|Home | About | Journals | Submit | Contact Us | Français|
Developmental lung biology is a field that has the potential for significant human impact: lung disease at the extremes of age continues to cause major morbidity and mortality worldwide. Understanding how the lung develops holds the promise that investigators can use this knowledge to aid lung repair and regeneration. In the decade since the “molecular embryology” of the lung was first comprehensively reviewed, new challenges have emerged—and it is on these that we focus the current review. Firstly, there is a critical need to understand the progenitor cell biology of the lung in order to exploit the potential of stem cells for the treatment of lung disease. Secondly, the current familiar descriptions of lung morphogenesis governed by growth and transcription factors need to be elaborated upon with the reinclusion and reconsideration of other factors, such as mechanics, in lung growth. Thirdly, efforts to parse the finer detail of lung bud signaling may need to be combined with broader consideration of overarching mechanisms that may be therapeutically easier to target: in this arena, we advance the proposal that looking at the lung in general (and branching in particular) in terms of clocks may yield unexpected benefits.
The concept that lung organogenesis is instructed by coordinated mesenchymal-to-epithelial crosstalk originates in the classical recombination experiments of Alescio and Cassini (1962), in which replacing tracheal mesenchyme with mesenchyme from the lung periphery induced ectopic branching of tracheal epithelium in murine embryonic lung organ culture. This idea was extended in an early review by Warburton and Olver (1997) to include the coordination of genetic, epigenetic, and environmental factors in lung development, injury, and repair. Thereafter, a molecular basis of lung morphogenesis was attempted by Warburton et al. (2000). Over the last decade, significant progress has been made in this field as reviewed by Cardoso and Lu (2006), Maeda et al. (2007), and others. Nevertheless, the ultimate goal remains as stated by Warburton and Olver (1997), “to devise new rational and gene therapeutic approaches to ameliorate lung injury and augment lung repair … the ideal agent or agents would therefore mimic the instructive role of lung mesenchyme and would correctly induce the temporospatial pattern of lung-specific gene expression necessary to instruct lung regeneration.” To this overall strategy, we can now add (i) the modulation of lung mechanobiology to favor appropriate lung regeneration and (ii) the stimulation of endogenous stem/progenitor cells or supply of exogenous ones for lung regeneration. Therefore, the current review draws together three important strands of information on lung organogenesis as of April 2010: (i) molecular embryology of the lung, (ii) mechanobiology of the developing lung, and (iii) pulmonary stem/progenitor cell biology. Applying advances in these complementary areas of research to lung regeneration and correction of lung diseases remains the therapeutic goal of this field. With the recent human transplanation of a stem/progenitor cell-derived tissue-engineered major airway (Macchiarini et al., 2008), we can clearly see the potential of this field, while recognizing the many problems yet to be solved.
Before concentrating on the molecular biology, mechanobiology, and stem cell biology of the lung, a first step in regenerative strategies is to consider the developmental anatomy of the lung. From this, we can at least see what type of structures we need to generate.
A diagrammatic overview of lung morphogenesis is given in Fig. 3.1. Three lobes form on the right side and two lobes on the left side in human lung; in mice four lobes form on the right (cranial, medial, and caudal lobes, plus the accessory lobe) and one on the left. In contrast to humans, in the mouse, there are only 12 airway generations and alveolarization occurs entirely postnatally.
Histologically, lung development and maturation has been divided into four stages: pseudoglandular, canalicular, terminal saccular, and alveolar (Fig. 3.2).
During this, the earliest developmental stage, epithelial tubes lined with cuboidal epithelial cells undergo branching morphogenesis and resemble an exocrine gland (hence the nomenclature). However, this fluid-filled primitive respiratory tree structure is too immature to support efficient gas exchange.
The cranial part of the lung develops faster than the caudal part, resulting in partial overlap between this stage and the previous stage. During the canalicular stage, the respiratory tree is further expanded in diameter and length, accompanied by vascularization and angiogenesis along the airway. A massive increase in the number of capillaries occurs. The terminal bronchioles are then divided into respiratory bronchioles and alveolar ducts, and the airway epithelial cells are differentiated into peripheral squamous cells and proximal cuboidal cells.
There is substantial thinning of the interstitium during the terminal saccular stage. This results from apoptosis as well as ongoing differentiation of mesenchymal cells (Hashimoto et al., 2002; Lu et al., 2002). Additionally, at this stage, the alveolar epithelial cells (AECs) are more clearly differentiated into mature squamous type I pneumocytes and secretory rounded type II pneumocytes bearing lamellar bodies that contain surfactant. The capillaries also grow rapidly in the mesenchyme surrounding the saccules to form a complex network. In addition, the lymphatic network in lung tissue becomes well developed during this stage. The thick wall of these saccules, also called primary septae, comprises lining epithelial cells on both sides of a connective tissue core, within which there is a double parallel network of capillaries. Toward the end of this stage, the fetal lung can support air exchange in prematurely born human neonates. Maturation of surfactant synthesis and secretion is a key factor in determining whether the newborn lung can sustain gas exchange without collapsing.
Alveolarization is the last step of lung development. The majority of the gas exchange surface is formed during this stage.
Genome-wide expression profiling has measured developing human lung transcriptomes in pregnancies terminated between 7 and 22 weeks post conception (Kho et al., 2009). Within the 3,223 gene developing lung-characteristic subtranscriptome, transitions in gene expression correlated with some histological stages, as well as suggesting novel substages exist. For example, induction of surfactant gene expression identifies a “molecular transition” in the pseudoglandular phase.
Hence, the histological account of lung development is complimented by the molecular embryology that we consider in the next main section of the review.
Branching morphogenesis is a critical part of overall lung development and a crucial phenomenon in the development of several other organs. Understanding this key hurdle in lung regeneration strategies requires us to appreciate that despite the apparent and beautiful complexity of the lung, there are key simplicities that can help us in our task.
Fractal mathematics has revealed that relatively straightforward algorithms, when applied iteratively, could generate patterning of great complexity (Mandelbrot, 1982). Moreover, when the mathematical parameters were chosen appropriately, the approximation to living branched structures was striking. At first sight, intrapulmonary airway branching (distal to the primary bronchi) appears to become increasingly complicated as it proceeds distally and the number of individual branches increases into the millions. But, once the laryngotracheal complex and left–right laterality are established, distal airway branching is now thought to be driven by a relatively simple set of genetically encoded control routines. Echoing the fractal pattern formation achieved in mathematical models, these include the following: (i) a master branch generator routine, with three slave subroutines instructing a periodicity clock which times the appearance of subsequent branches, (ii) a rotational orientation subroutine which determines the orientation of the branches around the axis of the airway, and (iii) a branch tip division subroutine (Metzger et al., 2008; Warburton, 2008). Thus, branching morphogenesis of the bronchi in early mouse embryo lung can be parsed anatomically into three simple geometric forms, termed domain branching, planar bifurcation, and orthogonal bifurcation (Fig. 3.3). These basic forms are repeated iteratively to form different arrangements of branches. The planar or bottlebrush array describes the sequential proximal to distal emergence of secondary branches along the lateral axis of the primary bronchial airway. The bottlebrush mechanism is then reoriented around the branch axis to form a second row of branches at right angles to the first row. The terms planar array and rosette array describe the patterns formed by sequential bifurcation of the tips of secondary, tertiary, and subsequent buds at right angles to each other. Repetition of these simple branching modules, together with the hierarchical control and coupling of them, may therefore explain how the genome could possibly encode the highly complex yet stereotypic pattern of early bronchial branch formation, using a relatively simple toolbox of genetic modules. In a further illustration of how the mammalian lung uses simple routines and subroutines to construct itself, substantial homology has been identified between the genetic regulation of lung organogenesis and airway morphogenesis in Drosophila (Hacohen et al., 1998; Tefft et al., 1999). Despite the latter’s relative simplicity, it is striking to note not only the genetic homology but also the similar epistatic signaling hierarchy into which these regulators are arranged in the fly.
Using real-time microscopic cinematography, individual airway tip branching can be parsed temporally into a branch extension phase, a branch tip arrest phase, and a tip-splitting budding phase, followed once again when the branch budding phase is completed by branch extension until the next round of budding follows once more. A clock mechanism mediated by FGF–FGFR–Sprouty signaling plays a key role in timing the rate of bud extension and hence the inter-branch distance (Unbekandt et al., 2008; Warburton, 2008). Indeed a nested hierarchy of clock routines are likely to be present throughout lung development given the number of oscillating systems intrinsic to the lung (branching, airway peristalsis, calcium oscillations) or visited extrinsically upon it (fetal breathing, circadian rhythms). Branching morphogenesis is accompanied by contractile oscillations (airway peristalsis) that are themselves underpinned by periodic calcium waves (Featherstone et al., 2006; Jesudason et al., 2005). These oscillators appear to be coupled to lung growth, and their precise relation to the timing of branching remains to be determined. However, we postulate that clock routines underlying the linear process of somitogenesis are redeployed three-dimensionally for branching morphogenesis in the lung and other organs (Pourquie, 2003).
The airway is developed sequentially by early epithelial tube branching and later septation of terminal air sacs. Pulmonary vasculature develops within lung mesenchyme in close conjunction with epithelial morphogenesis. Airway and vascular smooth muscle also develop during early morphogenesis. Perturbation of these developmental processes results in abnormal lung structure, deficiency of gas exchange, and neonatal respiratory failure. Clinical examples of such disruption of normal lung growth include cystic adenomatoid malformation of the lung, bronchopulmonary dysplasia (BPD) (a sequel of premature human delivery), and hypoplasia of the lung (seen in congenital diaphragmatic hernia (CDH), a birth defect as common as cystic fibrosis). More subtle lung dysplasias that are not lethal neonatally may emerge later in life with asthmatic wheezing and perhaps predisposition to early onset of chronic obstructive pulmonary disease (COPD). One of the clearest examples of how early development can affect not only lung organogenesis but also long-term health is the ciliary dyskinesia encountered in Kartagener syndrome and primary ciliary dyskinesia (Storm van’s Gravesande, 2005). Early in embryogenesis, failure of ciliary function leads to randomization of organ situs and hence a supranormal rate of dextrocardia. This is accompanied by randomization of lung asymmetry. Persisting ciliary dysfunction impairs mucociliary clearance in the sinuses as well as the airways and predisposes to chronic lung disease in later life. Crucially one can note that disruption of lung asymmetry does not itself lead to lung malformation: hence the lung “bauplan” is conserved despite the lungs’ left–right asymmetry being the reverse of normal. This observation reiterates to us that the complexities of lung organogenesis may actually be broken down into nested routines and subroutines used to accomplish particular tasks in the overall process.
The implication for lung regeneration is that one need not understand the formation of every last alveolus, but rather that elucidating the iterative routines could suffice to promote pulmonary “self-assembly.”
This section of the review serves as a comprehensive reference source. For those with no requirement for such detail, the reader is directed to the summary Fig. 3.4.
We will first use a step-wise “process-driven” description of lung growth followed by a catalogue of the biochemical factors involved: many such factors are involved at multiple stages and do not map neatly on to the “process-driven” account. The biochemical factors are considered as follows: growth and transcription factors in order of first appearance and then other participating factors such as extracellular matrix (ECM) and miRNA.
Early lung induction is regulated by genes that act cooperatively to define the location of laryngotracheal groove and help specify the spatial axes of the developing organ.
Among the earliest endodermal signals essential for gut morphogenesis and gut tube closure are the GATA (zinc-finger proteins that recognize GATA DNA sequence) and hepatocyte nuclear factor (HNF/Fox) transcription factors. Foxa2 is required for gut tube closure, while Gata-6 is required for activation of the lung developmental program within the foregut endoderm. Hnf-3/Foxa2β is a survival factor for the endoderm; its expression is induced by Sonic hedgehog (Shh). Retinoids and their transcriptional factor receptors also play key roles in induction of early lung branching: retinoic acid (RA) deficiency and compound null mutation of retinoid receptors prevent induction of the laryngotracheal groove. Most recently, Wnt2/2b and β-catenin signaling have been shown to be necessary and sufficient to specify lung progenitors in the foregut (Goss et al., 2009; Harris-Johnson et al., 2009). Embryos lacking Wnt2/2b expression exhibit complete lung agenesis and do not express Nkx2.1, the earliest marker of the lung endoderm. This phenotype is recapitulated by an endoderm-restricted deletion of β-catenin. Conversely, conditional expression of an activated form of β-catenin leads to ectopic expansion of the Nkx2.1 expression domain into esophagus and stomach epithelium. Thus, gain or loss of trachea/lung progenitor identity is accompanied, respectively, by contraction or expansion of esophagus/stomach progenitor identity. Taken together, these findings suggest that Wnt2/2b signaling through the canonical Wnt pathway is required to specify lung endoderm progenitors within the foregut. Furthermore, ectopic lung bud formation can be induced in the esophagus by Tbx4 misexpression activating Fgf10 expression (Sakiyama et al., 2003). In addition, left–right asymmetry is controlled by several genes, including nodal, Lefty-1,2, and Pitx-2. For example, single-lobed lungs are found bilaterally in Lefty-1−/− mice, and bilateral isomerism of the lung is found in Pitx2-null mutants.
The processes whereby trachea and esophagus form from primitive foregut is of clinical interest due to the common birth defect, tracheoesophageal fistula (TEF) (Fig. 3.5).
Usually encountered in conjunction with esophageal atresia (EA), the combined sequence is sometimes found together with other anomalies of heart, vertebrae, anorectum, and limbs. Genetic defects identified in patients with EA-TEF have recently been comprehensively reviewed (Felix et al., 2009). Transgenic murine mutants with deletions in RA receptors or Gli2/Gli3 feature a form of EA-TEF. Moreover, the transcriptomic changes associated with budding of the lung from the foregut have recently been enumerated. Alongside identifying the known regulators described above, further candidates will need experimental evaluation (Millien et al., 2008). Illustrating that environmental factors may play a role, a EA-TEF phenotype can be generated by exposure of murine embryos to adriamycin (Diez-Pardo et al., 1996). Interestingly, despite the major anomaly of foregut development, lung formation in EA-TEF patients is usually grossly normal. Their respiratory tract morbidity tends to derive from tracheomalacia and, more chronically, reactive airways disease. Whilst the latter is traditionally attributed to gastroesophageal reflux and pulmonary aspiration, it remains possible that some of this pulmonary morbidity stems from subtly abnormal early lung development.
Children with EA-TEF may also suffer from tracheal weakness (tracheomalacia) in which inadequate formation of the tracheal cartilages results in potentially life-threatening airway closure during expiration.
Dorsoventral patterning of the trachea during embryonic development is associated with formation of C-shaped cartilage rings ventrally and trachealis muscle dorsally. Ventral mesenchyme segregates into successive cartilaginous and noncartilaginous domains, providing a compromise between flexibility and rigidity.
Tracheomalacia describes weakness of the walls of the trachea and it may result in life-threatening episodes and/or recurrent hospitalizations for lower airway infections (Austin and Ali, 2003; Boogaard et al., 2005; Carden et al., 2005; McNamara and Crabbe, 2004). It can be an isolated idiopathic anomaly or associated with EA-TEF, primary defects of cartilage synthesis (e.g., dyschondroplasia), cartilage degeneration due to trauma (e.g., long-term tracheal intubation), or extrinsic compressive lesions such as vascular rings or tumors (Berdon, 2000). Tracheal stenosis is narrowing of the trachea: it can follow prolonged intubation or accompany a cartilaginous sleeve malformation of the trachea or may again be associated with extrinsic compressive lesions. Tracheal cartilaginous sleeve comprises fusion of the ventral cartilage rings. It is a rare malformation associated with craniosynostosis syndromes like Crouzon syndrome, Pfeiffer syndrome, Goldenhar syndrome or Apert syndrome (Lin et al., 1995).
Tracheal maldevelopment can be modeled: nitrofen administration to pregnant dams results in tracheal malformations as well as CDH in offspring (Diez-Pardo et al., 1996; Xia et al., 1999). Although genetic control of the regulation of tracheal cartilage versus smooth muscle cell (SMC) formation remains unclear, relevant transgenic murine phenotypes have been observed. Miller et al. (2004) showed that partial Shh inactivation causes tracheobronchial cartilage abnormalities indicative of tracheomalacia. Park et al. (2009) demonstrated Shh augments Sox9 expression: Sox9 induces type II collagen (Col2a1) expression and promotes the chondrocyte lineage amongst mesenchymal cells. Bone morphogenic protein 4 (BMP4) also regulates Sox9 to induce chondroprogenitors amongst mesenchymal cells (Hatakeyama et al., 2004). Impaired BMP signaling induces tracheal cartilage defects with EA-TEF (Que et al., 2006). β-Catenin also interacts with Sox9 but to inhibit differentiation of tracheal chondroprogenitor cells (Akiyama et al., 2004).
Recently, FGF18 and FGF10 have also been described to play important roles in tracheal cartilage ring formation (Elluru et al., 2009; Tiozzo et al., 2009; Whitsett et al., 2002). Specific targeted inactivation of Fgf18, using the SP-C promoter driving Cre, induced malformation of the cartilage rings (Whitsett et al., 2002). Overexpression of Fgf18 also resulted in malformation of the cartilage rings, possibly via Sox9 upregulation (Elluru et al., 2009). Tiozzo et al. (2009) reported that ectopic fibroblast growth factor receptor (FGFR)2b expression in tracheal mesenchyme renders this hyper-responsive to FGF10, resulting in cartilaginous sleeve formation reminiscent of the Apert syndrome tracheal phenotype (Fig. 3.6). This abnormal cartilage structure arises secondary to increased proliferation of cartilage progenitor cells within tracheal mesenchyme.
Despite incomplete understanding of such genetics, tissue engineered airway (formed using stem cells and cadaveric scaffold) has been successfully transplanted into adult and a pediatric patients to replace damaged bronchus and trachea, respectively (Macchiarini et al., 2008)
A multiplicity of factors are required for normal airway branching morphogenesis. Much of the research in this area is focused on epithelial morphogenesis: this is discussed in detail in subsequent sections dealing with individual signaling pathways. A key insight has been that epithelial morphogenesis proceeds interdependently with vascular development. Indeed tight coupling of endothelial with epithelial development is required for efficient gas transport and fluid clearance at birth. Vascular endothelial growth factor (VEGF) signaling from the epithelium to the developing endothelium is essential for the primitive hemangioblasts to develop into mature capillary networks. Likewise, the endothelium probably signals back to coordinate epithelial morphogenesis. There is a stereotyped anatomical relationship between the developing pulmonary capillaires, arteries, and veins (Fig. 3.7). The arteries run along the superior surface of the developing lobules, while the veins run along the interior surface.
Septation of terminal sacs generates alveoli and involves interacting mesenchymal myofibroblasts, epithelial cells, and endothelial cells. Myofibroblasts are smooth muscle precursors with fibroblast morphology that migrate within nascent septa and deposit elastin (particularly at tips) as the first step of secondary septa development (Bostrom et al., 1996; Lindahl et al., 1997). Alveolar myofibroblast differentiaion requires Lunatic fringe and Notch signaling, and their elastin deposition is PGFa- and FGFR3/4−regulated (Xu et al., 2010). Septal thinning and maturation of the alveolar capillary network are also needed. Interstitial thinning proceeds with expansion of septal epithelial, vascular, and airspace compartments but also myofibroblast apoptosis (Awonusonu et al., 1999; Schittny et al., 1998) that affects lipid-filled interstitial fibroblasts (LFIF) rather than non-LFIF (NLFIF). This apoptosis is associated with downregulation of insulin-like growth factor I receptor (Igf-IR) mRNA and cell surface protein expression (Srinivasan et al., 2002).
Finally, the new alveolar septum differentiates into a functional respiratory membrane that consists of type I AECs (AECI), basement membrane, and capillary endothelial cells. The respiratory membrane provides a short distance for diffusion thereby facilitating gas exchange. It is estimated that about 50 million alveoli are present in neonatal lung. However, by age 7–8 years, when alveolarization is largely complete, the number of alveolar units in the lung has increased six-fold to about 300 million. Meanwhile the adut alveolar capillary bed is capable of accommodating the entire adult cardiac output of 5 L/min, rising five-fold to 25 L/min during maximal exercise.
Alveolarization can be adversely affected by premature delivery, hyperoxia, postnatal steroid exposure, and prolonged mechanical ventilation, even with room air (Mokres et al., 2010). Thus, prematurity and hyperoxia plus pressure plus time are well-recognized risk factors for the hypoalveolarization characteristic of BPD in human premature infants. It is postulated that a cascade of events including endotoxin exposure, inflammation, and expression and activation of excessive amounts of transforming growth factor (TGF)-β ligand inhibit aleolarization in BPD and therefore portend an adverse outcome of this disease (reviewed in Shi et al., 2009).
Having considered the process of building the lung, we next turn to catalogue the factors required for lung growth and maturation. Transgenic mouse technology has allowed us to evaluate roles in organogenesis by, for example, overexpressing or knocking out a specific genes (Costa et al., 2001). Some murine pulmonary phenotypes resulting from a loss or gain of gene function are listed in Table 3.1, modified from Cardoso and Lu (2006).
At least four groups of transcription factors, forkhead box, Nkx homeodomain, RA receptors, and Gli family play important roles in lung development.
Members of the forkhead box family transcription factors, such as Foxa1, Foxa2, HFH8, and HFH4, share homology in the winged-helix DNA-binding domain and regulate pulmonary cellular proliferation and differentiation.
HNF-3α (Foxa1) and HNF-3β (Foxa2) share 93% homology in amino acid sequences and were first identified as factors in hepatocyte differentiation (Qian and Costa, 1995). However, Hnf-3β is expressed in developing lung, with higher levels in proximal epithelial cells and lower levels in distal type II epithelial cells (Zhou et al., 1996b). Overexpression of Hnf-3β under control of epithelial specific SP-C promoter inhibits lung branching morphogenesis and vasculogenesis in vivo (Zhou et al., 1997). HNF-3α and HNF-3β also regulate expression of CCSP and surfactant proteins in bronchiolar and type II epithelial cells (Bingle et al., 1995; Bohinski et al., 1994; He et al., 2000). Hnf-3β is inducible by interferon, and regulates in turn the expression of the Nkx homeodomain transcription factor Nkx2.1 (also termed Ttf-1 and CebpI), which in turn regulates transcription of the surfactant protein genes in peripheral lung epithelium (Ikeda et al., 1996; Samadani et al., 1995).
HFH8 is restricted to splanchnic mesoderm contacting embryonic gut and presumptive lung at E9.5 suggesting that Hfh-8 may participate in lung induction. HFH-8 expression continues in lateral mesoderm-derived tissue during development. By E18.5, Hfh-8 expression is restricted to distal lung mesenchyme and bronchial muscle (Peterson et al., 1997). The level of Hfh-8 expression is important for normal development: alveolar hemorrhage is observed in Hfh8(+/−) mice, while Hfh8(−/−) mice die in utero. Reduced Hfh-8 expression in Hfh-8+/− mutants is accompanied by decreased expression of VEGF and its receptor 2 (Flk-1), bone morphogenetic protein 4 (BMP-4), and the transcription factors of the Brachyury T-Box family (Tbx2–Tbx5) and Lung Kruppel-like factor (Kalinichenko et al., 2001). HFH8 regulates mesebchymal Pdgf receptor (Bostrom et al., 1996; Shinbrot et al., 1994; Souza et al., 1996). HFH8 binding sites are also found in the promoter region of genes, such as Bmp4, Hgf, and Hoxa5, that are critical regulators of lung morphogenesis (Ohmichi et al., 1998; Weaver et al., 1999).
Hfh4 (Foxj1) regulates ciliated epithelial cell differentiation. It is expressed in E15.5 airway epithelium just before ciliated cells appear (Hackett et al., 1995) and Hfh4−/−-null mutant mice feature defective ciliogenesis in airway epithelial cells and randomized left–right asymmetry (mimicking human Kartagener syndrome). The latter can result in perinatal lethality, but in low penetrance it gives rise to situs inversus, sinusitis, bronchiectasis, and sterility, all caused by defects in ciliary beat (Brody et al., 2000; Chen et al., 1998). Interestingly, HFH4 and other proximal lung markers such as CCSP are upregulated by BMP antagonist Noggin in mesenchyme-free airway epithelial culture (Hyatt et al., 2002).
Foxp1, Foxp2, and Foxp4 are highly expressed in mouse lung and gut. Foxp1 and Foxp4 are expressed in both proximal and distal airway epithelium while Foxp2 is expressed primarily in distal epithelium. Foxp1 protein expression is also observed in the mesenchyme and vascular endothelial cells of the lung (Lu et al., 2002).
One of the most prominent homeodomain transcription factors in lung development is NKX2.1, also called TTF-1 (thyroid-specific transcription factor) or CEBP-1. Nkx2.1 is expressed in epithelial cells derived from foregut endoderm in lungs, thyroid, and pituitary, as well as restricted regions of fetal brain (Guazzi et al., 1990; Lazzaro et al., 1991). Hence, human Nkx2.1 mutants may feature benign hereditary chorea, congenital hypothyroidism, and neonatal respiratory distress at term (sometimes retrieved by the transactivating activity of Pax8) (Carre et al., 2009). Nkx2.1−/− mice exhibit impaired tracheoesophageal separation and early arrest of lung development featuring two main bronchi but no distal branches (Kimura et al., 1996; Minoo et al., 1999). In developing mouse airway epithelium, Nkx2.1 is initially expressed proximally and distally becoming restricted at later stages to distal AECs (Zhou et al., 1996b). Overexpression of Nkx2.1 causes dose-dependent morphological alterations in postnatal lung: modest overexpression raises type II pneumocyte proliferation and SP-B levels; greater overexpression disrupts alveolar septation with emphysema due to alveolar hypoplasia. The highest overexpression of Nkx2.1 in transgenic mice causes severe pulmonary inflammation, fibrosis, and respiratory failure, associated with eosinophil infiltration and increased eotaxin and IL-6 expression (Wert et al., 2002). Nkx2.1 signaling is critical for surfactant protein, T1a, and CC10 gene expression (Boggaram, 2003; Bruno et al., 1995; Guazzi et al., 1990; Ramirez et al., 1997; Whitsett and Glasser, 1998; Yan et al., 1995; Zhang et al., 1997). Nkx2.1-deficient pulmonary epithelial cells fail to express nonciliated marker genes, including differentiated Sp-B, Sp-C, and CC10. Bmp4 expression in these cells is also reduced. In addition to modulating expression of other lung-related genes, it is clear that NKX2.1 phosphorylation plays a crucial role in its signaling: mice with point mutation of seven serine phosphorylation sites of NKX2.1 died immediately following birth with malformation of acinar tubules, pulmonary hypoplasia, and reduced expression of surfactant proteins, CC10/secretoglobulin 1A, and Vegf (DeFelice et al., 2003). Whilst regulating expression of numerous genes, Nkx2.1 expression can itself be activated by transcription factors HNF-3β (Ikeda et al., 1996) and GATA-6 (Shaw-White et al., 1999) during lung morphogenesis.
Hox transcription factors are expressed with proximodistal polarity in developing lung: Hoxa5, Hoxb2, and Hoxb5 are restricted to distal lung mesenchyme, whilst Hoxb3 and Hoxb4 are expressed in proximal and distal mesenchyme (Aubin et al., 1997; Bogue et al., 1996; Volpe et al., 1997). Illustrating their functional role, Hoxa5−/−-null mutant mice have tracheal defects and occlusions, impaired lung branching morphogenesis, diminished surfactant protein expression, and alveolar wall thickening (Aubin et al., 1997).
GLI 1, 2, 3 are zinc-finger transcription factors and activated by SHH. All are mesodermally expressed, particularly in the distal lung (Grindley et al., 1997). Combined Gli2−/− and Gli3−/− mutant mice feature lung agenesis. Gli3−/− mice are viable but have small dysmorphic lungs (Grindley et al., 1997). Gli2 regulates normal lung asymmetry: Gli2−/− mice have a fused right and left lung (a small single lobe with defective primary branching in the right lung) and hypoplastic trachea and esophagus that are nevertheless distinct and retain normal proximal–distal differentiation (Motoyama et al., 1998).
Embryonic lung mesenchymal and epithelial cells communicate through autocrine and paracrine factors, as demonstrated by effects of added growth factors on cultured embryonic lung growth (Jaskoll et al., 1988; Warburton et al., 1992).
FGF family members are found throughout the vertebrates and invertebrates. Their functions in respiratory organogenesis are conserved from Drosophila to mammals (Glazer and Shilo, 1991; Sutherland et al., 1996). Based on protein sequence homology, FGFs have been divided into 23 subgroups. Similarly, their cognate transmembrane protein tyrosine kinase receptors (FGFRs) are classified into four types, contributing to the specificity of FGF ligand binding (Ornitz and Itoh, 2001). Heparan sulfate proteoglycan, an ECM glycoprotein, has been reported to be essential for FGF ligand–receptor binding and activation (Izvolsky et al., 2003a,b; Lin et al., 1999). FGFs play critical roles in cell proliferation, migration, and differentiation during development. Early inhibition of murine FGFR signaling shows it is required for early lung branching morphogenesis. Later FGFR inhibition in E14.5 lung decreases prenatal airway tubule formation and is associated with severe emphysema at maturity. At E16.5, FGFR inhibition causes mild focal emphysema. Murine mutants lacking FGFR3 and FGFR4 fail to undergo normal alveolarization, with poorly organized myofibroblasts and excessive amounts of poorly organized elastin. However, inhibition of FGFR signaling after birth did not appear to alter postnatal alveolarization (Hokuto et al., 2003).
FGF10 is one of the most-studied family members during lung development. Fgf10-null mice lack distal lung despite formation of larynx and trachea (Min et al., 1998). Fgf10 is expressed focally in E11–12 mouse peripheral lung mesenchyme and signals through adjacent distal epithelial FGFR2IIIb (whose loss also disrupts lung development) (De Moerlooze et al., 2000). These sites of expression change dynamically, compatible with the idea that FGF10 appears at sites of bud formation (Bellusci et al., 1997b). FGF10 has a chemotactic effect on nearby epithelium in culture: epithelial tips will proliferate and migrate toward FGF10 in mesenchyme or on beads (Park et al., 1998; Weaver et al., 2000). FGF10 controls epithelial differentiation, inducing Sp-C expression and downregulating Bmp4 expression (Hyatt et al., 2002). FGF10 dosage and signal transduction level is critical: mice with 20% of normal FGF10 expression (due to an enhancer trap bearing LacZ inserted 100Kb upstream in the FGF10 promoter) feature lung hypoplasia (Ramasamy et al., 2007); similarly, downstream signaling inhibition by misexpression of Sprouty2 under control of the Sftpc promoter induces lung hypoplasia (Mailleux et al., 2001). Several key regulatory molecules such as SHH, BMPs, and TGF-βs crosstalk with FGF10 during embryonic lung morphogenesis: their interactions will be discussed later.
FGF7 (KGF) is found in developing lung mesenchyme at late stages (Post et al., 1996). In early cultured mouse embryonic lung, addition of FGF7 promotes epithelial growth and formation of cyst-like structures with extensive cell proliferation. FGF7 can also contribute to distal airway epithelial cell differentiation (Cardoso et al., 1997; Deterding et al., 1996). Erm and Pea3 are ETS domain transcription factors known to be downstream of FGF signaling. FGF7 can induce Erm/Pea3 expression more effectively than FGF10. Erm is transcribed exclusively in the epithelium, while Pea3 is expressed in both epithelium and mesenchyme. When examined at E18.5, transgenic expression of a repressor form of Erm specifically in the embryonic lung epithelium shows that the distal epithelium of Sp-C-Erm transgenic lungs is composed predominantly of immature type II cells, while no mature type I cells are observed. By contrast, the differentiation of proximal epithelial cells, including ciliated cells and Clara cells, appears to be unaffected (Liu and Hogan, 2002; Liu et al., 2003). FGF7 does not seem to protect against hyperoxic inhibition of normal postnatal alveoli formation and early pulmonary fibrosis, but FGF7 consistently had a significant protective/preventive effect against the development of pulmonary hypertension during hyperoxia (Frank, 2003). However, Fgf7−/− mutant mice have no gross lung abnormalities (Guo et al., 1996), suggesting a FGF7 redundancy during lung development.
FGF9, which signals through FGFR2IIIc, also regulates branching morphogenesis. In E10.5 lung, Fgf9 is expressed in visceral pleura outlining the lung bud and in bronchial epithelium, while Fgfr2IIIc is predominantly expressed in lung mesenchyme. At E12.5 and E14.5, Fgf9 expression persists in visceral pleura but is no longer detected in epithelium (Colvin et al., 1999). Fgf9-null mice exhibit reduced mesenchyme and decreased airway branching but show significant distal airspace formation and pneumocyte differentiation. The reduction in the amount of mesenchyme in Fgf9−/− lungs limits expression of mesenchymal Fgf10 (Colvin et al., 2001). By contrast, addition of recombinant FGF9 protein inhibits the differentiation response of the mesenchyme to N-SHH, but does not affect proliferation (Weaver et al., 2003).
The signaling cascade activated by FGF10 and FGF9 involves FGFR2b and 2c, respectively, as well as Shp2, Raf, MAP ERK kinase (MEK), and extracellular-regulated kinases (ERK) 1 and 2 as signal transducers. MEK inhibition has been shown to reduce lung branching and epithelial cell proliferation, but increase mesenchyme cell apoptosis in fetal lung explants (Papadakis et al., 1997). FGF signaling is regulated at several levels. One of the key inducible negative regulators is the Sprouty family. There are four sprouty (Spry) genes in mouse (mSpry1–4) and human (hSpry1–4). Murine Spry2 is expressed in the distal tip of embryonic lung epithelial branches, but is downregulated between the sites of new bud formation. Murine Spry4 is predominantly expressed in the distal mesenchyme of the embryonic lung (Mailleux et al., 2001), and may play roles in branching morphogenesis. Sprouties (SPRY1, 2, 4) act as suppressors of Ras–MAP kinase signaling (Hacohen et al., 1998; Kramer et al., 1999; Reich et al., 1999). Overexpression of mSpry2 or mSpry4 can inhibit lung branching morphogenesis through reducing epithelium cell proliferation (Hadari et al., 1998; Perl et al., 2003; Tefft et al., 2002). SPRED-1 and SPRED-2 are two sprouty related proteins, which contain Enabled/VAsodilator-Stimulated Phosphoprotein (VASP) Homology-1 (EVH-1) domains. Spreds are predominantly expressed in mesenchymal cells. Expression of Spreds is especially strong in the peripheral mesenchyme and epithelium of new bud formation. After birth, Spreds expression decreases, while the expression of Sprouties expression remains high. Both Sprouties and spreds play important roles in mesenchyme– epithelium interaction during lung development (Hashimoto et al., 2002).
The TGF-β superfamily comprises numerous structurally related polypeptide growth factors including TGF-β, BMP, and activin subfamilies. TGF-β ligands bind to cognate cell surface receptors, and activate Smad proteins, which translocate to the nucleus and modulate target gene expression (Massague, 1998; Shi and Massague, 2003).
The TGF-β ligand subfamily comprises three isoforms, TGF-β1, 2, and 3. TGF-β1 is expressed in early embryonic lung mesenchyme, particularly underlying distal epithelial branch points; TGF-β2 is localized mainly in distal epithelium; TGF-β3 is mainly expressed in proximal mesenchyme and mesothelium (Bragg et al., 2001; Millan et al., 1991; Pelton et al., 1991a,b; Schmidt et al., 1991). Each TGF-β isoform has nonredundant roles revealed by isoform-specific knockouts. Mice lacking TGF-β1 develop apparently normally, but die within 2 months of life from aggressive pulmonary or gut inflammation, as a result of failure to negatively modulate the immune system (McLennan et al., 2000). TGF-β2−/− mutation results in embryonic lethality around E14.5 in mice featuring complex cardiac anomalies and lung dysplasia amongst others (Bartram et al., 2001). TGF-β3−/− mutant mice display cleft palate, retarded lung development, and neonatal lethality with difficulty swallowing and breathing (Kaartinen et al., 1995; Shi et al., 1999). Furthermore, blockade of TGF-β signaling by null mutation of TGF-β activated kinase-1 binding protein-1 (TAB1) results in lethal cardiovascular and lung dysmorphogenesis (Komatsu et al., 2002).
As with the FGFs, the timing and dosage of TGF-β signaling are critical during lung development. Optimal physiological levels of TGF-β-Smad3 signaling appear essential for secondary alveolar septa formation: abrogation of TGF-β type II receptor in lung epithelial cells reduces alveolar septation and allows emergence of AECI (Chen et al., 2008). However, TGF-β1 overexpression in early mouse embryonic lung epithelium inhibits branching morphogenesis (Zhao et al., 1999), whereas misexpression of Sp-C promoter-controlled TGF-β1 in embryonic lung epithelium arrests embryonic lung growth and epithelial cell differentiation whilst inhibiting pulmonary vasculogenesis (Zhou et al., 1996a, 2001). Suggesting a crucial role for optimal TGF-β1 levels in human lung maturation, excessive activated TGF-β1 has been reported in tracheal aspirates of human premature infants who develop more severe BPD (Lecart et al., 2000; Toti et al., 1997). Furthermore, misexpression of TGF-β1 in neonatal rat lung using recombinant adenoviral vectors resulted in neonatal alveolar hypoplasia and interstitial fibrosis; this histological picture closely phenocopies human BPD (Gauldie et al., 2003). By contrast, misexpression of TGF-β1 in adult rats results in chronic progressive interstitial pulmonary fibrosis with increased proliferation and matrix secretion by the mesenchyme (Sime et al., 1997; Zhao et al., 2002). In addition, TGF-β1 may be centrally involved in pulmonary fibrotic responses to bleomycin, or endotoxin and infection (Bonniaud et al., 2005). Blockade of TGF-β signaling via Smad3-null mutation strongly attenuates bleomycin-induced pulmonary fibrosis (Zhao et al., 2002).
The activity of TGF-β signaling is multiply regulated: β6-integrin, Latent transforming growth factor-beta binding proteins (LTBPs), and thrombospondin regulate TGF-β release, whilst β-glycan, endoglin, or decorin modulate TGF-β receptor binding affinity. As expected, mutation of the above genes causes some phenotypes similar to those of TGF-β mutants: loss-of-function mutation in human and mouse endoglin (whose protein binds TGF−β and Alk1, its type I receptor) causes hereditary hemorrhagic telangiectasia (Li et al., 1999; Massague, 2000; McAllister et al., 1994; Urness et al., 2000). Null mutation of LTBP-3 or LTBP-4 causes profound defects in elastin fiber structure and lung alveolarization similar to Smad3 knockout mouse lung (Sterner-Kock et al., 2002; Colarossi et al., 2005; Chen et al., 2005).
In addition, TGF-β signaling blockade has distinct impacts on lung branching morphogenesis and alveolarization depending on whether epithelial or mesenchymal cells are targeted. Mesenchymal TGF-β signaling blockade driven by Dermo1 retards branching after mid-gestation; by contrast, epithelial TGF-β signaling abrogation lacks prenatal impact and only disrupts postnatal lung alveolarization (Chen et al., 2008). Meanwhile, human TGF-β pathway mutations in, for example, TGF-β type II receptor (Alk5) or TGF-β binding proteins such as fibrillin underlie dysplastic matrix elastin defects that predispose to aortic dissection or sudden alveolar rupture in Marfan’s syndrome patients (Kaartinen and Warburton, 2003). Thus, TGF-β signaling has to be regulated “just right” with both deficiency and excess deleterious to normal alveolarization (a concept we term the Goldilocks hypothesis).
BMPs, with more than 20 ligand family members, regulate many processes, including lung development (Hogan, 1996). Expression of Bmp3, 4, 5, and 7 is detected in embryonic lung (Bellusci et al., 1996; King et al., 1994; Takahashi and Ikeda, 1996). BMP4 plays a central role in normal lung development (Hogan, 1996). Addition of BMP4 to whole embryonic lung explants stimulates branching (Bragg et al., 2001; Shi et al., 2001). However, BMP4 inhibited FGF10-induced growth of isolated E11.5 mouse lung endoderm cultured in Matrigel (Weaver et al., 2000). Transgenic overexpression of BMP4 in distal fetal lung endoderm, driven by a 3.7 kb human surfactant protein C (SP-C) promoter, causes abnormal morphogenesis with cystic terminal air sacs (Bellusci et al., 1996). Conversely, Sftp-C promoter-driven overexpression of BMP antagonists Noggin or Gremlin severely reduces distal epithelial cell phenotype whilst increasing proximal cell types (Lu et al., 2001; Weaver et al., 1999). Interestingly, blockade of endogenous BMP4 in embryonic mouse lung epithelial cells using a conditional gene knockout approach results in abnormal lung development with similar dilated terminal sacs as seen in BMP4 transgenic mouse lung (Eblaghie et al., 2006). This suggests optimal BMP4 levels are essential for normal lung development. As extracellular growth factors, BMPs bind heteromeric complexes of BMP serine/threonine kinase type I and type II receptors to activate intracellular signal pathway (Massague, 1998; Shi and Massague, 2003). Three cognate BMP type I receptors (Alk2, Alk3, and Alk6) have been identified. Among them, Alk3 is expressed predominantly in distal airway epithelial cells during mouse lung development. Alk3 abrogation in mouse lung epithelia either from early lung organogenesis or from late gestation resulted in similar neonatal respiratory distress phenotypes, accompanied with collapsed lungs (Sun et al., 2008). Early induction of Alk3 knockout in lung epithelial cells causes retardation of early lung branching morphogenesis and reduces cell proliferation and differentiation. But late gestation induction of Alk3 knockout also causes significant epithelial apoptosis accompanied by lack of surfactant secretion (Sun et al., 2008). Furthermore, canonical Wnt signaling was perturbed, possibly through reduced WIF-1 expression in Alk3 knockout lungs (Sun et al., 2008). Therefore, deficiency of appropriate BMP signaling in lung epithelial cells results in prenatal lung malformation, neonatal atelectasis, and respiratory failure.
In addition, BMP signaling is also important in lung vasculogenesis and angiogenesis. Mutations of BMP type II receptor (BMPRII) and change in expression of BMP antagonist Gremlin are associated with primary pulmonary hypertension (PPH) (Lane et al., 2000; Costello et al., 2008). Moreover, upregulation of Gremlin is also associated with pulmonary fibrosis and the severity of the fibrotic pathology (Koli et al., 2006; Myllarniemi et al., 2008)
Sonic hedgehog is a vertebrate homolog of Hedgehog (Hh) that patterns the segment, leg, wing, eye, and brain in Drosophila. Hh binds to patched (Ptc), a transmembrane protein, and releases its inhibitory effect on downstream smoothened (Smo), which is a G protein-coupled transmembrane spanning receptor. This leads to the activation of cubitus interruptus (Ci), a 155-kDa transcription factor that is cleaved to form a 75-kDa transcription inhibitor in cytosol. Elements of the Drosophila Hh signaling pathway and their general functions in the pathway are highly conserved in vertebrates, albeit with increased levels of complexity. Gli1, 2, and 3 are the three vertebrate Ci gene orthologs (van Tuyl and Post, 2000).
The SHH signal transduction pathway plays important roles in mesenchyme–epithelium interaction. In developing mouse lung, Shh is detected in the tracheal diverticulum, the esophagus, and later in the trachea and lung endoderm. Shh is expressed at low levels throughout the epithelium, whilst at higher level in the growing distal buds (Bellusci et al., 1997a; Urase et al., 1996). Null mutation of Shh produces profound lung hypoplasia and failed trachea–esophageal septation. Mesenchymal Ptc, Gli1, and Gli3 expression are all downregulated in Shh knockout lung. Nevertheless, proximodistal differentiation of airway epithelium is preserved (Litingtung et al., 1998; Pepicelli et al., 1998). Also, Fgf10 expression is dysregulated in Shh-null mutant lung compared to the precisely restricted expression seen normally. Lung-specific Shh overexpression results in severe alveolar hypoplasia and significant increase in interstitial tissue caused by increased epithelial and mesenchymal proliferation (Bellusci et al., 1997a). Defective hedgehog signaling may lead to EA and TEF (Spilde et al., 2003).
The membrane-bound Hedgehog interacting protein 1 (HIP1) directly binds mammalian Hedgehog (HH) proteins and attenuates their signaling (Chuang and McMahon, 1999). Hip1 is transcriptionally activated in response to HH signaling, overlapping the expression domains of Ptc1 (Chuang and McMahon, 1999; Goodrich et al., 1996). Targeted disruption of Hip1 results in upregulated Hedgehog signaling and lethal neonatal respiratory failure: left– right asymmetry persists but initial branching from the two primary buds is absent; Fgf10 expression is slightly downregulated at the tips of the primary buds in Hip1−/− lungs at E10.5 but completely absent from the mesenchyme where secondary branching normally initiates (Goodrich et al., 1996). Attenuated PTC1 activity in a Hip1−/− mutant lungs leads to an accelerated lethality. Hip1 and Ptch1 have redundant roles in lung branching control (Goodrich et al., 1996). Both of them can attenuate SHH signal in lung development and pancreas development (Goodrich et al., 1996; Kawahira et al., 2003).
Wnt signals are transduced via seven transmembrane Wnt receptors encoded by Frizzled (Fzd) genes to activate the β-catenin T Cell transcription Factor (TCF) pathway, the c-Jun N-terminal kinases (JNK) pathway, or the intracellular Ca2+-releasing pathway. The Wnt/β-catenin pathway plays a critical role in many developmental and tumorigenesis processes. Following Wnt binding to the receptor, β-catenin is dephosphorylated and translocates to the nucleus to activate downstream gene expression (Wodarz and Nusse, 1998).
TOPGAL and BATGAL reporter transgenes have been used to analyze patterns of β-catenin stabilization in developing lung. Within the respiratory precursor region, the TOPGAL reporter is expressed in the undivided proximal endodermal tube and then the lung buds as early as E9.5 (Okubo and Hogan, 2004). This pattern is maintained as the trachea and esophagus separate and the lung buds grow out between E10 and E11.5 (Dean et al., 2005; De Langhe et al., 2005; Okubo and Hogan, 2004; Shu et al., 2005). Between E12.5 and E18.5, analysis of TOPGAL and BATGAL transgene activity suggests a dynamic pattern of TCF/β-catenin-dependent gene expression. Reporter gene activity is found in the tracheal epithelium and cartilaginous condensations at E12.5 but is restricted to the bronchial mesenchyme at E13.5 (De Langhe et al., 2005; Shu et al., 2005). The distal lung epithelium expresses both reporters by E9.5. The pattern of TCF/β-catenin-dependent gene activity in the distal lung at later time points is somewhat variable and dependent on the reporter transgene analyzed. In general, transgene activity clears from the central airways between E13.5 and postnatal day 14 (Okubo and Hogan, 2004; Shu et al., 2005). At E14.5, expression in the distal tip epithelium is either extinguished (TOPGAL) (De Langhe et al., 2005) or restricted to a subset of early alveolar type 2 cells (BATGAL) (Shu et al., 2005). In the adult lung, the TOPGAL transgene is highly expressed in the distal trachea and in clusters of airway secretory and ciliated cells but rarely in the alveolar region (De Langhe, unpublished data).
β-catenin deletion in proximal airway epithelium during development resulted in no obvious alteration to lung structure (Mucenski et al., 2003). By contrast, embryonic deletion of β-catenin in the distal lung epithelium resulted in profound perturbation of normal epithelial, mesenchymal, and vascular development. The latter mice feature proximalization of lung epithelium with decreased expression of alveolar type 2 cell marker Sftpc, vascular endothelial marker PECAM, and α-smooth muscle actin; upper airway epithelial markers (Scgb1a1, FoxJ1, and β-tubulin) were unaltered.
Stabilization of β-catenin in proximal epithelium using the CatnbfloxedExon3 allele raised epithelial β-catenin levels, resulting in squamous, cuboidal, and goblet cell dysplasia in intrapulmonary conducting airways and the appearance of alveolar type 2-like cells in the bronchioles (Mucenski et al., 2005). Epithelial levels of Scgb1a1 immunopositive cells were low whilst SPC expression increased, indicating an increase in Scgb1a1/Sftpc double-positive cells. Similar expansion of Scgb1a1/Sftpc double-positive bronchioalveolar stem cells (BASCs) in response to increased canonical Wnt signaling has been shown in the lung epithelium upon Gata6 loss (Zhang et al., 2008). These authors also showed that canonical Wnt signaling is activated within the niche containing BASCs during lung epithelial regeneration, while forced Wnt activation greatly increases BASC numbers.
Li et al., (2009) stabilized β-catenin in the entire developing lung epithelium using Nkx2.1-cre and Catnb[+/lox(ex3)] mice: in trachea and main bronchi, polyp-like structures formed featuring intracellular β-catenin accumulation suggesting blocked differentiation of spatially-appropriate airway epithelial cell types, Clara cells, ciliated cells, and basal cells (BCs), while activating UCHL1, a marker for pulmonary neuroendocrine cells.
Alternatively, the method of using a Spc promoter-regulated Lef1-dN89β-catenin to stabilize β-catenin from about E10.5 was employed by Okubo and Hogan (2004) to generate mice with widened primary bronchial tubes opening directly into saccules (lined with simple cuboidal or columnar epithelium), decreased progenitor differentiation into secretory and ciliated cells, and absence of alveolar type 2 and type 1 cells. Thus, constitutive β-catenin signaling in developing foregut endoderm partially inhibited branching morphogenesis and blocked expression of lung-specific differentiation genes.
Using a hypomorphic Fgf10 allele, Ramasamy et al. (2007) showed that FGF10 signaling via FGFR2b controls the proliferation of the pulmonary epithelial progenitors in part by autoregulation of β-catenin signaling in the epithelium. This correlation of a reduction in epithelial FGF signaling and epithelial TOPGAL activity has also been demonstrated in lungs of a mouse Apert disease model (De Langhe et al., 2006). Intriguingly, the regulation of epithelial β-catenin signaling by FGF10 and concomitant upregulation of Fgfr2b receptor expression result in potentiating this signaling cascade locally, thus maintaining the distal epithelial progenitor state.
By contrast, the lack of significant activity of well-established Wnt reporters in mesenchyme (including TOPGAL and BATGAL mice) does not support an important role for mesenchymal Wnt signaling during organogenesis. However, expression of several mesenchymal Wnt receptors in the lung has been reported (De Langhe et al., 2005). Furthermore, Wnt5a overexpression either directly or indirectly regulates mesenchymal Fgf10 expression (Li et al., 2005), while Wnt7b acts on lung vascular SMCs via Frizzled 1 and LRP5 (Wang et al., 2005). Besides Lef1/TCF-mediated β-catenin signaling, β-catenin also acts via PITX family transcription factors (Kioussi et al., 2002), which are abundantly expressed in developing mesenchyme (Kitamura et al., 1999). Using Dermo1Cre/+-mediated conditional inactivation (CKO) of β-catenin, De Langhe et al. (2008) showed Dermo1-cre/β-catenin CKO embryos have multiple defects reminiscent of double knockout of Pitx1 and Pitx2 genes (Marcil et al., 2003). Combining fate analysis and global gene expression studies, mesenchymal β-catenin signaling was shown to have dual, lineage-dependant functions: it regulates formation and amplification but not differentiation of Fgf10-expressing parabronchial smooth muscle progenitors (in part via regulation of Fgfr2c expression) but is required for normal endothelial cell differentiation (De Langhe et al., 2008). Cohen et al. (2009) confirmed the role of Wnt in parabronchial smooth muscle development and showed Wnt pathway upregulation in experimental asthma.
EGF, TGF-α, and amphiregulin are EGF receptor (EGFR) ligands. Loss- or gain-of-function experiments in mouse, rat, or other animal models prove that EGF ligands positively modulate early mouse embryonic lung branching morphogenesis and cytodifferentiation through EGFR (Schuger et al., 1996a; Seth et al., 1993; Warburton et al., 1992). EGF is expressed in mature AECs and regulates type 2 cell proliferation via autocrine mechanism in culture and in vivo (Raaberg et al., 1992). However, epithelial TGF-α overexpression under Sp-C promoter control induces postnatal lung fibrosis (Korfhagen et al., 1994). TGF-α overexpression caused severe pulmonary vascular disease mediated via EGFR in distal epithelium, but reductions in VEGF may also contribute (Le Cras et al., 2003).
EGFR is a tyrosine kinase receptor whose deletion (Egfr−/−) causes abnormal branching, poor alveolarization, and aberrant matrix metalloprotease protein (MMP) expression (Kheradmand et al., 2002). EGFR phosphorylation in response to stretch induces, at least in part, fetal epithelial cell differentiation via ERK pathway activation. Specific EGFR or ERK pathway blockade reduces stretch-inducible Sp-C mRNA expression. Thus, EGFR may represent a mechanical signal sensor during lung development (Sanchez-Esteban et al., 2003).
Tumor necrosis factor-α (TNFα)-converting enzyme (TACE) is a transmembrane metalloprotease disintegrin that functions as a membrane sheddase to release the ectodomain portions of many transmembrane proteins, including the precursors of TNFα and several other cytokines, as well as the receptors for TNFα, and neuregulin (ErbB4) (Shi et al., 2003). Neonatal TACE-deficient mice had visible respiratory distress and their lungs failed to form normal saccular structures. Mouse embryonic lung explant cultures show that TGF-α and EGF can rescue the inhibition of TACE activity (Zhao et al., 2001).
PDGF-A and PDGF-B form homodimers (AA or BB) or heterodimers (AB). Two PDGF receptor types, α and β, occur in embryonic mouse lung and are differentially regulated in fetal rat lung epithelium and fibroblasts (Buch et al., 1994). PDGF-A regulates DNA synthesis and branching in cultured embryonic murine lung epithelium (Souza et al., 1995b). Pdgf-a−/− or Pdgf-α−/− mutants die perinatally with loss of alveolar myofibroblasts and SMCs and reduced parenchymal elastin fiber deposition; furthermore, failed alveolar septation is associated with emphysema (Bostrom et al., 1996, 2002). Antisense oligodeoxynucleotide abrogation of PDGF-B reduces the epithelial component of embryonic mouse lung explants but not branch number (Souza et al., 1994). PDGF-B and its receptor are crucial for vascular development during the alveolar phase (Lindahl et al., 1997). PDGF-C and D also dimerize and bind PDGF α or β receptor (LaRochelle et al., 2001; Li et al., 2000). PDGF-C mRNA expression shows a significant increase in bleomycin-induced lung fibrosis (Zhuo et al., 2003).
IGFs and their receptors are expressed in rodent and human fetal lung (Batchelor et al., 1995; Lallemand et al., 1995; Maitre et al., 1995; Retsch-Bogart et al., 1996; Schuller et al., 1995). Null mutant mice for the cognate type 1 IGF receptor (Igf1r) gene die at birth with respiratory failure and growth deficiency (45% of normal birth weight). Dwarfism is further exacerbated (70% of size reduction) in either Igf1 or Igf2 double null mutants or in Igf1r and Igf2 double null mutants. There does not appear to be a gross defect in branching morphogenesis per se; the lungs merely appear hypoplastic (Liu et al., 1993). IGF signaling may play a role in facilitating other peptide growth factor pathways during lung morphogenesis, e.g., IGF1R signaling is required for both mitogenic and transforming activities of EGFR (Coppola et al., 1994). Igf1-deficient mice have reduced airspaces, which are exacerbated in mice with additional leukemia inhibitory factor (Lif) deletion (featuring abnormal epithelium and decreased Sp3, Nkx2.1, and Sp-B expression) (Pichel et al., 2003). IGF1 is also trophic for fetal lung endothelial cells: in human fetal lung explants, IGF-IR inactivation results in endothelial cell loss, attenuates time-dependent increase in budding of distal airway, and increases mesenchymal cell apoptosis (Han et al., 2003).
Effective pulmonary gas exchange requires alignment between alveoli and pulmonary capillaries. VEGFs regulate pulmonary vascular development (reviewed extensively in Pauling and Vu, 2004 and in Warburton et al., 2000 and in Warburton et al., 2003; Shi et al., 2007) and signal through cognate receptors Flk-1 (fetal liver kinase-1, VEGFR2) and Flt-1 (fetal liver tyrosinase-1, VEGFR1) (Larrivee and Karsan, 2000). VEGF is regulated by hypoxia-inducible transcription factor-2α (Compernolle et al., 2002) and expression is controlled transcriptionally by hypoxia inducible factor-1 (HIF1) (see Dunwoodie, 2009 for an excellent review). Isoforms VEGF 120, 164, and 188 are expressed in E12.5 murine lung epithelial and mesenchymal cells and regulate endothelial proliferation and microvascular structure (Greenberg et al., 2002; Ng et al., 2001). As epithelial branching progresses, Vegf-A expression becomes restricted to distal lung (Healy et al., 2000), which may be partly due to the high affinity of VEGF-A for matrix components concentrated around branching tips (Acosta et al., 2001; Ng et al., 2001). Epithelial and vascular branching morphogenesis of cultured mouse embryonic lung features epithelial, mesenchymal, and endothelial crosstalk mediated in part by VEGF-A signaling via Flk-1 (Del Moral et al., 2006a). Inhibited VEGF signaling disrupts pulmonary endothelial survival and reduces postnatal alveolarization (Kasahara et al., 2000). Neonatal lungs treated with antibodies to Flt-1 are small with simplified alveoli (Gerber et al., 1999). VEGR3−/− mice have lymphatic hypoplasia and lethally delayed removal of lung liquid at birth. Excessive VEGF signaling also disrupts vascular and epithelial lungmorphogenesis: Vegf misexpression under SP-C promotercontrol yields decreased acinar tubules and mesenchyme (Zeng et al., 1998).
Roundabout (ROBO) is a receptor involved in repellent signaling and controlling axonal extension and with its ligand SLIT regulates non-neuronal cell migration (Wu et al., 2001). Suggesting interaction during perinatal lung development, Slit-2 is expressed in saccular mesenchyme whilst Robo is expressed in adjacent apical epithelium (Anselmo et al., 2003). A Robo knockout mouse loses alvelolar septation with thickened mesenchyme (Xian et al., 2001).
A class of noncoding RNAs called microRNAs (miRNAs) has been recognized for their abundance and conservation between species (Ambros, 2001; Lau et al., 2001; Lagos-Quintana et al., 2001). An intranuclear primary transcript (pri-miRNA) is cleaved by Drosha RNase III endonuclease to liberate the pre-miRNA, a 60–70 nt stem loop intermediate (Lee et al., 2002). Pre-miRNA is transported to the cytoplasm by Exportin-5 (Yi et al., 2003) where Dicer, another RNase III endonuclease, generates the mature miRNA. Mature miRNAs are incorporated as single-stranded RNAs into a ribonucleoprotein complex, the RNA-induced silencing complex (RISC) that downregulates gene expression by mRNA cleavage or translational repression. Mice lacking Dicer die before gastrulation indicating a fundamental developmental role of miRNAs. In lung, conditional epithelial Dicer inactivation arrests branching and disrupts Fgf10 expression (Harris et al., 2006).
Our studies (Carraro et al., 2009) and others’ (Lu et al., 2007) highlighted specific miRNAs’ importance in lung epithelial cell development. Ubiquitous developing epithelial overexpression of the miR-17–92 cluster augments multipotent Sox9-expressing epithelial cell number at the expense of delayed epithelial differentiation (Lu et al., 2007). In addition, miR-17 and its paralogs miR-20a and miR-106b target Stat3 and Mapk14 and thereby regulate E-Cadherin expression by modulating FGF10–FGFR2b downstream signaling. Thus, mir17 paralogs control a key pathway modulating the FGF10–FGFR2b–Sprouty-driven bud morphogenesis periodicity clock (Carraro et al., 2009; Warburton et al., 2008).
Differentially expressed protein components of extracellular basement membrane, laminins (LNs), entactin/nidogen, type IV collagen, perlecan, SPARC, and fibromodulin mediate cell–cell and cell–ECM interaction during lung morphogenesis. ECM components provide tissue support, may modulate cell proliferation and differentiation (Lwebuga-Mukasa, 1991), and serve as barrier and reservoir for growth factors. Preventing epithelial interaction with basement membrane disrupts lung development (Hilfer, 1996; Minoo and King, 1994).
LNs are glycoproteins involved in cell adhesion, migration, proliferation, and differentiation during tissue development and remodeling. LNs are composed of three chains, one central (α) and two lateral (β and γ), linked by disulfide bonds to form a cross-shaped molecule (Burgeson et al., 1994). To date five α, three β, and three γ chain isoforms have been identified, which suggests their combination can lead to approximately 30 LN variants (Bernier et al., 1995; Ehrig et al., 1990; Galliano et al., 1995; Iivanainen et al., 1995a,b, 1997a,b, 1999; Koch et al., 1999; Pierce et al., 1998; Vuolteenaho et al., 1994). The α1 chain is principally localized in basement membrane at the epithelial–mesenchymal interface with a predilection for specific zones. LN α1 chain also surrounds some mesenchymal cells. A domain in the cross-region of the α1 chain influences lung epithelial cell proliferation (Schuger et al., 1992). The α4 chain in LN8 and LN9 variants is highly expressed developing murine lung and heart (Frieser et al., 1997; Iivanainen et al., 1995a,b, 1997). LN α4 chain is localized around vessels in fetal lung and may assist organization of lung mesenchyme (Miner et al., 1997; Pierce et al., 1998). The α5 chain in LN10 and LN11 is abundantly expressed during lung morphogenesis (Miner et al., 1995, 1998; Pierce et al., 1998): mutated LN α5 chains impair lobe septation and bronchiolar branching in mutant murine lung.
With roles in cell adhesion, β1 and γ1 are constantly expressed during fetal lung development (Durham and Snyder, 1995): globular domains near their N-termini help regulate cell polarization (Schuger et al., 1995, 1996b). LN β2 chain isoform localizes to the basement membrane of prealveolar ducts, airways, SMCs of airways, and arterial blood vessels, as well as type II pneumocytes.
Nidogen (150 kDa) in basement membrane binds γ1 and γ3 chains helping link LN to collagen IV (Dziadek, 1995; Koch et al., 1999; Reinhardt et al., 1993). Nidogen is mesenchymally synthesized and appears to help basement membrane organization during lung morphogenesis (Senior et al., 1996). Blocking Nidogen’s interaction with LN disrupts lung development (Dziadek, 1995; Ekblom et al., 1994; Senior et al., 1996). Nidogen to degradation by matrix metalloproteinases (MMPs) may facilitate remodeling via basement membrane degradation (Mayer et al., 1993).
Proteoglycans comprise a protein core with sulfated carbohydrate side chains. They form flexible structures and function as a reservoir for growth factors, water, and ions. Inhibiting proteoglycan sulfation disrupts branching of E13 mouse lung explants and inhibits epithelial migration toward lung mesenchyme or FGF10-soaked beads (Shannon et al., 2003). Perlecan is a predominant basement membrane proteoglycan, composed of an approximately 450 kDa core protein with three heparan sulfate chains. Pulmonary perlecan synthesis rises sharply with increased fetal SMC proliferation (Belknap et al., 1999).
Fibronectin (FN) is essential for clefting during initial epithelial branching in salivary gland and is accumulated at sites of epithelial constriction and indentation in lung and kidney (Sakai et al., 2003), supporting roles for FN in lung branching morphogenesis (Roman, 1997). Indeed, FN is a Wnt target required in the control mechanism of lung bud tip splitting (De Langhe et al., 2005). Treating lung rudiments with anti-FN antibody or siRNA inhibited branching morphogenesis, while FN supplementation promoted branching (Sakai et al., 2003). Located in epithelium and mesenchyme, the EIIIA segment is one of the alternatively spliced FN segments, modulating FN’s cell proliferative effect: expression decreases from pseudoglandular to saccular stages, increases into the alveolar stage, and parallels changes in Proliferating Cell Nuclear Antigen (PCNA)-positive distal pulmonary cell number (Kikuchi et al., 2003).
MMPs are ECM-degrading enzymes that are inhibited by tissue inhibitors of metalloproteinases (TIMPs). MMPs may alter cell fate and behavior by ECM modulation and modulate signaling of bioactive molecules by their cleavage, by their release from bound stores, or by altering activity of their inhibitors (Vu and Werb, 2000). Timp3-null mutant mice have attenuated airway branching and alveologenesis (Gill et al., 2003) and develop progressive emphysema-like changes (Leco et al., 2001). MMP2, MMP9, and their tissue inhibitors (TIMP1 and TIMP2) are influenced by early postnatal dexamethasone exposure: Timp2 expression falls and Mmp9 expression increases. Such changes may contribute to steroids’ effects on neonatal lung structure (Valencia et al., 2003). MT1-MMP, a potent MMP2 activator, is a major downstream EGFR target. Egfr−/− mice had low MT1-Mmp expression with 10-fold reduction in active MMP2. Abnormal lung alveolarization in Mmp2−/− mice is similar to but less severe than that in neonatal Egfr−/− mice (Kheradmand et al., 2002).
Retinoids (all-trans, 9-cis, and 13-cis) are fundamental for development and homeostasis of numerous systems including lung, and there is a well-described mammalian RA synthesis and degradation system (Chambon, 1996). Retinaldehyde dehydrogenase-2 (RALDH-2) plays a prominent role in generating RA during organogenesis (Niederreither et al., 1997, 1999; Ulven et al., 2000). RA signaling is mediated by its nuclear receptors of the steroid hormone receptor superfamily: RAR (α, β, γ) and retinoid RXR (α, β, γ) (Chambon, 1996). RAR/RXR heterodimers also transduce RA signaling in vivo (Kastner et al., 1997). Within E13.5 lung, Rar-β isoform transcripts are localized to proximal airway epithelium and adjacent mesenchyme, whereas Rarα1, Rarα2, and Rarγ2 isoforms are ubiquitously expressed (Chazaud et al., 2003).
RA signaling is required for lung bud initiation. Acute vitamin A deprivation in pregnant rats at the onset of lung development results in blind-ending tracheae and lung agenesis in some embryos, which is similar to the case of Fgf10−/− mutant mice (Dickman et al., 1997; Sekine et al., 1999). Disruption of RA signaling in Rarα/β2 knockout mice causes left lung agenesis and right lung hypoplasia (Mendelsohn et al., 1994). Lung branching morphogenesis is characterized by dramatic downregulation of RA signaling. Preventing this by treating embryonic lung explants with high RA concentrations (10−6–10−5 M) disrupts distal budding with formation of proximal-like immature airways (Cardoso et al., 1995; Malpel et al., 2000). Continued RA activation by overexpression of constitutively activated Rara chimeric receptors meant lungs did not form saccules or identifiable type I cells: raised epithelial Sp-C, Nkx2.1, and Gata6 levels (but not Sp-A or Sp-B) at birth suggested differentiation was arrested early in these lungs. Downregulation of RA signaling is required to allow differentiation to form mature type I and II cells (Wongtrakool et al., 2003). RA inhibits expression and alters Fgf10 and Bmp4 distribution (Cardoso et al., 1995; Malpel et al., 2000). Pan-RAR antagonism alters Tgf-β3, Hnf-3β, and Cftr expression in proximal tubules and Bmp4, Fgf10, and Shh expression in distal buds (Chazaud et al., 2003).
In early E11–12.5 murine lung, Raldh-2 is concentrated in trachea (mesenchyme) and proximal lung (mesothelium) at sites of limited branching: this lack of overlap with Fgf10 suggests RA may restrict Fgf10 expression thereby defining the proximal–distal lung axis. However, during postnatal lung development, RA increases alveolar number, partially rescuing dexamethasone’s effects. In adult rats, RA reverses features of elastase-induced emphysema (Maden and Hind, 2003; Massaro and Massaro, 1996, 2000). RAR-γ-null mice have defective alveolar septation consistent with abnormal elastin deposition (Chytil, 1996). Additional deletion of a retinoid X receptor (RXRα) allele decreases alveolar surface area and number (McGowan et al., 2000). RA is a vitamin A metabolite. Vitamin A deficiency injures lung and impairs rat type II pneumocyte function (McGowan et al., 2000). This combined evidence suggests RA may have a complex role in alveolar development.
Mechanical stimuli to lung development have been long appreciated. Recent advances in molecular and stem cell biology allow these fields to be integrated with modern mechanobiology (Ingber, 2003). For example, ECM polymers, in addition to binding and presenting growth factors, provide resistance to deformation, fluid flow, and diffusion, and transmit force over surprisingly long distances. Adhesion molecules regulate cell motility and tissue structure, and these in turn interact through forces and deformations. Cytoskeletal components generate and transmit forces and conversely provide resistance to deformation. If we do not understand the relevant physics, we miss major factors in development, physiology, and regulation.
In an inflated rubber glove, raising internal pressure decreases curvature with simplification of form. By contrast, in lung increased internal pressure is associated with increased curvature and more complex morphology. Why this difference? Part of the answer is that the rubber glove is a thin elastic shell whilst embryonic tissue better resembles a viscoelastic fluid (Forgacs et al., 1998; Jakab et al., 2008). Over the timescale of growth, the elastic component of tissue viscoelasticity may be neglected; the tissue therefore behaves mechanically as a fluid, such that a simple mechanobiological model (Lubkin and Murray, 1995) predicts pulmonary pressure–morphology relationships (Unbekandt et al., 2008).
Mechanics also influence differentiation: in vitro mesenchymal stem cells differentiate toward neurons at 1 kPa, muscle at 10 kPa, and cartilage at 30 kPa (Engler et al., 2006). Hypothesizing that lung seeks to equilibrate tangential epithelial stress, a mechanobiological model of pseudoglandular lung (Lubkin and Murray, 1995) treated the epithelium as a viscous fluid with surface tension (Foty et al., 1994) to predict that branch size will be inversely related to pressure difference between external medium (and native mesenchyme) and lumen. Indeed embryonic lung epithelium appears to regulate tangential stress by modulating cytoskeletal tension via the Rho– ROCK system (Moore et al., 2005).
Human birth defects and in utero experiments have demonstrated lung development is subject to mechanics. For example, CDH (Smith et al., 2005) comprises a diaphragmatic defect, intrathoracic herniation of abdominal viscera and lung hypoplasia: affected newborns retain a high mortality rate due to inadequate lung growth. Traditionally, lung hypoplasia was attributed to lung compression by herniated abdominal viscera. Indeed lung growth is impaired when fetal CDH is created surgically (Starrett and de Lorimier, 1975). Similarly, human fetuses with renal agenesis or profound renal failure exhibit Potter’s syndrome, in which an underfilled amniotic cavity is thought to cause lung hypoplasia due to excessive lung fluid loss and/or fetal thorax compression. Certainly, bilateral fetal nephrectomy impairs ovine lung growth (Wilson et al., 1993). Alternatively, lung hypoplasia may result from developmental insults to the lung that precede or coincide with the origins of CDH and renal agenesis, respectively. For example, in the nitrofen-induced CDH model, early lung malformation precedes CDH (Jesudason et al., 2000). Similarly, lung hypoplasia emerges before fetal urine output normally contributes to amniotic fluid in a transgenic murine model of renal dysgenesis (Smith et al., 2006). Synthesizing these positions argues for an early developmental insult to the lung that is then compounded by unfavorable mechanical influences (Keijzer et al., 2000). In addition to extrinsic forces acting on fetal lung, a distending pressure is generated by lung liquid production. Draining this fluid by fetal tracheostomy is associated with lung hypoplasia (Fewell et al., 1983). Likewise, retention of this fluid in congenital laryngeal atresia is associated with lung overgrowth and distension (Harding and Hooper, 1996). This led to development of fetal tracheal occlusion to rescue hypoplastic lung growth in human CDH (Harrison et al., 2003; Hedrick et al., 1994). The normal fetal larynx appears to open only during diaphragmatic contraction (fetal breathing movements: FBMs), which restricts lung liquid efflux (Fewell and Johnson, 1983). Hence, failure of FBM in CDH may also contribute to lung hypoplasia. Experimental FBM abolition by phrenic nerve section is associated with lung hypoplasia (Miller et al., 1993). Loss of skeletal muscle formation also causes lung hypoplasia: thinned diaphragms in MyoD−/− mice cannot support FBM and the lungs are hypoplastic with reduced cell proliferation at E18.5 (Inanlou and Kablar, 2003). Neonatally, mechanical ventilation conspires with factors such as inflammation to generate BPD in premature newborns (Warburton et al., 2001). Mechanical factors appear influential beyond this period: compensatory lung growth follows lung resection (Thurlbeck, 1983) comprising lung distension and parenchymal growth. This postpneumonectomy effect suggests the lung responds to altered mechanics and that the organism to reduced alveolar surface area. At a smaller scale, airway smooth muscle (ASM) hypertrophy and hyperreactivity in asthma are associated with air trapping and acute lung distension; however, with time, this is associated with airway remodeling and chronic lung hyperexpansion. ASM-led airway occlusions in asthma may therefore have analogous effects to fetal tracheal occlusion (which distends and remodels prenatal lung) (Jesudason, 2007). Moreover, transient endogenous ASM-led airway occlusions occur in fetal lung (known as airway peristalsis), and this contractility may be an important regulator of lung growth (discussed below) (Jesudason, 2006a). With this in mind, we next focus on three areas of interest in lung mechanobiology: (i) lung liquid, (ii) airway contractility, and (iii) calcium signaling in this secretory, contractile environment.
Prenatal lung liquid is neither plasma ultrafiltrate nor “inhaled” amniotic fluid (Adamson et al., 1969). Lung liquid is produced throughout prenatal lung development by incompletely understood mechanisms that involve active Cl− transport from blood/interstitium into lumen (Olver and Strang, 1974). Intracellular Cl− accumulation is energized by the basolateral Na+/K+-ATPase (Bland and Boyd, 1986) and accomplished via Na+-linked cellular Cl− uptake through the Na+/K+/2Cl− co-transporter (Thom and Perks, 1990); indeed Cl− secretion rate depends on NKCC1 expression (Gillie et al., 2001). Movement of accumulated Cl− down its concentration gradient via apical Cl− channels results in accompanying Na+and water flux to generate fetal lung fluid (see Olver et al., 2004 for comprehensive review). Whilst active Cl− and fluid secretion are crucial to lung growth (Alcorn et al., 1977), they may not contribute to branching per se (Souza et al., 1995a).
The identity of the apical Cl− channel remains unclear. Numerous channels are demonstrated in fetal alveolar type II cells, including a G protein-regulated maxiCl channel (Kemp et al., 1994), cystic fibrosis transmembrane conductance regulator (CFTR) (McCray et al., 1993), at least one member of the Chloride Channel (CLC) channel family (Blaisdell et al., 2004; Murray et al., 1995), and a Ca2+-activated Cl− channel, TMEM16a (Rock et al., 2008). CFTR−/− mice have normal prenatal lungs (Wallace et al., 2008), suggesting CFTR plays no role in producing lung liquid or there is functional redundancy. Although a definitive link between CLC channels and lung liquid production remains to be established in vivo, there is evidence that CLC-2 contributes to fluid secretion and cyst expansion in vitro (Blaisdell et al., 2004). Interestingly, murine TMEM16a−/− mutants die of respiratory failure at an interval after birth with characteristic tracheomegaly and disruption of trachealis formation (Rock et al., 2008).
The rate of liquid production and the laryngeal valve function help determine hydraulic pressure in the lung. Obstructing the prenatal trachea increases intraluminal pressure two- to three-fold and airway branching three-fold; the rate of bud extension increases about two-fold whilst inter-bud distance is halved. These effects depend on FGF10–FGFR2b–Sprouty signaling (Unbekandt et al., 2008). Several studies have used tracheal obstruction to try to improve lung growth in human CDH (Harrison et al., 2003; Jani et al., 2005). However, clinical evidence of benefit of this potentially hazardous intervention remains limited. An alternative being explored is to exploit spontaneous airway occlusions that may be important for lung growth and perhaps avoid invasive fetal interventions (Jesudason, 2009).
Early mammalian airway exhibits spontaneous transient airway occlusions due to airway peristalsis. This is mediated by spontaneous ASM contractions that occur in birds and humans and which increase in frequency from embryonic stages to birth (Schittny et al., 2000). Peristaltic contractions and airway occlusions direct waves of fluid toward the lung’s tips. This results in rhythmic stretch and relaxation of growing buds (Fig. 3.8). Hence airway peristalsis and occlusions are well placed to regulate both pressure and stretch in the tips of developing lung (Jesudason, 2009). These ASM waves emanate from pacemaker areas in proximal airway before transmission distally (Jesudason et al., 2005). This pacemaker-driven airway contractility may even be important postnatally in asthma (Jesudason et al., 2006b). Thus, putative pulmonary pacemakers might be targeted for ablation by bronchial thermoplasty for asthma (Jesudason, 2009). Studying frequency of peristalsis in embryonic lung culture revealed that it is amenable to acceleration by cholingergic agents as well as growth factors (FGF10). These accelerated rates accompany enhanced in vitro lung growth. Similarly, in vitro inhibition of peristalsis is associated with reduced lung growth (Jesudason et al., 2005). This apparent coupling raised interest in mechanisms linking morphogenesis and peristalsis-led airway occlusions. In particular, Ca2+-imaging studies revealed that prenatal lung features spontaneous regenerative intercellular ASM calcium waves that propagate along main airways immediately prior to the wave of peristaltic contractility (Featherstone et al., 2005). Using pharmacological inhibitors, we showed that ASM calcium waves depend on extra- and intracellular calcium as well as gap junction integrity. Moreover, these calcium waves are abnormal in experimental lung hypoplasia (Featherstone et al., 2006). Thus, if peristaltic airway contractions do regulate lung growth, it means that underlying calcium oscillations govern lung development.
Airway peristalsis is coupled to lung growth, responsible for phasic lung stretch and underpinned by calcium oscillations. Transduction of such mechanical activity involves key modulators and sensors of serum Ca2+. For example, stretching alveolar type II cells produces parathyroid hormone-related protein (PTHrP) (Torday et al., 2002), which binds in paracrine fashion to its cognate receptor on adjacent adepithelial fibroblasts (Torday and Rehan, 2002); the latter are induced to differentiate into a lipofibroblast lineage (Schultz et al., 2002). The lipofibroblasts, discovered by Vaccaro and Brody (1978), protect against oxidant injury (Torday et al., 2001) and produce leptin, which stimulates synthesis of surfactant phospholipids and proteins, by ligating its receptor on alveolar type II cells (Torday et al., 2002a,b). Alveolar stretch releases PTHrP, which dilates alveolar capillaries (Gao and Raj, 2005) and coordinates surfactant production with alveolar perfusion. The Torday group showed that fluid distension permits cultured lung to differentiate like lung in vivo, agreeing with the concept discussed elsewhere that rhythmic mechanical distension serves as a clock or “zeitgeber” mechanism for organogenesis (Gross et al., 1978). Conversely, inhibiting lung fluid in explants inhibits the putative “zeitgeber,” while PTHrP treatment rescues the temporal molecular development of the lung. Fluid distension also stimulates endodermal Shh expression, which in turn stimulates the mesodermal Wnt pathway. Wnt increases endodermal PTHrP expression, which feeds back negatively on the Wnt pathway by stimulating protein kinase A. Wnt/β-catenin signaling has recently been linked to both PTHrP and the CFTR (Cohen et al., 2008).
Stretch-induced PTHrP regulates Ca2+ and lung growth, and stretch-inducing airway peristalsis is underpinned by Ca2+ signaling. However, the developing fetus is actually hypercalcemic compared to the adult, with free extracellular Ca2+ concentration ([Ca2+]o) of around 1.7 mM, while maternal [Ca2+]o is 1.1.–1.3 mM. Studies in mouse lung explant culture demonstrated fetal [Ca2+]o suppresses branching and cellular proliferation while enhancing Cl−-dependent fluid secretion (Finney et al., 2008). Since effects of fetal [Ca2+]o are mimicked by pharmacological activation of Ca2+-sensing receptor (CaSR), the G protein-coupled extracellular CaSR and fetal Ca2+ via CaSR may balance branching morphogenesis and lumen distension. CaSR expression has been documented in mouse (Finney et al., 2008) and humans (Finney, Wilkinson, Kemp, and Riccardi, unpublished observations), throughout the pseudoglandular stage. At the start of this phase, CaSR is restricted to the epithelium where its expression is maintained for at least 48 h in the organ culture. From E14.5 in mouse (and week 9 in humans) mesenchymal CaSR expression appears. CaSR expression wanes during the canalicular phase and is absent from adult mouse, rat, and bovine lung (Brown et al., 1993; Finney et al., 2008; Riccardi et al., 1995). CaSR-dependent regulation of branching and fluid secretion occurs via phospholipase C-dependent rises in intracellular Ca2+ concentration and phosphoinositide 3 (PI3) kinase activation (Fig. 3.9). Whether CaSR-dependent rises in Ca2+ affects lung development through peristaltic waves (discussed above) and/or controls growth by altering epithelial integrity via activation of PI3 kinase-dependent β-catenin signaling remains unclear. CaSR activation could also affect clock mechanisms of airway branching by interfering with FGF–FGFR–Sprouty signaling events: lungs cultured in 1.7 mM Ca2+ delay both bud extension and tip-splitting, when compared to lungs cultured in lower, adult levels of [Ca2+]o (1.2 mM) (Finney et al., 2008). Finally, patients with inactivating CaSR mutations have increased risk of interstitial lung disease (Auwerx et al., 1985a,b). Given that CaSR expression appears confined to the pseudoglandular phase, we need to test if this phenotype originates prenatally.
A stem cell describes a self-renewing, primitive, undifferentiated, multipotent source of multiple cell lineages. Such cells are critical for development and growth; pools of adult stem cells are hypothetical sources for tissue regeneration and repair as well as cancers. In contrast to embryonic stem cells and tumor cells, adult stem cells reduce telomere length with age (Warburton et al., 2008).
In the lung, there is limited knowledge about existence of self-renewing cells, whether such cells conform to classic or nonclassic stem cell hierarchies and whether a single stem/progenitor cell suffices to generate the more than 40 distinct cell types required in mature lung. At least five putative epithelial stem/progenitor cell niches are reported in adult mouse airway (Liu and Engelhardt, 2008), as well as endothelial stem cells in the pulmonary vasculature and ASM stem cells. Additionally, circulating stem/progenitors may lodge in the lung. Unlike skin and gastrointestinal tract, postnatal lung turns over slowly, which hampers putative progenitor identification. Specific markers and clonality assays also limit attempts at isolation. Herein we review new information regarding the development and function of lung stem/progenitor cells in organogenesis.
Failing regeneration and repair with age has been suggested to be due to stem cell failure. In pseudoglandular lung, tips of the branching tubules appear to contain undifferentiated multipotent epithelial progenitors. In adult lung, putative endogenous epithelial progenitors have been located in the basal layer of the upper airways, within or near pulmonary neuroendocrine cell rests as well as at the bronchoalveolar junction and in the alveolar epithelium (Engelhardt, 2001; Giangreco et al., 2002, 2004; Kim et al., 2005; Rawlins and Hogan 2006; Reddy et al., 2004; Reynolds et al., 2000, 2004). The distal-most epithelial cells were shown to be a multi-potent progenitor cell population during branching morphogenesis of the lung (Rawlins et al., 2009a). A recent model suggests, as the lung branches, descendents of distal tip progenitors are left to differentiate in the stalks, whereas self-renewing progenitors remain in the epithelial tips: distal epithelial cells have a unique gene expressionpattern, including high levels of Sox 9, Id2 (inhibitor of differentiation 2), N-myc, and Etv5/ERM (ets variant gene 5). In addition, they are exposed to high levels of FGF, Wnt, Shh, and BMP signaling (Bellusci et al., 1996; Liu et al., 2002; Shu et al., 2005). Many of these pathways including sox9, etv5, and FGFR signaling are associated with stem/progenitor cells in other endodermally derived organs (Seymour et al., 2007; Zhou et al., 2007). Distal epithelial cells also have different cell cycle kinetics compared with the rest of the epithelium; a higher proportion of them incorporate the thymidine analog bromodeoxyuridine (BrdU) during a short pulse (Okubo et al., 2005).
Candidate endogenous stem or progenitor cells have been identified in trachea and lung, using lung injury/repair models. For example, CC10/Scgb1a1+ Clara cells self-renew and proliferate after tracheal injury, but seem not the main source for tracheal epithelial regeneration (Rawlins et al., 2009b). However, subsets of Keratin-14 (K-14)-expressing BCs in the trachea (Hong et al., 2004a) can restore differentiated epithelium after injury and are distinct from bronchial BCs (Hong et al., 2004b). Lineage tracing in the adult mouse lung and trachea showed K-14-positive cells act as progenitors and ciliated cells cannot (Hong, et al., 2004a; Rawlins, et al., 2007). Human SpC and rat CC10 promoters have also been used for lineage tracing in lung (Perl et al., 2002, 2005a).
Since pseudostratified epithelium of mouse trachea and human airways contains a BC population expressing cytokeratin 5 (Krt5), a recent study using Krt5-CreERT2 transgenic mouse line for lineage tracing showed BCs of mouse trachea function as progenitors during postnatal growth, in adult homeostasis and also in epithelial repair of experimentally-induced SO2 damage (Rock et al., 2009). A clonality assay also found BCs of mouse and human airways self-renew and differentiate into mucus and ciliated lineages in the absence of stroma or columnar epithelial cells.
There is also a rare mixed population of pluripotent cells in lower respiratory tract characterized as a Hoechst dye effluxing side population (SP) cells. They express molecular markers of airway and mesenchymal origin (Giangreco et al., 2004). CD45− SP cells isolated from human tracheobronchial epithelium have proliferative potential. Increased numbers of these cells in asthmatic airways suggest that dysregulation of pluripotent cells may play a role in this chronic disorder (Hackett et al., 2008). In developing lung, some SP cells, which are CD45+ and CD45− have endothelial progenitor cell (EPC) potential in response to hyperoxia (Irwin et al., 2007).
The submucosal gland ducts in proximal airway are likewise suspected to contain stem cells (Liu and Engelhardt, 2008). However, relatively little is known about glandular stem/progenitor cells and their niche(s). Studies have suggested regenerating tracheal epithelium after naphthalene injury arises from cells migrating from gland ducts (Borthwick et al., 2001).
Rawlins et al. (2009b) used lineage tracing to circumvent obstacles that hampered earlier studies of lung stem/progenitor cells. Using the restricted expression of CC10/Scgb1a1, they generated a “knocking” transgenic mouse with a tamoxifen (TM) inducible Cre-recombinase (ScgB1a1-CreER™) that lineage-tags Clara cells of the airway. By varying dose and timing of TM administration, they discovered that epithelial reconstitution in the bronchioles involves Clara cell self-renewal and differentiation into ciliated cells. These data argue that bronchiolar ScgB1a1-expressing cells, largely mature Clara cells, are a self-renewing progenitor pool. The observation that lineage tags are chased into ciliated cells over time is consistent with early findings of Evans and colleagues (1978) that Clara cells are progenitors for ciliated cell renewal. On the other hand, Rawlins et al. (2009b) showed that lineage tags introduced into ScgB1a1-expressing cells of tracheobronchial airways were depleted within the ScgB1a1-expressing population over time. Collectively, these data suggest that ScgB1a1-expressing cells of proximal airways behave like transit amplifying (TA) cells, like those of intestinal epithelium, whereas ScgB1a1 cells of bronchiolar airways behave like self-renewing progenitors present in the interfollicular epidermis (reviewed by Chen et al., 2009).
A different approach by Giangreco and colleagues (2009) to investigate long-term behavior of airway progenitors in normal and injured airways showed in concordance with Rawlins et al. (2009b) that during homeostasis an abundant progenitor cell pool maintains the airway epithelium (rather than rare tissue stem cells). However, clonal patches of labeled cells emanate from tissue-specific stem cells located at airway branch points or bronchioalveolar duct junctions, after Clara cell depletion resulting from naphthalene exposure. In this naphthalene injury case, repairing bronchiolar airways more closely resemble the renewing epidermis after wounding, wherein stem cells are recruited from the hair follicle bulge to replace the depleted BC pool of the interfollicular epidermis (Zemke et al., 2009).
Varying dose and timing of TM administration, Rawlins et al. (2009a) discovered that reconstitution of bronchiolar epithelium involves Clara cell self-renewal and differentiation into ciliated cells and that Clara cells contribute to tracheal repair. Using lineage tracing, this study showed that a special population of BASCs which coexpress CC10 and SP-C, which have been proposed to contribute to both bronchioles and alveoli, has no apparent function during postnatal growth, adult homeostasis, or alveolar repair. Thus, they propose that trachea, bronchioles, and alveoli are maintained by distinct progenitor populations (Rawlins et al., 2009a). Currently, the significance that some Scgb1a1+ bronchiolar Clara cells express SftpC and some alveolar type 2 cells express Scgb1a1 is not understood. There is accumulating evidence the (Scgb1a1+, SftpC+) coexpressing cell population increases in number in murine lung cancer models (Ventura et al., 2007; Yang et al., 2008). However, it is unclear if this is due to preferential proliferation of preexisting (Scgb1a1+, SftpC+) cells or oncogenic upregulation of SftpC or Scgb1a1.
In a recent study (Tompkins et al., 2009), selective Sox2 deletion in Clara cells with Scgb1a1-Cre showed that Clara cell Sox2 is required for differentiation and/or maintenance of ciliated, Clara, and goblet cells in bronchiolar epithelium after birth and caused progressive loss of ciliated, Clara, and goblet cells and an inability to produce goblet cells in response to allergen. The findings indicate Clara cells can serve as common progenitors of ciliated, Clara, and goblet cells in a process requiring Sox2.
Epithelial progenitors of the alveoli have yet to be identified. An interesting model is that the alveolar progenitors are located in distal epithelial tips during the canalicular stage. However, there is no published evidence to support, or refute, this hypothesis. Presumably, because of its vast area a large number of AECs must function as a “ready reserve” to repair damaged alveolar surface. For instance, the expression of telomerase, a stem/progenitor cell marker, after acute oxygen injury is widely upregulated in AECs during recovery (Driscoll et al., 2000). This suggests that either AECs contain a progenitor cell subpopulation or that the majority of AECs undergo reactivation progenitor-like states after injury (Driscoll et al., 2000). In addition, without telomerase, resistance to injury and repopulation of damaged alveoli are compromised, indicating this pathway is likely critical for alveolar progenitor cell activity (Driscoll et al., personal communication). Moreover, Kim and colleagues (2005) described BASCs, which possess stem cell characteristics, are resistant to naphthaline injury and proliferate after airway or alveolar injury. Such BASCs reside near bronchioalveolar junctions and coexpress both alveolar (SP-C) and airway (CC10) epithelial cell markers, as well as coexpressing Sca-1. They are capable of self-renewal and differentiation into Clara cells and alveolar cells, and are also multipotent in clonal assays. Moreover, studies by Hong and coworkers (2001) identified variant Clara cells as endogenous lung stem cells, which infrequently proliferate during steady state but are held responsible for repopulating distal airway epithelium after injury. Variant Clara cells express Clara cell secretory protein, but survive naphthalene injury. As the lung continues to grow postnatally, Clara cells both self-renew and act as progenitors for ciliated cells, based on kinetics of cell labeling after a pulse of tritiated [3H]-thymidine (McDowell et al., 1985; Plopper, et al., 1992). This is supported by recent lineage labeling (Perl et al., 2005a). Whether all Clara cells have this capacity requires investigation. Moreover, type II cells proliferate and give rise to type I cells after adult alveolar injury, and this probably also occurs during postnatal growth (Evans et al., 1975). Several putative endogenous alveolar stem cell populations thus provide targets for directed regenerative therapies. Taking acute oxygen injury as an example, AECs undergo DNA and other forms of damage such as mitochondrial failure, glutathione depletion, and apoptosis (Buckley et al., 1998; Lee et al., 2006; Roper et al., 2004).
Several studies have shown that signals from lung mesenchyme are essential for branching morphogenesis. Mesothelium-derived FGF9 activates and controls FGF10 signaling from peripheral mesenchyme via FGFR2b, SHP2, Grb2, Sos, Ras, and Sprouty in epithelium (Del Moral et al., 2006b; Bellusci et al., 1997b; Tefft et al., 2002, 2005).
Distal Fgf10-expressing mesenchymal cells serve as progenitors for peripheral ASM (De Langhe et al., 2006; Mailleux et al., 2005; Ramasamy et al., 2007). Fgf10-lacz lineage tracing reveals ASM progenitors begin as Fgf10-expressing cells that, as the airway elongates, become distributed along peripheral airway. Transdifferentiation to express alpha–smooth muscle actin occurs under the control of SHH and BMP4, which are expressed proximal to the airway tip. Thus, increase in population size and localization of peripheral ASM progenitors occur early in development. Another population of ASM progenitors arise in proximal mesenchyme and advance peripherally (Shan et al., 2008).
Lung microcirculation is rich in progenitors, but our understanding of these is limited. Mesothelium overlying the lung contains progenitors that give rise to pulmonary vascular (but not airway) SMCs during embryonic development (Que et al. 2008). Endothelial progenitors arise from endogenous vascular wall or from circulating progenitors. Similar to lung epithelial cells, heterogenous pulmonary endothelial cells may require a site-specific niche (Clark et al., 2008); alternatively, putative resident endothelial progenitors may constitute a universal pool of progenitors that lack segmental specification (Blaisdell et al., 2009).
Distal airspace and vascular growth are coordinated so injury can affect both (Jakkula et al., 2000). Balasubramaniam et al. (2007) examined endothelial progenitors in BPD to show that hyperoxia disrupts alveolar and vascular growth, limiting surface area for gas exchange. In the lung, nitric oxide, VEGF, and erythropoietin contribute to mobilization and homing of EPCs. Several related developmental changes occur after hyperoxia in neonatal mice: expression of endothelial nitric oxide synthase, VEGF, and erythropoietin receptor and the number of EPCs in the blood, bone marrow, and lung were all reduced (Balasubramaniam et al., 2007).
Primitive capillaries surround the laryngotracheal groove as the lung buds from foregut and can be visualized by β-galactosidase expression under control of Flk1 promoter. This promoter is active and the earliest known marker of hemangioblasts. Under stimulation of epithelial VEGF, these hemangioblasts differentiate into a capillary network that surrounds bronchial, lobar, and segmental airways (Del Moral et al., 2006a; Ramasamy et al., 2007). Organization of this plexus appears essential for correct branching and perfusion. Thus, mesothelial–mesenchymal–epithelial–endothelial crosstalk matches epithelial and vascular progenitor function and will likely be essential for lung regeneration to succeed. Further studies are needed to define phenotypes of the pulmonary endothelial cell but also SMCs within the vasculature (Stevens et al., 2008).
Embryonic progenitors undergo symmetric and asymmetric divisions. To distinguish these, one can look at differences in spindle orientation or differential inheritance of cytoplasmic or membrane-bound proteins such as cell fate determinant Numb and atypical protein kinase C (PKC) (Huttner and Kosodo, 2005; Morrison and Kimble, 2006; Wang et al., 2009; El-Hashash and Warburton, unpublished data). Cells divide asymmetrically in response to extrinsic or intrinsic fate determinants: extrinsically, daughter cells placed in different microenvironments adopt different fates; intrinsically, cytoplasmic cell fate determinants (e.g., Numb) are asymmetrically localized within a cell and segregate differentially into daughters that adopt different fates (reviewed by Yamashita, 2009). Comparing progenitor numbers in mutant and sibling control lungs, we infer that certain molecules promote progenitor self-renewal or differentiation (Rawlins, 2008).
Several transcription factors and signaling molecules control lung growth and therefore probably affect progenitor cell proliferation. Thyroid transcription factor 1 (Ttf-1/Nkx2.1) expression marks lung lineage commitment in the early embryo and is critical for distal lung progenitor development (Kimura et al., 1999). Ttf1−/−-null mice have insufficiently differentiated lungs for survival (Kimura, et al., 1996). HMG box transcription factor, Sox9 is intensively expressed in distal epithelial progenitors from E11.5 to E16.5 (Liu and Hogan, 2002). However, lung-specific conditional deletion has no effect on progenitor cell behavior (Perl et al., 2005b). Sox9 may therefore act redundantly with other, as yet unknown, regulators: N-myc is also essential for maintaining a distal population of undifferentiated, proliferating progenitor cells, and may promote their self-renewal (Okubo et al., 2005).
In addition, several forkhead/winged helix (fox) family transcription factors have mutant knockout phenotypes and may promote lung epithelial progenitor proliferation. For example, conditional deletion of both foxa1 and foxa2 genes in lung results in small lungs with decreased cell division rates (Wan et al., 2005). A similar phenotype was reported after conditional deletion of both foxp1 and foxp2, which are enriched in the distal epithelial progenitors. In foxp22/2; foxp11/2 double mutants, the lungs are smaller than normal, with inhibited proliferation, but normal proximal–distal patterning (Shu et al., 2007). This suggests an essential role of fox transcription factors in the maintenance of the progenitor cell population and their self-renewing divisions.
Similarly, five key signaling molecules regulate many processes in embryonic development: Wnt, Notch, Hedgehog, FGF, and TGF-β family. Embryos lacking Wnt2/2b exhibit lung agenesis and do not express Nkx2.1, the earliest marker of lung endoderm. Endoderm-restricted deletion of β-catenin replicates this, suggesting canonical Wnt2/2b signaling is required to specify lung endoderm progenitors in the foregut (Goss et al., 2009, Harris-Johnson, 2009). FGF signaling plays an essential role in specification of distal lung lineages (De Langhe et al., 2008; Ramasamy et al., 2007) and others (Serls et al., 2005). FGF10 is expressed by lung mesenchyme and is a chemotaxin during morphogenesis. FGF10 overexpression maintains epithelial progenitor cell proliferation and leads to goblet cell metaplasia (Nyeng et al., 2008). In addition, FGF10 coordinates alveolar SMC formation and vascular development (Ramasamy et al., 2007). RA signaling is also essential for expansion of lung progenitors and formation of primary lung buds, by affecting Fgf10 expression through TGF-β signaling (Chen et al., 2007). Similarly, Shh in distal epithelium controls proliferation and branching and is believed to promote progenitor proliferation (Pepicelli et al., 1998). Autocrine Bmp signaling is likewise important for proliferation of the distal epithelial progenitor cell compartment. Wnt5a is also highly expressed around distal epithelial tips. Wnt5a−/− lungs have increased cell proliferation and an additional airway branch (Li et al., 2002), but it is unknown if this phenotype relates to defective progenitors. The details of how these signaling pathways regulate distal epithelial progenitor cells remain to be determined.
Recent studies suggest that Wnt and Bmp signaling controls proximal–distal lung patterning, but there is currently no evidence to confirm that this is mediated through progenitors. Shu et al. (2005) demonstrated that proximal–distal lung patterning depends on Wnt/β-catenin signaling and is mediated, in part, through regulation of N-myc, Bmp-4, and FGF signaling. Potentiation of β-catenin signaling in proximal airway results in arrested differentiation of immature bronchiolar stem cells, but β-catenin is unnecessary for adult bronchiolar stem cell maintenance (Zemke et al., 2009). Fortunately, reporters of Wnt pathway activity are highly active in distal lung epithelial cells. Recent studies suggested that Wnt signaling regulates proximal–distal patterning and progenitor proliferation independently, and that Wnt promotes distal airway fate at the expense of the proximal. (Mucenski et al., 2003; Shu et al., 2005). Shu and coworkers overexpressed Dickkopf-1 to inhibit Wnt pathway activity throughout developing epithelium: this expands proximal (conducting) airways at the expense of the distal, without effects on total levels of cell proliferation (Shu et al., 2005). Similarly, Mucenski et al. (2003) showed that lung-specific deletion of β-catenin abrogates distal epithelial differentiation. Notch signaling favors progenitor identity at the expense of differentiated phenotypes in different organs (Jadhav et al., 2006; Mizutani et al., 2007) and is also required for lung epithelial progenitors. Notch1 is highly expressed in distal epithelial progenitors during the pseudoglandular stage (Post et al., 2000). Notch controls cell fates in developing airways (Tsao et al., 2009), and arrests normal differentiation of distal lung progenitors before they initiate an alveolar program (Guseh et al., 2009). Notch misexpression in the distal lung prevented the differentiation of alveolar cell types (Guseh et al., 2009); expression of a constitutively active form of Notch3 throughout the developing lung epithelium prevents cell differentiation (Dang et al., 2003).
Furthermore, BMP signaling is also required for lung epithelium development, probably by promoting distal and repressing proximal cell fate. Inactivation of Bmp signaling by overexpression of a dominant-negative BMP receptor, or BMP antagonists Gremlin or Noggin, results in proximalization of lung epithelium (Weaver et al., 1999; Lu et al., 2001). Thus, reduction of BMP or Wnt signaling causes lung proximalization phenotypes (Eblaghie et al., 2006; Li et al., 2002).
At least 40 differentiated cell types emerge during lung organogenesis. Early trachea and esophagus are both lined with ciliated epithelium; following their septation, esophageal epithelium becomes squamous, while tracheal epithelium retains cilia. Primitive airway epithelium expresses several marker proteins including cGRP, Clara cell protein, and SP-A: its differentiation starts around E16 in mouse with emergence of pulmonary neurendocrine (PNE) cell rests, surrounded shortly after by Clara cells. In the periphery, AEC2 differentiation in E18 mouse is denoted by glycogen granules’ disappearance and emergence of surfactant-containing lamellar bodies with increased SP-C expression.
In mature lung, epithelial lineages are arranged proximodistally along the airways. Cartilage lies outside the submucosa and decreases in amount as bronchial caliber decreases; it is absent from bronchioles. The two major epithelial cell types in proximal bronchi are pseudostratified ciliated columnar cells and mucous (goblet) cells. Both arise from BCs, but ciliated cells predominate. Goblet cells begin to mature around 13 weeks’ gestation in humans (mature ciliated columnar cells are already present), express mucin markers (MUC5B, 5A, 5C), and release mucus granules into the airway which reduces drying and, through ciliary-driven cephalad mucus flow, cleanses the airway. In cystic fibrosis, mutation of the cystic fibrosis transmembrane conductance regulator (Cftr) gene disrupts expression of the encoded transmembrane Na+ ion transporter protein leading to thick mucus that overwhelms ciliary clearance and increases susceptibility to infection. In chronic airway injury, goblet cell hyperplasia may follow repair or experimental epithelial IL-9 exposure; the latter increases epithelial lysozyme and mucus production (Vermeer et al., 2003). IL-4, IL-13, and allergens enhance TGF-α release, which is a ligand for the EGFR that also stimulates goblet cell differentiation (Lordan et al., 2002).
There are three types of cells in bronchial submucosal glands. Myoepithelial cells surround the gland, while mucous cells (pale cytoplasm) and serous cells (basophilic cytoplasm) produce mucins. These secreted mucins mix with lysozyme and IgA on airway surface.
Kulchitsky cells are also found next to bronchial glands, but their function is unclear. It is believed they are pulmonary neuroendocrine cells (PNECs) producing peptides such as serotonin and calcitonin. Their cytoplasmic extensions usually reach the airway lumen. Kulchitsky cells expressing gastrin-releasing peptide (GRP), calcitonin gene-related peptide (CGRP), and chromogranin may be related to small cell carcinoma and carcinoid tumors. However, PNEC differentiate earlier by 10 weeks’ human gestation and are the first fully differentiated murine airway epithelial cells.
Clara cells reside in distal bronchiolar airway epithelium (normally lacking mucous cells) and produce mucus-poor, watery secretion. They emerge during the 19th week in humans and appear to assist with clearance, detoxification, and surface tension reduction in small airways. Clara cell-specific protein (CC10, CCSP, or uteroglobin) and cytochrome P450 reductase CC10 can be used as Clara cell markers. Whilst normal mice feature few mucin-positive cells in the airway, mucus metaplasia is associated with numerous Clara cell-derived mucous cells with excess mucin production or reduced secretion (Evans et al., 2004).
Most of the alveolar surface is covered by flat type I epithelial cells that are believed to be terminal differentiated and express several markers, such as T1a and aquaporin 5. T1a is developmentally regulated and encodes an apical membrane protein of unknown function. Absence of T1a protein blocks type I cell differentiation. Homozygous T1a null mice die at birth of respiratory failure with lungs that will not inflate normally (Ramirez et al., 2003). Aquaporin 5 is a water-channel in type I epithelial cells. Recently, a knock in of Cre-ERT2 into the Aqp5 locus has been reported (Borok, personal communication). These mice will be useful, not only to target gene deletion in type I cells but also for lineage tracing of the type I cells under development, injury, and repair.
Whilst type I epithelial cells cover more than 95% of the alveolar surface area, they account for only 40% of total AECs: the other 60% are rounded type II pneumocytes. These plump or cuboidal cells can regenerate and replace type I cells post injury and have finely stippled cytoplasm and surface microvilli. They manufacture surfactant phospholipids and proteins that modulate alveolar surface tension, such that despite varied size, alveoli remain open and end-expiratory atelectasis is reduced. Surfactant protein C (SftpC) is a commonly used type II cell marker. The four surfactant proteins, SP-A, B, C, and D play critical roles: SP-A and SP-D participate in airway host defense; SP-B and SP-C contribute to surfactant’s surface tension reduction (Whitsett et al., 2002).
Macrophages, although a small percentage of alveolar cells, are a major sentinel of host defense and derived primarily from blood monocytes; once in the lung, their turnover is extremely slow.
Stem and progenitor cells presumably help repair damaged lung, but identification of such cells remains problematic. The large surface area, numerous branches, and folded topography suggest lung harbors several stem or progenitor cell types. In trachea and bronchi, certain BCs and mucous-gland duct cells are believed to be stem/progenitor cells. Clara-like cells and type II pneumocytes are also thought to function as stem/progenitor cells in bronchioles and alveoli, respectively. Another population of stem/progenitor cells apparently lies at the bronchoalveolar duct junction (Kim et al., 2005). They have been proposed to function as bipotential precursors of both the SP-C- and CC10-expressing cell lineages.
It has recently been reported that bone marrow-derived mesenchymal stem cells can differentiate into airway epithelial cells and alveolar type I pneumocytes, particularly post-injury. By contrast, in vitro cell culture indicates Syrian hamster fetal lung epithelial M3E3/C3 cells differentiate into Clara cells and type II pneumocytes under different culture conditions. Whether CCSP-expressing cells with pre-Clara cell phenotypes are stem cells for the entire respiratory tract remains to be determined. In addition, the concept of a pluripotent stem cell for the whole lung needs to be further investigated due to the great differences between identified stem cell or progenitor candidates in proximal bronchi and distal alveoli. Recently, we discovered lung contains cells with stem or progenitor characteristics that can be FACS sorted from adult rat and mouse lung and which are relatively apoptosis-resistant and perhaps responsible for post-injury alveolar repopulation. Another such population sorted as “side cells” on FACS may repopulate several tissues including bone marrow.
ASM derives from at least two progenitor populations: one comes from the lung periphery and arises from Fgf10-expressing cells in subepithelial mesenchyme. By virtue of FGF10 expression, these cells first help mediate epithelial branching; however, as the airway elongates into peripheral mesenchyme, these progenitors remain to lie along the more proximal stalk of the distal bud. Here they differentiate into ASM, most probably under paracrine induction of BMP4 and Sonic hedgehog from the adjacent epithelium (DeLanghe et al., 2006). The second ASM progenitor population arises around the proximal large airways (Shan et al., 2008) and appears to meet with the distally-derived counterparts beyond major lobar and segmental branches. It is speculated that whilst size of such ASM progenitor populations are determined during embryonic airway branching, they may nevertheless determine susceptibility to later BPD and asthma. Moreover, maternal smoking may dysregulate ASM progenitors and their progeny via the cholinergic-agonist, nicotine.
Both FGF7 and inosine treatment ameliorate DNA damage in AECs, as well as enhancing mitochondrial protection and the ability of AEC to migrate and repair in an in vitro scratch assay (Buckley et al., 1997). FGF7 has also been evaluated by others in vivo as a treatment to enhance resistance to alveolar injury in animal models (Plantier et al., 2007; Ray et al., 2003). Also, FGF10 has a protective effect against lung injury and fibrosis (Gupte et al., 2009). We have also shown that inosine has protective properties against oxygen injury, including glutathione repletion, mitochondrial protection, decreased apoptosis, and increased VEGF expression (Buckley et al., 2005). Thus, it appears that protection or enhancement of alveolar progenitor cell function may be a viable therapeutic option that could possibly be evaluated in clinical trials of lung progenitor cell protection using small molecules such as inosine or FGF7 or FGF10.
Maturation of the surfactant system is one of two key steps to prepare fetal lung for air breathing. During the last 8 weeks of human gestation, fetal lung glycogen is converted into surfactant phospolipids, the most important of which is disaturated phosphatidylcholine (DSPC). This maturation is under the control of, and can be stimulated by, corticosteroids since it is blocked in mice with null mutations of glucocorticoid receptors or corticotrophin-releasing hormone. Human mutations have been found, such as surfactant protein B, that adversely affect stability of surfactant and hence the ability to maintain lung inflation.
The transition to air breathing occurs rapidly in mature neonatal lung. Immediately following severance of the umbilical circulation, a spike in catecholamine levels switches off chloride secretion and stimulates sodium/potassium ATPase (Brown et al., 1983; Olver and Strang, 1974; Olver et al., 1986). This replaces tracheal fluid production with its rapid absorption into lung interestiitum (and thence to lymphatic and capillary circulations). Null mutation of Na/K ATPase in mice leads to failure to absorb fetal lung liquid, which causes significant respiratory distress and even neonatal lethality (Hummler et al., 1996). In humans delayed lung liquid absorption manifests as transient tachypnea of the newborn.
From middle age in normal humans, an inexorable decline in lung function supervenes (illustrated by FEV1). By 120 years, FEV1 resembles that end-stage COPD in a younger person; hence, degenerating lung function appears currently to limit human life expectancy (Fletcher and Peto, 1977; Shi, W. and Warburton, D. 2010). Whilst some genetic mutations and/or environmental exposures fundamentally disrupt lung development and result in pre- or perinatal death, less critical leions may only be manifest as lung disease in infancy, childhood, or beyond. For example, minor genetic changes such as DNA polymorphisms may have very subtle impacts on lung organogenesis with apparently normal neonatal phenotype. Nevertheless, such lungs may have abnormal responses to subsequent environmental injury (e.g., cigarette smoke or vehicular pollution) that degrade lung anatomy and physiology faster than normal and predispose to, for example, COPD (Figure 3.10). Therefore, by understanding, protecting, and re-entraining developmental processes, amelioration or reversal of lung degeneration may permit enhanced duration and quality of life.
Appreciating that distal lung mesenchyme could trigger epithelial airway development has stimulated the search for controls of lung development. Given the mortality and morbidity of lung disease at all stages of life, lung regeneration is a global therapeutic priority. To achieve such goals, clinicians and scientists need to decipher how the lung is formed. Whilst this understanding began with histological analyses, advances in biology have allowed the “molecular embryology” of the lung to be elucidated. In parallel with this progress, lessons from human lung maldevelopment illustrate the importance of mechanical forces to normal lung growth. Such forces encompass both extrinsic factors (thoracic size, FBMs) and intrinsic ones (lung fluid, airway peristalsis, endogenous airway occlusions). Attempting to weave these diverse influences to facilitate regenerative lung growth appears a daunting task. Nevertheless, there are reasons for optimism: first, following Alan Turing’s insight, complex (lung) morphogenesis may arise via simple iterative biochemical signaling; secondly, Benoit Mandelbrot illustrated that simple mathematics can be applied to generate apparently complex form; thirdly, D’Arcy Thompson made clear that the set of genetically possible forms are vastly constrained by fundamental physical constraints; fourth, despite huge uncertainties about the regulation of lung development, regenerative medicine has already allowed transplantation of autologous tissue-engineered airway to aid patients. Hence, despite the structural complexity of the lung, its organogenesis is governed by simpler routines more readily susceptible to discovery and therapeutic exploitation. In pursuing the latter, we may similarly be reassured that physical constraints limit the possible structures we may engineer. Finally, despite all that we do not know, clinically important aspects of pulmonary regeneration can already be achieved. The challenge for the future will be the generation of more complex and vascularized structures that can ultimately support and/or replace impaired lung function.
We apologize to those colleagues whose important work in this field we have failed to cite.
Funding sources: National Heart, Lung and Blood Institute, National Institutes of Health, USA, National Science Foundation, USA, California Institute for Regenerative Medicine, Medical Research Council UK, Biotechnology and Biological Sciences Research Council, UK, Foreign and Commonwealth Office UK/USA stem cell collaboration grant, American Heart Association, American Lung Association, and Pasadena Guild of Childrens Hospital Los Angeles.
Editorial assistance: Zoe Ly and Theresa Webster.
Facilitation of US/UK collaborations on this review: UK Science & Innovation Network, British Consulate-General Los Angeles.