Activation of the Canonical Bone Morphogenetic Protein (BMP) Pathway during Lung Morphogenesis and Adult Lung Tissue Repair

Signaling by Bone Morphogenetic Proteins (BMP) has been implicated in early lung development, adult lung homeostasis and tissue-injury repair. However, the precise mechanism of action and the spatio-temporal pattern of BMP-signaling during these processes remains inadequately described. To address this, we have utilized a transgenic line harboring a BMP-responsive eGFP-reporter allele (BRE-eGFP) to construct the first detailed spatiotemporal map of canonical BMP-pathway activation during lung development, homeostasis and adult-lung injury repair. We demonstrate that during the pseudoglandular stage, when branching morphogenesis progresses in the developing lung, canonical BMP-pathway is active mainly in the vascular network and the sub-epithelial smooth muscle layer of the proximal airways. Activation of the BMP-pathway becomes evident in epithelial compartments only after embryonic day (E) 14.5 primarily in cells negative for epithelial-lineage markers, located in the proximal portion of the airway-tree, clusters adjacent to neuro-epithelial-bodies (NEBs) and in a substantial portion of alveolar epithelial cells. The pathway becomes activated in isolated E12.5 mesenchyme-free distal epithelial buds cultured in Matrigel suggesting that absence of reporter activity in these regions stems from a dynamic cross-talk between endoderm and mesenchyme. Epithelial cells with activated BMP-pathway are enriched in progenitors capable of forming colonies in three-dimensional Matrigel cultures. As lung morphogenesis approaches completion, eGFP-expression declines and in adult lung its expression is barely detectable. However, upon tissue-injury, either with naphthalene or bleomycin, the canonical BMP-pathways is re-activated, in bronchial or alveolar epithelial cells respectively, in a manner reminiscent to early lung development and in tissue areas where reparatory progenitor cells reside. Our studies illustrate the dynamic activation of canonical BMP-pathway during lung development and adult lung tissue-repair and highlight its involvement in two important processes, namely, the early development of the pulmonary vasculature and the management of epithelial progenitor pools both during lung development and repair of adult lung tissue-injury.


Introduction
Mammalian lungs are designed to optimize exposure of blood to oxygen. To achieve this, two intertwined and highly branched tree-like tubular systems, one conducting air and the other conducting blood must develop in a coordinated way to generate the millions of functional alveolar gas-exchange units [1,2,3].
Prominent role in the regulation of lung development and homeostasis is played by members of the Bone Morphogenetic Protein (BMP) family [8]. BMPs, like all other members of the TGFb superfamily, signal via specific membrane receptors that have serine-threonine kinase catalytic activity [9]. Functional BMP receptor units are composed of two Type-I and two Type-II receptor polypeptides. Four different Type-I BMP receptors (ALK2, ALK3/BMPRIa, ALK6/BMPRIb and ALK1), and three Type-II receptors (BMPRII, ActRIIA and ActRIIB) have been identified [10]. Upon ligand binding, the constitutively active Type-II receptors phosphorylate and thus activate their Type-I partners, which in turn phosphorylate their intracellular targets, the receptor-regulated Smad proteins 1, 5 and 8. Phosphorylated Smads form complexes with the ''common'' Smad4 and translocate to the nucleus where they regulate expression of their target genes, synergistically with other transcription factors [8,11]. BMPs can also signal via Smad-independent intracellular pathways that involve mitogen-activated protein (MAP) kinases [12,13].
Several studies using transgenic and conventional or conditional knock-out mice have clearly demonstrated the key role played by BMPs during early lung development [14,15,16,17,18,19,20,21]. Disruption of BMP signaling by ectopically expressing the BMP antagonists noggin or gremlin in the lung epithelium [15,22], inactivating BMP receptors [16] or expressing a dominant negative form of the BMP Type-I receptor (dnALK6) result in abnormal distal lung architecture. Remarkably, over-activation of the BMP pathway is also incompatible with normal lung development. Ectopic over-expression of Bmp4 in the epithelium leads to smaller lungs and to substantially reduced epithelial cell proliferation [14] and mice with deletion of the BMP antagonist Follistatin-Like 1 (Fstl1) gene die at birth from respiratory distress and show multiple defects in lung development [23,24].
Moreover, increasing evidence supports a key role for BMP signaling angiogenesis and vasculogenesis in the lung [25,26]. A large number of genetically modified mice with lesions in genes encoding either BMP ligands, receptors or antagonists exhibit defective angiogenesis. The importance of BMP signaling for the vascular system is demonstrated by the association of mutations in genes encoding for BMPRII and ALK1 with the development of two genetic vascular diseases, namely, pulmonary arterial hypertension and hereditary hemorrhagic telangiectasia.
The BMP signaling pathway has also been implicated in the regulation of adult lung homeostasis and tissue repair following injury [27,28,29,30,31]. Several BMP ligands have been found upregulated in allergen challenged lungs [27] and notably, ectopic expression of gremlin by adenovirus mediated gene transfer in the lung of adult rats causes severe pulmonary fibrosis [32] illustrating the importance of the BMP pathway for lung homeostasis as well. Consistently, over-expression of gremlin has been observed in humans suffering from Idiopathic Pulmonary Fibrosis (IPF) [28,29].
Despite the significant progress, the precise mechanism(s) by which BMPs regulate lung development have not been fully delineated [5]. Moreover, characterization of the mechanisms by which BMPs affect adult lung homeostasis and tissue repair is still rudimentary. The current study was based on the premise that identification of the actual cellular targets of BMP mediated signaling during lung development, homeostasis and adult lung tissue repair will facilitate interpretation of earlier genetic studies, guide rational targeting of BMP signaling components in future experiments and contribute to the clarification of the mechanisms of action of this signaling system. Therefore, utilizing a mouse transgenic line carrying the eGFP reporter gene under the control of a canonical BMP pathway responsive regulatory element [33] we have constructed a detailed spatiotemporal map of canonical BMP signaling in the lung and identified putative cellular targets of this signaling pathway during lung development and adult lung tissue-injury repair. Our studies suggest that canonical BMP pathway plays key roles during early development of the pulmonary vasculature and the management of lung epithelial cell progenitors during late lung development and repair of adult lung tissue injury.

Ethics Statement
Animals were housed in individually ventilated cages under specific pathogen free conditions, in full compliance with FELASA (Federation of Laboratory Animal Science Associations) recommendations, at the Animal House Facility of the Foundation for Biomedical Research of the Academy of Athens (Athens, Greece). All procedures for care and treatment of animals were approved by the Institutional Committee on Ethics of Animal Experiments and the Greek Ministry of Agriculture (Permit Number: K/1054).

Animals
BMP-responsive eGFP expressing (BRE-eGFP) mice [33] kept in the C57BL/6 background maintained on a 12:12 hours light:dark cycle at the Animal Facility of the Foundation for Biomedical Research of the Academy of Athens.

RNA isolation and real-time quantitative PCR analysis
Total RNA from mouse lungs was isolated using the Tri-Reagent protocol (Sigma, St. Louis, MO) and the yield and purity of RNA was determined electrophoretically and spectrophotometrically. After DNase treatment with RQ1 RNase-Free DNase (Promega, M160A), 2 mg of RNA were reverse transcribed into cDNA using M-MLV Reverse Transcriptase (Promega, M170B) and random primers (Invitrogen, 58875) according to the manufacturer's instructions. The primer pairs for real-time PCR were designed using Beacon Designer v7.01 software (Premier Biosoft International, Palo Alto, CA). Sequences for the primer pairs used are given in table S1. The PCR reactions contained SYBRH Green ER TM qPCR Super Mix Universal (Invitrogen, 11762-500), 200 nM of each primer (Invitrogen, Carlsbad), and 0.2 ml of cDNA template in a 20 ml reaction volume. RT-qPCR cycling parameters were initial incubation at 50uC for 2 minutes, denaturation at 95uC for 10 minutes followed by 50 cycles of 15 seconds at 95uC and 40 seconds at 60uC. Data collection was carried out using a Chromo4 Real-Time PCR detector (BioRad) and analyzed with Opticon Monitor 3 expression analysis software (BioRad). Relative levels of mRNA expression were calculated according to the DDCt method [34].
In-vivo lung tissue-injury/repair models Female BRE-eGFP animals, between 12-16 weeks of age were treated with naphthalene or bleomycin. For naphthalene treatment, 300 mg/kg naphthalene in corn oil (vehicle) or vehicle alone was administered by intra-peritoneal injection. Lung tissues were collected on days 2, 3, 6, 9 post naphthalene-treatment. For bleomycin treatment, bleomycin (0.033 Units/mouse, Nippon Kayaku co ltd, Bleocin) diluted in PBS or PBS alone (vehicle) was administered intra-tracheally in 30 ml total volume. Lungs were collected on days 2, 4 and 12 post-bleomycin-treatment.
In vitro cultivation of sorted epithelial cells in Matrigel was done as previously described [38] with minor modifications.  Selenium (Gibco, 51300, 1:100), Pen Strep (Gibco, 15070, 1:100), amphotericin B (Gibco, 15290, 1:1000) and 10 mM SB431542 (Sigma-Aldrich, S4317). Culture medium was changed every other day until day 8. SB431542 was then removed and the cultures were maintained for an additional four days before fixing overnight at 4uC with 4% PFA in PBS. The Matrigel was removed by freezing-thawing and rinsing with fresh PBS. The content of the Transwell inserts was embedded in OCT (Shandon Cryomatrix) and 10 mm sections were prepared by cryostat (Leica CM1950).
Although epithelial colonies developed in Matrigel even in the absence of SB431542 its presence resulted in higher number of larger colonies and therefore this culture system was preferred. Treatment of the Mlg cells with SB431542 for 24 hours led to substantial increase in FGF10 mRNA synthesis (data not shown) suggesting that stroma cell derived FGF10 may be part of the underlying mechanism.

Results and Discussion
Spatiotemporal pattern of expression of the BRE-eGFP reporter during lung development Previous studies utilizing two independent canonical BMP pathway reporter mouse lines, one harboring a BRE-eGFP [33] and another harboring a BRE-lacZ allele [39] have demonstrated expression of the BRE-eGFP and BRE-lacZ reporters respectively in the epithelium of trachea and primary-bronchi already at around E10-E12.5. To extend these findings and map in detail the spatiotemporal pattern of canonical BMP signaling in the lungs, we collected lung tissues from the BRE-eGFP reporter mice at embryonic days E11, E12, E13.5, E14.5, E16, E17.5, E19.5 and postnatal days 1, 15 and 2 months and analyzed tissue sections for eGFP expression. Our analysis demonstrated that the BRE-eGFP reporter was active already at E11, reached maximal levels around birth, thereafter declining to minimal levels in the adult ( Figure 1 and Figure S1).
Around E11, eGFP expression was confined to the endothelial cells of the large pulmonary vessels, and the developing vascular network in the parenchyma. A very small fraction of sub-epithelial smooth muscle cells (SMCs) expressed very low levels of eGFP at E12 ( Figure S1). However, from E13.5 onwards robust eGFP expression was detected in sub-epithelial SMCs located at the proximal regions of the developing airways (Figures 1 and Figures  S1 and S2).
From E16 onwards, expression of eGFP was detected in both proximal and distal epithelial compartments.In the proximal epithelium (i.e. the developing airways) eGFP expression exhibited a proximal-distal axis pattern, with high eGFP expression in bronchi and large airways and surprisingly no expression at the distal portion of the growing airway tree. Interestingly, isolated clusters of eGFP pos cells were observed at the tips of airway branch-points. In the distal lung compartment, eGFP expression was detected in cuboidal epithelial cells of the developing saccules ( Figure 1B). As lung morphogenesis approached completion, the eGFP signal was reduced and in adult mice,(,2 months old), few eGFP pos cells were detected scattered in the epithelium of bronchi and large airways, in the alveolar compartment and in a subpopulation of cardiac cells in the outer layer of the tunica media of the pulmonary veins ( Figure 1B) [40].

Expression of the BRE-eGFP reporter in the developing lung mirrors faithfully activation of the canonical BMP pathway
Although the capacity of BRE element to monitor BMP mediated transcriptional activation in-vivo has been demonstrated previously [33,39,41], additional studies were undertaken to further increase confidence with respect to transgene expression in the lung. The kinetics of mRNA expression for eGFP and the BMP target gene Id1 were analyzed by quantitative PCR. EGFP and Id1 mRNA levels followed similar kinetics to eGFP protein expression, reaching maximum levels around birth ( Figure 2A). P1 lungs were therefore enzymatically digested and lung cells were separated by cell sorting on the basis of eGFP expression into eGFP negative (eGFP neg ), eGFPlow (eGFP low ) and eGFP high (eGFP high ) cells ( Figure 2B). Western blot analysis of protein extracts from the isolated population demonstrated good correlation between expression of eGFP and levels of pSmad1/5/8 ( Figure 2C, D). Analysis of the mRNA levels for eGFP and BMPregulated target genes such as Smad6, Id1 and Id3 demonstrated very good correlation between the level of eGFP expression and the activation of known BMP targets ( Figure 2E). Interestingly, expression of Id2 mRNA did not correlate with levels of eGFP expression. Furthermore, treatment of embryonic lung explants exvivo with either BMP-or TGFb-receptor inhibitors resulted in either suppression or augmentation of BRE-eGFP transgene expression, respectively (see below Figure 3B). Moreover, treatment of embryonic lung explants with FGF10, a known inducer of BMP4 production by distal endodermal cells, or recombinant BMP4 resulted in substantial up-regulation of eGFP expression ( Figure S3). Treatment of the lung explants with purmorphamin, a Shh agonist, resulted in redistribution of the eGFP staining pattern that involved a small increase of eGFP expression in a narrow zone of sub-mesothelial cells and a reduction of eGFP expression in the more central parts of the mesenchyme. This pattern of expression could stem from a Shh mediated reduction in FGF10 synthesis by mesenchymal cells [35] that in turn leads to a more confined BMP expression zone.
Too further validate the BRE-eGFP reporter animals and assess its sensitivity in comparison to staining tissues for nuclear pSmad1/5/8 by immuno-histochemistry, we stained adjacent tissue sections of P1 lungs, the developmental stage demonstrating maximal BRE-eGFP reporter activity, with anti-eGFP and anti-pSmad1/5/8 antibodies. As shown in Figure S4, nuclear pSmad1/5/8 was detected in regions with intense eGFP immuno-staining demonstrating consistency between BRE-eGFP reporter activity and pSmad1/5/8 immuno-staining ( figure S4). Unfortunately, in our hands the anti-pSmad1,5,8 reagents we have analyzed so far were unable to detect pSmad1/5/8 in lung tissues from embryos before stage E19.5 and thus we were not able to confirm correlation of eGFP and pSmad1/5/8 at earlier developmental stages. A number of studies have indicated that the value of measuring pSmad1/5/8 as the sole indicator for canonical BMP pathway activation is questionable since presence of pSmad1/5/8 does not necessarily lead to transcriptional activity of BMP target genes. Canonical BMP pathway activity is known to be regulated by cell type-specific Smad-interacting transcription factors, co-activators and co-repressors. For example Tbx1 can bind to Smad1, interfere with Smad1/Smad4 interaction and decrease BMP transcriptional activity without affecting the amount of pSmad1/5/8 in the responding cell [42]. Furthermore, Leeuwis et al. [41] demonstrated that in the kidney of BRE-eGFP animals, whereas the expression pattern of BRE-eGFP exhibited excellent correlation with the expression of the BMP7 and the BMP target genes Id1, and Smad6, in contrast pSmad1/5/8 exhibited a brought expression pattern that did not correlate with the expression of BRE-eGFP, BMP7, Id1 or Smad6. Therefore, a safer assessment of canonical BMP pathway should combine analysis of several known BMP target genes and pSmad1/5/8 immuno-stainings when this is technically possible. Collectively, the analysis by Monteiro et al. [33], Leeuwis et al. [41] and the results described herein indicate that the BRE-eGFP transgenic line reports quite faithfully activation of the canonical BMP pathway in-vivo and ex-vivo and thus it provides an additional method for assessing potential canonical BMP pathway activation.
Canonical BMP-pathway is activated in the developing vascular and sub-epithelial smooth muscle networks during early lung development To identify the putative cellular targets of canonical BMP signaling, tissue sections collected from lungs at different developmental stages were stained with antibodies specific for appropriate cellular markers. Double immunostaining for eGFP and CD31 (PECAM-1), a surface marker of mature endothelium, or VEGFR2, a surface marker of vascular progenitor cells ( Figure  S1), demonstrated that the majority of CD31 pos or VEGFR2 pos cells at E12, E14.5 and E17 were also eGFP pos ( Figure 3A), confirming that the network of eGFP pos cells in the parenchyma is composed of vascular cells and that the canonical BMP pathway is activated during the establishment of the vascular network in the developing lung. At P1, ,40% of the CD31 pos cells were eGFP pos , from P15 onwards the number of eGFP pos -CD31 pos cells declined and in the adult they were undetectable. EGFP expression with similar kinetics was observed in the endothelial cells lining the large blood vessels ( Figure 3A). Very few eGFP pos alpha smooth muscle actin (aSMA) double positive cells were found in the walls of the vessels ( Figure 3A, white arrow in P1 vessel).
To validate the functional significance of BMP signaling for the development of the vascular network of the developing lung, ex-vivo cultures of E12 lung explants were treated with the respective BMP-or TGFb/Activin-receptor inhibitors LDN193189 and SB431542. Treatment of lung explants for 24 hours with SB431542 led to a substantial increase in eGFP, Id1, and FGF10 mRNA expression, whereas treatment with LDN193189 led to a decrease ineGFP, Id1 and FGF10 mRNA expression and an increase, probably compensatory, in BMP4 and VEGFa mRNA expression ( Figure 3B and D). Remarkably, even 8 hours of treatment with the SB431542 led to an increase in the density of the CD31 pos network and a more pronounced increase in the eGFP pos network, whereas, similar treatment with LDN193189 led to reduction in the density and connectivity of the eGFP pos and CD31 pos cells ( Figure 3C).
Collectively, these findings indicate that during the pseudoglandular stage, in the course of which most of the stereotypical branching morphogenesis is thought to take place, an important target of canonical BMP signaling in the lung is the developing pulmonary vasculature.
Strong eGFP expression was detected in the developingsubepithelial SMC layer of the proximal airways from E14.5 to P1 ( Figure 3A and Figures S1 & S2). The eGFP expression, which interestingly involved only a portion of the sub-epithelial SMCs, was undetectable by P15. It has been proposed that sub-epithelial SMCs originate from a pool of sub-mesothelial, FGF10-expressing, mesenchymal cells. Under the influence of several signalling pathways, such as those mediated by FGF, BMP, Shh, and Wnt, these cells proliferate and passively translocate to surround the proximal portion of the developing airway tree [35,43,44,45,46]. http://www.sciencedirect.com/science/article/pii/S0012160611 010189 -bb0125To map more precisely the temporal activation of the BRE-eGFP reporter during development of the subepithelial SMC compartment we isolated P11, P12 and P13.5 lungs from the BRE-eGFP reporter animals and analysed coexpression of eGFP and SM22a, a marker of both primitive and mature SMCs [47], or aSMA a marker of more differentiated SMCs. As shown in Figure S1B, SM22a expression was observed in a sub-epithelial population of cells around the developing proximal airways already at embryonic stage E11, aSMA expression was evident only in the most proximal portion of the SM22a pos zone and eGFP expression was clearly detected in the aSMA pos zone only after embryonic stage E13.5 (very few aSMA pos cells expressed very low levels of eGFP at E12). These findings are consistent with the notion that canonical BMP signalling does not involve primitive SMCs but rather mature aSMA expressing sub-epithelial SMCs.
Canonical BMP pathway activation in developing airway epithelial cells Although strong BRE activity was detected in the trachea and primary bronchi as early as E10-E12.5 [33], consistent with the demonstrated role of BMP signaling in trachea development [48,49], BMP-driven eGFP-expression was not detected in the epithelial compartments of the interlobular airways before the canalicular stage of lung development. As shown in Figures 1 and  4A, minimal eGFP expression was detected around E14.5 in the large airways, and only after E17 strong eGFP signal was detected in cartilaginous and large airways following a proximal-distal axis pattern, with more intense signal in the proximal part of the bronchial tree. Immunostaining of BRE-eGFP lung tissue sections with an antibody specific for the Clara cell specific differentiation marker Scgb1a1/CCSP/CC10 (hereafter referred to as CC10), revealed expression of the eGFP reporter from E17.5 and onwards, in luminal cells of the developing airways. Strong CC10 expression was detected around E19.5, approximately two days after the onset of eGFP expression in the airway epithelium ( Figure 4A) CC10 and eGFP exhibited partially overlapping zones of expression, with eGFP being highly expressed in the proximal portion and CC10 highly expressed in the distal portion of the developing airway tree. A zone of eGFP pos -CC10 pos cells appeared to separate the CC10 pos epithelial population from the eGFP pos airway regions. Postnatal expansion of CC10 pos regions in the conducting airways was accompanied by a gradual reduction of the eGFP pos regions ( Figure 4A).
Interestingly, regions of airways exhibiting enhanced eGFP expression were detected close to the tips of airway branch-points in close proximity to neuro-epithelial bodies (NEBs). Analysis of the eGFP expression in relation to CC10 and CGRP, a differentiation marker of neuro-epithelial cells, revealed that eGFP pos -CC10 low cells formed ''caps'' over clusters of CGRP pos cells. From E19.5, the eGFP pos -CC10 low ''cap'' cells (green-arrows in Figure 4B) were separated from the eGFP neg -CC10 pos cells by a zone of double positive epithelial cells (yellow-arrows in Figure 4B). Notably, the zone of eGFP pos -CC10 pos cells coincided with zones richin NEBs ( Figure S5). From P15 and beyond, the only NEBassociated eGFP pos cells were CGRP pos neuro-epithelial cells (blue-arrows in Figure 4B).
As the majority of eGFP pos cells did not express the marker for secretory airway epithelium (CC10), their possible relationship to ciliated airway epithelial cells was investigated by staining with antibodies for Foxj1 a ciliated cell specific differentiation marker. As shown in Figure S6, a remarkably small portion of eGFP pos cells were FoxJ1 pos (yellow-arrows in Figure S6).
Two alternative models have been proposed regarding the role of BMP signalling during lung branching morphogenesis. One model suggests that mesenchymal derived FGF10 induces Bmp4 expression by the distal bud epithelium to limit, in an autocrine manner, Fgf10-mediated bud outgrowth [45,50]. The other model, based on the differential response of the distal epithelial buds to BMP4 in the presence or absence of mesenchyme, proposes that Bmp4 produced by the distal epithelium acts in an autocrine manner to limit proliferation of distal epithelial buds, and in a paracrine manner, on the adjacent mesenchyme, to induce production of a mesenchymal signal that enhances proliferation of distal epithelial buds [51]. Thus, according to the latter model the negative or positive effect of BMP signalling on branching is the outcome of a dynamic balance between negative autocrine and positive paracrine mechanism.
The absence of eGFP reporter activation in the distal buds of the branching airway tree (Figures 1 and 4) prompted us to investigate whether this is the result of an intrinsic inability of the distal epithelium to activate the canonical BMP pathway or, in accordance to the second model, the result of a dynamic interplay with the adjacent mesenchyme. Therefore, mesenchyme free distal epithelial buds and epithelial free mesenchyme were isolated as previously described [35,36] and cultured in Matrigel. Remarkably, within 8 hours of incubation, the BRE-eGFP reporter was strongly up-regulated in the mesenchyme-free epithelial buds and conversely it exhibited a tendency for decline in the isolated mesenchyme ( Figure 5). These findings are consistent with the model of Bragg et al. [51] and moreover suggest that the elusive, mesenchyme derived, epithelial growth promoting signal could simply act by negatively regulating activation of the canonical BMP pathway in the distal epithelium buds.
Collectively, the above findings demonstrate that activation of the canonical BMP pathway in airway epithelial cells coincides with the beginning of the canalicular stage, when the developmental plan of the lung shifts from branching morphogenesis to the development of distinct respiratory epithelial cell compartments [52]. Keeping in mind that gas conducting airways and blood conducting vessels need to develop in a coordinated manner and be properly juxtaposed to support optimal gas-exchange in the adult, it is understandable that the tips of developing airways should reciprocally exchange growth, differentiation or guidance signals with the developing vasculature and other mesenchymal cells to ensure coordinated development and spatial integration of these two branching systems. The finding that disruption of the vascular network in growing lung explants affects branching morphogenesis by reducing primarily ''orthogonal bifurcations'' of the epithelial tubules [53,54] and the findings presented herein demonstrating the influence of the adjacent mesenchyme on the activation of the canonical BMP pathway reporter ( Figure 5) illustrates the importance of the crosstalk between epithelial and vascular cells during early lung development.
It should be pointed out that due to our failure to successfully apply anti-pSmad1/5/8 immuno-histochemistry in tissues from embryos earlier than stage E19.5, we cannot conclusively rule out for the moment activation of the canonical BMP pathway at the tips of the early branching endoderm. Further studies will be required to assess whether the BRE-eGFP reporter is not active in the distal endoderm because: a) non-canonical BMP pathways are primarily involved; b) the canonical BMP pathway is active however, does not reach the activation threshold required to activate the BRE-eGFP reporter; or c) that the dynamic balance between the mesenchyme and endoderm which keeps the BRE-eGFP reporter inactive most of the time (Figure 5), allows at critical points along the branching process transient spikes of canonical BMP pathway signaling to reach the nucleus and regulate expression of target genes.

Canonical BMP pathway activation in alveolar epithelial cells
Over-expression of BMP ligands or inactivation of BMP receptors in the distal part of the developing lung leads to dramatic defects in alveolar development [14,16,18,19]. Since eGFP expression was highly increased in lung parenchyma of the BRE-eGFP mice during the saccular and alveolar stages, the period during which functional alveoli are formed, we characterized the nature of the eGFP pos cells in this part of the developing lung. Tissue sections from different developmental stages were analyzed by co-staining for eGFP and pro-surfactant protein C (pro-SpC), a marker of type-II pneumocytes. As shown in figure 6 and Figure S2, eGFP staining during the pseudoglandular stage was confined to endothelial cells and peri-bronchial smooth muscle cells of the proximal airways. Pro-SpC pos epithelial cells in the distal airways were uniformly eGFP neg at this developmental stage ( Figure S2).In contrast, starting at the canalicular stage (E17) and persisting until approximately P15, a population of eGFP pospro-SpC pos cells emerged in the distal lung compartments ( Figure 6A).The SpC pos cells at this stage of development are thought to be immature type-II pneumocytes [55]. Morphometric analysis demonstrated that during the saccular and alveolar stages, approximately half of the pro-SpC pos alveolar cells were eGFP pos . Adult tissues, in which alveolarization was complete, contained a very small number of eGFP pos -pro-SpC pos alveolar cells (,0.5-1%) ( Figure 6B). High resolution confocal microscopy analysis revealed that at all the developmental stages analyzed only a very small portion of type-I pneumocytes were eGFP pos ( Figure S7). Interestingly, ectopic over-expression of a constitutive ALK3 in cultures of adult primary pulmonary cells resulted in increased pro-SpC (,6-fold) and Id1 (,15 fold) mRNA levels, in comparison to the untreated or ALK5ca treated groups ( Figure 6C) further implicating canonical BMP-signaling in the physiology of type-II pneumocytes.
The sequential appearance of eGFP neg -SpC pos (immature type-II pneumocytes), eGFP pos -SpC pos and eGFP neg -SpC pos (mature type-II pneumocytes) may suggest that the eGFP pos -SpC pos population is an intermediate step through which immature type-II pneumocytes must pass before developing into mature functionally competent type-II pneumocytes. This conclusion is compatible with the finding that mice with canonical BMP pathway disrupted due to deletion of the Smad1 gene accumulate in their developing lungs high numbers of periodic acid-Schiff (PAS) positive immature type-II pneumocytes [19].
The above findings indicate that during the cannalicular stage, an important target of canonical BMP signaling in the distal lung compartment is the developing type-II pneumocyte population. This could explain the dramatic distal lung phenotype of animals with disrupted BMP signaling in the distal lung epithelium [15,16,18,19].
It is very intriguing that with the beginning of the canalicular stage, when the developmental plan of the lung shifts from branching morphogenesis to the development of distinct respiratory epithelial cell compartments [52], the BMP-responsive eGFP reporter is activated in two epithelial domains, namely, the proximal airways and the distal developing alveolar sacs. The   conclusion that canonical BMP signaling becomes crucial for lung epithelial cell development at the cannalicular stage is consistent with earlier genetic studies demonstrating that deletion of Bmpr1 [16,18] or Smad1 [19] and ectopic expression of negative regulators of BMP signaling such as XNoggin and dominantnegative ALK6 [15] by SpC-promoter-driven expression, resulted in defects in lung development visible only around E15-E16, i.e. after completion of the branching morphogenesis.

The eGFP pos epithelial compartment is enriched for epithelial progenitors capable in forming colonies in Matrigel
The localization of eGFP pos cells in the vicinity of NEBs and the minimal overlap between cells expressing eGFP and markers for specialized secretory (CC10) or ciliated cells (FoxJ1) prompted us to investigate whether active canonical BMP signaling may define cells with an immature progenitor phenotype. Thus, a single two hour pulse of BrdU was given to pregnant mothers at E17 to label cycling epithelial cells which were subsequently detected by staining with anti-BrdU antibodies. BrdU pos cells accounted for ,6.7% of the surface airway epithelium in conducting airways and were uniformly distributed throughout the airway tree ( Figure 7A). This pattern is compatible with the findings of Giangreco et al. [56] and Rawlins et al. [57] demonstrating that randomly distributed progenitor cells contribute to normal epithelial homeostasis. Of the BrdU pos cells ,77.8% were eGFP neg and ,22.2% were eGFP pos , indicating that both populations contribute to the expansion of the airway epithelial compartment ( Figure 7B). Despite the dominance of the eGFP neg -BrdU pos cells (in absolute numbers), ,13.4% of the eGFP pos were BrdU pos, compared to ,5.5% for the eGFP neg cells ( Figure 7C).
To further compare the proliferative potential of the eGFP pos and eGFP neg cells different epithelial populations were isolated from dissociated lung tissue by cell sorting and analyzedin-vitro for colony-forming ability. Following a previously described protocol [38], EpCAM pos , Lineage negative (i.e. CD45 neg -CD31 neg -CD34 neg ) cells from BRE-eGFP transgenic animals were subdivided into Sca1 neg and Sca1 low cells. Finally the Sca1 neg and Sca1 low cells were separated by sorting into eGFP neg , eGFP low and eGFP high cells respectively ( Figure 7D). Isolated cells were cocultured with MLg feeder cells in a 3-dimensional Matrigel culture system [38]with a modification developed by some of us (HC and BRS) that involved co-culture of the cells in Matrigel for eight days in the presence of the TGFb receptor inhibitor SB431542 (10 mM), followed by an additional four days in the absence of inhibitor.
Lin neg -EpCAM pos -Sca1 low cells cultured in this way, in agreement with previous reports [58], gave rise primarily to large cystic epithelial colonies and Lin neg -EpCAM pos -Sca1 neg cells gave rise almost exclusively to smaller compact epithelial colonies ( Figure 7E). The clonogenic capacity of both Sca1 low and Sca1 neg cells correlated with the expression levels of the BRE-eGFP reporter with the corresponding eGFP high populations exhibiting highest clonogenic capacity ( Figure 7F). Previous studies have suggested that cystic colonies derived from Lin neg -EpCAM pos -Sca1 low cells contain progenitors of airway epithelial cells (Muc5Ac pos ), whereas, the compact colonies derived from Lin neg -EpCAM pos -Sca1 neg cells contain primarily SpC pos alveolar progenitors [58]. Consistently, as shown in Figure 7G, whereas all compact colonies derived from the Sca1 neg subpopulations exhibited a similar SpC pos -Muc5Ac neg staining pattern, the large cystic colonies derived from the Sca1 low subpopulations exhibited a more heterogeneous pattern of staining with approximately half of them expressing Muc5Ac and half of them expressing SpC.
It is worth noting that adult Lin neg -EpCAM pos -Sca1 neg epithelial cells have very low clonogenic capacity in Matrigel [38]. The development of colonies fromLin neg -EpCAM pos -Sca1 neg cells in the current study is most probably due to the fact that the analysis was done with P3 lungs which were still undergoing intense alveolar development.
Previous studies have provided evidence for two types of airway epithelial progenitors, one randomly distributed in the airways that maintains normal epithelial homeostasis and responds to minor injuries of the epithelium, and a second, associated with previously described progenitor cell niches that can renew depleted Clara cells upon severe epithelial injury [38,56]. The randomly distributed eGFP neg -BrdU pos cells detected in the developing lungs ( Figure 7A) could represent the former type whereas the eGFP pos epithelial cells could represent the latter type of progenitors. Once lung development is completed, the homeostatic, low turnover, maintenance of the epithelial population can be supported by the distributed Clara cells [38,56] and thus active BMP signaling may not be required any longer. The decline in eGFP expression in the lungs of the adult BRE-eGFP transgenics may reflect just that. Reactivation of the BRE-eGFP reporter in adult lung after tissue injury suggests an active role of BMP-signaling in adult lung repair after injury The dramatic decline of eGFP expression in the adult lung of the BRE-eGFP reporter mice and the association of eGFP expression with the epithelial progenitor pool prompted us to investigate whether the reporter was reactivated in adult lung during injury and repair. Adult BRE-eGFP animals were analysed in two established models of lung tissue injury and repair, namely the naphthalene and bleomycin induce lung injury models. Naphthalene treatment causes selective depletion of Clara cells in the airways [59]. In contrast, bleomycin causes epithelial cell death, acute inflammation and fibro-proliferative remodeling of the peripheral lung [60]. Immuno-staining for CC10 and CGRP demonstrated that by day 3, naphthalene treatment had caused a large reduction in CC10 pos cells in the airways, an increase in the number of CGRP pos cells per NEB, and a substantial increase in BRE-eGFP reporter activity at airway branch-points and terminal bronchioles ( Figure 8A).This pattern of CC10 localization and eGFP reporter expression corresponds with previously determined anatomic locations that maintain chemically resistant epithelial progenitor cells [61,62,63,64].
Analysis at different time points after naphthalene-treatment demonstrated a gradual increase in the number of eGFP pos bronchial epithelial cells that was evident on day 2 and maximum increase of 15-fold over control levels by day-9 post-naphthalene treatment ( Figure 8B). Up to day-2, the increase in eGFP pos airway epithelial cell numbers involved CGRP pos and CGRP neg -CC10 lowcells ( Figures 8C and D). Thereafter, although the total number of eGFP pos -CGRP pos cells remained stable from days 2-9, a wave of eGFP pos -CC10 low -CGRP neg cells that reached maximum around day 3 was followed by a wave of eGFP pos -CC10 pos cells that reached plateau around day-9 ( Figure 8D).
Interestingly, the distribution of eGFP and CC10 expression recapitulated the pattern observed during early lung development i.e. eGFP pos -CC10 pos cells separating eGFP pos -CC10 low from apparently newly formed eGFP neg -CC10 pos cells. The kinetics of appearance and relative localization of these cells within the tissues are consistent with the notion that they may represent sequential stages of development from the eGFP pos -CC10 low to the eGFP neg -CC10 pos phenotype.
Bleomycin is an anti-cancer drug, which induces lung injury and fibrosis [60]. As shown in Figures 9A and 9B, treatment of adult BRE-eGFP animals by intra-tracheal administration of bleomycin resulted in substantial reactivation of the eGFP reporter in type-II pneumocytes (Pro-SpC pos ). During the first days of treatment the eGFP pos cells were primarily cuboidal SpC pos cells. From day-4 onwards both cuboidal and squamous eGFP pos cells were found in the alveolar regions. Interestingly, the squamous eGFP pos cells were T1a-expressing type-I pneumocytes and were always localized in the borders between affected and apparently normal tissue ( Figure 9C). We postulate that the cuboidal eGFP pos cells observed in the early stages of the bleomycin induced response represent a reparatory progenitor population that expands and differentiates into the eGFP pos -T1a low cells observed in the later stages to repair the damaged alveolar epithelium. Further studies will be required, however, to validate this hypothesis.
Collectively, our findings demonstrate that the canonical BMP pathway is re-activated during adult lung injury in a manner that bears resemblance to the activation of this pathway during early lung development, strongly supporting its importance during adult lung tissue injury repair.

Conclusions
Utilizing a transgenic reporter mouse line harboring a BMPresponsive eGFP reporter allele we were able to construct a detailed spatiotemporal map of canonical BMP signalling during early lung development and adult lung tissue injury repair. Our studies demonstrated that during the pseudoglandular stage, when branching morphogenesis characterises lung development, canonical BMP pathway is active mainly in the vascular network and the airway smooth muscle layer. Only after the completion of branching morphogenesis and the initiation of epithelial cell differentiation canonical BMP pathway activation commences in airway and alveolar epithelial cell that are enriched in progenitors that can form epithelial colonies in Matrigel in-vitro. The BRE-eGFP reporter is reactivated in mesenchyme-free distal epithelial buds cultured in Matrigel suggesting that the low reporter activity in intact lungs is the result of a dynamic interplay between the endoderm and the mesenchyme and moreover suggest that the elusive, mesenchyme derived, epithelial growth promoting signal postulated by Bragg et al. [51] may act by regulating negatively activation of the canonical BMP pathway in the distal epithelium buds.
BRE-eGFP expression, in agreement with earlier reports describing the temporal expression pattern of some BMP pathway components during late lung development [65], peaks around birth and returns to very low levels upon completion of lung development. Remarkably, severe depletion of Clara cells in the adult lung by naphthalene treatment, leads to re-expression of the eGFP reporter around NEBs and terminal bronchioles, areas known to harbour airway progenitor cells. Likewise, injury of the alveolar epithelium in the adult lung by bleomycin treatment, leads to re-expression of eGFP initially in SpC pos cuboidal epithelial cells, which we postulated to be alveolar epithelial progenitors and subsequently in squamous T1a low cells which we postulated to be derived from the former population and on the way to differentiate to Type-I pneumonocytes. Our findings are compatible with the notion that during the late stages of lung development the role of canonical BMP signalling pathway is to maintain the undifferentiated state of the airway and alveolar epithelial progenitors preventing premature exhaustion of their pools and securing the continuous supply of the differentiated epithelial cells that are required for a fully developed lung. The reactivation of the BMP responsive reporter in adult animals undergoing repair of severe epithelial injury, where the same requirements regarding proper management of the undifferentiated epithelial progenitor pools apply, is also compatible with this conjecture. An analogous function has been previously proposed for BMP signaling regarding the maintenance of the undifferentiated state of pluripotent mouse embryonic stem (ES) cells [66,67]  and the maintenance of the undifferentiated, multipotent state of distal lung progenitor cells [45,68].
The construction of a spatiotemporal map of canonical BMP signalling during lung development and adult lung tissue injury repair (summarised in Figure 10 and table S2) will facilitate interpretation of earlier genetic studies and guide rational targeting of BMP signaling components in future experiments. Moreover, the ease isolation of living eGFP pos i.e. BMP responding, subpopulations of lung cells by cell sorting will greatly facilitate definitive clarification of the mechanisms of action of this signaling system during early lung development and repair of lung tissue injury in the adult.  Figure S3 FGF10, BMP4 and a Sonic Hedgehog agonist affect expression of the BRE-eGFP reporter. A) Whole E12 lung-explants were cultured on Nuclepore membranes for eight hours in the presence of vehicle, FGF10 (250 ng/ml), BMP4 (100 ng/ml) or Sonic Hedgehog agonist Purmorphamin (500 nM). Upper panel shows bright field images of representative explants. Lower panel shows eGFP expression in the same explants. B) Confocal images of tissue sections prepared from the explants described above stained for eGFP (green stain). Nuclei were counterstained with DAPI. Note the strong upregulation of eGFP expression in the FGF10 and BMP4 treated explants, and the  Figure S4 Correlation between pSmad1/5/8 and BRE-eGFP immune-staining in P1 lung tissue sections. Adjacent tissue section of lungs from BRE-eGFP reporter animals were stained with anti-GFP or anti-pSmad1/5/8 antibodies as described in materials and methods. A) Staining of adjacent sections of a large airway with anti-GFP and anti pSmad1/5/ 8antibodies or normal rabbit IgG fraction as isotype control. B) Representative images from the indicated tissue regions demonstrating that tissue areas with intense BRE-eGFP expression coincide with regions exhibiting pSma1/5/8 immuno-staining. The scale bar corresponds to 25 mm. (TIF) Figure S5 The zone of eGFP pos CC10 pos cells coincides with the NEB-rich portion of the airway tree. A) Confocal image of a lung tissue section derived from a P1 BRE-eGFP animal stained for eGFP (green staining), CC10 (red staining) and CGRP (white staining). The images illustrate the zone in the airway-tree where co-expression of eGFP and CC10 occurs. The lower images, showing the CGRP and DAPI channels of the upper images, illustrate the coincidence of eGFP pos -CC10 pos zone with the NEB-rich regions of the airway tree. B) Confocal images of the transitional zone between the eGFP pos -CC10 neg and eGFP neg -CC10pos domains of the airways demonstrating the preferential association of eGFP pos -CC10 low cells with NEBs. The image depicting NEBs is obtained from a section sequential to the ones depicting CC10 and eGFP expression. Nuclei were counterstained with DAPI (blue staining). (TIF) Figure S6 Minimal eGFP expression in the Foxj-1 pos airway ciliated cells. Representative confocal images demonstrating minimal activation of the BRE-eGFP transgene in FoxJ1 pos ciliated airway epithelial cells. Tissue section collected from E17.5, E19.5, P1, P15 and adult BRE-eGFP transgenic lungs were stained for eGFP (green staining), FoxJ1 (red staining) and CGRP (white staining). Nuclei were counterstained with DAPI (blue staining). The images demonstrate the presence of remark-ably low number of eGFP pos -FoxJ1 pos epithelial cell (depicted with yellow arrows). (TIF) Figure S7 A small, however, detectable number of BRE-eGFP developing type-I pneumocytes (T1a pos ) express eGFP. Representative confocal images of lung sections of E14.5, E17.5, E19.5, P1, P15 and adult BRE-eGFP transgenic lungs stained for eGFP (green staining), CD31 (red staining) and T1a (white staining). Nuclei were counterstained with DAPI (blue staining). The Scale bars are 20 mm. The images demonstrate that whereas the majority of the developing endothelial cells express eGFP, only a small number of type-I pneumocytes, shown in the inserts of the P1 and P15 image, are eGFP pos . (TIF)