1. Introduction
Branching morphogenesis is characteristic of the development of many organs. Mammalian lung, kidney, mammary gland, pancreas and vasculature share common processes in organ formation. The process includes reiterative steps of duct outgrowth, bifurcation and formation of new branches (Hogan, Reference Hogan1999; Metzger & Krasnow, Reference Metzger and Krasnow1999; Hogan & Kolodziej, Reference Hogan and Kolodziej2002; Affolter et al., Reference Affolter, Bellusci, Itoh, Shilo, Thiery and Werb2003). At embryonic day (E) 9·5, mouse lung starts to develop as a diverticulum from the ventral side of the foregut. It has two layers: outside mesenchyme and underlying epithelium. The epithelial layer evaginates into surrounding mesenchyme and elongates rapidly giving rise to the trachea. At the very tip it splits into two buds, and they will emerge into the primary bronchus. The primary bronchi buds undergo outgrowth, extension and branching to give rise to the secondary bronchus. The airway goes through million times branching and finally fills the pleural cavity. Three simple branching modes are adopted including domain branching, planar bifurcation and orthogonal bifurcation (Metzger et al., Reference Metzger, Klein, Martin and Krasnow2008). The respiratory system is finally generated with conducting airways and a maximized gas exchange surface two days after birth.
The elaborate branching mechanism is genetically programmed. Each branching cycle has three sequential steps including bud outgrowth, tip arrest and branching. Common genes and signalling pathways, known as the “branching module”, are activated (Davies, Reference Davies2002). Embryonic lung development is regulated by several intercellular signalling pathways, such as the fibroblast growth factors (FGFs), bone morphogenetic protein (BMP) and Sonic hedgehog (SHH) (Bellusci et al., Reference Bellusci, Furuta, Rush, Henderson, Winnier and Hogan1997 a; Bellusci et al., Reference Bellusci, Grindley, Emoto, Itoh and Hogan1997 b; Celli et al., Reference Celli, LaRochelle, Mackem, Sharp and Merlino1998; Arman et al., Reference Arman, Haffner-Krausz, Gorivodsky and Lonai1999; Lebeche et al., Reference Lebeche, Malpel and Cardoso1999; Weaver et al., Reference Weaver, Dunn and Hogan2000; Mailleux et al., Reference Mailleux, Tefft, Ndiaye, Itoh, Thiery, Warburton and Bellusci2001). These signalling pathways interact with each other and form a network that controls the programming and accuracy of the reiterated branching cycle. Mesenchymal Fgf10 plays the dominant role to guide the epithelial layer outgrowth and branching (Sutherland et al., Reference Sutherland, Samakovlis and Krasnow1996; Bellusci et al., Reference Bellusci, Grindley, Emoto, Itoh and Hogan1997 b; Park et al., Reference Park, Miranda, Lebeche, Hashimoto and Cardoso1998; Lebeche et al., Reference Lebeche, Malpel and Cardoso1999). Moreover, other signal pathways, such as SHH and BMP4, are important to regulate lung morphogenesis (Bitgood & McMahon, Reference Bitgood and McMahon1995; Bellusci et al., Reference Bellusci, Henderson, Winnier, Oikawa and Hogan1996; Bellusci et al., Reference Bellusci, Grindley, Emoto, Itoh and Hogan1997 b; Lebeche et al., Reference Lebeche, Malpel and Cardoso1999; Miller et al., Reference Miller, Wert and Whitsett2001). Shh is expressed at the highest level at the distal tip of the epithelium bud, and regulates lung mesenchymal cell proliferation. Remarkably, over expression of Shh caused an increase in the ratio of interstitial mesenchyme to epithelial tubules (Bellusci et al., Reference Bellusci, Furuta, Rush, Henderson, Winnier and Hogan1997 a; Lebeche et al., Reference Lebeche, Malpel and Cardoso1999). Another key player, Bmp4, is localized at high levels in the distal tips of the epithelium and at lower levels in the adjacent mesenchyme (Bitgood & McMahon, Reference Bitgood and McMahon1995). Moreover, Bmp4 functions upstream to Shh by regulating Shh expression in developing mouse lung and tooth germ (Bellusci et al., Reference Bellusci, Henderson, Winnier, Oikawa and Hogan1996; Zhang et al., Reference Zhang, Zhang, Zhao, Yu, Hu, Geronimo, Fromm and Chen2000 a). These results suggest that reciprocal signalling between mesenchyme and the underlying epithelium is critical for lung development.
Sprouty (spry), the first identified FGF signalling antagonist, functions as an intracellular inhibitor of the receptor tyrosine kinase pathway in Drosophila (Hacohen et al., Reference Hacohen, Kramer, Sutherland, Hiromi and Krasnow1998). Interestingly, Sprouty2 is an intracellular protein localized to the cell membrane and binds Gap1 and Drk to inhibit Ras activation (Casci et al., Reference Casci, Vinos and Freeman1999). Upon growth factor stimulation, Spry2 translocates to the plasma membrane and becomes phosphorylated, which induces binding to the adaptor protein Grb2 to inhibit the recruitment of the Grb2–Sos complex to either the FGF receptor (FGFR) docking adaptor protein FRS2 or to Shp2. Membrane translocation of Spry is necessary for its phosphorylation, which is essential for its inhibitor activity (Hanafusa et al., Reference Hanafusa, Torii, Yasunaga and Nishida2002). Many studies have demonstrated that Sprouty genes play an important role in early organ formation, such as kidney development, auditory epithelium differentiation, regulation of odontogenesis, and limb and craniofacial development (Minowada et al., Reference Minowada, Jarvis, Chi, Neubuser, Sun, Hacohen, Krasnow and Martin1999; Hansen et al., Reference Hansen, Floss, Van Sloun, Fuchtbauer, Vauti, Arnold, Schnütgen, Wurst, von Melchner and Ruiz2003; Basson et al., Reference Basson, Akbulut, Watson-Johnson, Simon, Carroll, Shakya, Gross, Martin, Lufkin, McMahon, Wilson, Costantini, Mason and Licht2005; Shim et al., Reference Shim, Minowada, Coling and Martin2005; Klein et al., Reference Klein, Minowada, Peterkova, Kangas, Yu, Lesot, Peterka, Jernvall and Martin2006; Welsh et al., Reference Welsh, Hagge-Greenberg and O'Brien2007). As a common antagonist of FGF and EGF signalling pathways, Spry2 has been demonstrated to regulate airway branching in Drosophila (Kramer et al., Reference Kramer, Okabe, Hacohen, Krasnow and Hiromi1999). Mouse Spry2 is expressed in a domain adjacent to Fgf10 and is down regulated in the clefts between new lung branches (Tefft et al., Reference Tefft, Lee, Smith, Leinwand, Zhao, Bringas, Crowe and Warburton1999). FGF10 beads can up regulate the expression of mSpry2 in adjacent epithelium (Mailleux et al., Reference Mailleux, Tefft, Ndiaye, Itoh, Thiery, Warburton and Bellusci2001). Inhibition of mSpry2 using antisense oligonucleotides stimulates branching and increases epithelial proliferation (Tefft et al., Reference Tefft, Lee, Smith, Leinwand, Zhao, Bringas, Crowe and Warburton1999). In addition, over expression of mSpry2 leads to less branching (Mailleux et al., Reference Mailleux, Tefft, Ndiaye, Itoh, Thiery, Warburton and Bellusci2001). These results suggest that Spry2 might play important role during mouse lung formation.
The periodic outgrowth, tip arrest and branching of lung buds is precisely controlled by signalling networks from both mesenchyme and epithelium. The role of Spry2 within the lung branching signalling network remains unclear. In this study, we report that lung branching morphogenesis is disrupted in mice that have lost the Spry2 locus. Abnormal lung phenotype with reduced branching and enlarged terminal cystic structures was observed. We demonstrate that without Spry2, Fgf10 spatial distribution was disrupted and expression levels of Fgf10 and other signalling factors were altered. Furthermore, this finding provides novel evidence that a reiteratively formed signalling centre guides the development of embryonic lung and the signalling balance is critical to coordinate each branching cycle during respiratory airway formation.
2. Materials and methods
(i) Mice and production of Spry2-BAC transgene
36Pub deletion mice were maintained on a C57BL/6J genetic background. 36Pub mice were genotyped in a PCR assay using the deletion flanking markers D14Mit265 and D14Mit177 as previously described (Roix et al., Reference Roix, Hagge-Greenberg, Bissonnette, Rodick, Russell and O'Brien2001; Welsh et al., Reference Welsh, Hagge-Greenberg and O'Brien2007). Spry2ΔORF mice were genotyped as described (Shim et al., Reference Shim, Minowada, Coling and Martin2005). Spry2 BAC transgenic mice were genotyped and maintained as described (Welsh et al., Reference Welsh, Hagge-Greenberg and O'Brien2007).
(ii) Silicon cast injection and MicroCT data analysis
Silicon casts were injected into E18·5 wild-type and mutant lungs. The lungs were imaged via MicroCT. The image analysis microview of the CT data allowed the internal branch structure to be visualized two-dimensionally. Three-dimensional reconstructs of the branched architecture were made from the CT images using MATLAB software.
(iii) Histology and whole mount in situ hybridization
Embryonic lung tissue was dissected in PBS (4 °C) and fixed overnight in 4% paraformaldehyde and strained with Hematoxylin and Eosin following standard protocols. Whole mount in situ hybridization was performed as standard protocol and a detailed protocol is available upon request.
(iv) Quantitative RT-PCR analysis
E14.5 lung tissue was dissected and frozen in RNAlater (Ambion) and stored at −80 °C. Following the Invitrogen TRIzol instruction, total RNA was extracted using TRIzol. Quantitative RT-PCR analysis was performed on ABI7500 using TaqMan gene expression assays.
3. Results
(i) 36Pub mice exhibit lung branching defects and complementation test with Spry2-BAC transgene rescue with a hypomorphic phenotype
The piebald deletion mouse model is a collection of overlapping Mb-scale chromosomal deletions centred around the endothelia receptor B (Ednrb) coat colour spotting locus (O'Brien et al., Reference O'Brien, Metallinos, Chen, Shin and Tilghman1996). In a previous study, we have used these mouse models to annotate the function of a distal 6Mb region of mouse chromosome 14. Ednrbs-36Pub (hereafter 36Pub) has been show to lose several loci essential for mammalian development: Spry2 and two other genes, Ndfip2 and Rbm26 (Roix et al., Reference Roix, Hagge-Greenberg, Bissonnette, Rodick, Russell and O'Brien2001; Peterson et al., Reference Peterson, King, Hagge-Greenberg, Roix, Bult and O'Brien2002). Mice homozygous for Ednrbs-36Pub die from postnatal respiratory distress and craniofacial malformations (Rice et al., Reference Rice, Spencer-Dene, Connor, Gritli-Linde, McMahon, Dickson, Thesleff and Rice2004). Spry2 has been reported to play an important role in craniofacial differentiation, and auditory and tooth development (Shim et al., Reference Shim, Minowada, Coling and Martin2005; Klein et al., Reference Klein, Minowada, Peterkova, Kangas, Yu, Lesot, Peterka, Jernvall and Martin2006).
To annotate if Spry2 is the gene causing postnatal respiratory failure, we explored the development status of 36Pub homozygous lungs. At E18·5, 36Pub mutant lung was smaller in volume and round in shape missing the sharp curved edge of each lobe (Fig. 1(A), (C), (D) and (E)). To better characterize this phenotype we injected silicon casts into E18·5 wild-type and mutant lungs. Wild-type lungs exhibit curved edge and organized distribution of the terminal alveolar sacs, whereas 36Pub mutant lungs show apparent shape changes and characteristically dysmorphic terminal alveolar sacs (Fig. 1(a’) and (b’)). Additionally, cross sections show the main bronchus is larger than the wild-type lung (data not shown). Silicon casted lungs were imaged via MicroCT. The image analysis microview of the CT data allowed the internal branch structure to be visualized two-dimensionally. Comparison of wild-type and mutant lungs indicated the presence of large cystic structures throughout the bronchial tree of the mutant lungs. Three-dimensionally reconstructed branched architecture from the CT images confirmed the presence of large cystic structures within the mutant lungs observed in the two-dimensional CT slices (Fig. 1(F) and (G)). Interestingly, we found that the cystic structure is located at the termini of the main bronchi that define each of the five lobes of the mutant lung (data not shown). We next compared frontal sections of left lobe from E14·5 embryos. Clear morphologic differences of mutant lung were observed in the earlier embryonic stage with distal enlarged structure. These results imply that the abnormal structures might be related to the altered branching behaviour of 36Pub lungs.
36Pub mutant exhibits altered lung structure and Spry2-BAC trangene partially rescues the defects. Hypomorphic rescue by Spry2-BAC transgene of lung defects observed in 36Pub mutant mice (A–C). 36Pub mutant shows a round blunt edge losing the thin sharp transparent edge of wild-type (arrow) (C). Silicon casted 36Pub mutant lung exhibits altered lung structure (D–E). E18.5 wild-type (D) show normal lung structure with curved edge (arrow) and surface pattern of the terminal alveolar sacs (white box enlarged in a’), whereas E18·5 36Pub mutant lung loses those features (E and b’). Three dimensional reconstruction of MicroCT silicon casted lung reveals the presence of cystic structure (F and G). Five lateral secondary bronchi are labelled with 1–5. Cystic structure was also observed on a Hematoxylin and Eosin stained cross section of E14.5 36Pub lung (H–J).
The 36Pub deletion removes Spry2 and two other genes (Peterson et al., Reference Peterson, King, Hagge-Greenberg, Roix, Bult and O'Brien2002). As an FGF signalling antagonist, Spry2 is critical for craniofacial differentiation, and auditory and tooth development (Shim et al., Reference Shim, Minowada, Coling and Martin2005; Klein et al., Reference Klein, Minowada, Peterkova, Kangas, Yu, Lesot, Peterka, Jernvall and Martin2006; Welsh et al., Reference Welsh, Hagge-Greenberg and O'Brien2007). Based on the respiratory phenotype of 36Pub and our previous study, we prioritized Spry2 as a candidate gene (Rice et al., Reference Rice, Spencer-Dene, Connor, Gritli-Linde, McMahon, Dickson, Thesleff and Rice2004). To further investigate whether Spry2 was the gene, a Spry2-BAC transgene approach was carried out. In our previous study, the transgenic lines (Spry2-BAC-69) have a similar expression pattern to that of the endogenous locus; however, the level of the expression was as low as 25% of wild-type lung and palate levels (Welsh et al., Reference Welsh, Hagge-Greenberg and O'Brien2007). We tested if each transgene line could rescue the lung defects in 36Pub deletion mutant mice. Spry2-BAC transgenic mice showed rescue of the postnatal breath failure (data not shown). In addition, analysis of E18·5 lungs showed that Spry2-BAC transgenic lines partially rescued the lung defects associated with size, edge and terminal alveolar sac organization, which are less obvious severe phenotypes compared to 36Pub homozygous lungs (Fig. 1(B) and (I)). However, we observed small enlarged bronchioles at the distal region of each lobe (data not shown). The phenotype of Spry2-BAC transgenic lung is hypomorphic. This might be explained by the dose of the transgene.
To further evaluate if Spry2 is the candidate of the lung phenotype, a complementary cross of the 36Pub mouse with the Spry2 knockout mouse (Spry2ΔORF) was carried out. The results showed that Spry2ΔORF mutation fails to complement the lung branch defects associated with the 36Pub deletion. Using a Spry2 allelic series to study the effect of the dosage of Spry2 on lung development by comparing the organization and distribution of the terminal alveolar sac, we found the ranking of the phenotype to be from severe to light: Del/Del, Del/ORG, ORF/ORF, wild-type (data not shown). The results supported that Spry2 contributed to the 36Pub deletion lung branching phenotype and is likely the candidate gene within the critical interval that responds to the cystic structures. However, respiratory phenotype of Spry2 knockout is weaker compared to the phenotype of 36Pub mice. Spry2 knockout lungs exhibit the defect on the domain branching with an increased number of ventral branches (Metzger et al., Reference Metzger, Klein, Martin and Krasnow2008). It is possible that in the deletion region there are other genes or functional sequences also involved in lung development.
(ii) Cystic growth arises early in lung development
The defects of 36Pub lungs were further studied in detail by temporally reconstructing the branching process associated with the caudal extension of primary bronchus of the left lobe with serially staged (E12.5–E13.5) wild-type and mutant lungs (Fig. 2). In this study we used Etv5, a protein specifically expressed in the branching epithelium, as an epithelium marker to visualize the branching process. The asymmetric division of the terminal bud produces a major and minor bud. The major bud will elongate the main bronchus and the minor bud will grow into a lateral branch, respectively. The five lateral branches produced via this process correspond to the five secondary bronchi seen in the E18·5 lung (numbered 1–5 in Fig. 1 and Fig. 2). Comparison of terminal bud branching in wild-type and mutant lungs through the same developmental interval (E12.5–E13.5) demonstrates an abnormal and progressive expansion of the mutant terminal bud. The expansion of the terminal bud is accompanied with less branching. Over the same interval the wild-type terminal bud gives rise to two branches whereas the mutant undergoes expansion (Fig. 2(e–h) and (i–l)). Besides left lobe, similar cystic structure and delayed branching was also observed in other lobes (Fig. 2(m–r)). We were able to match the abnormalities in morphology observed in the E18.5 mutant lung to altered branching behaviour of the terminal bud in earlier developmental stage.
Cystic buds appear in the early stages of lung development. (A) Bud growth was observed using in situ hybridization on the left lobe of WT and 36Pub mutant lung (from E12.5 through E13.5) for the detection of Etv5 expression. Top panel shows branching process associated with the caudal extension of primary bronchus of the left lobe (a–d). The WT left main bronchi bud undergoes a series of five asymmetric branching events (numbered 1–5). Middle panel focuses on the fourth and fifth branching events and TB formation (e–h). Lower panel shows the branching process of 36Pub mutant lung, mainly focused on the expansion of the terminal bud (i–l). (B) Reduced lung buds were observed on in situ hybridization of caudal lobe of WT and 36Pub mutant lung (from E12.5 through E14.5) for the detection of Etv5 expression (m–r).
(iii) Delayed branch cycle was characterized with reduced branching event
Mesenchymal FGF10 signalling has been proven as a dominant force in leading the epithelial bud elongation during lung development (Bellusci et al., Reference Bellusci, Grindley, Emoto, Itoh and Hogan1997 b; Park et al., Reference Park, Miranda, Lebeche, Hashimoto and Cardoso1998; Lebeche et al., Reference Lebeche, Malpel and Cardoso1999). Spry2 is known to function as an antagonist of the FGF signalling pathway. Therefore, we expect to observe elevated FGF10 signalling in 36Pub lungs. Since FGF10 leads the lung bud growth and elongation we expected to observe more branches and larger lung bud formation. However, bud expansion with delayed growth was observed in 36Pub mutant lung (Fig. 2(a) and (b)). Quantification of branching events on whole-mount lung explants is carried out by counting the number of epithelial buds around the periphery of the lung explants. Interestingly, 36Pub deletion lungs have a significantly lower number of branches (Fig. 3). At E12.5 there are an average of seven lung edge buds in wild-type lungs versus four buds in the caudal lobe of 36Pub lung. By E14.5, wild-type caudal lobes have 25 edge buds whereas mutant only have 16. Reduced branching and cystic expansion suggested FGF10 signalling was altered and an unbalanced signalling network leads to a delayed branch cycle.
Cystic growth results in a reduced branch cycle. Counts of the outside edge buds from caudal lobe of WT and Spry2 mutant lung at E12.5, E13.5 and E14.5. We only count the lung buds localized to the outside edge of the lobe (data collecting from 30 WT and 31 36Pub mutant lungs).
(iv) FGF10 and FGF10 responsive genes were altered in the absence of Spry2
The reduction in number and increase in size of the 36Pub lung buds reveals that without Spry2, FGF10 signalling is changed more dynamically than simple reduction. To further evaluate the role of Spry2 during lung development we assessed the Fgf10 distribution pattern and expression level by double labelled in situ hybridization. At E13.5, Fgf10 localized to a restricted domain with an ordered lattice pattern in the mesenchyme surrounding the underlying epithelial buds (Fig. 4(A) and (C)). A similar pattern persists in the E14.5 (Fig. 4(E) and (G)). Surprisingly, the 36Pub deletion lung exhibits reduced FGF10 signal and loses the stereotypic lattice distribution pattern (Fig. 4(B), (D), (F) and (H)). Additionally, both Shh and Bmp4, two epithelial negative regulators of Fgf10, were extremely elevated compared to wild-type (Fig. 4(K) and (L)). To further evaluate the level of Fgf10 and other factors we carried out quantitative analysis using RT-PCR. Surprisingly, 36Pub mutants exhibited reduced Fgf10 but elevated Shh and Bmp4. The data reflects a dynamic change among the signalling pathways. When there is loss of antagonism of Spry2, Fgf10 is elevated in the earlier stage of lung development (data now shown), which leads to elevated Bmp4, Shh and other factors. Stronger local inhibition from SHH and BMP4 signalling might cause a weak FGF10 signal in the later stages. An altered balance between FGF10 and other signalling factors due to the loss of Spry2 causes delayed branch cycling in the developing lung and results in cystic growth.
Loss of Spry2 alters Fgf10 and other signalling factors. (A–H) 36Pub lung shows reduced Fgf10 (blue) production and changed spatial distribution pattern compared with wild-type lungs (E13.5 and E14.5). Double labelled in situ hybridization Fgf10 (blue) and NKX2.1 (pink) (A–H). Altered Shh and Bmp4 expression of caudal lobe in E13.5 lungs (I–L). (M) Quantitative RT-PCR analysis of Fgf10, Etv5, Shh and Bmp4 expression in wild-type and 36Pub mutant lung (E13.5). 36Pub mutants exhibit reduced expression of Fgf10 and elevated expression of Etv5, Shh and Bmp4 (error bars represent standard error of the mean).
4. Discussion
The embryonic respiratory tree is generated by million times branching events. Each branching event is dependent on restricting cellular responsiveness to signalling. Reciprocal signalling from the epithelium and surrounding mesenchyme interact to ensure the precise control of the branching program in both space and time during development. Signalling factors from the FGF10, BMP4 and SHH pathways are the most critical ones functioning in the branching program. In this study we have shown that Spry2 plays an essential role in lung development. Mice homozygous for 36Pub deletion exhibit postnatal respiratory failure and the mutant lung shows abnormal architecture of the bronchial tree and disorganized alveolar sac distribution. The Spry2 knockout allele fails to complement the 36Pub deletion phenotype. A Spry2-BAC transgene greatly rescues the deletion lung phenotype. Absence of Spry2 in the 36Pub deletion mutant lung causes cystic structures to form at the distal end of secondary bronchioles of each of the five lobes and significantly reduces branching events. The mouse model therefore represents a unique way to understand the role of epithelial Spry2 in the developing lung. Further analysis of mutant lungs showed cystic growth arises early in lung development, around E12.5. Consistent with the loss of SPRY2 antagonism of FGF signalling, the deletion mice showed an altered expression pattern of Ffg10, and also of FGF responsive genes such as Shh and Bmp4. The results in this study highlight the role of Spry2 in the regulation of embryonic lung development.
(i) Branch cycle is programmed by the mesenchyme–epithelium signalling network
Given the complexity of the architecture of the lung is remarkably stereotypic, the core developmental mechanism is reused many times to build up the bronchial tree (Metzger et al., Reference Metzger, Klein, Martin and Krasnow2008). The reiterative outgrowth and branching of an epithelial tube into the surrounding mesenchymal bed by coordinating the sequential steps of bud extension, tip arrest and branching, which is the three-step branching cycle (Fig. 5(A)), is the basic simplified reused mechanism. The process is tightly coordinated and controlled by the factors within the signalling centre at the distal tip of the lung bud. Mesenchymal FGF10 signalling from the distal signalling centre directs the outgrowth of the lung epithelial bud (Fig. 5(A)). In response to FGF10, epithelium cells at the tip of lung bud express multiple signals to antagonize FGF10, such as SHH, BMP4 and SPRY2. Two antagonistic strengths reach certain levels and cause tip arrest. The strongest FGF10 inhibition is at the distal bud tip, and this causes a local temporary relatively low level of Fgf10 compared to the adjacent mesenchyme area. Therefore, epithelium cell growth in this area is relatively slow and a branch cleft is formed. The distal signalling centre splits into two separate centres. The two newly developed buds will continue the branching cycle (Fig. 5(A)). The two newly formed FGF10 signalling centres lead to the outgrowth of the primary bud in two opposite directions while the bud is branching. Symmetric branching or asymmetric partitioning is determined by the relative location and strength of the two newly formed distal signal centres. The accuracy of the branching program is determined by the dynamic integration of inductive and antagonistic interactions among these pathways. The spatial localization of gene expression directs morphogenesis events. Many studies have suggested that disrupted signalling balance will result in lung malformation. Defects in Fgf10 or its epithelial FGF receptor 2 (Fgf2b) cause the trachea to be terminated as a blind sac (Celli et al., Reference Celli, LaRochelle, Mackem, Sharp and Merlino1998; Arman et al., Reference Arman, Haffner-Krausz, Gorivodsky and Lonai1999; Sekine et al., Reference Sekine, Ohuchi, Fujiwara, Yamasaki, Yoshizawa, Sato, Yagishita, Matsui, Koga, Itoh and Kato1999). Mutants of Shh and Bmp4 have been found with severe lung branching defects (Bellusci et al., Reference Bellusci, Henderson, Winnier, Oikawa and Hogan1996; Bellusci et al., Reference Bellusci, Furuta, Rush, Henderson, Winnier and Hogan1997 a; Litingtung et al., Reference Litingtung, Lei, Westphal and Chiang1998; Pepicelli et al., Reference Pepicelli, Lewis and McMahon1998; Lebeche et al., Reference Lebeche, Malpel and Cardoso1999; van Tuyl & Post, Reference van Tuyl and Post2000; Zhang et al., Reference Zhang, Zhang, Zhao, Yu, Hu, Geronimo, Fromm and Chen2000 a; Zhang et al., Reference Zhang, Yu, Zhang, Geronimo, Lovlie, Fromm and Chen2000 b; Miller et al., Reference Miller, Wert and Whitsett2001). Spatial localization and gene expression of signalling factors are critical in directing morphogenesis events.
Models of distal signalling centre regulation on branch cycle in the developing lung. Mesenchymal FGF10 signalling directs bud outgrowth and also induces the expression of Shh and Bmp4 from the epithelium. Shh, Bmp4 and Spry2 function together to antagonize the FGF10 signal and control the distribution pattern of FGF10. In the 36Pub mutant lung, absence of negative control of FGF10 signal by Spry2 results in an altered signalling balance. This leads to decreased and delayed branch cycles and cystic terminal buds. WT, Wild-type.
(ii) The signalling network is modulated by Spry2 in the developing lung
FGF10 signalling directs epithelium layer outgrowth and meanwhile Spry2 was found to be expressed at the corresponding epithelium adjacent to the distal signalling centre. Without Spry2, 36Pub homozygous mice have dramatically reduced and disorganized Fgf10 expression. Consistent with the branching cycle model (Fig. 5(A)), our data supported that during lung development the FGF10 signal is tightly regulated and organized. Multiple signalling pathways function together to provide spatial and temporal restriction upon FGF10 signalling. The cystic structure of 36Pub lung is caused by the altered expression and distribution of Ffg10 at the distal signalling centre.
The Spry2 gene suppresses branching during lung development, antagonizing Fgf10. Without the antagonism from Spry2, Fgf10 loses its tight control and the balance between bud outgrowth and branching is disturbed. At an early stage, loss of Spry2 and the elevation of the Fgf10 activated MAPK pathway results in a negative feedback loop that is activated by inducing it's negative regulators such as Shh and Bmp4 from the epithelium. From our gene expression data we observed that without Spry2, Fgf10 induced elevated expression of Bmp4 and Shh from E9.5 to E13.5. Elevated BMP4 and SHH negatively regulate Fgf10 and lead to reduced FGF10 signalling at E14.5. Continual inhibition of Fgf10 from elevated Shh and in the later lung development stage causes the delayed branch cycle. The fewer buds of 36Pub homozygous lung are the evidence of delayed branch cycle. The data also provided evidence that Spry2 is important for branching cleft formation (Fig. 5(B)). Branching morphogenesis relies on signalling thresholds that FGF10 and anti-FGF10 signalling will reach. Without Spry2, there is an altered signalling balance and it is relatively hard to reach the threshold and form the branching cleft. Therefore, the terminal bud continually grows and undergoes expansion. The branching cycle is delayed with cystic growth and reduced branching. Further studies are needed to test if the proliferation state of 36Pub mutant lung is altered. The prediction is that a less active proliferation state might be observed in 36Pub mutant lung. It is interesting to note that the branching cycle is delayed but not arrested. One possible explanation is that the altered branch cycle needs a longer time to reach the branching threshold.
Our data also support that domain branching is not equally bifurcated. We provide evidence that the cystic structure phenotype of 36Pub is a result of accumulation of delayed branch cycle. Domain branching, which affects the proximal–distal axis of each of the five lobes, is different from planar and orthogonal bifurcation and is not simply equal bifurcation (Metzger et al., 2008). The embryonic lung epithelial and mesenchymal stem cells, which mainly give rise to secondary branches of the respiratory tree, have memory. In the other words, during the domain branching L1, two newly formed branches are not exactly the same, one still remains as the elongation stem of the respiratory tree and this feature will also remain after L2 is formed. In our mouse model, as domain branching continues, 36Pub deletion lung phenotype accumulated in the elongation stem with relatively normal lateral branches. Several runs of the branching cycle in the early stage of development attempt to overcome the loss of Spry2. Lungs at E10.5 and E11.5 only show the delayed growth. The signalling balance, which is the axis that drives the branching cycle, is altered in the early stage because of loss of Spry2 and also further disrupted in the later stage due to altered Shh and Bmp4. In situ hybridization of E13.5 mutant lungs show evidence of growth of the cystic bud. The phenotype is accumulated as the cystic structure locates near to L5 and D4 of the left lobe (Metzger et al., 2008).
(iii) Timing and spacing of branching morphogenesis is coordinated by Spry2 in the developing lung
36Pub homozygous lung exhibits abnormal lung phenotype of cystic enlarged buds at the early stage of development and cystic structures located at the termini of the main bronchi of the five lobes with changed alveolar sac distribution. Many previous studies on the complexity and diversity of highly organized branched organs, such as lung, kidney and mammary gland, raise the question of what is the basic mechanism to develop these organs. One possibility is that a branch program is encoded in the genome and drives growth and branching. In other words, the morphologic complexity can be developed by repetitively using a simple branching mechanism (Metzger et al., Reference Metzger, Klein, Martin and Krasnow2008).
Based on our data, without the antagonism from Spry2, domain branching was affected more severely than planar bifurcation and orthogonal bifurcation. Similar to our findings, Metzger and colleagues found Spry2 null lung to show the effect on domain branching. However, interestingly, our data show that without Spry2 the branching events are reduced in the early stage of development and cystic growth accumulated at the terminus of each of the five lobes, whereas they found that the Spry2 knockout lung has a relatively weaker phenotype: the number of ventral branches is significantly increased in Spry2−/− relative to Spry2+/+ wild-type controls. Ectopic branches form earlier and proximal to normal ventral secondary branch (V1). The different phenotype of knockout and deletion lungs could be explained considering more function sequence is removed in 36Pub mutant mice. However, as a mouse model to study lung morphogenesis, 36Pub uncovered the general role of Spry2 to maintain the balance of the branching cycle during lung development and also the specific role that Spry2 plays in the domain branching. Our results support that the branching mechanism for each branch event is not exactly the same. The core program is similar but each parameter will be subtly tuned to produce the morphogenetic differences.
Acknowledgements
We thank Dr. Gail Martin for providing Spry2ΔORF mice. We also thank Ian Welsh and Randal Babiuk for technical assistance. This work was supported by National Institutes of Health (NIH) (T. P. O., grant number HD41066) and a Presidential Life Sciences fellowship from Cornell University (Y. Z., 2006–2007).
Declaration of interest
None.
