Crucial requirement of ERK/MAPK signaling in respiratory tract development

Lung development is an orchestrated process directed in part by mesenchymal-epithelial interactions that rely on paracrine signaling mediated by secreted ligands and cell-cell contacts. How signal instructions are interpreted to ultimately control cell fate and morphogenesis remains an open question in lung biology. The mitogen-activated protein kinase (MAPK) pathways are involved in signal transduction through transmembrane receptors. They consist of protein kinase cascades linking extracellular stimuli to targets scattered throughout the cell to generate specific cellular responses to environmental changes. The classical extracellular signal-regulated kinase ERK/MAPK pathway constitutes the major cascade for the control of cell proliferation, differentiation and survival. It transduces the effects of growth factors implicated in cell fate determination and developmental processes. In mammals, the ERK/MAPK pathway contains the MEK1 and MEK2 kinases that activate ERK1 and ERK2. The high sequence homology between MEK1 and MEK2 and the sharing of common activation mechanisms suggest that they are functionally redundant (Raman et al., 2007). However, the loss of Mek1 (Map2k1 – Mouse Genome Informatics) function in mice results in embryonic death due to placenta defects, whereas Mek2 (Map2k2 – Mouse Genome Informatics) mutant mice do not show a phenotype (Bélanger et al., 2003; Giroux et al., 1999). Mek1^+/−;Mek2^+/− embryos die from abnormal placenta development, revealing the importance of Mek2 for proper placentation (Nadeau et al., 2009). The embryonic death of Mek1;Mek2 mutants precludes the study of their function at later stages. Using a Mek1 conditional allele and lineage-specific Cre mice, the function of both MEK1 and MEK2 proteins in numerous processes has been unveiled, highlighting the broad role of the ERK/MAPK pathway throughout life (Li et al., 2012; Newbern et al., 2008; Scholl et al., 2007; Yamashita et al., 2011).

In mice, lung formation starts with the outgrowth of the ventral foregut endoderm into the flanking mesenchyme to form the laryngotracheal groove. The tracheal rudiment then elongates caudally and bifurcates to form the two primary bronchi. During the pseudoglandular stage, numerous rounds of dichotomous branching produce the bronchial tree (Metzger et al., 2008). The subsequent canalicular and saccular phases lead to the thinning of lung mesenchyme, the differentiation of the distal epithelium and the formation of the vascular network, with the establishment of the air-blood barrier. After birth, alveoli formation increases the gas exchange surface to meet the requirements of the organism (Morrisey and Hogan, 2010). In vivo and in vitro experiments have revealed the importance of growth factors trafficking via MAPK pathways throughout lung development, including members of the fibroblast growth factor (FGF) family, such as FGF1, -2, -7, -9, -10 and -18, the platelet-derived growth factor A (PDGFA) and the epidermal growth factor (EGF). Despite the accumulating knowledge about the role of growth factors in lung morphogenesis, the downstream signaling events remain largely uncharacterized. The p38/MAPK pathway has been identified as a key player in the control of bud morphogenesis (Liu et al., 2008). Conversely, the implication of the ERK/MAPK pathway in lung morphogenesis has been suspected without ever being clearly demonstrated. In vitro treatment of fetal rat lung explants with a MEK inhibitor reduced branching and caused mesenchymal cell apoptosis that impaired explant growth (Kling et al., 2002). Moreover, the targeted expression in the developing lung epithelium of the activated forms of K-Ras or B-Raf, two kinases located upstream of MEK, affected normal airway shape in mice (Tang et al., 2011).

To define the role of the ERK/MAPK pathway in lung development, we specifically ablated Mek1 function in the mesenchyme and epithelium on a Mek2 null background. Deletion of all four Mek alleles in the mesenchyme caused multiple phenotypes, including omphalocele, kyphosis, lung hypoplasia, abnormal trachea patterning and death at birth. The rescue of the branching defect in mutant lung explant cultures suggests that reduced intrathoracic space contributes to the lung phenotype. Wild-type lung explants exposed to MEK inhibitor exhibited less branching, thereby suggesting a role for MEK in lung epithelium. This was directly addressed by the Mek conditional deletion in the respiratory epithelium with the Shh^cre strain. Mek1^flox/flox;Mek2^−/−;Shh^+/Cre embryos presented lung agenesis and no tracheal epithelial cell differentiation, phenotypes also seen in Erk1^−/−;Erk2^flox/flox;Shh^+/Cre specimens. Thus, the ERK/MAPK pathway integrates distinct signals in the lung epithelium and the flanking mesenchyme that are essential for the correct respiratory tract formation.

Ablation of Mek1 function in the mesenchyme was produced with the Dermo1^cre deleter mouse line (supplementary material Fig. S1A-D). The Mek1 mesenchymal deletion on a Mek2 null background caused death at birth of all Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre pups, which also showed reduced body size (Table 1). Comparison of the body weight of embryonic (E) day 18.5 embryos showed that the mesenchymal loss of all Mek alleles led to a ∼38% decrease in weight, thus indicating intrauterine growth restriction, a disorder due to placental underdevelopment (Fig. 1A; supplementary material Fig. S2A-F). Mek1^+/flox;Mek2^−/−;Dermo1^+/Cre and Mek1^flox/flox;Mek2^+/−;Dermo1^+/Cre mice did not show obvious phenotypes. Conversely, Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos displayed severe fully penetrant malformations, such as eyelid closure defect and giant omphalocele, a protrusion of the gut and liver outside the abdomen resulting from the incomplete closure of the ventral body wall (Fig. 1B-G; Christison-Lagay et al., 2011). Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre mice died at birth, most likely from the cannibalization of the omphalocele by the mother. Alcian Blue staining of E15.5 mutants revealed the important disorganization of the axial skeleton, including kyphosis, which might contribute to the reduced space in the thoracic cavity (Fig. 1D; supplementary material Fig. S2G-R).

Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre pups had smaller lungs with narrower airspaces and a thicker mesenchyme indicative of a delayed development (Fig. 2A-D). Lung hypoplasia was confirmed at E18.5, as the lung weight/body weight ratio corresponded to ∼67% of that of controls (Fig. 2E). To identify the factors contributing to hypoplasia, we measured lung cell proliferation and apoptosis. At E15.5, the number of phospho-histone H3 (pHH3)-positive mesenchymal cells was significantly reduced in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre mutants, whereas no difference was observed in epithelium (Fig. 2F-H). Similarly, the number of apoptotic cells was significantly augmented in the mesenchyme of Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre specimens (Fig. 2I-K). The imbalance in the proliferation/apoptosis ratio in lung mesenchyme of Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos underlined the reduced mesenchymal area observed (Fig. 2L). As decreased lung branching may also contribute to hypoplasia, we evaluated the number of acini at E15.5, as, at this stage, the ramifying bronchial tree gives rise to epithelial tubular branches ending in terminal acini (Kotecha, 2000). Lungs from Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos displayed a decreased number of acini, indicating a defective branching process (Fig. 2M). Thus, insufficient proliferation, augmented apoptosis and decreased branching contributed to lung hypoplasia in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos.

To test whether lung cell differentiation requires ERK/MAPK mesenchymal activity, we monitored the expression of cell-specific markers at E18.5. Although Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos exhibited abnormally compact lungs with poorly developed saccules, no evident difference was observed for the secretory club (Clara) cells and the alveolar type I and type II pneumocytes with the markers CC10, T1α (podoplanin) and pro-SPC, respectively (supplementary material Fig. S3A-F). The distribution of capillary endothelial cells detected with PECAM also appeared normal. Smooth muscle cells positive for α-SMA were correctly localized around the bronchi and the large blood vessels (supplementary material Fig. S3G-J). We investigated by quantitative RT-PCR (qRT-PCR) the expression levels of stromal cell markers: desmin (Des) and transgelin (Tagln) for smooth muscle cells, Pdgfrα for myofibroblasts and Pdgfrß for pericytes (Lindahl et al., 1997; Rock et al., 2011; Tada et al., 2007). Expression of Hoxa5 and Wnt2 genes, two mesenchyme-expressed regulators of lung development, was also analyzed (Boucherat et al., 2013; Morrisey and Hogan, 2010). No difference was observed except for Des and Pdgfrα expression (supplementary material Fig. S3K). Thus, cell differentiation appeared to occur normally along the respiratory tract in absence of a functional ERK/MAPK cascade in lung mesenchyme.

We validated Mek1 inactivation in lung mesenchyme from Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre specimens by analyzing MEK protein expression (supplementary material Fig. S4A-D). In E15.5 controls, MEK1 and MEK2 proteins showed ubiquitous expression in the epithelial and mesenchymal compartments along the entire respiratory tract. In mutants, no MEK1 mesenchymal expression was detected, except for a few cells, and MEK1 epithelial expression remained unchanged. No MEK2 expression was seen in these mutants. To assess the impact of the Mek mutation on ERK activation, we monitored phospho-ERK (pERK) expression (supplementary material Fig. S4E,F). In controls, immunostaining was detected in mesenchymal cells flanking tubules undergoing branching and in a few epithelial cells. Conversely, in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre specimens, pERK was barely detectable in the mesenchyme, supporting the notion that MEK1 and MEK2 are the only kinases activating ERK1 and ERK2 in lung mesenchyme. Surprisingly, a strong pERK staining was detected in lung epithelium, suggesting deregulated cross-talk of the ERK/MAPK pathway between the tissue layers.

Microarray analysis of RNA extracted from lungs of E15.5 Dermo1^+/Cre, Mek1^+/flox;Mek2^−/−;Dermo1^+/Cre and Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos was performed to evaluate the molecular impact of the mesenchymal loss-of-all Mek alleles on lung development. Venn diagrams displaying the total number of up- or downregulated transcripts with P<0.05 revealed that the number of differentially expressed transcripts between the Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre experimental group and the Dermo1^+/Cre control group was higher than that obtained when Mek1^+/flox;Mek2^−/−;Dermo1^+/Cre and Dermo1^+/Cre groups were compared (Fig. 3A). Thus, the mesenchymal loss of all Mek alleles had a greater influence on the lung transcriptome than the absence of three Mek alleles, which correlated with the phenotype observed.

A DAVID (Database for Annotation, Visualization and Integrated Discovery) functional annotation chart algorithm was used to identify the major biological processes modulated in mutant lungs (Huang et al., 2009). Upregulated genes in the Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre group were associated with chromatin organization, metabolism, cell structure and motility, cell signaling, cell proliferation, differentiation and development, whereas downregulated genes were enriched for metabolism, cell proliferation, cell organization, biosynthetic processes and protein folding (supplementary material Fig. S5). In the Mek1^+/flox;Mek2^−/−;Dermo1^+/Cre group, only genes associated with chromatin organization were upregulated, whereas genes related to metabolism and RNA processing were downregulated, indicating a broader impact of the four Mek alleles mutation on biological processes (data not shown).

A heat map was established with genes exhibiting a ≥1.5-fold change (Fig. 3B). As expected, Mek2 transcripts were downregulated in Mek1^+/flox;Mek2^−/−;Dermo1^+/Cre and Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre groups. Downregulation of genes involved in cell proliferation, such as Cdca3, Taf15, Egr1 and Ccl21, and upregulation of Ddit4, a gene associated with apoptosis, were detected in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre lungs, concurring with the lung phenotype observed (Ayad et al., 2003; Yoshida et al., 2010). Expression of genes implicated in lung development was also affected, such as R-spondin 2 homolog (Rspo2 – Mouse Genome Informatics), a ligand acting through the Wnt canonical pathway, and TRIC-B (Marveld2 – Mouse Genome Informatics), a monovalent cation-specific channel crucial for lung maturation (Bell et al., 2008; Yamazaki et al., 2009). The expression of selected candidates was confirmed by qRT-PCR and/or immunohistochemistry (IHC), validating the microarray results. For instance, expression of the Scgb3a2 gene, encoding a growth factor and anti-apoptotic agent expressed in airways that promotes fetal lung development, was decreased in mutants (Fig. 3C-E; Kurotani et al., 2008). Expression of Klf5, encoding an epithelial transcription factor required for perinatal lung morphogenesis and known to activate ERK/MAPK signaling, was augmented in mutant lungs (Fig. 3C; Wan et al., 2008; Yang et al., 2007).

Expression of Mkk6 (Map2k6 – Mouse Genome Informatics), encoding the upstream activator of the p38/MAP kinase (Mapk14 – Mouse Genome Informatics), was diminished in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre specimens (Fig. 3C). It impacted on p38/MAPK activation as shown by the reduced phospho-p38 (p-p38) expression in lung epithelium from mutants (Fig. 3F,G). As p38/MAP kinase is involved in lung branching, decreased p38 activation might contribute to the branching defect observed in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre mutants (Liu et al., 2008).

Many genes with altered expression in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre mutants are expressed in lung epithelium, implying that the loss of mesenchymal Mek function influenced epithelial behavior. However, the presence of thoracic cage deformities accompanied by omphalocele in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos made it difficult to distinguish between the impact on gene expression due to the complete loss of Mek function and the genes dysregulated following physical interferences.

Impairment of fetal breathing movements, inadequate volume of amniotic fluid and reduced intrathoracic space can cause lung hypoplasia (Kotecha, 2000). To examine whether impaired lung development in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre was due to the physical repercussions of omphalocele and abnormal skeleton, we cultured whole-lung explants from E12.5 embryos, as, at this age, omphalocele was not apparent. At day (D) 0, lung primordia from controls and Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos had similar sizes. As expressed by the percentage in increase of distal branching between D0 and D1, D2 and D3, Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre explants showed an increased branching and complexity comparable to controls (Fig. 4A-I). The efficiency of Cre-mediated Mek1 deletion in explants was assessed by IHC. No MEK1 staining was detected in lung mesenchyme from mutants (Fig. 4J,K). Moreover, the proportion of proliferative cells in lung mesenchyme and epithelium remained unchanged in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre explants, indicating that the phenotype observed in vivo was not recapitulated in vitro (Fig. 4L-N).

Treatment of lung explants with FGF9 causes increased mesenchymal growth and luminal dilation of epithelial tubules (del Moral et al., 2006; Yin et al., 2011). We investigated the response of Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre explants to FGF9. Both control and mutant lung explants cultured with FGF9 showed bud dilation and reduced epithelial branching, suggesting that the FGF9 effect on lung mesenchyme is mediated by MEK-independent intracellular signaling pathways (supplementary material Fig. S6).

Rat lung explant experiments have shown that pharmacological MEK inhibition reduces in vitro lung branching (Kling et al., 2002). We repeated this study with lung explants from E12.5 wild-type mice treated with the MEK inhibitor PD98059. In agreement with the previous analyses, PD98059 treatment diminished lung explant branching (Fig. 4O-U). Thus, lung hypoplasia in Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos might be a consequence of the omphalocele and/or the abnormal skeleton, both participating in the reduced intrathoracic space. Our data also point towards a role of the ERK/MAPK pathway in the lung epithelium for a normal branching process.

To directly address the epithelial role of MEK1 and MEK2 in lung branching, Mek1 was conditionally inactivated in Mek2^−/− lung epithelium using Shh^Cre mice (supplementary material Fig. S1E-H; Harfe et al., 2004; Harris et al., 2006). All Mek1^flox/flox;Mek2^−/−;Shh^+/Cre newborns became cyanotic and died at birth (Fig. 5A). E18.5, Mek1^floxflox;Mek2^−/−;Shh^+/Cre embryos presented a shortened trachea with a reduced number of cartilage rings (Fig. 5F,G). The trachea split into truncated primary bronchi without further branching, failing to form lungs (Fig. 5B-E). No lung phenotype was observed in Mek1^+flox;Mek2^−/−;Shh^+/Cre and Mek1^flox/flox;Mek2^+/−;Shh^+/Cre embryos or mice. Thus, the presence of one functional Mek allele either in lung mesenchyme or in lung epithelium is sufficient for correct lung development, maturation and function.

The efficiency of Cre-mediated Mek1 deletion was confirmed by the epithelial loss of MEK1 and pERK expression in Mek1^flox/flox;Mek2^−/−;Shh^+/Cre specimens (supplementary material Fig. S4G-J). To identify the causes of lung agenesis, cell proliferation was assessed at E12.5 using cell cycle progression markers. Reduced cyclin D1 immunostaining was observed in lung epithelium from Mek1^flox/flox;Mek2^−/−;Shh^+/Cre mutants in contrast to the strong expression seen in controls (Fig. 5H,I). Ki67 labeling in lung epithelium and mesenchyme was also diminished in mutants (Fig. 5J,K). Finally, a significantly decreased number of BrdU-positive cells was detected in the lung epithelium and mesenchyme of Mek1^flox/flox;Mek2^−/−;Shh^+/Cre mutants (Fig. 5L-N). The reduced epithelial cell proliferation correlated with an increased number of epithelial cells positive for the cell cycle inhibitor p27^Kip1 (Fig. 5O,P). Thus, the ERK/MAPK pathway in lung epithelium is essential for cyclin D1 expression and cell cycle progression.

Cell apoptosis was examined. Compared with controls, E12.5 Mek1^flox/flox;Mek2^−/−;Shh^+/Cre mutants presented substantial cleaved caspase 3 immunostaining along the epithelium of the truncated bronchi, with the strongest signal at the distal tip. Apoptosis was also observed in the distal mesenchyme of lung rudiments (Fig. 5Q,R). The ERK/MAPK pathway inhibits caspase 9 activity by direct phosphorylation at Thr 125, blocking caspase 3 activation and subsequent apoptosis (Allan et al., 2003). In Mek1^flox/flox;Mek2^−/−;Shh^+/Cre mutants, phosphorylation of caspase 9 was reduced in bronchial buds when compared with controls (Fig. 5S,T). These data provide compelling evidence that the ERK/MAPK pathway is essential for lung branching by directly controlling cell proliferation and survival.

Lung agenesis in Mek1^flox/flox;Mek2^−/−;Shh^+/Cre mutants resembled Fgf10 and Fgfr2b mutant lung phenotypes (De Moerlooze et al., 2000; Min et al., 1998; Sekine et al., 1999). In vitro experiments have shown that Bmp4 expression is induced at the tip of growing buds in response to mesenchymal FGF10 participating in branching regulation (Weaver et al., 2000). The Bmp4-reduced expression in lungs from E12.5 Mek1^flox/flox;Mek2^−/−;Shh^+/Cre mutants demonstrated that a functional ERK/MAPK pathway in lung epithelium is required to transduce FGF10-induced Bmp4 expression in vivo (supplementary material Fig. S7).

Cartilage rings constitute the supporting wall of the upper airways during breathing, thereby maintaining the patency of the lumen. E18.5 Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos displayed a thinner trachea without cartilage rings along the trachea and primary bronchi. Only a few isolated nests of cartilage cells were seen (Fig. 6A,B). Cre activity from the Dermo1^Cre allele was detected in the cartilage nodules, demonstrating that the tracheal phenotype of Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos was cell-autonomous (Fig. 6C). In controls, activated ERK was found in cartilage nodules, thus further supporting the requirement of the ERK/MAPK pathway in tracheal cartilage formation (Fig. 6D).

We analyzed the expression of SRY (sex-determining region Y)-box 9 (SOX9), a master regulator of chondrogenesis essential for patterning and formation of the tracheal cartilage (Park et al., 2010). In E14.5 controls, the segmented SOX9 expression showed the pattern of the future cartilage rings. Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos displayed a continuous SOX9 expression, indicating the lack of formation of the precartilage nodules (Fig. 6E,F). Accumulation of cartilage-specific type II collagen, which normally takes place following mesenchymal condensation, did not occur in mutant airways (Fig. 6G,H). qRT-PCR analyses revealed reduced expression levels of the chondrogenic regulator trio, Sox5, Sox6 and Sox9, and their downstream targets, aggrecan (Acan) and collagen type II alpha 1 (Col2a1), in upper airways from E18.5 Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos (Fig. 6I,J). Thus, the lack of ERK/MAPK function in the upper airway mesenchyme of Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre mutants perturbed the expression of major regulators of chondrogenesis, resulting in failure of mesenchymal condensation and subsequent tracheal cartilage formation, thus causing the flaccidity of the trachea.

The trachea of Mek1^flox/flox;Mek2^−/−;Shh^+/Cre mutants was shorter with fewer cartilage rings (Fig. 5F,G). We assessed whether Mek epithelial deletion had direct effects on the tracheal epithelium integrity. In E12.5-E18.5 controls, the trachea was lined by a pseudostratified columnar epithelium. At E18.5, the epithelium is normally composed of ciliated cells, club cells and few goblet cells, covering a layer of p63 (Tcp1 – Mouse Genome Informatics)-positive basal cells (Fig. 7; supplementary material Fig. S8). In Mek1^flox/flox;Mek2^−/−;Shh^+/Cre mutants, cells failed to maintain their stratification around E14.5 (supplementary material Fig. S8D,H). E18.5 mutants exhibited a simple squamous epithelium virtually devoid of basal, goblet and ciliated cells, with only a few club cells (Fig. 7). The lack of basal cells was also observed at E14.5, indicating a crucial role for the ERK/MAPK pathway in basal cell differentiation (supplementary material Fig. S8). This correlated with the diminution in basal cells in mutants overexpressing an FGFR2b dominant negative form (Volckaert et al., 2013).

No staining was detected in Mek1^flox/flox;Mek2^−/−;Shh^+/Cre mutants with the stage-specific embryonic antigen 1 (SSEA1; Fut4 – Mouse Genome Informatics), a marker of pluripotent stem cells (Fig. 7G,H; Henderson et al., 2002). To explain the lack of tracheal epithelial cell differentiation, we hypothesized that the loss of Mek function could lead to a downregulation of SOX2 expression, preventing differentiation of embryonic airway progenitors into specific lineages (Que et al., 2009). At E18.5, Mek1^flox/flox;Mek2^−/−;Shh^+/Cre mutants presented rare SOX2-positive cells along the trachea. At earlier stages, all tracheal epithelial cells expressed SOX2 but at a lower intensity, suggesting that the ERK/MAPK pathway is required to maintain SOX2 expression at levels sufficient to allow cell differentiation (Fig. 7I,J; supplementary material Fig. S8). Together, our findings demonstrate that the ERK/MAPK pathway is essential for the maintenance of a tracheal progenitor cell population and for the correct tracheal epithelial cell differentiation. They also emphasize the absolute requirement of mesenchymal and epithelial expression of Mek1 and Mek2 in trachea formation.

We wanted to determine whether the respiratory tract phenotype arising from the epithelial inactivation of both Mek genes would be recapitulated upon inactivation of the downstream effectors Erk1 (Mapk3 – Mouse Genome Informatics) and Erk2 (Mapk1 – Mouse Genome Informatics). E18.5 Erk1^−/−;Erk2^flox/flox;Shh^+/Cre embryos exhibited the same phenotypes as Mek1^flox/flox;Mek2^−/−;Shh^+/Cre embryos: lung agenesis, abnormal tracheal epithelial differentiation and death at birth (Fig. 8). These data underscored that the MEK-ERK module acts as a linear pathway, in which ERK1 and ERK2 are the sole effectors of MEK kinases in the lung epithelium.

MEK isoforms possess unique functions as reflected by the distinct phenotypes of the Mek1 and Mek2 single mutants (Bélanger et al., 2003; Giroux et al., 1999). This specificity can be explained by the exclusive roles and/or the different spatial distribution or levels of MEK proteins. Our genetic studies on Mek gene function in respiratory tract formation demonstrated that the presence of only one functional Mek allele in the mesenchyme or in the epithelium is sufficient to support normal development and to allow survival of mutant pups. Thus, Mek1 and Mek2 genes display redundant functions during lung morphogenesis. A similar observation could be made for the role of Erk1 and Erk2 in lung epithelium.

The phenotypes resulting from the mesenchymal ablation of Mek genes were anticipated due to the large spectrum of actions of the ligands trafficking via the ERK/MAPK pathway. For instance, abnormal body wall formation and omphalocele were reported in mice carrying conditional mutations in both Fgfr1 and Fgfr2 genes (Nichol et al., 2011). Normal lung growth requires adequate space in the thoracic cavity and appropriate tonic and cyclic distending forces originating in part from fetal breathing movements (Kotecha, 2000). Mesenchymal Mek mutation caused kyphosis and omphalocele, mimicking human pathological conditions that can impose a significant physical restriction on lung growth and function (Argyle, 1989; McMaster et al., 2007). In addition to Fgfr1;Fgfr2 compound mutants, abnormal body wall formation was seen in mice carrying conditional mutations in the Alk3 gene encoding a BMP type I receptor, and the Alk5 gene encoding a Tgfβ type I receptor (Matsunobu et al., 2009; Sun et al., 2007). BMP4 and TGFβ were shown to activate the ERK/MAPK pathway in various experimental conditions (Hough et al., 2012; Lee et al., 2007). Therefore, it is possible that the ERK/MAPK pathway acts as a convergence point for the FGF-, BMP- and TGFβ-driven signaling cascades in the mesenchyme to control ventral body wall closure.

Lung explant cultures demonstrated that defective branching and cell proliferation of Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre specimens can be rescued in vitro. This indicates that the lung has the capacity to correctly develop in absence of a functional ERK/MAPK pathway in lung mesenchyme at the pseudoglandular stage. These results identify the mechanical restriction due to dysmorphic rib cage and severe omphalocele as a probable cause of the pulmonary hypoplasia that challenged the neonatal survival of Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre pups. However, a role of Mek genes in lung mesenchyme at later stages of lung development cannot be excluded.

In Mek1^flox/flox;Mek2^−/−;Shh^+/Cre embryos, the specification of the primary lung field occurred with the formation of the trachea and its bifurcation to produce primary bronchial buds. However, the lack of Mek function in lung epithelium did not allow further development resulting in lung agenesis. Thus, even though the early steps of the specification of the trachea and lung progenitors in the ventral foregut were reported to necessitate WNT and FGF signals, the latter was not transduced via the ERK/MAPK pathway as suggested by our data (Goss et al., 2009; Serls et al., 2005). Although recently challenged, the dynamic pattern of Fgf10 expression to sites of prospective bud formation and the FGF10 ability to induce epithelial proliferation and budding in lung organ cultures have led to the hypothesis that FGF10 controls the directional outgrowth of lung buds during branching morphogenesis. In agreement with this, lung agenesis was reported in Fgf10 and Fgfr2b mutant embryos, which both failed to develop primary bronchial buds (Bellusci et al., 1997; De Moerlooze et al., 2000; Harris et al., 2006; Min et al., 1998; Sekine et al., 1999; Volckaert et al., 2013). Our results support the notion that the tracheal bifurcation mediated by the FGF10-FGFR2b interaction is also ERK/MAPK-independent, as the epithelial mutation with the Shh^Cre allele occurs before lung formation onset (Harris-Johnson et al., 2009). Thus, trachea formation and its branching into primary bronchi do not depend on the epithelial ERK/MAPK pathway, whereas the subsequent branching events seem to do. This concurs with the idea that trachea bifurcation and the following ramifications arise in distinct anatomical fields that require different genetic networks (Metzger et al., 2008).

Decreased cell proliferation combined with augmented apoptosis underlie lung agenesis in Mek1^flox/flox;Mek2^−/−;Shh^+/Cre embryos, two cell processes that are under the direct control of the ERK/MAPK pathway. It has been shown that the ERK/MAPK cascade regulates progression through the cell cycle by stimulating increased cyclin D1 levels that signal continued proliferation (Lavoie et al., 1996). This necessitates the phosphorylation by ERK of the MYC transcription factor, which, once activated, promotes the transcription of the cyclin D1 gene Ccnd1 (Mouse Genome Informatics) (Daksis et al., 1994). In parallel, the ERK/MAPK pathway can suppress the levels of the cdk inhibitor p27^Kip1 via the proteasome pathway, which impacts on the rate of cell cycle progression (Schepers et al., 2005; Stacey, 2010). Consequently, in absence of functional MEK proteins in lung epithelium, Ccnd1 expression was reduced and p27^Kip1 levels were increased in mutant lungs, leading to cell cycle arrest and decreased proliferation.

The ERK/MAPK pathway is also involved in the control of apoptosis by directly controlling the activity of the initiator protease caspase 9 via its phosphorylation at Thr 125 by ERK. Phosphorylated caspase 9 is inefficient to process caspase 3 proteolytic activation and the ensuing apoptosis (Allan et al., 2003). In Mek1^flox/flox;Mek2^−/−;Shh^+/Cre embryos, the levels of phosphorylated caspase 9 were diminished and, consequently, those of cleaved caspase 3 were augmented, thereby triggering increased apoptosis. Together, our findings provide a mechanistic insight into how the absence of Mek function in the lung epithelium directly affects cell proliferation and survival. It thus causes a profound deregulation of the proliferation/apoptosis balance, a likely explanation for the observed lung agenesis (Fig. 9).

The loss of Mek function either in the mesenchyme or the epithelium impacted on trachea formation, thus affecting the targeted cell layer and revealing the importance of the cell-autonomous role of the ERK/MAPK pathway during trachea development. Tracheas from Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos manifest a near-complete absence of C-shaped cartilage rings along the entire length of the upper airways, thereby providing a putative mouse model for primary tracheomalacia, a weakness of the tracheal walls causing life-threatening conditions (Carden et al., 2005). Tracheal cartilage development results from a multistep process, involving mesenchymal cell proliferation, aggregation and condensation into pre-cartilaginous nodules, which subsequently differentiate into chondrocytes to form cartilage. Several genes, including Hoxa5, Shh, Fgf10, Fgfr2, Gli2 and Gli3, Fstl1 and Tmem16a (Ano1 – Mouse Genome Informatics), were identified as key players in tracheal cartilage development based on mutant mice characterization (Aubin et al., 1997; Geng et al., 2011; Motoyama et al., 1998; Rock et al., 2008; Sala et al., 2011). As most of these genes encode growth factors or transcriptional regulators, the mesenchymal deletion of Mek genes represents one of the first examples demonstrating the crucial role of intracellular signaling molecules in tracheal chondrogenesis. Our results strongly suggest that this action is mediated by SOX9. FGF as well as a MEK constitutive active form enhance Sox9 expression and SOX9 transcriptional activity on a SOX9-dependent enhancer in primary chondrocytes (Murakami et al., 2000). The twofold decrease in Sox9 transcript levels in the upper airways of Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos specimens supports a direct effect of ERK/MAPK signaling on Sox9 expression. This reduction in gene dosage might be sufficient to perturb cartilage formation, as Sox9^+/− embryos die perinatally with severe cartilage hypoplasia (Bi et al., 2001). Similarly, SOX9 haploinsufficiency in humans causes campomelic dysplasia characterized by severe skeletal dysmorphologies (Wunderle et al., 1998). Alternatively, the absence of Mek function in tracheal mesenchyme might affect SOX9 phosphorylation, known to increase its DNA binding and transcriptional activity (Huang et al., 2000). In support of this notion, activated ERK1 and ERK2 can physically interact with SOX9 in chondrocyte nuclei, raising the possibility that the ERK/MAPK cascade directly promotes SOX9 phosphorylation, thus modulating its transcriptional role (Shakibaei et al., 2006). Therefore, it is not surprising that the loss of Mek function in lung mesenchyme with its negative impact on Sox9 expression resulted in the downregulation of SOX9 targets, such as SOX5 and SOX6, two transcriptional regulators that act as a trio together with SOX9 to control chondrogenesis (Akiyama et al., 2002). Consequently, the diminished expression of the three regulators led to the downregulation of cartilage-specific genes, like the proteoglycan aggrecan (Acan) and the extracellular matrix protein collagen Col2a1, resulting in absence of mesenchymal condensation and subsequent tracheal cartilage formation, which causes the flaccidity of the trachea (Fig. 9).

Conversely, tracheas from Mek1^flox/flox;Mek2^−/−;Shh^+/Cre and Erk1^−/−;Erk2^flox/flox;Shh^+/Cre embryos exhibit a correct patterning of the cartilage rings, but a profound impairment of tracheal epithelial differentiation. This contrasts with the phenotypes of Fgf10^−/− and Fgfr2b^−/− mutants, which both exhibited a shortened trachea with defective cartilage patterning. It has been shown that the mesenchymal Fgf10 expression activates FGFR2b signaling in the tracheal epithelium to drive cartilage formation via expression of Shh. However, the trachea from Fgf10^−/− embryos did not present major alterations in epithelial differentiation (Sala et al., 2011). The divergence in phenotypic traits raises questions about the identity of the ligand-receptor pair(s) trafficking via the ERK/MAPK pathway in the tracheal epithelium (Fig. 9).

The formation of an undifferentiated squamous epithelium lacking the expression of specific markers for basal, club, ciliated and mucus cells in Mek1^flox/flox;Mek2^−/−;Shh^+/Cre and Erk1^−/−;Erk2^flox/flox;Shh^+/Cre embryos can be attributed to the absence of expression of SOX2, a key transcriptional regulator of tracheal epithelium differentiation (Que et al., 2009). The loss of SOX2 expression in tracheal epithelial cells in Mek1^flox/flox;Mek2^−/−;Shh^+/Cre and Erk1^−/−;Erk2^flox/flox;Shh^+/Cre mutants is intriguing, as it indicates new, unsuspected roles for the ERK/MAPK pathway in progenitor cell maintenance and SOX2-mediated airway epithelial differentiation. Further works are needed to fully elucidate the underlying signaling mechanisms (Fig. 9).

In conclusion, we present new evidence that the ERK/MAPK pathway plays specific roles in the epithelium and the mesenchyme of the developing respiratory tract. Whereas its direct function in lung mesenchyme remains to be clearly established, the absence of the ERK/MAPK pathway in lung epithelium is crucial for lung formation, and it phenocopies lung agenesis also observed in Fgf10 and Fgfr2b mutant mice. Decreased epithelial cell proliferation coupled with increased cell death appears to be the mechanism responsible for lung agenesis. In the trachea, the ERK/MAPK pathway plays a cell-autonomous role in each cell layer, transducing signals in the mesenchyme for correct tracheal chondrogenesis and in the epithelium for progenitor cell maintenance and differentiation. These phenotypes diverge from those observed in Fgf10 and Fgfr2 mutants, indicating that other ligand-receptor pair(s) trafficking via the ERK/MAPK pathway are involved in trachea formation.

The Mek1^flox/flox and Mek2^−/− mouse lines were described previously (Bélanger et al., 2003; Bissonauth et al., 2006). The Erk1^−/− and Erk2^flox/flox mouse lines were obtained from Dr Meloche (IRIC, Université de Montréal, Montréal, Canada; Pagés et al., 1999; Voisin et al., 2010). The Rosa26 reporter line (Gt(ROSA)26Sor^tm1Sor) and the Shh^Cre deleter strain (Shh^{tm1(EGFP/cre)Cjt}) were purchased (Jackson Laboratory; Harfe et al., 2004; Soriano, 1999). Dermo1^Cre mice were provided by Dr Ornitz (Washington University, St. Louis, USA; Yu et al., 2003). For the mesenchymal ablation of Mek1 on a Mek2-null background, Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre specimens were generated by breeding Mek1^flox/+;Mek2^+/−;Dermo1^+/Cre with Mek1^flox/flox;Mek2^−/− mice. Similar breeding steps were carried out to produce Mek1^flox/flox;Mek2^−/−;Shh^+/Cre and Erk1^−/−;Erk2^flox/flox;Shh^+/Cre specimens for the Mek1 and Erk2 epithelial deletions, respectively. As only individuals carrying the Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre, Mek1^flox/flox;Mek2^−/−;Shh^+/Cre and Erk1^−/−;Erk2^flox/flox;Shh^+/Cre genotypes presented defects, all other genotypes are referred to hereafter as controls. The age of the embryos was estimated by considering the morning of the day of the vaginal plug as E0.5. Experimental specimens were genotyped by Southern blot and PCR analyses. Lungs from control and mutant embryos were collected at E12.5, E14.5, E15.5, E18.5 and D0. For adult specimens, lungs were dissected, instilled with 4% cold paraformaldehyde and paraffin-embedded. For RNA extraction, lungs were snap-frozen in N₂. Experiments were performed according to the guidelines of the Canadian Council on Animal Care and approved by the institutional animal care committee.

Lungs were collected from E12.5 mouse embryos and cultured for 72 h on porous membranes (8 μm pore size) in 24-well plates (del Moral and Warburton, 2010). The number of terminal buds at 0, 24, 48 and 72 h at the periphery of explants was counted and the branching increase (%) at each time point was calculated as:

(Branches_{24, 48 or 72 h}−Branches_0 h)/Branches_0 h×100.

Exogenous recombinant mouse FGF9 (rmFGF9) was added to explants at a concentration of 200 ng/ml (del Moral et al., 2006). Stock solutions of PD98059 (20 mM) were prepared in DMSO and added to medium at a final concentration of 50 μM. Medium was changed daily.

Paraffin-embedded embryos or lungs were sectioned at 4 µm for E12.5-E18.5 specimens and at 6 µm for older specimens. Morphology was analyzed by hematoxylin and eosin (H&E) staining. Immunostaining experiments were performed (Boucherat et al., 2012). Slides were counterstained with either Alcian Blue (for detection of acid mucus-producing cells) and nuclear Fast Red, methyl green or hematoxylin. For IF studies, nuclei were visualized by DAPI staining. Primary and secondary antibodies used are listed in supplementary material Table S1.

Proliferation and apoptotic indices were obtained by counting the number of pHH3-, BrdU- and cleaved caspase 3-immunoreactive cells, respectively, divided by the total cell number in lung epithelium or mesenchyme for each section analyzed. Five to six random fields were taken for an average number of 300 cells per field, from four to five embryos per genotype. For BrdU analyses, pregnant females were injected intraperitoneally with 100 µg/g body weight, and embryos were collected 1 h later.

Whole-mount detection of β-galactosidase activity in E8.5-E12.5 embryos and in dissected lung from E12.5 embryos was performed (Trainor et al., 1999). Cryostat sections (10 µm) of E12.5 embryos were processed for X-gal staining (Boucherat et al., 2012).

RNA in situ hybridization was performed on paraffin sections of E12.5 embryos (Simmons et al., 2008). A 486-bp SmaI-BstEII fragment containing 5′ non-coding and coding sequences from the mouse Bmp4 gene was used for synthesizing the digoxigenin-labeled riboprobe. The experiment was performed on four specimens per genotype.

Total RNA was isolated from the trachea/primary bronchi and the lungs of individual E15.5 and E18.5 embryos, and qRT-PCR experiments were performed (Boucherat et al., 2012). Four to six specimens were used for each genotype tested. Primer sequences are listed in supplementary material Table S2.

Total RNA was isolated from lungs of E15.5 Dermo1^+/Cre, Mek1^+/flox;Mek2^−/−;Dermo1^+/Cre and Mek1^flox/flox;Mek2^−/−;Dermo1^+/Cre embryos (n=4/genotype). RNA quality and quantity assessment, cDNA probe preparation, hybridization to the Illumina MouseWG-6 v2.0 Expression BeadChip and image scan were performed at the Genome Quebec Innovation Centre at McGill University (Montréal, Canada). Data were normalized using log2 transformation followed by quantile normalization with the R/lumi package in BioConductor (Du et al., 2008). Raw and normalized data were uploaded in the NCBI Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/projects/geo) with the accession number GSE51643 according to MIAME standards (Edgar et al., 2002). Significantly modulated probes were identified using the empirical Bayes statistics available in limma (Smyth, 2004). Probes were considered to be significantly modulated when the Benjamini–Hochberg-adjusted P<0.05.

Total lung surface and the number and area of epithelial tubules were measured using Adobe Photoshop CS3 software. Total lung mesenchymal surface was obtained by subtracting the airway tubule area from the total lung area. The number of acini was normalized according to the total lung area.

Student's t-test was performed for comparative studies. A significance level inferior to 5% (P<0.05) was considered statistically significant. *P<0.05, **P<0.01, ***P<0.001.

We thank Dr F. Costantini for insightful comments on the manuscript, Drs S. Meloche and D. Ornitz for mouse lines, Drs S. Kimura and G. Singh for antibodies, Dr L. Robertson for in situ riboprobe and E. Paquet for the analysis of microarray data.