|Home | About | Journals | Submit | Contact Us | Français|
While Kras/mitogen-activated protein kinase (MAPK) and canonical Wnt/β-catenin are critical for lung morphogenesis, mechanisms integrating these important signaling pathways during lung development are unknown. Herein, we demonstrate that the Foxm1 transcription factor is a key downstream target of activated KrasG12D. Deletion of Foxm1 from respiratory epithelial cells during lung formation prevented structural abnormalities caused by activated KrasG12D. Kras/Foxm1 signaling inhibited the activity of canonical Wnt signaling in the developing lung in vivo. Foxm1 decreased T-cell factor (TCF) transcriptional activity induced by activated β-catenin in vitro. Depletion of Foxm1 by short interfering RNA (siRNA) increased nuclear localization of β-catenin, increased expression of β-catenin target genes, and decreased mRNA and protein levels of the β-catenin inhibitor Axin2. Axin2 mRNA was reduced in distal lung epithelium of Foxm1-deficient mice. Foxm1 directly bound to and increased transcriptional activity of the Axin2 promoter region. Foxm1 is required for Kras signaling in distal lung epithelium and provides a mechanism integrating Kras and canonical Wnt/β-catenin signaling during lung development.
Lung formation is controlled by receptor-mediated signaling events between mesenchyme and foregut endoderm-derived epithelial cells that regulate the gene expression required for branching lung morphogenesis (34, 44). Stereotypic branching results in the formation of conducting airways and peripheral saccules which form the alveoli that mediate gas exchange after birth (33). The respiratory tract is lined by distinct epithelial cell types that vary along the proximal-peripheral axis of the lung. Alveoli are highly vascularized by an extensive capillary bed that is in close apposition to alveolar epithelial cells, facilitating gas exchange. During branching lung morphogenesis, cellular proliferation and differentiation are regulated by various signaling pathways, including tyrosine kinases, G protein-coupled receptors, fibroblast growth factors (FGFs), tumor growth factor beta (TGF-β), bone morphogenetic proteins (BMPs), Sonic hedgehog (SHH), retinoic acid, glucocorticoid, and Wnt signaling pathways (3, 44, 55, 56). Previous studies demonstrated that FGF9/10, Wnt2/7b, BMP4, and SHH activate Erk1/2 (Mapk3 and Mapk1) and p38 (Mapk14), which phosphorylate proteins critical for cell cycle progression, cell migration, and cell survival (34). Expression of an activated KrasG12D in distal lung epithelium caused severe defects in branching morphogenesis that were associated with increased mitogen-activated protein kinase (MAPK) activity and increased expression of Sprouty-2, a Ras/MAPK antagonist (45). Reduced epithelial branching was observed in fetal lung explants treated with MEK1/2 inhibitor (18).
Canonical Wnt/β-catenin signaling is required for initiation of lung morphogenesis; loss of Wnt2 and Wnt2b in splanchnic mesenchyme or endoderm-specific deletion of the β-catenin gene caused lung agenesis (13). Deletion of β-catenin from respiratory epithelial cells after initiation of lung development impaired branching lung morphogenesis, causing enlarged bronchiolar tubules and loss of peripheral lung (36). Expression of activated β-catenin in respiratory epithelial cells disrupted alveolarization (35). While Kras/MAPK and canonical Wnt/β-catenin signaling pathways are critical for normal lung morphogenesis, the molecular mechanisms integrating these two important signaling pathways in the developing lung are unknown.
Forkhead Box m1 (Foxm1; previously known as HFH-11, Trident, Win, or MPP2) is a member of the FOX family of transcription factors that play an important role in organ morphogenesis and development of cancer (6, 21, 23, 47). The Ras/MAPK (extracellular signal-regulated kinase [ERK]) signaling activates cyclin/cyclin-dependent kinase (CDK) complexes, which phosphorylate proteins critical for cell cycle progression (32). In vitro studies of neoplastic cells and mouse embryonic fibroblasts (MEFs) demonstrated that both ERK and CDK-cyclin complexes phosphorylate the FOXM1 protein, contributing to its transcriptional activation (30, 31). Phosphorylation of FOXM1 by CDK1 and CDK2 enhanced binding of FOXM1 with CBP, a transcriptional coactivator protein (31). Foxm1 is a direct transcriptional activator of various cell cycle regulatory genes, including Cyclin B1, JNK1, ATF2, c-Myc, Cdc25B, Aurora B, Plk1, Survivin, and Cenp-A, Cenp-B, and Cenp-F isoforms (5, 23, 49). Although these in vitro studies support the concept that Foxm1 is an important downstream target of Kras, whether Foxm1 is required for Kras signaling in vivo remains unknown.
Loss of Foxm1 in mouse causes abnormalities in liver, lung, and heart morphogenesis, which result in embryonic lethality (20, 21, 42). Conditional deletion of Foxm1 from precursors of cerebellar granule neurons delayed brain development and altered SHH signaling (43). Deletion of Foxm1 from pancreatic β cells reduced cell proliferation and impaired islet function, causing early onset of diabetes in mice (60). Foxm1 inactivation in smooth muscle cells resulted in perinatal pulmonary hemorrhage and structural abnormalities in blood vessels and the esophagus (48). Deletion of Foxm1 from the T lymphocyte lineage caused a decrease in proliferation of early thymocytes and activated mature T cells (57). Thus, Foxm1 plays an important role in cell proliferation in multiple organs and tissues. Surprisingly, deletion of Foxm1 from embryonic respiratory epithelial cells did not influence epithelial proliferation but impaired lung maturation, decreasing surfactant production by type II cells and preventing differentiation of alveolar type I cells (17). Expression of the activated FOXM1 mutant (FoxM1-delN) in embryonic epithelial cells of conducting airways caused epithelial hyperplasia (53). While precise regulation of both intercellular signaling and Foxm1 are critical for lung development, mechanisms by which Foxm1 activity is regulated by Kras/MAPK or Wnt/β-catenin signaling are unknown.
Generation of SPC-rtTA/tetO-Cre/Foxm1fl/fl mice was described previously (17). The tetO-KrasG12D mice were purchased from The Jackson Laboratory (11) and backcrossed with Foxm1fl/fl mice to generate tetO-KrasG12D/Foxm1fl/fl mice. SPC-rtTA/tetO-KrasG12D/tetO-Cre/Foxm1fl/fl quadruple transgenic embryos were obtained from breeding SPC-rtTA/tetO-Cre/Foxm1fl/fl and tetO-KrasG12D/Foxm1fl/fl mice (mixed C57BL/6 and FVB/N genetic background). SPC-rtTA/tetO-GFP-FoxM1-delN (epFoxM1-delN) mice were generated and characterized in our laboratory (53). The epFoxM1-delN mice and SPC-rtTA/tetO-KrasG12D mice were bred with TOPGAL mice (2) to generate TOPGAL/SPC-rtTA/tetO-FoxM1-delN and TOPGAL/SPC-rtTA/tetO-KrasG12D embryos, respectively. TOPGAL/SPC-rtTA/tetO-Cre/Foxm1fl/fl embryos were generated by breeding TOPGAL/Foxm1fl/fl mice with SPC-rtTA/tetO-Cre/Foxm1fl/fl mice. Mice were maintained in pathogen-free, filtered vivarium cages at the animal facility of Cincinnati Children's Hospital Medical Center. Sentinel mice were free of common viral and bacterial pathogens. Oral doxycycline (Dox) was given in mouse chow as described previously (53). Animal studies were reviewed and approved by the Animal Care and Use Committee of the Cincinnati Children's Hospital Medical Center.
Mouse embryos were fixed in 4% paraformaldehyde and paraffin embedded. Five-micrometer-thick lung sections were stained with hematoxylin and eosin (H&E) for morphological examination or used for immunohistochemistry (IHC) as described previously (51, 53). The following antibodies were used for IHC: transgenic (human) FoxM1-delN mutant (1:2,000; H-300) and endogenous (mouse) Foxm1 (1:1000; K-19) were from Santa Cruz Biotechnology; proSP-C (1:500) (53), TTF-1 (1:1,500; number 8G7G3-1), and Foxa2 (1:2,000; number WRAB-FOXA2) were from Seven Hills Bioreagents (Cincinnati, OH); and β-galactosidase (β-Gal; 1:500) and Axin2 (1:3,000; ab32197) were from Abcam. We also used antibodies against CRE (1:10,000; number 69050; Novagen), PECAM-1 (1:50,000; number 553370; Pharmingen), and β-catenin (1:100; number 9582; Cell Signaling). Immunofluorescent staining (IF) of cultured cells and lung paraffin sections was described previously (49, 53). Fluorescence was detected using a Zeiss Axioplan 2 Imaging Universal Microscope with an Axiocam MRm digital camera (Axiovision, release 4.3). Sizes of epithelial and mesenchymal areas in H&E-stained lung sections were measured using the scaling function in Axiovision Rel software. Staining for β-Gal activity was performed as described previously (48, 53). Sections were counterstained with Nuclear Fast Red (Vector Laboratories).
Distal lung epithelial tubules were obtained from cryosectioned E15.5 embryonic lungs by laser capture microdissection (LCM). The trunks of mouse embryos were embedded in OCT, snap-frozen in isopentane, and stored at −80°C. Tissue was sectioned at −20°C in the cryostat. Thin sections (7 μm) were collected on 1:20 poly-l-lysine-coated slides (P8920; Sigma-Aldrich) and stored at −80°C. Prior to laser capture microscopy, slides were fixed and counterstained using a HistoGene LCM frozen section staining kit (number KIT0401; Arcturus). Distal lung epithelial cells were captured from cryosectioned E15.5 embryonic lungs by a Veritas microdissection instrument (model 704; Molecular Devices, Sunnyvale, CA) with a laser set at 15 μm. Total RNAs were purified by the Arcturus PicoPure RNA isolation kit (Molecular Devices). RNAs were then subjected to purification and DNase treatment by an RNeasy minikit (Qiagen), followed by cDNA amplification using the high-capacity cDNA reverse transcription kit (Applied Biosystems). cDNA samples were preamplified using a TaqMan PreAmp Master Mix kit (number 4384267; Applied Biosystems) followed by qPCR analysis using inventoried TaqMan primers (Applied Biosystems) (Table 1). Reactions were analyzed in triplicate, and expression levels were normalized to β-actin (ACTB) mRNA levels and presented as means ± standard deviations (SD).
Small interfering RNA (siRNA) duplexes specific to human (siFoxM1; 5′-GGA CCA CUU UCC CUA CUU U-3′) and mouse Foxm1 (siFoxm1; 5′-GGA CCA CUU CCC UUA CUU U-3′) with 2-uracil (U) 3′ overhangs were synthesized by Dharmacon Research (Lafayette, CO). As a control for human siFoxM1, we used mutant FoxM1 siRNA (siCtrl; 5′-GGA CCU GUA UGC GUA CAU U-3′; mutated nucleotides are underlined) as described previously (50, 59). Transfection was performed using Lipofectamine 2000 (Invitrogen) in serum-free tissue culture medium by following the manufacturer's protocol. Four hours after transfection, serum-free medium was replaced with complete medium containing 10% fetal bovine serum as described previously (51, 52).
Total cell extract from A549 cells was harvested 72 h after siRNA transfection as described previously (50). Extracts from nuclear and cytosolic protein fractions were prepared using a nuclear/cytosol fractionation kit (K266100; BioVision, Mountain View, CA). Protein extracts were subjected to immunoblot analysis using antibodies against FOXM1 (1:1,000; clone C-20; Santa Cruz Biotechnology), AXIN2 (1:250; number 2151; Cell Signaling), activated β-catenin (1:500; clone 8E7; Millipore), β-actin (AC-15; 1:5,000; Sigma), phospho-p44/42 MAPK (Erk1/2) (1:2,000; number 4370; Cell Signaling), p44/42 MAPK (Erk1/2) (1:1500; number 4695; Cell Signaling), phospho-SAPK/JNK (Thr183/Tyr185) (1:250; number 9251; Cell Signaling), SAPK/JNK (1:1,000; number 9258; Cell Signaling), phospho-c-Jun (Ser63) (1:500; number 9261; Cell Signaling), and c-Jun (1:500; number 9165; Cell Signaling). MEK1/2 inhibitor U0126 (LC Laboratories) was used at a concentration of 10 μM (diluted in dimethyl sulfoxide [DMSO]) for 24 h.
TOPflash plasmid was purchased from Upstate Biotechnology (number 21-170). A mouse Axin2 promoter-luciferase (Luc) construct was obtained from Frank Costantini (Columbia University) (15). Cytomegalovirus (CMV)-β-catenin (CNNTB1) was kindly provided by Nickolai Dulin (The University of Chicago) (46). Forty-eight hours after transfection, cells were harvested for dual-luciferase assay (Promega) as described previously (51). CMV-Renilla luciferase plasmids were used as internal controls to normalize transfection efficiency. Promoter sequences for human and mouse Axin2 genes (−5 kb from the start of transcription) were obtained from the NCBI database and analyzed for regions of homology by mVISTA analysis (http://genome.lbl.gov/vista/index.shtml).
A chromatin immunoprecipitation (ChIP) assay was performed using in situ cross-linked mouse MLE-12 cells transfected with either Foxm1-specific siRNA or CMV-FoxM1b plasmid as described previously (42) (51). The amount of promoter DNA associated with IP chromatin was quantified by real-time PCR. The following sense (S) and antisense (AS) PCR primers were used to amplify mouse Axin2 promoter DNA fragments in ChIP assays: Axin2 −3061(S) (5′-CGT GCT TAT TGC TGG CGT TC-3′) and Axin2 −2920(AS) (5′-GTT CCT TTG TTG AGA GGG AGA AGA G-3′) (the annealing temperature [Ta] was 62.5°C); Axin2 −1885(S) (5′-GAG AGA GAG AGA GAA GAA TCA GGG G-3′) and Axin2 −1752(AS) (5′-GCC ACA GAG GTA TTG TCA CAG AAA C-3′) (Ta, 61°C); and Axin2 −659(S) (5′-AGG AAG GTG CTG ACT GAT GTG TAA G-3′) and Axin2 −554(AS) (5′-TGT TTG GAT GAA GAT GGG CG-3′) (Ta, 62.5°C).
Student's t test was used to determine statistical significance. P ≤ 0.05 was considered significant. Values for all measurements were expressed as the means ± SD.
Previous studies demonstrated that Foxm1 expression in respiratory epithelial cells is required for lung development (17). To examine molecular mechanisms regulated by Foxm1 in the embryonic lung, mouse embryos with a conditional deletion of Foxm1 from respiratory epithelium (SPC-rtTA/tetO-Cre/Foxm1fl/fl or epFoxm1−/−) were used. Embryos were treated with doxycycline (Dox) from embryonic day 7.5 (E7.5) to E15.5 to induce Cre recombinase. Cre staining was confined to distal respiratory epithelial cells and not detected in either proximal epithelial cells or pulmonary mesenchyme in Foxm1-deleted embryos (Fig. 1A) (17). mRNA analysis was performed on epithelial cells from E15.5 distal lung tubules obtained by laser capture microdissection (Fig. 1B). Dox-treated Foxm1fl/fl embryos were used as controls. Foxm1 mRNA was markedly decreased in distal lung tubules from epFoxm1−/− embryos (Fig. 1C), consistent with the loss of FOXM1 staining from distal epithelial cells (Fig. 1A). Deletion of Foxm1 markedly reduced Jnk1 and Polo-like kinase 1 (Plk1) mRNAs (Fig. 1C and andD),D), both known Foxm1 target genes (22, 49, 50). mRNAs encoding Fgf9 and noncanonical Wnt5a ligand (9, 15, 26) were reduced by deletion of Foxm1 (Fig. 1D and andE).E). In contrast, Gata6, Wnt7b, Sox4, Tcf3, and Tcf4 mRNAs were increased (Fig. 1D and andE),E), suggesting a compensatory activation of canonical Wnt signaling in Foxm1-deficient distal lung epithelium. Fgf1, Sfrp1, Sox2, Lef1, and cyclin D1 mRNAs were not altered (Fig. 1D and andE).E). Decreased mRNA of Axin2, an inhibitor of the Wnt/β-catenin signaling pathway, was consistent with reduced AXIN2 staining in distal lung epithelium of epFoxm1−/− mice (Fig. 1E and andH).H). AXIN2 staining was not changed in proximal epithelial tubules (Fig. 1H). Thus, deletion of Foxm1 from distal lung epithelial cells altered expression of genes associated with canonical Wnt signaling.
We next determined whether deletion of the mouse Foxm1 gene influences canonical Wnt/β-catenin activity in respiratory epithelium. TOPGAL reporter mice were bred with epFoxm1−/− mice to generate TOPGAL/SPC-rtTA/tetO-Cre/Foxm1fl/fl quadruple transgenic embryos. Embryos were treated with Dox from E7.5 to E15.5 to induce Cre recombinase and delete Foxm1 (Fig. 1F). At E15.5, β-Gal staining was detected in distal respiratory epithelial cells in control TOPGAL/Foxm1fl/fl and TOPGAL/SPC-rtTA/Foxm1fl/fl embryos (Fig. 1G). TOPGAL activity was dramatically increased in both distal and proximal lung epithelium of TOPGAL/SPC-rtTA/tetO-Cre/Foxm1fl/fl embryos (Fig. 1G), indicating that deletion of Foxm1 increased canonical Wnt signaling during lung development.
Activation of canonical Wnt signaling depends on the stabilization of β-catenin, its translocation to the nucleus, and formation of transcription complexes with nuclear partners, including T-cell factors (TCFs) and lymphoid enhancer-binding factors (LEFs) (4). To test whether Foxm1 inhibits β-catenin transcriptional activity, the TCF reporter plasmid TOPflash (19) was utilized. Cotransfection of CMV-β-catenin (CTNNB1) enhanced activity of TOPflash-induced Luciferase (Luc) in U2OS cells (Fig. 2A). CMV-FoxM1b decreased TOPflash reporter activity induced by CMV-β-catenin (Fig. 2A). Furthermore, depletion of Foxm1 by siRNA restored the ability of β-catenin to induce TOPflash (Fig. 2A). Thus, Foxm1 decreases transcriptional activity of β-catenin in vitro.
Since β-catenin is a key downstream effector in the canonical Wnt signaling, β-catenin was assessed in nuclei of Foxm1-depleted cells in vitro. Nuclear and cytoplasmic proteins were isolated from A549 cells transfected with either Foxm1 siRNA (siFoxM1) or control siRNA. siFoxM1 decreased Foxm1 mRNA and cyclin B1 mRNA, a known Foxm1 target (25, 54) (Fig. 2E). Activated β-catenin (dephosphorylated at Ser37 or Thr41) was increased in nuclear fractions isolated from Foxm1-depleted cells (Fig. 2B). Likewise, an increase in nuclear localization of β-catenin was detected by immunofluorescent staining after depletion of Foxm1 mRNA in both A549 and MLE-12 cells (Fig. 2C and andD).D). Thus, depletion of Foxm1 increased nuclear β-catenin and its activity in lung epithelial cells in vitro.
Depletion of Foxm1 in A549 cells increased c-Myc and cyclin D1 mRNAs, both of which are known targets of canonical Wnt/β-catenin signaling (4). Sox2 and Sox17 mRNA levels were also increased (Fig. 2E), consistent with previous studies of epFoxM1-delN transgenic mice (53). GSK3a and Axin1 mRNAs were not influenced by depletion of Foxm1 (Fig. 2E). Consistent with changes in mRNAs seen in laser capture microdissected epFoxm1−/− epithelium (Fig. 1D and andE),E), depletion of Foxm1 significantly increased TCF-4 mRNA and decreased JNK1 and Axin2 mRNAs (Fig. 2E and and3A3A and andC).C). Consistent with the decrease in Axin2 mRNA, a decrease in AXIN2 protein was detected in Foxm1-depleted cells (Fig. 3B). GSK3β protein was not influenced by Foxm1 depletion (Fig. 3B). Phosphorylated JNK1/2 and total JNK1/2 proteins were reduced after depletion of Foxm1 in A549 cells (Fig. 3B). Phosphorylation of c-Jun, a known target of activated phospho-JNK1/2, was also reduced, whereas total c-Jun was not altered (Fig. 3B). These results are consistent with published studies demonstrating that Foxm1 directly induces JNK signaling (50). Altogether, Foxm1 depletion caused nuclear translocation of activated β-catenin, expression of Wnt/β-catenin target genes, and reduced levels of the β-catenin inhibitor Axin2.
Four putative FOXM1 binding sites (−3380, −1980, −551, and −343) were identified in the mouse Axin2 promoter region (Fig. 3D). When human and mouse Axin2 gene kb −5 promoter sequences were compared, the potential binding site at −1980 was found to be in a highly conserved DNA region (data not shown). A chromatin immunoprecipitation (ChIP) assay was performed to determine whether endogenous FOXM1 protein binds to Axin2 promoter regions. Depletion of Foxm1 by siRNA reduced FOXM1 binding to the DNA fragment −2057/−1932, which contained the conserved FOXM1 binding site (Fig. 3D). Furthermore, transfection of CMV-FoxM1b plasmid significantly increased binding of the FOXM1b protein to the −2057/−1932 Axin2 promoter (Fig. 3D). FOXM1 binding was not detected in the Axin2 promoter region −3061/−2920 or −659/−554 (Fig. 3D).
In promoter luciferase assays, cotransfection of CMV-FoxM1b activated a luciferase reporter gene driven by the region at bp −2883 of the mouse Axin2 promoter (Fig. 3E) (15). The 6×CDX2-Luc plasmid, a known Foxm1 reporter (50, 58), served as a positive control for CMV-FoxM1b transcriptional activity (Fig. 3E). Thus, FOXM1 protein bound to and induced transcriptional activity of the Axin2 promoter, indicating that Axin2 is a direct target gene of FOXM1.
We used U0126, a pharmacological inhibitor of MEK1/2, to inhibit Ras/MEK/ERK signaling in vitro. Treatment of A549 cells with U0126 decreased phosphorylation of ERK1/2 but did not influence total ERK protein levels (Fig. 3F). Reduced protein levels of FOXM1 and AXIN2 were observed in U0126-treated cells (Fig. 3F), suggesting that FOXM1 and AXIN2 are downstream of MEK/ERK signaling. Reduced FOXM1 was associated with decreased phospho-JNK1/2 p46 and total JNK1/2 p46, whereas phospho-JNK1/2 p54 remained unaltered (Fig. 3F). Thus, pharmacological inhibition of the MEK/ERK pathway reduced FOXM1, AXIN2, and phospho-JNK1/2 p46 protein levels.
Expression of activated KrasG12D in the developing respiratory epithelium disrupts branching lung morphogenesis (45). Since Foxm1 is a known downstream target of the Ras/ERK signaling pathway in cultured tumor cells (30, 31), we tested whether Foxm1 is required for Kras signaling during lung development. Quadruple transgenic mouse embryos, SPC-rtTA/tetO-KrasG12D/tetO-Cre/Foxm1fl/fl (epKrasG12D/epFoxm1−/−), were produced in which activated KrasG12D was expressed in distal respiratory epithelium in the absence of Foxm1. In these mice, treatment of the dams with Dox from E7.5 to E15.5 induced simultaneous expression of KrasG12D and Cre DNA recombinase in respiratory epithelial cells of the developing fetuses, with the latter specifically deleting Foxm1 floxed alleles (Fig. 4A). Foxm1fl/fl embryos and transgenic embryos without Cre (epKrasG12D/Foxm1fl/fl) were used as controls. Consistent with previous studies, severe sacculation defects and enlarged distal epithelial tubules (cysts) were observed in Dox-treated epKrasG12D/Foxm1fl/fl mice at E15.5 (Fig. 4B) (45). Deletion of Foxm1 (epKrasG12D/epFoxm1−/−) significantly reduced the size of KrasG12D-expressing distal epithelial tubules (Fig. 4B) and improved the morphogenetic abnormalities in lung epithelium and mesenchyme (Fig. 4C).
FOXM1 protein was detected in both the lung epithelium and mesenchyme at E15.5 in control (Foxm1fl/fl) embryos (Fig. 5A). FOXM1 staining was maintained in the dilated lung tubules of epKrasG12D/Foxm1fl/fl embryos (Fig. 5A). In contrast, FOXM1 protein was absent and nuclear staining for Cre recombinase was present in distal respiratory epithelial cells of epKrasG12D/epFoxm1−/− mice (Fig. 5A and andB).B). These findings are consistent with efficient deletion of Foxm1 by Cre recombinase (Fig. 1C). Expression of FOXM1 in mesenchymal cells was not altered (Fig. 5A). TTF-1, a transcription factor expressed in respiratory epithelial cells, was present in both epKrasG12D/Foxm1−/− and control lungs and was not influenced by Foxm1 deletion (Fig. 5C). The presence of staining for distal respiratory epithelial cell markers proSP-C and SOX9 (Fig. 5D and andF)F) and the lack of proximal epithelial marker SOX2 (Fig. 5E) was consistent with the concept that epithelial cells lining the dilated lung cysts in epKrasG12D/Foxm1fl/fl mice were distal lung cells. Pulmonary capillary endothelium was present in lungs of epKrasG12D/epFoxm1−/− and control mice as indicated by expression of PECAM1 (Fig. 5H). Deletion of Foxm1 did not alter cell proliferation in KrasG12D-expressing mice, as indicated by Ki-67 staining (Fig. 5G). While AXIN2 staining was increased in distal (cystic) lung epithelium after overexpression of KrasG12D, Foxm1 deletion selectively reduced AXIN2 staining in distal lung epithelial tubules without influencing proximal epithelial tubules (Fig. 5I and andJ).J). These results are consistent with a critical role for Foxm1 in regulation of the Axin2 gene during development of distal lung epithelium. Altogether, deletion of Foxm1 attenuated the branching defects caused by activated KrasG12D without influencing cell proliferation or proximal-distal patterning.
To determine whether Foxm1 deletion prevents the sacculation defects caused by activated KrasG12D, dams were treated with Dox from E7.5 to E17.5. Consistent with previous findings (45), activation of KrasG12D caused large intrapulmonary cysts (Fig. 4D). Deletion of Foxm1 in KrasG12D-expressing embryos significantly decreased the number and size of intrapulmonary cysts and increased the number of small peripheral saccules (Fig. 4E). Thus, deletion of Foxm1 from the respiratory epithelium attenuated the severe lung abnormalities caused by activated KrasG12D, supporting the concept that Foxm1 is an important downstream target of Kras signaling during lung development.
Kras/MAPK and canonical Wnt/β-catenin signaling pathways play important roles in lung development (13, 27, 36, 45). To examine the relationship between these two signaling pathways, we bred SPC-rtTA/tetO-KrasG12D transgenic mice with mice containing the TOPGAL transgene, a reporter for canonical Wnt signaling in vivo (2, 7). TOPGAL/SPC-rtTA/tetO-KrasG12D triple transgenic embryos were used to determine whether activation of Kras influences canonical Wnt signaling during lung development (Fig. 6A). Dams were treated with Dox from E7.5 to E15.5 to induce KrasG12D expression in developing lung epithelium. In control (TOPGAL/SPC-rtTA) lungs, β-Gal was detected in epithelial cells throughout lung tubules (Fig. 6A). Reduced β-Gal activity was found in dilated distal epithelial cysts of TOPGAL/SPC-rtTA/tetO-KrasG12D embryos (Fig. 6A), indicating KrasG12D inhibited canonical Wnt signaling in distal respiratory epithelium. β-Gal activity was not detected in SPC-rtTA/tetO-KrasG12D lungs that lacked the TOPGAL transgene (Fig. 6A).
A transgenic mouse line expressing a constitutively active form of FOXM1 (FoxM1-delN) under the control of the lung epithelium-specific SFTPC promoter (SPC-rtTA/tetO-FoxM1-delN) was recently generated (53). Expression of the FoxM1-delN transgene during embryogenesis disrupted branching lung morphogenesis (Fig. 6B) (53), similar to effects caused by oncogenic KrasG12D (Fig. 6A). Given our previous data that Foxm1 is required for Kras signaling (Fig. 4), we tested whether expression of the FoxM1-delN mutant influenced canonical Wnt activity in TOPGAL/SPC-rtTA/tetO-FoxM1-delN triple transgenic embryos. While β-Gal activity was observed in respiratory epithelial cells of control TOPGAL/tetO-FoxM1-delN embryos, expression of activated FOXM1 mutant by Dox markedly decreased β-Gal staining in enlarged distal epithelial tubules of TOPGAL/SPC-rtTA/tetO-FoxM1-delN embryos (Fig. 6B and andD),D), consistent with the hypothesis that Kras/Foxm1 signaling inhibits the canonical Wnt pathway during the development of distal respiratory epithelium.
Finally, since the FoxM1-delN transgene was expressed in mosaic fashion (53), we determined whether FoxM1-delN colocalizes with β-Gal in the same cell. Transgene-derived FoxM1-delN protein in Dox-treated TOPGAL/SPC-rtTA/tetO-FoxM1-delN mice was visualized with an antibody specific to human FOXM1 (53). Activated human FoxM1-delN mutant protein was found in distal lung epithelium of TOPGAL/SPC-rtTA/tetO-FoxM1-delN embryos but was absent from proximal lung epithelium (Fig. 6C and data not shown). The FoxM1-delN protein was not detected in control TOPGAL/SPC-rtTA lungs (Fig. 6C). FoxM1-delN did not colocalize with β-Gal in distal respiratory epithelial cells (Fig. 6C). Altogether, FoxM1-delN inhibited the TOPGAL reporter, an indicator of canonical Wnt signaling in distal respiratory epithelium.
In this study, we demonstrated that expression of the activated FoxM1-delN mutant decreased TOPGAL expression, whereas elevated TOPGAL activity was observed in distal respiratory epithelium of Foxm1-deficient embryos. Thus, Foxm1 is a negative regulator of canonical Wnt signaling in the developing respiratory epithelium. Using TOPGAL transgenic mice, previous studies found that canonical Wnt activity was increased in branching epithelial tubules at early stages of lung development (8, 55). Later in development, canonical Wnt activity was decreased, preceding perinatal lung maturation and the development of alveolar saccules (55). Overexpression of activated β-catenin caused epithelial dysplasia and sacculation defects (35), indicating that downregulation of canonical Wnt activity is critical for lung maturation. Since Foxm1 is expressed in epithelial cells of embryonic lungs (53), Foxm1 may contribute to the downregulation of canonical Wnt activity in the developing respiratory epithelium.
Canonical Wnt signaling has been implicated in proximal/distal patterning in developing lungs. Deletion of β-catenin from lung epithelial cells reduced development of peripheral alveolar saccules but induced formation of enlarged and elongated epithelial tubules expressing markers of bronchiolar epithelial cells (36). Ectopic differentiation of alveolar type II-like cells was found in bronchioles of mice expressing activated β-catenin (35). Interestingly, reduced canonical Wnt activity was insufficient to alter the proximal/distal differentiation in mice expressing either activated FoxM1-delN (53) or activated KrasG12D transgenes (described above). Our results also suggest that activation of the Kras/Foxm1 pathway does not influence proximal/distal patterning in branching lung epithelium.
Axin2 is a direct transcriptional target of canonical Wnt/β-catenin signaling and has been shown to promote β-catenin proteasomal degradation as a part of the APC/GSK3/axin/β-catenin complex, providing a negative feedback mechanism for inhibition of Wnt signaling (15, 29). Surprisingly, we found that Axin2 was reduced after activation of canonical Wnt signaling in Foxm1-deficient distal lung epithelium. The reduction in Axin2 mRNA and protein seen in epFoxm1−/− distal tubules was consistent with the mRNA and protein data from Foxm1-depleted A549 cells. Interestingly, Axin2 staining was unaltered in proximal lung epithelium of epFoxm1−/− embryos, illustrating the complexity and cellular specificity in regulation of Axin2 by Foxm1 in vivo. These results suggest that Foxm1 is a more potent transcriptional activator of Axin2 than β-catenin. Reduced Axin2 expression in Foxm1-deficient distal epithelial cells may impair the negative feedback mechanism, resulting in stabilization of β-catenin and abnormal activation of Wnt signaling (Fig. 7). Consistent with this hypothesis, in vitro studies demonstrated that Foxm1 directly bound to and induced the transcriptional activity of the Axin2 promoter. Alternatively, it is possible that regulation of Axin2 by either Foxm1 or β-catenin is dependent on cell specificity or biological context.
Decreased expression of JNK1 was found in Foxm1-depleted A549 cells and laser capture microdissected epithelial cells from Foxm1-deficient mice. These results are consistent with previous studies demonstrating that Jnk1 is a direct transcriptional target of Foxm1 (39, 50). JNK1 antagonizes the canonical Wnt pathway by regulating the nucleocytoplasmic transport and proteasomal degradation of the β-catenin protein (14, 28). JNK1 reduced expression of Wnt6 and Wnt4 in embryonic stem cells, reducing the activity of the canonical Wnt/β-catenin pathway (1). In intestinal tissue and tumors of Jnk1-deficient mice, β-catenin expression was increased (14). Likewise, both expression levels and transcriptional activity of β-catenin were significantly increased in mouse embryonic fibroblasts from Jnk1-deficient mice (14), further demonstrating the negative regulation of the canonical Wnt/β-catenin pathway by Jnk1. Our results suggest that reduced expression of Jnk1 in Foxm1-depleted respiratory epithelium contributes to stabilization of the β-catenin protein, thus activating the canonical Wnt signaling pathway (Fig. 7).
Published studies demonstrated that transgenic overexpression of KrasG12D in distal lung epithelium disrupts branching lung morphogenesis by causing abnormal localization of phospho-ERK (45). In this study, we found that deletion of Foxm1 from distal epithelial cells attenuated branching defects caused by activated KrasG12D. These results suggest that Foxm1 is critical for Kras/ERK signaling during branching lung morphogenesis and are consistent with published in vitro studies demonstrating that ERK and CDK, both downstream targets of Kras, phosphorylate FOXM1 protein and induce its transcriptional activity (30, 31). Given the important roles for both Kras and Foxm1 in regulation of the cell cycle, it is surprising that no proliferation defects were found in distal lung epithelium of either KrasG12D transgenic embryos (45) or SPC-rtTA/TetO-Cre Foxm1−/− embryos (17). Our results suggest a proliferation-independent role for Kras/Foxm1 signaling during epithelial branching in the developing lung. Kras and Foxm1 may influence branching lung morphogenesis by regulating cell migration, cell-to-cell contacts, and epithelial polarity, all of which are critical for epithelial branching (34). It is also possible that other signaling pathways maintain epithelial proliferation after the loss of Kras/Foxm1 signaling. Published studies demonstrate that the canonical Wnt/β-catenin pathway is a major regulator of cell cycle regulatory genes (4). Since canonical Wnt activity was increased in Foxm1-depleted epithelium, the Wnt/β-catenin pathway may compensate for the loss of Kras/Foxm1 signaling by inducing cell cycle regulatory genes. Interestingly, proliferation defects were not found in either β-catenin-deficient (36) or Foxm1-deficient distal lung epithelium (17). Thus, it is possible that either Kras/Foxm1 or Wnt/β-catenin is sufficient to maintain cellular proliferation in embryonic respiratory epithelium. Our results are consistent with previous studies demonstrating that activation of the canonical Wnt signaling pathway bypasses the requirement for RTK/Ras signaling during vulval development in C. elegans (12).
Kras/MAPK and Wnt/β-catenin signaling are both well-described oncogenic pathways in mice and humans (4, 10, 16, 24, 38, 40, 41, 45). In contrast to our findings in developing respiratory epithelium, recent studies demonstrate that Wnt/β-catenin synergizes with oncogenic KrasG12D to induce lung tumors in adult mice (37). Wnt/β-catenin signaling was induced by FoxM1, accelerating the tumor growth during glioma tumorigenesis (61). These results strongly suggest that relationships between Kras/Foxm1 and canonical Wnt/β-catenin pathways are different in lung carcinogenesis compared to lung embryonic development. Therefore, it is possible that neoplastic transformation influences Foxm1-mediated cross talk between Kras/MAPK and canonical Wnt/β-catenin signaling pathways, providing additional advantages for tumor growth.
In summary, we demonstrated that Foxm1 transcription factor is required for Kras signaling in distal respiratory epithelium. During lung development, Foxm1 inhibits canonical Wnt signaling through transcriptional activation of Jnk1 and Axin2. Downregulation of canonical Wnt activity by Foxm1 is critical for proper lung morphogenesis. Foxm1 mediates cross talk between Kras/MAPK and canonical Wnt/β-catenin signaling pathways in the developing respiratory epithelium.
We thank Frank Costantini (Columbia University) for mouse Axin2 promoter luciferase plasmid and Nickolai Dulin (The University of Chicago) for CMV-β-catenin (CTNNB1) plasmids. We also thank Craig Bolte, Masato Nakafuku, Aaron Zorn, and John Shannon for critically reviewing the manuscript and Ann Maher for editorial services.
These studies were supported by NIH grants RO1 HL84151 (V.V.K.) and R01 CA142724 (T.V.K.), Research Scholar grant RSG-06-187-01 from the American Cancer Society (V.V.K.), and the Career Development Award from the National Lung Cancer Partnership (I.-C.W.).
Published ahead of print 23 July 2012