? The lung arises from the ventral foregut at embryonic day (E) 9.5 in the mouse as two primary buds. The lung buds consist of three cell layers: the inner epithelium (endodermal origin), which is surrounded by loosely packed mesenchyme and a thin outer mesothelial (squamous epithelial) layer. Lung morphogenesis has been the subject of several excellent recent reviews (Cardoso and Lu, 2006; Morrisey and Hogan, 2010) and is described briefly here. During embryonic development the epithelium of the lung bud undergoes branching morphogenesis to elaborate the structure of the bronchioles and alveoli (Fig. 2A–D). The inner, endodermally derived, epithelium of the lung buds consists of a blind-ended hollow tube. Initial branches are mostly side-branches, which form by budding along the length of the tube, this type of budding has been termed domain branching (Metzger et al., 2008). The branches elongate and subsequently bud again, this time by bifurcation of the distal-most epithelial tip, termed planar, or orthogonal, bifurcation depending on the orientation of the two new branches (Metzger et al., 2008). Metzger and Krasnow have proposed a conceptual framework for lung branching that includes genetic subroutines that control timing, orientation, and birfurcation of the branches. In this model, the branch timing subroutine would be the program that regulates overall growth of the lung and also controls progenitor cell proliferation. However, these ideas have yet to be specifically tested. The branching epithelium initially generates the future bronchioles (∼E9.5–E16.5, embryonic and pseudoglandular stages of lung development). Between ∼E16.5 and E17.5 (canalicular stage), the distal epithelium continues to branch, although budding morphology and frequency changes, and it now gives rise to a framework for future alveolar development, the terminal sacs (Fig. 2, compare A,B, with C,D). These sacs increase in number from E17.5 to postnatal day (P) 5 (terminal sac stage). During the alveolar phases (P5–P30), the terminal sacs are remodeled into mature alveoli and alveolar ducts. Alveolarization involves the generation of septae, which initially form as ridges along the terminal sac walls, and microvascular remodeling. The trachea forms from the ventral side of the foregut just caudal to the lung buds (Que et al., 2006). The trachea and lung both arise from an Nkx2-1+ region of the foregut and presumably from a common progenitor population (reviewed in Cardoso and Lu, 2006). However, the trachea does not undergo branching morphogenesis, rather it separates from the future esophagus by means of a process of septation (Ioannides et al., 2010). Indeed after the initial specification of the Nkx2-1+ anterior foregut domain, the lung and trachea have somewhat different patterns of gene expression. Specific tracheal progenitor cell populations are not discussed in this article.
Mammalian lung development has largely been studied by analysis of gene-targeted mice. These studies have identified genes and cellular mechanisms that regulate progenitor cell fate and overall lung morphogenesis. More recently, various methods of transient and long-term lineage-labeling have been used to identify specific embryonic lung progenitor cell populations. These studies have successfully identified a multipotent epithelial progenitor cell, parabronchial smooth muscle progenitors, and a subset of vascular smooth muscle progenitors. They provide a framework for analyzing the genetic regulation of individual populations of lung progenitor cells. Other lung embryonic progenitor cell populations are less well characterized and lineage-tracing studies have either not been performed, or are just beginning. Here I consider the evidence for the existence of various embryonic lung progenitor cell populations in the branching lung and their molecular regulation. This synthesis of the current literature highlights hypotheses to be tested and additional areas where further research is urgently needed. I also discuss the relationship between lung progenitor cell proliferation and branching morphogenesis, and the embryonic and adult lung progenitor populations.
EPITHELIAL PROGENITOR CELL POPULATIONS
Multipotent Epithelial Progenitors
Embryonic progenitor cells can self-renew during development and give rise to one or more cell types. They do not necessarily persist in the postnatal animal. The best characterized progenitor population in the embryonic lung is an epithelial progenitor cell located at the distal tip of the endodermally derived branching epithelium. The work of many laboratories has suggested that these cells are a multipotent epithelial progenitor population (reviewed in Cardoso and Lu, 2006). Compared with the epithelial stalk cells, the distal tip cells have a unique pattern of gene expression including high levels of Bmp4, Shh, Nmyc, Sox9, Id2, and Etv4/5 (reviewed in Rawlins, 2008), faster cell cycle kinetics (Okubo et al., 2005), and a subtly different morphology (Liu et al., 2004). Moreover, lung epithelial-specific deletion, or overexpression, of Nmyc results in phenotypes consistent with a premature loss, or overproliferation, of epithelial progenitor cells (Okubo et al., 2005).
To directly test the hypothesis that the distal epithelial tip cells are a multipotent epithelial progenitor population, an Id2-CreER2T knock-in mouse line was generated and used to lineage-label the distal lung epithelium at different stages of development. When lineage-labeled during the pseudoglandular stage, the distal Id2+ epithelial cells both self-renewed during development and produced descendents which differentiated into all of the identifiable bronchiolar and alveolar epithelial cell types: Clara, ciliated, neuroendocrine, type 2, type 1, including the adult progenitor cells (Figs. 3A,C, 5; Rawlins et al., 2009a). These data definitively show that the Id2+ distal epithelial cells are a multipotent epithelial progenitor population. To test the hypothesis that individual Id2+ distal tip cells are multipotent, single epithelial cells were lineage-labeled at E11.5 by exposing pregnant dams to a lower dose of tamoxifen. In >80% of the clones examined, the descendents of these single labeled Id2+ cells differentiated into both bronchiolar and alveolar cell types, strongly supporting the hypothesis that they are multipotent (Rawlins et al., 2009a). (However, NE cell differentiation was not specifically tested in the single cell lineage-tracing experiments.) The remaining ∼20% of single labeled cells gave rise to small (up to 5 cells) bronchiolar, or alveolar-restricted clones. Further work is needed to determine if this 20% of more lineage-restricted clones actually represent lineage-restricted Id2+ progenitor cells. Or, alternatively, if the Id2+ distal epithelial cells are a homogenous population, but by chance some of the labeled cells did not divide many times in the single cell lineage-tracing experiments. Id2 expression is maintained at the lung distal epithelial tip throughout embryogenesis. When the Id2+ distal tip cells were lineage-labeled during the canalicular stage, their descendents differentiated as type 1 or type 2 alveolar cells (Fig. 3B). Taken together, these data suggest a model in which the Id2+ distal epithelial cells are multipotent progenitors for lung embryonic development. They self-renew throughout embryogenesis, initially generate bronchiolar daughters, and then alveolar daughters (Figs. 3C, 5). A similar distal epithelial multipotent progenitor population has been identified in the endodermally derived pancreas (Zhou et al., 2007). Many questions remain to be addressed. For example, how do the multipotent lung progenitor cells switch from making bronchiolar to alveolar descendents? The mechanism must be lung-intrinsic because isolated E11.5 lungs will branch in culture and generate both bronchiolar and alveolar cell types. Transplantation experiments will be required to determine if the epithelial progenitors themselves undergo an intrinsic change in competence, or if signaling from the surrounding mesenchyme determines daughter fate. These studies may benefit from comparison with the developing cortex and retina where transplant experiments have shown that progenitors undergo cell intrinsic changes in competence to produce a sequence of different daughters (Livesey and Cepko, 2001). One hypothesis to explain how this is achieved in the nervous system is a link between the time of cell cycle exit and fate determination, perhaps regulated by cyclin kinase inhibitors (Dyer and Cepko, 2000, 2001).
Lineage-restricted Epithelial Progenitors
As cells leave the Id2+ pool they continue to proliferate. One model is that as cells exit the distal tip they become more-restricted bronchiolar, or alveolar, progenitors (Fig. 3D). Directly testing this hypothesis will require a better understanding of the molecular mechanisms which are responsible for cell fate specification in the developing lung. Differentiation of the airway epithelium has been observed to move down the airways from the proximal to the more distal regions (McDowell et al., 1994; Toskala et al., 2005). Molecular markers for NE cell fate specification (for example, the transcription factor Ascl1) and Clara and ciliated cell differentiation (Secretoglobin1a1 and the transcription factor Foxj1) are first observed at E12.5 (NE cells) and E14.5, or 15.5 (ciliated and Clara cells), always at least one airway generation from the multipotent progenitors in the distal tip (Rawlins et al., 2007; McGovern et al., 2010). These results suggest that cells are not being assigned to specific fates as soon as they leave the multipotent progenitor pool, consistent with the presence of more restricted progenitors. The reported effects of loss or gain of Notch signaling in the developing lung also support the restricted progenitor hypothesis. Ectopic expression of the active Notch intracellular domain throughout the developing airways, resulted in increased numbers of secretory (mucous-producing) cells and a decrease in ciliated cell specification (Guseh et al., 2009). Conversely, when Notch signaling was ablated throughout the entire lung epithelium by conditional deletion of Pofut1 (which attaches O-fucose to the EGF repeats of Notch), or Rbpjk (the Notch transcriptional effector), the bronchiolar epithelium largely differentiated as ciliated cells at the expense of the secretory (Clara) cells (Tsao et al., 2009; Morimoto et al., 2010). In the loss-of-function experiments, the multipotent progenitors were unaffected, alveolar development continued normally, and the mice were born with normal-sized lungs. These data are consistent with Notch signaling occurring between undifferentiated bronchiolar progenitor cells to restrict ciliated cell fate. The existence of these bronchiolar-restricted epithelial progenitors, and a similar putative alveolar progenitor, needs to be tested directly.
Lineage-labeling the Id2+ distal epithelial cells showed that their descendents can differentiate as NE cells. However, no completely labeled NE cell clusters (NEBs) were identified (Rawlins et al., 2009a). By contrast, when a human SftpC (Surfactant Protein C) promoter fragment was used for doxycycline-dependent, Cre-mediated lineage-labeling throughout the developing lung epithelium, no labeled NE cells were identified (Perl et al., 2002). These results could be partially reconciled if a subset of the NE cells, possibly the founding cells of each cluster, are neural crest-derived. This hypothesis needs to be formally tested.
MESENCHYMAL PROGENITOR CELL POPULATIONS
The mesenchymal cells of the lung are not as well characterized as the epithelial cells. There is still no general consensus around the number of different differentiated mesenchymal cell types in the adult lung, or the molecular markers which can be used to distinguish between them. Indeed, a recent study of adult human lung mesenchymal cells suggests that no single molecular marker yet identified is specific for an individual mesenchymal cell type (Singh et al., 2010). This makes identification of the mesenchymal progenitor cells difficult, although progress has been made. The initial lung buds are composed of endodermal epithelium surrounded by mesenchyme derived from the splanchnic mesoderm, which itself comes from the lateral plate. Two different populations of mesenchymal cells can be identified morphologically at E12.5 (White et al., 2006). The subepithelial mesenchyme is tightly packed and oriented around the epithelial tubes, whereas the submesothelial mesenchyme is more loosely packed and not oriented. These morphological differences are subtle (Fig. 4A). However, they do seem to coincide with some gene expression differences. For example, Fgf10 is specifically expressed in the submesothelial mesenchyme (Bellusci et al., 1997b), whereas ptch1 (patched1) is expressed throughout the distal lung mesenchyme (Weaver et al., 2003).
Bronchiolar Smooth Muscle Progenitors
Fibroblast growth factor-10 (FGF10) is highly expressed in the submesothelial mesenchyme, several cell diameters from the branching tip of the lung epithelium (Bellusci et al., 1997b) and is required for branching morphogenesis (for example, Sekine et al., 1999; Weaver et al., 2000). The distal FGF10+ mesenchymal cells were followed for a short time by taking advantage of the perdurance of the β-galactosidase protein in an Fgf10-LacZ strain. In this strain, the β-galactosidase protein remains in the cells after Fgf10 transcription has been turned off. Although formal lineage-tracing experiments remain to be done, the descendents of the distal FGF10+ mesenchymal cells have been partially characterized. In these experiments, the FGF10+, SMA− (α-smooth muscle actin) distal mesenchyme cells gave rise to FGF10−, SMA+ parabronchial smooth muscle (Mailleux et al., 2005). The authors confirmed the result by grafting Fgf10-LacZ mesenchyme onto cultured wild-type lungs and vice versa. These data conclusively show that the distal tip mesenchyme includes progenitors for the parabronchial smooth muscle. It is not yet clear if all of the FGF10+ distal mesenchymal cells become parabronchial smooth muscle or, if they can also acquire other mesenchymal fates. Similarly, is all of the bronchiolar smooth muscle derived from this cell population? And how do the cells exit the submesothelial mesenchyme and move to a subepithelial position? Do similar submesothelial mesenchymal progenitor cells give rise to the alveolar smooth muscle at later stages of lung development?
Using retrovirus-mediated lineage-tracing in E11.5 lung/tracheal explants, it was shown that labeled proximal mesenchymal cells, just adjacent to the trachea and mainstem bronchi, gave rise to descendents which were dispersed throughout the lung mesenchyme after 11 days culture. These included cells with the morphological appearance of parabronchiolar smooth muscle (Shan et al., 2008). Laser capture microdissection and sequencing of the retroviral genome showed clonal origins for some of these widespread cells. This result suggests that there might be a second, more proximal, progenitor for the bronchiolar smooth muscle (and possibly other mesenchymal cell types). The existence of two populations of parabronchial smooth muscle progenitors (proximal bronchi region, and distal tip region) is also supported by lineage-tracing using the Id2-CreER mouse strain (E.L.R., unpublished data, Rawlins et al., 2009a). Id2-CreER drives recombination in small numbers of both proximal and distal mesenchymal cells at E11.5, these subsequently differentiate as smooth muscle (Fig. 4B,C). Id2 is not expressed at high levels in the developing lung mesenchyme and the Id2-CreER strain is not useful for further characterization of mesenchymal progenitor populations. It will be important to better-characterize the molecular profile of different mesenchymal cell types throughout lung development to allow for an improved analysis of the cellular and molecular regulation of the different mesenchymal progenitors.
Vascular Smooth Muscle Progenitors
In the heart and gut, lineage-tracing of the outer mesothelial layer, using mesothelial-specific Wt1-Cre mouse strains, has shown that the mesothelium is the source of ∼80% of the vascular smooth muscle cells during embryonic development (Wilm et al., 2005; Zhou et al., 2008). Similar lineage-tracing experiments in the lung demonstrated that the mesothelium contributes ∼30% of the vascular smooth muscle in both arteries and veins, but that it makes no contribution to the parabronchial smooth muscle (Que et al., 2008; Morimoto et al., 2010; Fig. 5). Where do the remaining vascular smooth muscle cells originate? Lineage-tracing of the endothelium using Tie1-Cre; Rosa26R-YFP showed that overall 30% of the vascular smooth muscle at E16.5 was derived from the endothelium. The proportion of endothelial-derived vascular smooth muscle cells varied along the proximal–distal axis of the lung. Most of the vascular smooth muscle in the proximal vessels, but almost none in the distal vessels, was of endothelial origin (Morimoto et al., 2010). It would be desirable to repeat this experiment with an inducible Cre, and also to confirm that the Tie1-Cre transgene was never expressed in the vascular smooth muscle cells. Nevertheless, there is also indirect evidence for an endothelial origin of some lung vascular smooth muscle cells (reviewed in Arciniegas et al., 2007). Together the mesothelial and endothelial progenitor populations account for only approximately 60% of the lung vascular smooth muscle. Where does the remaining 40% originate? Does it come from another progenitor located within the mesenchyme? In support of a separate mesenchymal origin, morphological studies of human embryonic lungs have concluded that both the endothelium and mesenchyme contain vascular smooth muscle progenitors (Hall et al., 2000). However, this needs to be confirmed in an animal model using a lineage-labeling technique. Other important questions include, does the origin of the vessel (angiogenesis vs. vasculogenesis), as well as its location, affect the origin of its smooth muscle? Do the different vascular smooth muscle progenitor populations share aspects of their molecular regulation? How much flexibility is there between the contributions made by the different vascular smooth muscle progenitor types? Are these multiple vascular smooth muscle progenitor populations relevant to lung vascular maintenance or repair during postnatal life, particularly to the development of lung vascular conditions such as hypertension?
Cellular Origin of the Vasculature
Two opposing models of lung vasculature development exist: angiogenesis (sprouting from existing vessels) and vasculogenesis (de novo formation of blood vessels within the lung mesenchyme from angioblasts). Recent descriptive studies, in which mouse embryos were processed for histology with an intact blood circulation, support the hypothesis that at least some of the blood vessels form by angiogenesis. At the earliest stages of branching morphogenesis (E9.5–E10) simple vessels composed of Tie2-LacZ+ and Pecam+ endothelial cells, but no vascular smooth muscle, were observed to contain primitive erythrocytes, suggesting that they were already connected to the overall circulation and had likely arisen by angiogenesis (Parera et al., 2005). Moreover, the budding endodermal epithelial tip of the developing lung was surrounded by a capillary network (Parera et al., 2005). Subsequent studies have confirmed and extended these results. A fluorescent lectin-tracer was used to demonstrate that the embryonic lung vasculature is connected to the circulation during the pseudoglandular stage, and mitotic figures were observed at the tips of the endothelium only, indicative of angiogenesis (Schwarz et al., 2009).
Multiple reports have provided evidence for the occurrence of vasculogenesis within the pseudoglandular stage lung mesenchyme. For example, using a Flk1-LacZ reporter mouse, vascular progenitor cells were detected from E10.5 onward, arranged in vessels without a lumen (Schachtner et al., 2000; Yamamoto et al., 2007). It is likely that both vasculogenesis and angiogenesis occur during lung development. Indeed, this has been suggested by careful anatomical studies of both mouse and human fetal lungs (deMello et al., 1997; deMello and Reid, 2000). Further molecular genetic studies will be needed to clarify the relative contributions of the two processes at different stages of development and in different regions of the lung. Vascular development continues as the lung grows and the vasculature is extensively remodeled during alveolarization, although no new vessels are formed at this stage (reviewed in Stenmark and Abman, 2005).
Other Mesenchymal Progenitor Populations
The Wt1-Cre lineage-labeling experiments showed that other lung mesenchymal cells (in addition to a subset of the vascular smooth muscle) were also derived from the mesothelium. These were not specifically identified and could have been alveolar myofibroblasts, microvascular pericytes, or even endothelial cells (Que et al., 2008). This finding warrants further investigation, preferably with an inducible Cre line, to identify all of the mesothelial-derived cells and the developmental stage at which they are produced; moreover, to determine how many different progenitor populations the embryonic lung mesothelium contains and whether any of these also function in the adult.
Other mesenchymal cell types (various fibroblasts, pericytes, alveolar smooth muscle) are thought to be produced by progenitors within the splanchnic mesoderm which surrounds the lung buds. It is not known how homogenous the cells within the mesoderm are at the early stages of lung development. For example, are there multiple populations of lineage-restricted mesenchymal progenitor cells as the lung buds initiate? Or, is there a multipotent mesenchymal progenitor at this stage? Answering these questions will require new genetic tools and an improved knowledge of gene expression profiles within the lung mesenchyme.
It is now incontrovertible that epithelial mesothelial cells delaminate and undergo EMT (epithelial–mesenchymal transitions) during lung development (Que et al., 2008; Morimoto et al., 2010). By contrast, all of the available evidence suggests that the endodermally derived epithelium does not undergo EMT during normal development. For example, in lineage-tracing experiments in which the entire lung bud endoderm was labeled, using a human Surfactant protein C (SftpC) promoter fragment to drive Cre expression, the lineage label was never observed in the mesenchyme (Perl et al., 2002).
GENETIC MECHANISMS OF SELF-RENEWAL IN EMBRYONIC LUNG PROGENITOR POPULATIONS
Lung morphogenesis is controlled by reciprocal signaling between all cell layers: the epithelium, mesenchyme (including developing blood vessels), and the mesothelium (reviewed in, Cardoso and Lu, 2006; Morrisey and Hogan, 2010). Multiple interacting signaling pathways and transcription factors regulate the proliferation of lung progenitor cells (Fig. 6). The direct targets of most of the genes implicated in embryonic lung progenitor proliferation have not been identified. This makes it hard to determine if the effect observed, for example proliferation, is directly downstream of the gene/pathway under investigation.
Epithelial Progenitor Regulation
The distal lung epithelial progenitors express a repertoire of transcription factors. Nmyc is both necessary and sufficient for the division of multipotent lung epithelial progenitors and likely promotes self-renewal (Okubo et al., 2005). Similarly, the forkhead genes Foxp1 and Foxp2 are enriched in the multipotent epithelial progenitors and Foxp2−/−; Foxp1+/− mutants have smaller lungs with decreased levels of proliferation and Nmyc expression, but normal proximal–distal patterning (Shu et al., 2007). Sox9 is a useful marker for the multipotent epithelial progenitors, but lung epithelial specific deletion did not result in a phenotype (Perl et al., 2005).
Fgfr signaling mediates branching morphogenesis of the lung. In vitro FGF10 is both a chemoattractant and a mitogen for the lung epithelium during branching (Bellusci et al., 1997b; Park et al., 1998). Fgf10 hypomorphs have decreased levels of epithelial proliferation during the pseudoglandular stage and this is associated with decreased levels of distal epithelial phosphorylated-ERK MAPK and decreased expression of a canonical Wnt pathway reporter (Ramasamy et al., 2007). By contrast, overexpression of FGF10 in the developing lung apparently arrests epithelial progenitors in an undifferentiated state (Nyeng et al., 2008). These results are consistent with a role for mesenchymal FGF10 in promoting self-renewal, and inhibiting differentiation, of the multipotent lung epithelial progenitors. The transcription factors Etv4/5 and Elf5 mediate some of the effects of Fgfr signaling in these cells (Liu et al., 2003; Metzger et al., 2007).
Wnt7b is expressed in the distal tips of the developing lung epithelium and acts by means of a canonical signaling pathway to promote both epithelial and mesenchymal proliferation (Rajagopal et al., 2008). Wnt7b null embryos have a normal overall body size, but severely hypoplastic lungs. Lung progenitor proliferation is significantly decreased, particularly in the distal tip epithelium and subepithelial mesenchyme, but the number of Sox9+ multipotent progenitor cells per tip is normal and differentiation proceeds at the normal rate (Rajagopal et al., 2008). The effects of Wnt7b in the epithelium are possibly mediated through loss of distal epithelial Bmp4 and Id2 expression (Rajagopal et al., 2008). However, Id2 null lungs have normal rates of BrdU incorporation during the pseudoglandular stage (E.L.R., unpublished data). Wnt2 is expressed in the mesenchyme surrounding the epithelial buds and also promotes lung progenitor proliferation. Similar to Wnt7b null animals, Wnt2 null mice have smaller lungs with both epithelial and mesenchymal proliferation defects, but normal overall patterning (Goss et al., 2009). Wnt5a is highly expressed in the distal lung epithelium and surrounding mesenchyme. The null phenotype suggests that it may act antagonistically to Wnt7b and Wnt2 and repress epithelial and mesenchymal progenitor proliferation (Li et al., 2002). Wnt signaling has also been implicated in proximal–distal epithelial patterning and differentiation during lung development and the roles of other Wnt ligands remain obscure (Shu et al., 2005; Zhang et al., 2008). Interestingly, Wnt signaling is also absolutely required for the initial specification of the respiratory epithelial lineage from the foregut endoderm (Goss et al., 2009; Harris-Johnson et al., 2009; Chen et al., 2010).
The role of Bmp signaling in lung epithelial progenitor regulation has been difficult to define. In vitro and overexpression studies gave conflicting results, suggesting that the responses of the different progenitor populations to Bmp signaling are highly dose-specific (for example, Bellusci et al., 1996; Weaver et al., 2000). Bmp4 is expressed in the distal epithelial tip cells, as well as the mesenchyme surrounding the developing bronchioles, whereas Bmpr1a is found throughout the epithelium and mesenchyme (Weaver et al., 2000, 2003). Lung epithelial-specific deletion of either Bmpr1a or Bmp4 results in hypoplastic lungs with significantly reduced rates of epithelial proliferation and a decrease in the levels of Shh and Nmyc (Eblaghie et al., 2006). Analysis of the phenotype suggested that Bmp signaling may be important for the self-renewal of the distal epithelial progenitor cells, but that it also affects cell morphology and survival. The specific roles of other Bmp signaling components have not been determined. Tgfβ signaling has been implicated in lung epithelial progenitor proliferation in vitro, perhaps acting antagonistically to FGF10 in the multipotent progenitors through effects on Pten transcription (Xing et al., 2008). The lung epithelial-specific disruption of individual transforming growth factor-beta (TGFβ) superfamily ligands, receptors and signaling components is in process, but specific effects on progenitor populations have yet to be identified (for example, Chen et al., 2005). Deletion of Pten throughout the developing lung epithelium led to epithelial hyperplasia and changes in cell fate allocation (Tiozzo et al., 2009). However, Pten expression is widespread in the lung and it is not clear in which specific embryonic progenitor populations its function is required.
The phenotypes of transgenic overexpression or targeted deletion of the miR17-92 microRNA cluster are consistent with a normal role in promoting self-renewal, and inhibiting differentiation, of the multipotent lung epithelial progenitors (Lu et al., 2007; Ventura et al., 2008). Direct targets of the miR17 family in the lung include Rbl2, Mapk14, and Stat3 mRNAs (Lu et al., 2007; Carraro et al., 2009).
Mesenchymal Progenitor Regulation
Specific transcription factors expressed in the different mesenchymal progenitor cell populations have not been identified, even for the well-established parabronchial smooth muscle progenitors. Nevertheless, tremendous progress has been made in identifying molecules which regulate overall lung embryonic mesenchymal proliferation, rather than specific progenitor populations. FGF9 is highly expressed in both the mesothelium and the distal endoderm during the pseudoglandular stage of lung development. Fgf9 null lungs have a smaller mesenchymal compartment than wild-type and, as a consequence, branching morphogenesis is reduced (Colvin et al., 2001). This initial observation has lead to the elucidation of a complex signaling loop, involving the FGF, Shh, and Wnt pathways, which regulates the proliferation and differentiation of mesenchymal progenitor cell populations in the pseudoglandular stage lung. The cross-talk between these pathways is in the process of being extensively characterized. However, the detailed consequences of signaling for specific mesenchymal progenitor populations continues to lag behind. It is clear that the different progenitor populations respond differently to the same signaling molecules, even though they are developmentally related and in close proximity. For example, an increase in the levels of FGF9 protein inhibits the development of bronchiolar smooth muscle, but not vascular smooth muscle (Weaver et al., 2003; White et al., 2006). The cell biology underlying these varying responses is likely to include, first, the different developmental histories of the progenitor populations (different transcription factors, signaling pathway components, epigenetic marks). Second, slight differences in progenitor location within the organ resulting in varying levels of signaling, or a different combinatorial pattern of signals.
Genetic evidence suggests that FGF9 is itself a mitogen for the subepithelial mesenchyme where it signals through Fgfr1 and Fgfr2. The submesothelial mesenchyme was absent in the Fgf9 null lungs and expanded when Fgf9 was overexpressed. Moreover, adding FGF9 to cultured lungs caused a preferential increase in submesothelial mesenchymal proliferation (White et al., 2006). By contrast, the effects on proliferation of the subepithelial mesenchyme in these conditions were much weaker and most likely mediated by Shh. Shh expression was decreased in the Fgf9 null lungs. Moreover, simultaneously adding FGF9 and the Shh pathway inhibitor cyclopamine to cultured lungs blocked the effects of FGF9 on the subepithelial mesenchyme (White et al., 2006). Consistent with a role for Shh in subepithelial mesenchymal proliferation, it is highly expressed in the distal epithelial endoderm budding tips and its receptor, Patched1 (Ptch1), is highly expressed throughout the distal mesenchyme (Weaver et al., 2003). In vitro and in vivo overexpression experiments also support the idea that Shh functions as a mitogen for the mesenchymal progenitors (Bellusci et al., 1997a; Weaver et al., 2003). There is evidence that TGFβ signaling negatively regulates the expression of Shh signaling components in the lung mesenchyme (Li et al., 2008).
Subsequent studies identified an FGF9–Wnt regulatory loop that also promotes mesenchymal proliferation. FGF9 in the mesothelium induces Wnt2a expression in the submesothelial mesenchyme. Blocking Wnt signaling throughout the mesenchyme and mesothelium by deleting a floxed β-catenin allele with Dermo1-Cre resulted in lower levels of proliferation, decreased mesenchymal levels of Fgfr1 and 2, and was sufficient to prevent the mesenchyme from responding to exogenous FGF9 (De Langhe et al., 2008; Yin et al., 2008). The model suggested by Yin et al. is that Wnt2 (downstream of FGF9 and previously known as Wnt2a) functions to mediate submesothelial mesenchyme proliferation. Whereas, Wnt7b (which is expressed independently of FGF9) in the distal epithelium acts on the subepithelial mesenchyme (Shu et al., 2002; Rajagopal et al., 2008; Yin et al., 2008). Loss of function studies have also shown that Wnt7b signaling from the epithelium by means of the canonical β-catenin–mediated pathway is required for smooth muscle precursor cell proliferation (Cohen et al., 2009).
FGF10+ cells in the distal mesenchyme at the pseudoglandular stage are parabronchial smooth muscle progenitors and moreover, Fgf10 hypomorphic lungs have a smaller number of smooth muscle cells (Mailleux et al., 2005). However, the hypomorphs do not display changes in proliferation or apoptosis of the mesenchymal progenitors and FGF10 elicited no direct response (neither activation of Erk or Akt) on primary lung mesenchymal cells (Mailleux et al., 2005). These data suggest that, unlike the epithelial progenitors, the function of FGF10 is not to promote progenitor self-renewal, rather it is required for smooth muscle specification. Similarly, no requirement for canonical Notch signaling in proliferation of the mesenchymal progenitor populations has been identified (Morimoto et al., 2010). The roles of other pathways, including Egfr, Bmp, cytokine, in the different populations of lung mesenchymal progenitors remain to be tested.
Vascular Progenitor Regulation
In addition to their roles in mesenchymal proliferation and differentiation, FGF9 and Shh also affect vascular development in the lung. Fgf9 null lungs have a smaller capillary network around the budding epithelial tips at E11.5. This could be partially phenocopied by deletion of Fgfr1 and Fgfr2 throughout the developing mesenchyme and mesothelium, but not by endothelial-specific Fgfr deletion (White et al., 2007). The effects of FGF9 on the developing vasculature are therefore indirect and the authors show that it most likely regulates mesenchymal Vegfa expression. Loss of Shh in the lung epithelium resulted in a simplification of the lung vascular network (Miller et al., 2004). Similarly, lung-specific deletion of the Shh receptor, Smoothened, also resulted in a decreased vascular network (White et al., 2007). However, Shh and FGF9 were unable to compensate for one another in vitro, suggesting that they act independently in the embryonic lung mesenchyme to regulate vascular development (White et al., 2007).
LUNG PROGENITORS AND BRANCHING MORPHOGENESIS
Lung development involves the coordinated proliferation, migration, and differentiation of epithelial, mesenchymal, and endothelial cell types during branching morphogenesis. The multipotent epithelial progenitor cells are located in the distal budding tip and these cells also mediate epithelial branching. The detailed cell biology of lung branching is poorly characterized. However, it is clear that during branching the distal epithelial cells migrate, alter their shape, make changes in cell–substratum and cell–cell adhesion, and possibly also undergo oriented cell division and local cell rearrangements (Andrew and Ewald, 2010). Many of the signals that regulate lung embryonic progenitor proliferation also mediate branching. It is tempting to speculate that a master-regulator controls both processes coordinately. One candidate for a master-regulator is FGF10. In Fgf10, or its receptor Fgfr2-IIIb, null animals the lungs do not branch (Sekine et al., 1999; De Moerlooze et al., 2000). Moreover, as we have seen, the Fgf10 hypomorphic lungs have defects in epithelial progenitor proliferation and mesenchymal, including vascular, differentiation (Ramasamy et al., 2007). Testing the hypothesis that FGF10 is a master-regulator for lung development will require both the targeted deletion of its receptor in specific progenitor cell populations and a full analysis of the gene regulatory network controlling progenitor proliferation and morphogenesis. A lung epithelial-specific deletion of Fgfr2 has been performed and the results are consistent with roles in branching, survival, and proliferation (Abler et al., 2009). As yet, there has been no attempt to build a gene regulatory network for lung morphogenesis. Gene Regulatory Networks are systems-level models of the molecular circuitry governing cell fate decisions. When complete, these include all of the transcription factors, their cis-binding sites and the intercellular signaling molecules required for these decisions (Stathopoulos and Levine, 2005; Oliveri et al., 2008). The establishment of a gene regulatory network for lung progenitor proliferation and branching morphogenesis will be the work of many labs over the course of many years. Nevertheless, the tools to build it are now available and the process of building the network will generate testable predictions and provide new insights into these highly complex cellular events (Materna and Oliveri, 2008).
An alternative hypothesis is that a signaling network centered on the distal budding endoderm and surrounding mesenchyme/mesothelium controls both progenitor proliferation and lung morphogenesis. If this is the case, one prediction is that the as yet unidentified mesenchymal progenitor populations will also be located in this area. Indeed, this may be the most-likely location of any putative multipotent mesenchymal progenitor.
One relatively unexplored area of lung development is the effect of physical factors, such as mechanical forces and pressure, on progenitor proliferation and branching morphogenesis. A recent study showed that increasing the internal pressure in mouse lungs branching in culture by sealing the trachea, increases both proliferation and branching rates and that these changes are dependent on FGF10 signaling (Unbekandt et al., 2008). It has previously been speculated that the epithelial dilation phenotype of Fgf9 null lungs could result from a change in physical constraint on the epithelium due to mesenchymal thinning in the mutant, but this has not been tested (Colvin et al., 2001). Moreover, these authors suggest that Fgf9 and Fgf10 expression are normally regulated through changes in pressure (Colvin et al., 2001; Unbekandt et al., 2008). Such mechanical forces may be an important upstream regulator of both progenitor proliferation and morphogenesis, but this area of lung development is in much need of further study.
EMBRYONIC AND ADULT LUNG PROGENITOR CELLS: DISTINCT LINEAGE-RELATED POPULATIONS
Can the study of embryonic lung progenitors contribute to our understanding of adult lung stem and progenitor cells? The identity of the adult stem and progenitors is still controversial. As in the embryo, the epithelial progenitor populations are much better characterized than their mesenchymal counterparts. Lineage-tracing studies have shown that the adult lung progenitors are descended from the Id2+ multipotent distal epithelial embryonic progenitors. When small numbers of Id2+ epithelial progenitors were lineage-labeled at E11.5, these formed patches (or clones) of labeled cells in either the bronchioles, or the alveoli, by postnatal age 8 months (Rawlins et al., 2009a). However, Id2+ cells themselves have not so far been detected in the steady-state, or injured, adult lung (Rawlins et al., 2009a). Together these data show that the embryonic multipotent epithelial progenitor cell is not retained in the adult lung, rather some of its descendents become the adult lung progenitors.
Which specific cells are the adult lung epithelial stem and progenitors? The most recent in vivo postnatal lineage-tracing experiments support the idea that each compartment of the adult lung (trachea, bronchioles, alveoli) is maintained by its own separate stem/progenitor population (Giangreco et al., 2009; Rawlins et al., 2009b; Rock et al., 2009). Of interest, adult epithelial cells have been identified that have the ability to generate both bronchiolar and alveolar daughters in vitro (Kim et al., 2005; McQualter et al., 2010). The prospective isolation of adult lung stem cells is still in its early stages, and there is so far little consensus around the methods for cell isolation and culture, making it very hard to directly compare findings from different labs (see also, Summer et al., 2007; Teisanu et al., 2009, 2010; McQualter et al., 2009; Rock et al., 2009). None of these prospectively isolated, putative adult stem cells has been reported to express any of the known embryonic multipotent epithelial cell markers. So, is there any relevance of the embryonic progenitors to the adult stages? In addition, how do we resolve the conflicting in vitro and in vivo adult results? One possibility is that the adult cells are not multipotent progenitors in vivo and the in vitro results are tissue culture artifacts. A second, more interesting, possibility is that a specific, small population of adult lung epithelial cells retains the ability for multi-lineage differentiation, but is normally quiescent in vivo.
Current evidence suggests that the embryonic and adult lung progenitors are distinct, different cell types. In addition, the molecular regulation of the adult lung progenitors may be different to that of their embryonic counterparts. Wnt signaling is required for embryonic progenitor proliferation (Rajagopal et al., 2008; Goss et al., 2009). However, specific deletion of β-catenin, which is necessary for canonical Wnt signaling, in a known adult lung progenitor population did not have any effect on airway epithelial growth, maintenance or repair (Zemke et al., 2009). This suggests that Wnt signaling is dispensable for regulating postnatal bronchiolar stem or progenitor cells. One possibility is that the molecular regulation of adult and embryonic lung epithelial progenitors is different, maybe because the adult progenitors are not located within a signaling centre. Alternatively, interpreting the roles of signaling pathways in adult progenitors may require more subtle cell-type specific manipulation of signaling levels. It would be very surprising if there were no common regulatory mechanisms between the embryonic and adult lung progenitors, particularly during lung repair when the adult progenitors take on some of the morphogenetic functions performed by the embryonic cells (for example, migration, cell shape changes, cell–substrate interaction changes, more rapid proliferation; see Kida et al., 2008). However, even if the regulation is different, the published lung embryonic progenitor experiments provide a framework for adult experimental design.
Much progress has been made in identifying lung embryonic progenitors. Nevertheless, many of these populations have only been superficially described, or their existence is inferred from mutant phenotypes. To aid in their characterization, we urgently need more markers for lung differentiated and progenitor cell types, especially for the mesenchyme. Cell lineage-specific transcription factors would be particularly useful. These reagents would allow the development of improved genetic tools for lineage-labeling, and manipulating, lung progenitor cells. Improved characterization of mutant phenotypes with respect to specific cell populations would also be possible. Other important areas of future research include analysis of the molecular regulation of the progenitors at the canalicular stage of development and the specific cell–cell interactions which regulate progenitor behavior.
Thanks to Gayan Balasooriya for technical assistance.