The forkhead box m1 transcription factor is essential for embryonic development of pulmonary vasculature.

Transgenic and gene knock-out studies demonstrated that the mouse Forkhead Box m1 (Foxm1 or Foxm1b) transcription factor (previously called HFH-11B, Trident, Win, or MPP2) is essential for hepatocyte entry into mitosis during liver development, regeneration, and liver cancer. Targeted deletion of Foxm1 gene in mice produces an embryonic lethal phenotype due to severe abnormalities in the development of liver and heart. In this study, we show for the first time that Foxm1(-/-) lungs exhibit severe hypertrophy of arteriolar smooth muscle cells and defects in the formation of peripheral pulmonary capillaries as evidenced by significant reduction in platelet endothelial cell adhesion molecule 1 staining of the distal lung. Consistent with these findings, significant reduction in proliferation of the embryonic Foxm1(-/-) lung mesenchyme was found, yet proliferation levels were normal in the Foxm1-deficient epithelial cells. Severe abnormalities of the lung vasculature in Foxm1(-/-) embryos were associated with diminished expression of the transforming growth factor beta receptor II, a disintegrin and metalloprotease domain 17 (ADAM-17), vascular endothelial growth factor receptors, Polo-like kinase 1, Aurora B kinase, laminin alpha4 (Lama4), and the Forkhead Box f1 transcription factor. Cotransfection studies demonstrated that Foxm1 stimulates transcription of the Lama4 promoter, and this stimulation requires the Foxm1 binding sites located between -1174 and -1145 bp of the mouse Lama4 promoter. In summary, development of mouse lungs depends on the Foxm1 transcription factor, which regulates expression of genes essential for mesenchyme proliferation, extracellular matrix remodeling, and vasculogenesis.

The lung mesenchyme undergoes vasculogenesis (formation of blood vessels de novo) and angiogenesis (branching of preexisting blood vessels) in a process requiring appropriate levels of vascular endothelial growth factor (VEGF), which stimulates mesenchyme proliferation and differentiation toward endothelial cell lineage (16,17). Targeted disruption of the VEGF or VEGF receptor type I (Flt1) or type II (Flk1) causes an embryonic lethal phenotype displaying impaired blood-island formation and delayed endothelial cell differentiation, leading to abnormal blood vessel development (18 -21). Overexpression of VEGF in the respiratory epithelium stimulates vasculogenesis in transgenic mouse lungs but results in aberrant vessel formation and increased expression of Flk1 (22). Differentiation of pulmonary mesenchyme also depends on proper expression of extracellular matrix proteins, including laminins, collagens, and integrins (1). Mice deficient in the laminin ␣4 (Lama4) gene exhibit impaired microvessel maturation and hemorrhage (23). Likewise, the disruption of Lama2 protein function causes defects in cell adhesion of a subpopulation of embryonic mesenchymal cells bearing a myofibroblast phenotype (24).
The Forkhead Box (Fox) proteins are an extensive family of transcription factors that share homology in the Winged Helix/ Forkhead DNA binding domain (25)(26)(27). Fox proteins Foxa2, Foxj1, Foxf1, and Foxp play important roles in regulating transcription of genes involved in branching lung morphogenesis and vasculogenesis during lung development (11, 12, 28 -33). Expression of the Foxm1 transcription factor (previously known as HFH-11B, Trident, Win, or MPP2) is induced during cellular proliferation and extinguished in terminally differentiated cells (34 -37). Partial hepatoctomy experiments demonstrated that mice with postnatal hepatocyte-specific deletion of the Foxm1 fl/fl (LoxP-targeted) allele exhibited a significant reduction in hepatocyte DNA replication and mitosis, which was associated with altered expression of proteins that limit Cdk1 and Cdk2 activity required for normal cell cycle progression (38). We recently demonstrated that Alb-Cre Foxm1 fl/fl hepatocytes are highly resistant to developing hepatocellular carcinoma following diethylnitrosamine/phenobarbital liver tumor induction (39). The mechanism of resistance to hepatocellular carcinoma development is associated with defects in cellular proliferation due to an aberrant increase in hepatocyte nuclear levels of Cdk inhibitor p27 Kip1 protein and diminished expression of the M-phase-promoting Cdc25B phosphatase (39). Interestingly, Foxm1 Ϫ/Ϫ embryos die in utero between 13.5 and 18.5 dpc due to severe defects in development of the embryonic liver and heart (40,41). Foxm1 Ϫ/Ϫ livers displayed abnormal accumulation of polyploid hepatoblasts resulting from diminished DNA replication and a failure to enter mitosis. This was associated with diminished protein levels of the Pololike kinase 1 (Plk-1) and Aurora B kinase (40), both of which phosphorylate regulatory proteins essential for orchestrating mitosis and cytokinesis (42,43). Foxm1 is required for differentiation of hepatoblast precursor cells toward biliary epithelial cell lineage, because Foxm1 Ϫ/Ϫ livers fail to develop intrahepatic bile ducts (40). Although it is well established that the Foxm1 protein is essential for hepatocyte proliferation and differentiation, the role of Foxm1 during embryonic lung development remains uncharacterized.
We previously reported on generating transgenic mice in which the Rosa26 promoter drives Foxm1 expression in all cell types and demonstrated that the Foxm1 is essential for cell proliferation required to repair lung injury in response to butylated hydroxytoluene treatment (44). Premature Foxm1 expression was associated with increased proliferation levels of pulmonary endothelial, epithelial, and smooth muscle cells as well as earlier expression of the cell cycle promoting cyclin A2 and cyclin B1 (44). In this paper, we demonstrated that Foxm1 Ϫ/Ϫ lungs displayed severe abnormalities in the development of pulmonary microvasculature that were associated with diminished pulmonary levels of the platelet endothelial cell adhesion molecule 1 (Pecam-1), TGF-␤ receptor type II, a disintegrin and metalloprotease domain 17 (ADAM-17) protein, VEGF receptors, Plk-1, Aurora B kinase, Lama4, and the Foxf1 transcription factor. Foxm1 is essential for proliferation of lung mesenchyme and vascular smooth muscle cells during embryonic lung development. Cotransfection experiments demonstrated that the Lama4 gene is a direct transcriptional target for Foxm1 transcription factor. These results suggest that Foxm1 regulates pulmonary genes essential for mesenchyme proliferation, extracellular matrix remodeling, and vasculogenesis during lung development.

Foxm1
Ϫ/Ϫ Mice-We previously described the generation of the Foxm1 ϩ/Ϫ mice in which the targeted allele lacked the DNA binding and transcriptional activation domains (40). These Foxm1 ϩ/Ϫ mice were bred for four generations into the C57BL/6 mouse genetic background. Foxm1 ϩ/Ϫ mice were mated in the evening, and vaginal plugs were checked in the morning. The noon of the day of appearance of a vaginal plug was designated as 0.5 dpc. Dams were killed by carbon dioxide asphyxiation to dissect Foxm1 Ϫ/Ϫ embryos at various days of gestation. Internal cartilage tissue samples were used for genotyping the Foxm1 allele by PCR analysis as described previously (40). 2 h before sacrifice, an intraperitoneal injection of phosphate-buffered saline containing 10 mg/ml 5-bromo-2Ј-deoxyuridine (BrdUrd, Sigma; 50 g/g body weight) was administered to pregnant females.
Foxa2 staining was performed using mouse monoclonal antibodies against Foxa2 (1:5, clone 4C7, University of Iowa Developmental Studies Hybridoma Bank, Iowa City, IA) followed by anti-mouse antibody conjugated with biotin, avidin-horseradish peroxidase complex, and diaminobenzidine substrate (all from Vector Laboratories). For immunofluorescent detection of ␣SM, antibody-antigen complexes were detected using fluorescein isothiocyanate-conjugated secondary antibodies and then counterstained with DAPI as described previously (39). To measure apoptosis in WT and Foxm1 Ϫ/Ϫ embryonic lungs, a TUNEL assay was performed using ApoTag Red in situ apoptosis detection kit from Intergen (Purchase, NY) according to the manufacturer's recommendations (39).
RNase protection assay was performed with [ 32 P]UTP-labeled antisense RNA synthesized from Flk1 or cyclophilin plasmid template with T7 RNA polymerase as previously described (48). RNA probe hybridization, RNase One (Promega, Madison, WI) digestion, electrophoresis of RNA protected fragments, and autoradiography were performed as described previously (49). Quantitation of expression levels was determined from scanned x-ray films by using the BioMax 1D program. The cyclophilin hybridization signals were used for normalization control among different lung RNA samples.
For EMSAs, Foxm1-expressing U2OS cells were used to prepare nuclear protein extract as described previously (51)(52)(53)(54). EMSAs were performed with 5 g of the U2OS nuclear extract and 5Ј end radioactively labeled double-stranded oligonucleotides from the mouse Lama4 promoter using binding reaction conditions as described previously (51,53,54). We used the following double-stranded oligonucleotides made to the Lama4 promoter region: Ϫ1180/Ϫ1139 bp, 5Ј-ATTGCTTGTTTT-GTTTATTTTGTTATTGTTTGTTTTGCCAAA-3Ј, and Ϫ891/Ϫ917 bp, 5Ј-GGCATGAAATTAAATAAAATAACTCAT-3Ј. The Foxm1 protein-DNA complex was separated from unbound labeled DNA using native polyacrylamide electrophoresis and this protein-DNA complex was visualized by autoradiography. For DNA competitions in EMSA, we added 500-fold molar excess of either cold competitor double-stranded oligonucleotide or nonspecific Sp1 oligonucleotide to the binding reaction. For antibody reactions, we preincubated nuclear extracts with 1 l of either mouse monoclonal Foxm1 antibody (55) or rabbit polyclonal Foxm1 antibody against C-terminal region of the human Foxm1 protein (amino acids 365-748) for 30 min at room temperature prior to adding the nuclear extract to the binding reaction as described previously (46,56). For antibody specificity control, the nuclear extract was preincubated with 1 l of monoclonal antibody against platelet-derived growth factor receptor ␣ chain (clone APA5, Pharmingen).
Statistical Analysis-Student's t test was used to determine statistical significance. p value Յ 0.05 was considered significant. Values for all of the measurements were expressed as the mean Ϯ S.D.

Foxm1
Ϫ/Ϫ Embryonic Lungs Displayed Hypertrophy of Pulmonary Arteries-Foxm1 ϩ/Ϫ mice were bred to generate Foxm1 Ϫ/Ϫ embryos, which were used to examine whether Foxm1 deficiency causes gross morphological defects in lung development. Embryonic 18.5-dpc Foxm1 Ϫ/Ϫ and WT lungs were fixed, paraffin-embedded, sectioned, and stained with H&E. Although Foxm1 Ϫ/Ϫ 18.5-dpc lungs exhibited normal size, lobular architecture, and sacculation compared with WT or Foxm1 ϩ/Ϫ littermates (Fig. 1, A-B, and data not shown), Foxm1 Ϫ/Ϫ lungs exhibited hypertrophy of vascular smooth muscle cells of the pulmonary arteries ( Fig. 1, C-D) as evidenced by the immunohistochemical staining with one of the follows antibodies specific for smooth muscle cells: ␣SM (Fig. 2, A-B), smooth muscle calponin (Fig. 2, C-D), and smooth muscle myosin heavy chain (Fig. 2, E-F). We also compared the size of arterial muscle layer in WT and Foxm1 Ϫ/Ϫ 18.5-dpc lungs to demonstrate that Foxm1 deficiency is associated with a 5-fold increase in arterial muscularity (Fig. 1I). Interestingly, many extrapulmonary arteries including the carotid artery and aorta in Foxm1 Ϫ/Ϫ embryos exhibited severe hypertrophy of smooth muscle cells (Fig. 2, G-H, and data not shown).
These Foxm1 Ϫ/Ϫ vascular smooth muscle cells displayed abnormally large DAPI-stained nuclei (Fig. 2, B and H) and diminished pulmonary expression of Plk-1 and Aurora B kinase (Fig. 1, E-H and J), a finding consistent with a polyploid genotype resulting from inhibition of mitosis with premature initiation of DNA replication (38,40). Interestingly, smooth muscle cells of the pulmonary bronchi in Foxm1 Ϫ/Ϫ embryos displayed normal morphology and ␣SM staining (Fig. 2, I-J, and data not shown), suggesting that there is a compensation for Foxm1 function in airway smooth muscle cells. Embryonic lungs from WT (A and C) and Foxm1 Ϫ/Ϫ (B and D) 18.5-dpc embryos were fixed, paraffin-embedded, sectioned, and then H&Estained. Pulmonary arteries (Ar) of the Foxm1 Ϫ/Ϫ embryos displayed severe hypertrophy of vascular smooth muscle cells (D). E-H, immunohistochemical staining of lung sections with antibodies specific to Plk-1 (G-H) or Aurora B proteins (E-F). Paraffin section from WT (E and G) and Foxm1 Ϫ/Ϫ (F and H) embryos were stained with primary antibody and then visualized by anti-mouse antibody conjugated with alkaline phosphatase and BCIP/NBT substrate followed by counterstaining with nuclear fast red. I, increased muscularity of Foxm1 Ϫ/Ϫ arteries. H&E-stained sections were used to measure the thickness of smooth muscle layers in WT and Foxm1 Ϫ/Ϫ 18.5-dpc pulmonary arteries (arrowheads in C and D) with vessel diameter 100 -300 m. Mean Ϯ S.D. was calculated using ten arteries in three different WT or Foxm1 Ϫ/Ϫ embryos. J, decreased numbers of Plk-1-positive and Aurora B-positive cells in Foxm1 Ϫ/Ϫ lungs. Plk-1-or Aurora B-stained lung sections were used to count the number of cells with positive immunostaining in five random ϫ400 microscope fields. Three different WT or Foxm1 Ϫ/Ϫ lungs were used to calculate the mean Ϯ S.D. A p value Յ 0.05 is shown with an asterisk. Abbreviations: Br, bronchi. Magnifications: ϫ100 (A and B); ϫ400 (C-F); and ϫ200 (G and H). Foxm1 Ϫ/Ϫ Lungs Exhibit Defects in Development of Pulmonary Microvasculature-Mouse 17.5-dpc lung mesenchyme undergoes extensive migration and differentiation to form peripheral pulmonary capillaries (1,15). Therefore, we used lung sections from 17.5-dpc Foxm1 Ϫ/Ϫ or WT mouse embryos to perform immunostaining with endothelial-specific Pecam-1 antibody to determine the role of Foxm1 in lung capillary development. Although a similar Pecam-1 staining was observed in pulmonary arteries of WT and Foxm1 Ϫ/Ϫ embryos, we observed a significant reduction in the number of large pulmonary vessels compared with aged-matched WT embryonic lungs (Fig. 3, A-B and E-F). In contrast, significant decreases in Pecam-1 staining in the peripheral capillaries of Foxm1 Ϫ/Ϫ lungs suggested that their lung microvasculature was disrupted (Fig. 3, A-B). Consistent with these results, pulmonary levels of the endothelialspecific VEGF receptor Flk1 mRNA were significantly reduced in Foxm1 Ϫ/Ϫ lungs as determined by the RNase protection assay (Fig. 3G). Interestingly, cell death did not contribute to decreased vasculogenesis in Foxm1 Ϫ/Ϫ lungs because they do not display elevated levels of apoptosis compared with WT littermates as determined by the TUNEL assay (Fig. 3, C-D). These results suggest that Foxm1 is essential for proper development of pulmonary vasculature from lung mesenchyme.
Foxm1 Ϫ/Ϫ Lungs Exhibited Defects in Mesenchyme Proliferation-Foxm1 Ϫ/Ϫ 15.5-dpc lungs exhibited normal tubular structure and the expression levels of epithelial-specific Foxa2 protein compared with WT lungs (Fig. 4, G-H). Although the pulmonary mesenchyme in Foxm1 Ϫ/Ϫ embryos displayed normal staining for the mesenchyme-specific protein vimentin (Fig. 4, I-J), diminished numbers of mesenchymal cells were observed in distal regions of the Foxm1 Ϫ/Ϫ lungs (Figs. 4, A-D, and 5A) without significant change in the number of epithelial cells (Fig. 5A). To determine the role of Foxm1 in mesenchyme proliferation, we measured DNA replication rates in 15.5-dpc Foxm1 Ϫ/Ϫ and WT lungs by injecting BrdUrd into pregnant female mice 2 h prior to harvesting the embryos. Immunohistochemistry with BrdUrd antibody demonstrated that Foxm1 Ϫ/Ϫ lungs exhibited decreased BrdUrd staining compared with aged-matched WT embryos (Fig. 4, E-F). Br-dUrd incorporation was significantly decreased in mesenchyme of Foxm1 Ϫ/Ϫ lungs, yet normal BrdUrd levels were still found in Foxm1 Ϫ/Ϫ epithelial cells (Fig. 5B). Consistent with reduced expression of M-phase promoting Plk-1 and Aurora B proteins in Foxm1 Ϫ/Ϫ 18.5-dpc lungs (Fig. 1, E-H and J), RT-PCR analysis displayed a significant decrease in pulmonary levels of  To detect apoptotic cells, paraffin lung sections were prepared from WT or Foxm1 Ϫ/Ϫ 17.5-dpc embryos and used for TUNEL assay. E-F, Foxm1 Ϫ/Ϫ lungs displayed a reduced number of pulmonary blood vessels. H&E-and Pecam-1-stained lung sections from 15.5-and 17.5-dpc Foxm1 Ϫ/Ϫ and WT embryos were used to count the number of pulmonary vessels (arteries and veins with diameter Ն25 m) in ϫ400 microscope field. Five random lung sections were counted in four different mouse lungs to calculate the mean number of large vessels Ϯ S.D. A p value Յ 0.05 is shown with an asterisk. G, Foxm1 Ϫ/Ϫ lungs exhibit diminished Flk1 expression. Total lung RNA was prepared from WT and Foxm1 Ϫ/Ϫ 17.5-dpc mouse embryos and used to analyze for Flk1 and cyclophilin mRNA levels by RNase protection assay. Each individual sample was normalized to its corresponding cyclophilin level as described under "Materials and Methods." Numbers represent averages of Flk1 mRNA levels with respect to WT embryonic lungs. Abbreviations: Br, bronchi; Ar, artery. Magnification: ϫ200 (A-D).

Foxm1 Is Essential for Lung Vascular Development
these genes in Foxm1 Ϫ/Ϫ 14.5-dpc embryos (Fig. 5C). These results demonstrate that Foxm1 deficiency reduces proliferation of mesenchymal cells in the developing lung.

Affymetrix Microarray Identified Diminished Expression of Genes Essential for DNA Replication and Mitosis in Foxm1
Ϫ/Ϫ Lungs-To determine additional target genes regulated by Foxm1 in embryonic lungs, we performed Affymetrix gene array analysis. Affymetrix cDNA probes were synthesized from total RNA prepared from WT or Foxm1 Ϫ/Ϫ 14.5-dpc lungs, which represent a developmental stage preceding the major vascular defects in Foxm1 Ϫ/Ϫ lungs. This analysis allowed the identification of 23 genes in which expression levels were altered Ն2.5-fold in Foxm1 Ϫ/Ϫ lungs compared with WT lungs (Table I). Altered Foxm1 Ϫ/Ϫ expression levels of nine genes were verified by RT-PCR analysis. Affymetrix and RT-PCR analysis confirmed undetectable levels of Foxm1 mRNA in Foxm1 Ϫ/Ϫ lungs (Table I and Fig. 5D). Consistent with decreased cellular proliferation in Foxm1 Ϫ/Ϫ lungs (Fig. 4, E-F), we observed a reduced expression of S-phase promoting cyclin A2 and M-phase promoting Cdc25C phosphatase and cyclin B1 genes (Table I), the latter of which is a known transcriptional target for Foxm1 protein (57).

Foxm1 Ϫ/Ϫ Lungs Exhibited Diminished TGF-␤ Signaling and Reduced Expression of Genes Essential for Lung Vascular Development-Vascular defects in Foxm1
Ϫ/Ϫ lungs were associated with decreased expression of the Pecam-1, VEGF receptor type I (Flt1), integrin ␤1, and the Foxf1 transcription factor as demonstrated by Affymetrix analysis and RT-PCR (Table I and Fig. 5D). These genes are essential for differentiation and assembly of endothelial cells into functional blood vessels (21, 31, 58 -60). Interestingly, Foxm1 Ϫ/Ϫ lungs displayed normal expression levels of midkine and Wnt7b genes ( Fig. 5D and data not shown), both of which were implicated in the development of smooth muscle hypertrophy of pulmonary vessels (61,62). Affymetrix analysis and RT-PCR demonstrated that the Foxm1 Ϫ/Ϫ lungs displayed a 90% reduction in the expression levels of a disintegrin and metalloprotease domain 17 (ADAM-17, Table I and Fig. 5D), a protein essential for enzymatic activation of multiple cell surface proteins including the tumor necrosis factor ␣ (63). Diminished levels of ADAM-17 may contribute to the vascular defects seen in Foxm1 Ϫ/Ϫ lungs, because ADAM-17 Ϫ/Ϫ mice displayed lung hypoplasia, delayed vasculogenesis, and respiratory distress (64).
Foxm1 Ϫ/Ϫ lungs exhibited severe reduction in expression levels of the bone morphogenetic protein-5 and the Frizzled homolog-6 (Fzd6) (Table I), the latter of which is a negative regulator of the Wnt/␤-catenin signaling pathway (65). Con-  (Table I), we observed a 60 -80% reduction in pulmonary levels of procollagens type V ␣2 and type XII ␣1 and cadherin 11 genes (Table I and Fig. 5D), all of which are known downstream targets for TGF-␤ signaling pathway (66 -68). Interestingly, we also observed a 60% reduction in Foxm1 Ϫ/Ϫ expression levels of Smad3 (Table I), an intracellular mediator of the TGF-␤ signaling pathway (69). These results suggest that Foxm1 Ϫ/Ϫ lung phenotype is associated with diminished TGF-␤ signaling and reduced expression of extracellular matrix proteins that are regulated by the TGF-␤ signaling pathway. Finally, increased expression of several genes was observed in Foxm1 Ϫ/Ϫ lungs including the DNA repair enzyme AP endonuclease 2, intracellular signaling proteins adenylyl cyclase 3, and phospholipase C␦, and the homeobox transcription factors HoxC9 and paired mesoderm homeobox 2b (Pmx2b) ( Table I).
Foxm1 Ϫ/Ϫ Lung Phenotype Is Associated with Reduced Expression of Laminin ␣2 and ␣4 Genes-Affymetrix gene array demonstrated that Foxm1 Ϫ/Ϫ lungs displayed severely reduced pulmonary levels of Lama2 and Lama4 genes (Table I), a result confirmed by RT-PCR analysis (Fig. 5D). An analysis of the cellular expression pattern of the Lama4 protein by immunohistochemical staining revealed that Lama4 was abundantly expressed in the distal lung regions including subsets of both mesenchymal and epithelial cells of WT 15.5-dpc lung (Fig. 6, A  and C). In 18.5-dpc WT lungs, the expression of the Lama4 protein was also detected in endothelial cells of pulmonary blood vessels (Fig. 6E). Lama4 expression was severely reduced in Foxm1 Ϫ/Ϫ lung regions within both the mesenchyme (Fig. 6,  B, D, and G) and mesenchyme-derived endothelial cells of Foxm1 Ϫ/Ϫ pulmonary arteries (Fig. 6F). Vascular defects in Foxm1 Ϫ/Ϫ lungs were associated with reduced pulmonary levels of Lama2 and Lama4, both of which are essential for epithelial-mesenchymal signaling and mesenchyme differentiation in the developing lung (23,24).
Foxm1 Directly Regulates Mouse Lama4 Promoter-Because Foxm1 Ϫ/Ϫ lungs exhibited significantly reduced Lama4 mRNA and protein levels (Table I and Figs. 5D and 6), we investigated whether Foxm1 transcriptionally regulates the Lama4 promoter region. Two potential Foxm1 DNA binding sites were identified in Ϫ1.2-kb promoter region of the mouse Lama4 gene (Ϫ897/ Ϫ911 and Ϫ1174/Ϫ1145 bp) (Fig. 7A), the latter of which contains two overlapping Foxm1 binding motifs (Fig. 7C). Cotransfection experiments were performed in mesenchymal U2OS cells using CMV-Foxm1 expression vector and LUC reporter driven by the Ϫ1200-bp Lama4 promoter region. Cotransfection of the Foxm1 expression vector increased the expression of the Ϫ1200-bp Lama4-LUC reporter plasmid (Fig. 7B), suggesting that the Foxm1 protein is a transcriptional activator of Lama4 gene. Furthermore, deletions of Ϫ1140/Ϫ1200-, Ϫ920/Ϫ1200-, or Ϫ880/Ϫ1200-bp Lama4 regions were equally sufficient to reduce the ability of Foxm1 to activate transcription of the Lama4 promoter in cotransfection experiments (Fig. 7B), indicating that the Ϫ1140/Ϫ1200-bp Lama4 region contains a functional Foxm1 binding site.
To determine whether Foxm1 protein directly binds to DNA in the Lama4 promoter region, we performed EMSA with nuclear protein extract from Foxm1-expressing U2OS cells and two distinct oligonucleotides (Ϫ1180/Ϫ1139 and Ϫ891/Ϫ917 bp) that contain potential Foxm1 binding sites in the promoter region of the mouse Lama4 gene. Although the Ϫ891/Ϫ917-bp oligonucleotide did not bind to endogenous Foxm1 protein (data not shown), two overlapping Foxm1 binding sites in the Ϫ1180/ Ϫ1139 bp region formed specific Foxm1 protein-DNA complexes (Fig. 7C), as demonstrated by the disruption of the Foxm1 protein-DNA complex with either Foxm1 antibody or cold competitor oligonucleotide (Fig. 7C). The formation of Foxm1 protein-DNA complexes was not inhibited by the addition of either antibody to the platelet-derived growth factor receptor ␣ chain or nonspecific oligonucleotide, which contains a binding site for SP1 transcription factor (Fig. 7C). These results indicate that the endogenous Foxm1 protein specifically binds to DNA in Ϫ1180/Ϫ1139-bp Lama4 promoter region, which is required for transcriptional induction of the Lama4 promoter by Foxm1.

Foxm1
Ϫ/Ϫ embryos die in utero between 13.5 and 18.5 dpc because of severe defects in liver morphogenesis due to diminished hepatoblast DNA replication and a failure to enter mitosis causing polyploid phenotype (40). In this study, we demonstrate that embryonic Foxm1 Ϫ/Ϫ lungs exhibit a severe hypertrophy of arteriolar smooth muscle cells and a paucity of distal mesenchymal cells resulting from diminished proliferation of lung mesenchyme. The hypertrophy of vascular smooth muscle cells is consistent with decreased expression of the Polo-like kinase 1 and Aurora B kinase (Ref. 40 and this report), which contribute to a failure to complete mitosis and the reinitiation of the S-phase, resulting in a polyploid phenotype (42,43). An alternative explanation for muscle hypertrophy may result from severe heart defects in Foxm1 Ϫ/Ϫ embryos, which cause a diminished circulatory output and altered signaling during development of blood vessels. The finding that many extrapulmonary arteries in Foxm1 Ϫ/Ϫ embryos exhibited the hypertrophy of smooth muscle cells provides further support for this concept.
Foxm1 Ϫ/Ϫ lungs displayed reduced levels of Cdc25C phosphatase and cyclin B1, the latter of which is a direct transcriptional target for Foxm1 (57). Progression into mitosis requires the activation of Cdk1 through assembly with cyclin B regula-tory subunit and the removal of Cdk1 inhibitory phosphates at Thr-14 and Tyr-15 by the Cdc25B and Cdc25C phosphatases (70 -72). Therefore, diminished pulmonary levels of cyclin B1 and Cdc25C can decrease Cdk1 activation and M-phase progression, causing the hypertrophy of vascular smooth muscle cells in Foxm1 Ϫ/Ϫ lungs.
Proliferation defects in Foxm1 Ϫ/Ϫ mesenchyme were also associated with reduced expression of S-phase promoting cyclin A2, which activates Cdk2 protein required for DNA replication (73,74). Cdk2 complexes with either cyclin E or cyclin A cooperate with cyclin D-Cdk4/6 to phosphorylate the retinoblastoma protein, which releases the bound E2F transcription factor and allows it to stimulate expression of genes required for DNA replication (75,76). Therefore, defects in DNA replication in Foxm1 Ϫ/Ϫ lungs may be a direct consequence of the reduced pulmonary levels of cyclin A2. This result is consistent with previously published data (44) demonstrating that transgenic overexpression of Foxm1 in Rosa26-Foxm1 transgenic mice is associated with increased cyclin A2 levels and premature proliferation of different lung cell types following butylated hydroxytoluene lung injury. In the lung mesenchyme of Foxm1 Ϫ/Ϫ embryos, we also found significant reduction in the expression of the mitotic regulators Aurora B kinase and Pololike kinase 1, a finding consistent with diminished progression into mitosis as reported in our previous developmental studies of embryonic Foxm1 Ϫ/Ϫ livers (40). Our results suggest that Foxm1 controls the expression of pulmonary cell cycle regulatory pathways required for proliferation of pulmonary mesenchymal cells and vascular smooth muscle cells during lung morphogenesis (Fig. 8).
Foxm1 Ϫ/Ϫ lungs displayed normal branching and sacculation, and the pulmonary levels of epithelial-specific Foxa2 protein were normal in Foxm1 Ϫ/Ϫ embryos. Furthermore, detectable BrdUrd levels were observed in Foxm1 Ϫ/Ϫ pulmonary epithelium, suggesting that the function of Foxm1 protein in proliferation may be compensated, thus allowing normal epithelial proliferation during branching lung morphogenesis. Interestingly, Foxm1 Ϫ/Ϫ lungs displayed reduced expression of the TGF-␤ receptor type II and Smad3 genes, suggesting that Foxm1 Ϫ/Ϫ lung phenotype is associated with reduced TGF-␤ signaling. Consistent with this hypothesis, we observed diminished pulmonary levels of procollagens type V ␣2 and type XII ␣1 and cadherin 11 genes, all of which are known downstream targets for TGF-␤ signaling pathway (66 -68). Published studies demonstrated that the TGF-␤ inhibits pulmonary branching morphogenesis in culture (1), and abrogation of the TGF-␤ receptor II signaling, either with antisense oligonucleotides or with blocking antibodies, stimulates pulmonary epithelial proliferation and lung morphogenesis (77,78). Furthermore, abrogation of Smad2 and Smad3 expression results in a strong gain of function phenotype for epithelial proliferation and lung branching in embryonic lung culture experiments (79). Because lung branching morphogenesis was not affected in Foxm1 Ϫ/Ϫ embryos, diminished TGF-␤ signaling may provide a compensatory mechanism to stimulate epithelial proliferation in Foxm1 Ϫ/Ϫ lungs. Whether Foxm1 directly or indirectly regulates the expression of the TGF-␤ receptor II and Smad3 in pulmonary epithelial cells remains to be determined.
In the canalicular stage (16.5-17.5 dpc) of lung morphogenesis, the lung mesenchyme undergoes extensive differentiation toward endothelial cell lineage. In addition to proliferation defects in 17.5-dpc Foxm1 Ϫ/Ϫ lung mesenchyme, we found severe defects in the formation of pulmonary microvasculature. Foxm1 Ϫ/Ϫ lungs exhibited reduced Pecam-1 staining, diminished number of pulmonary blood vessels, and decreased mRNA levels of VEGF receptor type II (Flk1), which is essen- Foxm1 Is Essential for Lung Vascular Development tial for vessel formation and differentiation of endothelial cells (20). Interestingly, the mesenchyme of Foxm1 Ϫ/Ϫ 15.5-dpc lungs displayed normal levels of mesenchyme-specific protein vimentin but it failed to form peripheral pulmonary microvasculature at later stages of lung development, indicating that Foxm1-deficient mesenchyme is unable to differentiate into capillary endothelial cells (Fig. 8). Furthermore, diminished vasculogenesis in Foxm1 Ϫ/Ϫ lungs was accompanied by an 80% reduction in the expression of Foxf1 transcription factor. A previous study (31) demonstrated that the abnormal lung capillary development, pulmonary edema, and perinatal lethality were found in the ϳ55% newborn Foxf1 ϩ/Ϫ mice, which exhibited an 80% reduction in wild-type pulmonary Foxf1 levels. These defects in newborn Foxf1 ϩ/Ϫ lungs were associated with diminished expression of Flk1, Pecam-1, and other genes required for development of pulmonary vasculature (31), which is partially similar to gene expression defects in Foxm1 Ϫ/Ϫ lungs (Table I and Figs. 3 and 5D). Interestingly, the development of alveolar capillaries and expression of these lung genes was unchanged in 40% newborn Foxf1 ϩ/Ϫ mice that had near WT pulmonary levels of Foxf1 mRNA, indicating that the WT Foxf1 levels are critical for lung vasculogenesis (31). These studies suggest that reduction of Foxf1 levels in Foxm1 Ϫ/Ϫ embryos may be contributing to vascular defects during lung morphogenesis.
Foxm1 Ϫ/Ϫ lungs displayed severe reduction in pulmonary levels of Lama2 and Lama4 genes, both of which are essential for mesenchyme differentiation during lung development (23,24). Lama4 levels were severely reduced in both lung mesenchyme and mesenchyme-derived endothelial cells of Foxm1 Ϫ/Ϫ lungs, and this was associated with severe vascular defects. Because Lama4 Ϫ/Ϫ mice displayed impaired microvessel maturation and vascular leakage (23), diminished Lama4 expression may contribute to vascular abnormalities and edema seen in Foxm1 Ϫ/Ϫ embryos (Ref. 40 and this report). We also demonstrated that Foxm1 protein induces transcriptional activity of the mouse Lama4 promoter in cotransfection experiments, suggesting that the Lama4 gene is a direct transcriptional FIG. 7. Foxm1 directly regulates mouse Lama4 promoter through ؊1174/؊1145-bp Lama4 DNA region. A, schematic drawing of Lama4 promoter constructs. Schematically shown are LUC reporter constructs that use either Ϫ1200-bp Lama4 or one of the Lama4 deletion mutants to drive expression of the LUC reporter. B, the Foxm1 induces Lama4 promoter activity in cotransfection assays. We transiently transfected U2OS cells with CMV-Foxm1 expression vector and the Ϫ1200-bp Lama4, Ϫ1140-bp Lama4, Ϫ920-bp Lama4, or the Ϫ880-bp Lama4 LUC reporter plasmids. Cells were harvested at 36 h after transfection and processed for dual luciferase assays to determine luciferase activity. Transcriptional induction is expressed as a fold increase relative to CMV-empty vector (ϮS.D.). A p value Յ 0.05 is shown with an asterisk. C, EMSAs show that endogenous Foxm1 protein binds to its potential binding sites in Ϫ1174/Ϫ1145-bp Lama4 DNA region. Nuclear protein extract was prepared from U2OS cells, which express endogenous Foxm1 protein. EMSA was performed with the Ϫ1180/Ϫ1139-bp double-stranded oligonucleotide containing two overlapping Foxm1 binding sites. Specificity of the Foxm1 protein-DNA complexes was demonstrated by the ability of the cold competitor DNA (C, 500-fold molar excess) but not nonspecific Sp1 oligonucleotide to interfere with the formation of Foxm1 protein-DNA complexes (shown with arrows). For antibody reactions, we preincubated nuclear extracts with 1 l of either mouse monoclonal Foxm1 antibody (Ab1) or rabbit polyclonal Foxm1 antibody (Ab2) prior to adding the nuclear extract to the binding reaction. Monoclonal antibodies against platelet-derived growth factor receptor ␣ chain were used for antibody specificity control (nsAb). Abbreviation: NE, control without nuclear extract.
FIG. 8. Schematic diagram showing the regulation of lung mesenchyme development by Foxm1. Immature lung mesenchyme undergoes proliferation and differentiation toward endothelial, vascular smooth muscle, and airway smooth muscle cell lineages. Foxm1 is essential for proliferation of endothelial and vascular smooth muscle cells but not for epithelial or airway smooth muscle cells. Foxm1 positively regulates differentiation of pulmonary mesenchyme during the formation of lung microvasculature. target of Foxm1 during lung development. Further support for this concept provides the fact that endogenous Foxm1 protein specifically binds to the Ϫ1180/Ϫ1139-bp region of the mouse Lama4 promoter and that retention of these sequences are essential for transcriptional activation by Foxm1. This is the first evidence that proliferation-specific Foxm1 transcription factor may directly regulate the expression of extracellular matrix proteins.
In summary, Foxm1 Ϫ/Ϫ embryonic lungs exhibited diminished mesenchyme proliferation, hypertrophy of smooth muscle cells of pulmonary arteries, and severe abnormalities in development of pulmonary microvasculature. The Foxm1 Ϫ/Ϫ lung phenotype was associated with diminished pulmonary levels of Pecam-1, ADAM-17, Foxf1, Plk-1, Aurora B, and Lama4 proteins, and reduced TGF-␤ signaling. Cotransfection experiments demonstrated that Lama4 gene is a direct transcriptional target for Foxm1 protein. These results suggest that Foxm1 is essential for lung morphogenesis by regulating the expression of genes required for extracellular matrix remodeling, mesenchyme proliferation, and vasculogenesis.