|Home | About | Journals | Submit | Contact Us | Français|
The Forkhead Box m1 (Foxm1 or Foxm1b) transcription factor (previously called HFH-11B, Trident, Win, or MPP2) is expressed in a variety of tissues during embryogenesis, including vascular, airway and intestinal smooth muscle cells (SMC). Although global deletion of Foxm1 in Foxm1−/− mice is lethal in the embryonic period due to multiple abnormalities in the liver, heart and lung, the specific role of Foxm1 in SMC remains unknown. In the present study, Foxm1 was deleted conditionally in the developing SMC (smFoxm1−/− mice). The majority of smFoxm1−/− mice died immediately after birth due to severe pulmonary hemorrhage, and structural defects in arterial wall and esophagus. Although Foxm1 deletion did not influence SMC differentiation, decreased proliferation of SMC was found in smFoxm1−/− blood vessels and esophagus. Depletion of Foxm1 in cultured SMC caused G2 arrest and decreased numbers of cells undergoing mitosis. Foxm1-deficiency in vitro and in vivo was associated with reduced expression of cell cycle regulatory genes, including cyclin B1, Cdk1-activator Cdc25b phosphatase, Polo-like 1 and JNK1 kinases, and cMyc transcription factor. Foxm1 is critical for proliferation of smooth muscle cells and is required for proper embryonic development of blood vessels and esophagus.
Embryonic development of blood vessels depends on proper vasculogenesis (formation of blood vessels de novo) and angiogenesis (branching of preexisting blood vessels). While the majority of vascular cell types develop from the mesoderm (Saint-Jeannet et al., 1992), smooth muscle cells (SMC) may arise from multiple origins, including mesodermal cells, neural crest cells, hematopoietic stem cells, pericytes and endothelial cells (Thayer et al., 1995; Gittenberger-de Groot et al., 1999; Hao et al., 2003). In contrast to terminally differentiated skeletal and cardiac muscle cells, mature SMC retain the ability to change their phenotype in response to changing environment conditions (Halayko and Solway, 2001). Several signaling pathways are critical for proper development of blood vessels, including vascular endothelial growth factor (VEGF), angiopoietins, transforming growth factor β (TGF-β), fibroblast growth factors (FGFs), platelet-derived growth factor (PDGF), Ephrin/Eph receptor, Sonic hedgehog (Shh)/ Patch, and Delta-ligand/ Notch-receptors (Carmeliet et al., 1996; Rossant and Howard, 2002; Chi et al., 2007), as well as both canonical (Wnt7b) and Ca2+-dependent (Wnt4) Wnt-signaling pathways (Shu et al., 2002; Itaranta et al., 2006). These signaling proteins regulate proliferation, differentiation and migration of SMC in the vascular wall by inducing expression of transcription factors critical for regulation of SM-specific gene promoters. These transcription factors include the serum response factor (SRF), myocardin and myocardin-related transcriptional factors that act as SRF activators to induce expression of SRF target genes (Wang et al., 2001; Camoretti-Mercado et al., 2003; Pipes et al., 2006). Development of blood vessels also depends on MEF2B, P311 (C5orf13), MRF2 (ARID5B) and GATA4 transcription factors that are essential for differentiation of mesenchymal cells toward SMC cell lineage (Katoh et al., 1998; Belaguli et al., 2000; Pan et al., 2002; Watanabe et al., 2002; Kumar and Owens, 2003).
The Forkhead Box (Fox) proteins are an extensive family of transcription factors that share homology in the Winged Helix/Forkhead DNA binding domain (Clark et al., 1993; Clevidence et al., 1993; Kaestner et al., 1993). Foxm1 transcription factor (previously known as HFH-11B, Trident, Win, or MPP2) is expressed in all tissues during embryogenesis but its expression in adult mice is restricted to intestinal crypts, thymus and testes (Korver et al., 1997; Ye et al., 1997). During organ injury, Foxm1 expression is induced in a variety of cell types, including epithelial, endothelial and smooth muscle cells (Ye et al., 1997; Kalinichenko et al., 2003). Foxm1 expression is increased in tumor cells during progression of liver, lung, colon and prostate cancers (Kalinichenko et al., 2004; Kalin et al., 2006; Kim et al., 2006; Yoshida et al., 2007). In our previous studies we demonstrated that Foxm1−/− mice die in utero between embryonic day 13.5 (E13.5) and E16.5 due to multiple abnormalities in development of the embryonic liver, lung, and heart (Krupczak-Hollis et al., 2004; Kim et al., 2005a; Ramakrishna et al., 2007). Abnormal accumulation of polyploid cells, resulting from diminished DNA replication and failure to enter mitosis, was observed in these Foxm1−/− mouse organs (Korver et al., 1998; Krupczak-Hollis et al., 2004; Ramakrishna et al., 2007). Foxm1 is required for differentiation of hepatoblast precursor cells toward the biliary epithelial cell lineage, and Foxm1−/− livers fail to form intrahepatic bile ducts (Krupczak-Hollis et al., 2004). Likewise, Foxm1−/− embryos exhibit defects in differentiation of pulmonary mesenchyme into mature capillary endothelial cells during the canalicular stage of lung development (Kim et al., 2005a).
Given the importance of Foxm1 for differentiation and proliferation of many cell lineages during embryonic development, organ injury and carcinogenesis, several recent studies began to address cell-specific roles of Foxm1 using conditional knockout mouse models. Hepatocyte-specific deletion of Foxm1 in Albumin-Cre Foxm1fl/fl mice prevented formation of hepatocellular carcinoma (HCC) in adult mice by inducing p27kip1 tumor suppressor in HCC tumor cells and normal hepatocytes (Kalinichenko et al., 2004). In response to the inflammatory mediator LPS, mice with endothelial-specific Foxm1 deletion (Tie2-Cre Foxm1fl/fl mice) displayed significantly increased lung vascular permeability and markedly increased mortality (Zhao et al., 2006). A conditional deletion of Foxm1 from precursors of cerebellar granule neurons caused a delay in brain development by interfering with Shh-induced neuroproliferation (Ueno et al., 2008). Mice with pancreatic-specific deletion of Foxm1 displayed severe abnormalities in postnatal β-cell mass expansion, causing impaired islet function and diabetes by 9 week of age (Zhang et al., 2006). These mice also exhibited specific impairments in β-cell mass regeneration and islet growth after partial pancreatectomy, with reduced proliferation of α- and β-cells but no impairments in acinar or ductal cell proliferation (Ackermann Misfeldt et al., 2008). In our recent studies (Kalin et al., 2008), Foxm1 was deleted conditionally in the respiratory epithelium during lung development (SPC-rtTA/ TetO-Cre Foxm1fl/fl mice). Deletion of Foxm1 in the respiratory epithelium did not alter epithelial proliferation but inhibited lung maturation and caused respiratory failure after birth (Kalin et al., 2008). Maturation defects in Foxm1-deficient lungs were associated with a delay of type I cell differentiation and decreased expression of surfactant-associated proteins A, B, C, and D, indicating that Foxm1 is critical for surfactant homeostasis and lung maturation prior to birth and is required for adaptation to air breathing (Kalin et al., 2008). Although Foxm1 plays distinct roles in differentiation and proliferation in various cell lineages, the specific role of Foxm1 in smooth muscle cells remains unknown.
In this study, we used mouse embryos to demonstrate that Foxm1 is expressed in vascular smooth muscle and airway smooth muscle cells as well as in smooth muscle layers of esophagus, stomach and intestine. To investigate the role of Foxm1 in smooth muscle cells in vivo, we generated transgenic mice in which Foxm1-floxed allele was conditionally deleted using Cre recombinase transgene driven by a smooth muscle myosin heavy chain promoter (smMHC-Cre Foxm1fl/fl, or smFoxm1−/− mice). We demonstrated that a majority of smFoxm1−/− mice died immediately after birth due to severe pulmonary hemorrhage, structural defects in arterial wall, and esophageal abnormalities. Although Foxm1 deletion did not influence differentiation of smooth muscle cells, decreased cellular proliferation was observed in muscle layers of embryonic blood vessels and esophagus. In vitro experiments demonstrated that siRNA-mediated depletion of Foxm1 in cultured smooth muscle cells caused G2 arrest and decreased numbers of cells undergoing mitosis. Depletion of Foxm1 in vitro and in vivo was associated with reduced expression of cell cycle regulatory genes, including cyclin B1, Cdk1-activator Cdc25b phosphatase, Polo-like 1 and JNK1 kinases, and cMyc transcription factor. Our studies suggest that Foxm1 is required for proper development of blood vessels and esophagus by regulating smooth muscle genes essential for the cell cycle regulation.
We previously described the generation of Foxm1flox/flox (Foxm1fl/fl) mice, which contain LoxP sequences flanking DNA binding and transcriptional activation domains of the Foxm1 gene (Krupczak-Hollis et al., 2004). The Foxm1fl/fl mice were bred for 10 generations into C57Bl/6 mouse genetic background. The Foxm1fl/fl female mice were bred with smMHC-Cre-GFPtg/− C57Bl/6 male mice (obtained from Dr. Kotlikoff, Cornell University, (Xin et al., 2002)) to generate the smMHC-Cre-GFPtg/− Foxm1fl/fl double transgenic mice (smooth muscle-specific Foxm1 knockout mice or smFoxm1−/−). Although a majority of smFoxm1−/− embryos died immediately after birth, 13% of smFoxm1−/− mice survived through birth and were capable of mating. Foxm1fl/fl littermates lacking the smMHC-Cre-GFP transgene were used as controls. Further controls included double-heterozygous smMHC-Cre-GFPtg/− Foxm1fl/+ mice. No embryonic abnormalities were observed in control mice. smMHC-Cre-GFPtg/− Foxm1fl/fl mice were bred with either TOPGAL transgenic mice (obtained from J. Whitsett's lab, Cincinnati Children's Hospital Medical Center) or R26R reporter mice (Rosa26-LoxP-stop-LoxP-β-galactosidase mice; purchased from Jackson Lab) to study canonical Wnt signaling or Cre-mediated recombination, respectively. Animal studies were reviewed and approved by the Animal Care and Use Committee of Cincinnati Children's Hospital Research Foundation.
smFoxm1−/− and control Foxm1fl/fl embryos were harvested, fixed overnight with 10% buffered formalin, and then embedded into paraffin blocks. Paraffin 5 μm sections were either used for immunohistochemistry or stained with hematoxylin and eosin (H&E) for morphological examination as described (Kim et al., 2005b). The following antibodies were used for immunostaining: Cre (1:25,000; #69050-3; Novagen); Ki-67 (1:500; clone Tec-3; Dako); GFP (1:15,000; clone JL-8; Clontech); activated caspase 3 (1:200; 5A1; Cell signaling), α-smooth muscle actin (αSM; 1:15,000; clone 1A4; Sigma), γ-smooth muscle actin (γSM; 1:500; clone B4; Seven Hills Bioreagents, Cincinnati, OH), smooth muscle calponin (1:5000; clone hCP; Sigma), smooth muscle myosin heavy chain (1:250; clone HSM-V; Sigma), Pecam-1 (1:300, clone MEC 13.3, Pharmingen), Foxa2 (1:2,000, Seven Hills Bioreagents, Cincinnati, Ohio), proSPC (1:1,500, AB-3428, Chemicon International), CCSP (1:1,000, a gift from Dr. Stripp, Duke University), and MyoD (1:20, NCL-MyoD1, Vision Bio Systems Novocastra). Antibody-antigen complexes were detected using biotinylated secondary antibody followed by avidin-horseradish peroxidase (HRP) complex, and DAB substrate (Vector Labs, Burlingame, CA). Sections were counterstained with nuclear fast red. For co-localization experiments, paraffin sections were stained with FITC-labeled αSM antibodies (green fluorescence) followed by immunostaning with antibodies against either cleaved caspase 3 or Ki-67 that were detected by TRITC-labeled secondary antibodies (red fluorescence). DAPI was used as a counterstain. Fluorescent images were obtained using a Zeiss Axioplan2 microscope equipped with an AxioCam MRm digital camera and AxioVision 4.3 Software (Carl Zeiss Microimaging, Thornwood, NY).
Paraffin sections were also used for in situ hybridization with 35S-labeled antisense riboprobe specific to 1649 – 1947 bp region of the mouse Foxm1 mRNA as described (Kalin et al., 2008). We used frozen E16.5 sections to perform laser capture microdissection of aortic tissue in smFoxm1−/− and control Foxm1fl/fl mice. The Veritas Microdissecting System (Molecular Devices, Sunnyvale, CA) was used for the laser capture microdissection experiments accordingly to protocols supplied by the manufacturer.
Human aortic vascular smooth muscle cells (ATCC cell line T/G HA-VSMC) were cultured in F-12K medium with supplements recommended by ATCC. In order to inhibit Foxm1 expression in HAVSMC cells, we used a 21-nucleotide short interfering RNA (siRNA) duplex specific to 1066-1084 nucleotide region of the human Foxm1 cDNA (siFoxm1, 5’-gga cca cuu ucc cua cuu u-3’). Foxm1 siRNA containing symmetric 2-Uracil (U) 3’ overhangs was designed and synthesized using Dharmacon Research algorithm (Kim et al., 2006). We transfected 100 nM of either siFoxm1 or mutant siFoxm1 (mutFoxm1; mutated nucleotides are indicated by capital letters, 5’-gga ccU GuA uGc Gua cAu u-3’) duplexes into human HAVSMC cells using Lipofectamine™ 2000 reagent (Invitrogen) in serum free tissue culture media as described previously (Wang et al., 2005a; Kim et al., 2006). HAVSMC cells were harvested at 72 hr after transfection for total RNA preparation or propidium iodide (PI) staining. PI-stained HAVSMC cells were analyzed by flow cytometry to determine numbers of cells in different stages of the cell cycle as described previously (Wang et al., 2005a). The flow cytometry and analysis were performed in the Flow cytometry core facility of Cincinnati Children's Hospital Research Foundation. In separate experiments, HAVSMC cells undergoing either G0/ G1 or G2/ M phase of the cell cycle were identified using Vybrant DyeCycle™ Violet stain (Invitrogen) according to manufacturer's recommendations. Stained cells were sorted using a 5-laser FACSAria II flow cytometer (BD Biosciences). The cells in G0/G1 and G2/M phase were harvested into collection tubes filled with RLT RNA lysis buffer, and then used to prepare total RNA. For PH3 staining, siRNA-transfected HAVSMC cells were fixed with 10% buffered formalin and immunostained with antibodies against phospho-Histone H3 (PH3; 1:100 dilution; #06-570; Upstate) to visualize cells undergoing mitosis as described (Wang et al., 2005a).
Total RNA was prepared from either siFoxm1-transfected or mock-transfected HAVSMC cells using RNA-STAT-60 (Tel-Test “B” Inc. Friendswood, TX). Synthesis of human HAVSMC cDNAs with CyDye nucleotides (Cy3 and Cy5), hybridization of Affymetrix GeneChip® Human Genome U133 plus 2.0 array, scanning, and analysis of cDNA Microarrays were performed by the Functional Genomics Facility at the University of Chicago (Chicago, IL). Data were deposited to GEO public database (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE17543; accession # GSE17543). Data were analyzed using Genespring 7.2 (Silicon Genetics). A volcano plot was used to identify significance and magnitude of gene expression changes with a minimum set as 2-fold as previously described (Dave et al., 2006). Gene expression profile was also subjected to an additional filter and classified according to Gene Ontology classification on Biological Process using the publicly available web-based tool David (database for annotation, visualization, and integrated discovery) (Dennis et al., 2003). The Fisher exact test was used to calculate the probability of each gene ontology category among all genes included in the array. To avoid individual variations, we combined 10 μg of RNA from three distinct HAVSMC cell cultures for each group. Expression levels of 421 genes were decreased > 2.0-fold in siFoxm1-transfected cells compared to mock-transfected HAVSMC cells. Expression levels of 27 genes were increased > 2.0-fold in siFoxm1-transfected cells.
To verify expression levels of selected genes, we performed qRT-PCR analysis using a StepOnePlus Real-Time PCR system (Applied Biosystems, Foster City, CA) as described (Kalin et al., 2008). Samples were amplified with TaqMan Gene Expression Master Mix (Applied Biosystems) combined with inventoried TaqMan gene expression assays for the gene of interest (Table 1). Reactions were analyzed in triplicates and expression levels were normalized to β-actin. Total RNA was also prepared from laser-captured E16.5 aortic tissue of either smFoxm1−/− or control Foxm1fl/fl embryos and then analyzed by qRT-PCR. Three embryos were used in each group.
Student's T-test was used to determine statistical significance. P values ≤ 0.05 were considered significant. Values for all measurements were expressed as the mean ± standard deviation (SD).
Previous studies have demonstrated that Foxm1 is expressed in a variety of cell types during embryogenesis (Ye et al., 1997; Kalin et al., 2008). To detect Foxm1 expression in developing smooth muscle cells, in situ hybridization was performed with Foxm1-specific anti-sense riboprobe. In E15.5 mouse embryos, Foxm1 mRNA was detected in vascular smooth muscle cells of arteries (Fig. 1A-C) as well as in smooth muscle cells underlying the developing esophagus, trachea, bronchi, stomach and intestine (Fig. 1A-F and data not shown). In adult mice, Foxm1 expression was observed in epithelium of intestinal crypts (Fig. 1G-H). Foxm1 mRNA was not detected in smooth muscle cells of intestine, bronchi or blood vessels of the adult mice (Fig. 1G-H and data not shown). These data demonstrate that Foxm1 is expressed in different populations of smooth muscle cells during embryonic development but its expression is extinguished in adult smooth muscle cells.
To determine whether Foxm1 is required for smooth muscle cells during embryonic development, we generated double-transgenic mice containing LoxP-flanked exons 4-7 of the Foxm1 gene (Foxm1fl/fl) and Cre recombinase transgene driven by smooth muscle myosin heavy chain promoter (smMHC-Cre-GFP, (Xin et al., 2002)). Previous studies demonstrated that using this smMHC-Cre-GFP transgenic mouse line, Cre-mediated recombination occurs in all populations of smooth muscle cells beginning from E12.5 (Xin et al., 2002). The smooth muscle-specific expression of Cre recombinase causes deletion of exons 4, 5, 6 and 7, which encode the DNA binding and transcriptional activation domains of the Foxm1 protein (Fig. 1I), to produce smooth muscle-specific Foxm1 knockout mice (smMHC-Cre-GFP tg/−/ Foxm1fl/fl or smFoxm1−/−). In smFoxm1−/− newborn mice, body weight, and lung to body weight ratio were unchanged (data not shown). However, deletion of Foxm1 caused a perinatal lethality in 87% of smFoxm1−/− mice within the first 24 hours after birth (Table 2). No perinatal lethality was observed in control Foxm1fl/fl littermates, or double-heterozygous smMHC-Cre-GFP tg/−/ Foxm1fl/+ newborn mice (data not shown). When assessed prior to birth, smFoxm1−/− embryos were present as Mendelian inheritance (Table 2), indicating that the Foxm1 function in smooth muscle cells was not required for fetal survival in utero, but was critical for survival immediately after birth.
To determine whether Cre recombination in smFoxm1−/− embryos occurred prior to the perinatal lethality, we bred smFoxm1−/− mice with R26R reporter mice to generate smFoxm1−/−/ R26R embryos, which were stained for β-galactosidase (β-gal) activity. Although β-gal activity was not detected in control Foxm1fl/fl/ R26R embryos at E15.5 (Fig. 1L), β-gal activity was observed in Foxm1-deficient smooth muscle cells of E15.5 arteries, esophagus, stomach, trachea, bronchi, and intestine (Fig. 1M-P and data not shown). Furthermore, to detect the presence of Cre protein in smooth muscle cells, smFoxm1−/− and control Foxm1fl/fl embryos were immunostained with antibodies against Cre recombinase. Although Cre was not expressed in Foxm1fl/fl embryos (Fig. 1T), Cre protein was detected in smooth muscle cells surrounding esophagus, stomach, intestine, and pulmonary airways of E15.5 smFoxm1−/− embryos (Fig. 1U-W and data not shown), which is consistent with β-gal activity in smFoxm1−/−/ R26R embryos (Fig. 1L-P). Finally, specific expression of Cre transgene in smFoxm1−/− smooth muscle cells was also confirmed by immunostaining with antibodies against GFP (Fig. 1R-S and data not shown), a reporter protein expressed under control of smMHC promoter in smFoxm1−/− embryos (Fig. 1I). Altogether, these results demonstrate that both Cre-mediated recombination and expression of Cre transgene occurred in smooth muscle cells of smFoxm1−/− embryos prior to perinatal lethality. Interestingly, although a mosaic Cre recombination was detected in atriums of Foxm1 mutant hearts as wells as in cardiac regions directly adjacent to aorta and pulmonary trunk, no structural abnormalities were observed in these heart regions (Fig. 1Q and data not shown).
Since Foxm1 is expressed in esophageal, bronchial and intestinal muscle layers (Fig. 1A-F), histological analysis of these organs was performed in smFoxm1−/− embryos. Embryonic esophageal muscle is composed of outer longitudinal and inner circumferential layers, both of which express Cre recombinase transgene in smFoxm1−/− mutant mice (Fig. 1M and 1V). Histological structure of esophagus was similar in smFoxm1−/− and control Foxm1fl/fl littermates at E15.5 (Fig. 2A-B). At E18.5, esophageal muscle in smFoxm1−/− mutants appeared underdeveloped, and a progressive loss of mesenchymal cells became apparent (Fig. 2E and 2J). Significantly diminished thickness of esophageal muscle was found in smFoxm1−/− E18.5 embryos (Fig. 2V). No structural abnormalities were observed in either Foxm1fl/fl or smMHC-Cre-GFPtg/− embryos (Fig. 2C-D and 2H-I). In smFoxm1−/− newborn mice, the entire esophagus contained very thin muscular layers fused together (Fig. 2F-G). The structural abnormalities of the esophagus can interfere with proper functions of the gastrointestinal system, contributing to perinatal lethality in smFoxm1−/− newborn mice. Interestingly, no intestinal, bronchial or stomach defects were found in smFoxm1−/− embryos or newborn mice (Figs. 2K-O and P-R), indicating different requirements in Foxm1 function in different populations of developing smooth muscle cells. Finally, normal expression of CCSP (Clara cell marker) and proSPC (type II cell marker) was observed in smFoxm1−/− E18.5 airways when compared to either Foxm1fl/fl or smMHC-Cre-GFPtg/− embryos (Fig. 2P-U). These results suggest that Foxm1 deficiency does not influence the development of different populations of epithelial cells in pulmonary airways.
Histological examination of the smFoxm1−/− newborn mice revealed extensive pulmonary hemorrhage with red blood cells present in bronchioles and peripheral pulmonary saccules (Fig. 3A-B). Pulmonary arteries from smFoxm1−/− mice displayed multiple hernias in arterial walls, disruption of vascular muscle layers, and accumulation of large cells with round nuclei (Fig. 3G-L), the latter of which is a hallmark of Foxm1 deficiency (Korver et al., 1998; Krupczak-Hollis et al., 2004; Ramakrishna et al., 2007). Morphological evidence of vascular leakage from perforated pulmonary arteries was also detected in smFoxm1−/− newborn mice (Fig. 3E-F). Thus, the structural abnormalities of the arterial wall can compromise vascular integrity and contribute to perinatal lethality in smFoxm1−/− newborn mice. Interestingly, hemorrhage was not observed in the liver, gut, brain and skeletal muscle of the smFoxm1−/− newborn mice, and the heart structure in smFoxm1−/− mice was normal (data not shown). Since lung hemorrhage was not detected in smFoxm1−/− mutants prior to birth (data not shown), these results indicate that the hemorrhaging phenotype was concomitant with the increased pulmonary arterial blood flow that occurs following birth.
Published studies demonstrated that Wnt7b−/− mice displayed extensive pulmonary hemorrhage, perinatal lethality and structural abnormalities in vascular smooth muscle (Shu et al., 2002), a phenotype similar to smFoxm1−/− mice (Fig. 3A-L). The vascular defects in Wnt7b−/− mice were attributed to a disruption of canonical Wnt signaling in vascular smooth muscle cells (Shu et al., 2002; Wang et al., 2005b). Therefore, the vascular abnormalities in smFoxm1−/− mutants could occur due to diminished activity of canonical Wnt signaling pathway. To visualize the canonical Wnt signaling in Foxm1-deficient embryos, we bred smFoxm1−/− mice with TOPGAL transgenic mice to produce TOPGALtg/−/ smFoxm1−/− embryos. The TOPGALtg/−/ smFoxm1−/− embryos were stained for β-gal activity and then compared to control TOPGALtg/−/ Foxm1fl/fl littermates. In E12.5 and E13.5 embryos, β-gal activity was detected in pulmonary arteries, bronchi, stomach, cartilage, and aortic outflow and pulmonary trunks of the developing heart (Fig. 3M-T and data not shown). Similar β-gal activity was observed in control and Foxm1 mutant embryos (Fig. 3M-T). Thus, Foxm1 deletion from smooth muscle cells does not alter canonical Wnt signaling during embryogenesis.
Recent studies demonstrated that Foxm1 regulates differentiation of hepatoblasts, endothelial cells and neural precursor cells in mouse and Xenopus embryos (Krupczak-Hollis et al., 2004; Kim et al., 2005a; Ueno et al., 2008). To determine the role of Foxm1 in differentiation of smooth muscle cells, we used smFoxm1−/− embryos to examine expression of smooth muscle marker proteins at E18.5. smFoxm1−/− and control Foxm1fl/fl embryos displayed similar expression of both the early (α-smooth muscle actin, α-SM and smooth muscle myosin heavy chain, smMHC) and the late markers of smooth muscle cells (γ-smooth muscle actin, γ-SM and smooth muscle-specific calponin, smCalponin) in embryonic blood vessels and pulmonary airways (Fig. 4A). Furthermore, smFoxm1−/− and control Foxm1fl/fl esophagi displayed similar expression pattern for both α-SM and γ-SM, both of which were detected in three esophageal layers: an inner submucosal layer directly adjacent to the epithelium, and two smooth muscle layers representing longitudinal and circumferential layers (Fig. 4B). Expression of epithelial-specific marker Foxa2 and endothelial-specific marker Pecam-1 was not changed in smFoxm1−/− esophagus and artery, respectively (Fig. 4A-B). These results suggest that Foxm1 deficiency does not influence the differentiation of smooth muscle cells during embryonic development.
During embryogenesis, skeletal muscle contributes to the development of esophageal smooth muscle layers (Kablar et al., 2000; Rishniw et al., 2003). To determine whether there was a normal contribution of skeletal muscle to smFoxm1−/− esophagus, E18.5 embryos were stained for the presence of skeletal muscle cells using an antibody to the skeletal-muscle specific transcription factor MyoD. Although smFoxm1−/− esophagi contained less MyoD-positive cells compared to control Foxm1fl/fl embryos (Fig. 4B), no significant differences were observed in percentages of MyoD-positive cells among total numbers of muscle cells in the developing esophagus (data not shown). Thus, Foxm1 deficiency does not influence differentiation of skeletal muscle in smFoxm1−/− esophagus.
Immunohistochemistry was used to compare cellular proliferation and apoptosis in smFoxm1−/− and control Foxm1fl/fl embryos. Rare apoptosis was observed in newborn Foxm1fl/fl mice as demonstrated by immunostaining with an antibody specific to activated form of caspase 3 (Fig. 5A). In contrast, smFoxm1−/− newborn mice displayed an increased apoptosis in the lung (Fig. 5B-C), pulmonary blood vessels (Fig. 5E-F) and esophagus (Fig. 5H-I). Apoptosis was observed in both myocytes (Fig. 5H) and other cell types directly adjacent to muscle layers (Fig. 5F and I). Total numbers of caspase 3 positive cells were significantly increased in the lung of smFoxm1−/− mice when compared to either Foxm1fl/fl or smMHC-Cre-GFPtg/− lungs (Fig. 5J-L).
Cellular proliferation in smFoxm1−/− embryos was assessed by immunostaning using antibody against the cell proliferation marker Ki-67. In control Foxm1fl/fl and smMHC-Cre-GFPtg/− embryos, Ki-67 positive cells were readily detected in endothelial and smooth muscle layers of the developing aorta, as well as in all layers of the developing esophagus (Fig. 6A). Although cellular proliferation in epithelial cells was normal, diminished Ki-67 staining was observed in smooth muscle cells of smFoxm1−/− embryos (Fig. 6A-B). In comparison to control embryos, total numbers of Ki-67 positive smooth muscle cells were significantly decreased in smFoxm1−/− blood vessels and esophagi (Fig. 6C). These results demonstrated that Foxm1 deficiency reduces cellular proliferation in developing smooth muscle cells.
In order to determine whether Foxm1 directly induces proliferation of smooth muscle cells in vitro, human aortic vascular smooth muscle cells (HAVSMC) were transfected with short interfering RNA (siRNA) duplex specific to the human Foxm1 mRNA (siFoxm1) or with mutant siFoxm1 duplex (mutFoxm1) containing five mutations in the siRNA structure (Wang et al., 2005a; Kim et al., 2006). Seventy-two hours later, cells were stained with propidium iodide (PI) and examined for DNA content by flow cytometry analysis. Foxm1-depletion in HAVSMC cells caused a significant reduction in G0/G1-phase cells and a decrease in mitotic progression as evidenced by a 2-fold increase in G2/M-phase (4N DNA content) with accumulation of polyploid cells (Fig. 6D). To distinguish between G2 and M phases of the cell cycle and to visualize cells undergoing mitosis, siRNA-transfected cells were subjected to immunostaining with antibody specific to phospho-Histone H3 (PH3), a marker for mitotic cells. Depletion of Foxm1 caused a 75% reduction in the number of PH3-positive HAVSMC cells compared to either untransfected HAVSMC cells or cells transfected with mutFoxm1 (Fig. 6E). These results suggest that depletion of Foxm1 in cultured smooth muscle cells caused a G2 arrest with accumulation of polyploid cells due to a significant decrease in mitotic progression.
In order to identify potential Foxm1-target genes in smooth muscle cells, we performed analysis of mouse Affymetrix microarrays that were hybridized with cDNA probes synthesized from either siFoxm1-transfected or mock-transfected HAVSMC cells. This analysis identified 448 genes in which expression was either significantly decreased (> 2.0-fold, 421 genes) or increased (> 2.0-fold, 27 genes) after the Foxm1-depletion (see GEO database for a complete list of genes with altered expression levels in siFoxm1-transfected HAVSMC cells; http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE17543; accession # GSE17543). Differentially expressed genes in siFoxm1-transfected HAVSMC cells were functionally classified according to Gene Ontology (GO). Gene Ontology analysis was performed using David (database for annotation, visualization, and integrated discovery) (Dennis et al., 2003). “Mitosis” was the most significantly over-represented process, which accounted for 22% of down-regulated genes with a Fisher Exact Test P-Value of 1.54E-37 (Table 3). “Negative regulation of cellular processes “ and “negative regulation of apoptosis” were significantly enriched in subset of up-regulated genes (Table 3). In Table 4, we summarized our focus on several known Foxm1 target genes as well as novel genes regulated directly or indirectly by the Foxm1 protein.
Consistent with an efficient Foxm1 knockdown, a 10-fold reduction in Foxm1 mRNA levels was observed in siFoxm1-transfected HAVSMC cells (Table 4). A subset of mRNAs, whose expression was decreased in Foxm1-depleted cells, included centromere proteins A and F, cyclin B1, aurora B kinase, polo-like kinases (Plk) 1 and 4, JNK1 and cyclin-dependent kinase 1 (cdk1). Since these genes regulate G2/M transition of the cell cycle and are required for cytokinesis (Glover et al., 1998; Adams et al., 2001; Wang et al., 2008), their diminished levels can contribute to the cell cycle defects seen in Foxm1-depleted HAVSMC cells. Interestingly, numerous genes essential for G1/S transition were also decreased in siFoxm1-transfected HAVSMC cells (Table 4). These include cyclins A2, D1 and E2, c-Myc transcription factor, cdk2-inhibitor 3, and S-phase kinase-associated protein 2 (Skp2), the latter of which is a specificity subunit of the Skp1-Cullin 1- F-box (SCF) ubiquitin ligase complex that targets both p21 (Cip1) and p27 (Kip1) Cdk2 inhibitors for degradation during G1/S transition (Kossatz et al., 2004). Significantly decreased mRNA levels of IAP repeat-containing survivin protein, a known transcriptional target of Foxm1 (Gusarova et al., 2007), were found in siFoxm1-transfected cells (Table 4). We also found several new genes, expression of which was altered in siFoxm1-transfected HAVSMC cells. These include calmodulin 1, retinoic acid receptor α, interferon receptor 1, collagen type I receptor, placental growth factor, ADAM 10 protein, and other genes required for extracellular and intracellular signaling (Table 4). Altogether, these results suggest that Foxm1 directly or indirectly regulates expression of numerous cell cycle regulators, receptors, enzymes and adaptor proteins.
In order to verify gene expression levels in siFoxm1-transfected HAVSMC cells, quantitative real-time RT-PCR (qRT-PCR) was performed using primers specific to several Foxm1 target genes (Table 1). Approximately 10-fold reduction in Foxm1 mRNA levels was observed in siFoxm1-transfected HAVSMC cells when compared to either untransfected cells or HAVSMC cells transfected with mutant siRNA (Fig. 7A). Consistent with Affymetrix microarray analysis (Table 4), Foxm1 depletion significantly decreased mRNA levels of cyclin B1, cyclin D1, Plk1, JNK1, and cMyc (Fig. 7A). Interestingly, decreased expression of the same genes was found in G0/ G1 population of HAVSMC cells isolated by flow cytometry-based cell sorting (Fig. 7A), indicating that Foxm1 regulates expression of these genes prior to DNA replication. In G2/ M population of sorted HAVSMC cells, Foxm1 depletion was associated with decreased mRNAs of cyclin B1, Plk1 and JNK1, whereas expression of cyclin D1 and cMyc was not significantly changed (Fig.7A). Interestingly, Foxm1 depletion did not induce apoptosis in the cultured HAVSMC cells (data not shown).
Finally, to determine whether Foxm1 regulates expression of the same cell cycle regulatory genes in vivo, we used a laser capture microdissection of frozen E16.5 sections to isolate total RNA from developing aortas of smFoxm1−/− and control Foxm1fl/fl embryos (Fig. 7B). Expression levels of the same cell cycle regulatory genes were determined by qRT-PCR analysis. While these RNA samples contained contributions from both the muscle and non-muscle cells (endothelium, pericytes and fibroblasts), mRNA levels of Foxm1 were significantly decreased in smFoxm1−/− aortas (Fig. 7C). While cyclin D1 mRNA was not changed in smFoxm1−/− aortas, mRNAs of cyclin B1, JNK1, cMyc and M-phase promoting Cdc25B phosphatase were significantly decreased (Fig. 7C), indicating that Foxm1 induces expression of the cell cycle regulatory genes in vivo. Taken together, these data suggest that Foxm1 depletion inhibits cellular regulatory pathways that stimulate smooth muscle cell proliferation during embryonic vascular development.
The requirement for Foxm1 activity in smooth muscle cells was examined in smFoxm1−/− mice that contain a smooth muscle-specific deletion of exons 4 through 7, encoding the Foxm1 DNA binding domain and C-terminal transcriptional activation domains, both of which are essential for Foxm1 transcriptional activity (Krupczak-Hollis et al., 2004). Most of the smFoxm1−/− mice exhibited a perinatal lethality within the first 24 hours after birth because of severe pulmonary hemorrhage, and structural defects in arterial wall and esophagus. Although Foxm1 deletion did not influence differentiation of smooth muscle cells, decreased proliferation of smooth muscle cells was found in smFoxm1−/− blood vessels and esophagus. Proliferation defects in smFoxm1−/− embryos were associated with decreased expression of M-phase promoting Cdc25B phosphatase and cyclin B1. Progression into mitosis requires the activation of Cdk1 through assembly with cyclin B1 regulatory subunit and the removal of Cdk1 inhibitory phosphates at Thr 14 and Tyr 15 by the Cdc25B and Cdc25C phosphatases (Borgne and Meijer, 1996; Wells et al., 1999; Nilsson and Hoffmann, 2000). Therefore, diminished levels of Cdc25B and decreased levels of cyclin B1 can decrease Cdk1 activation and M-phase progression causing delayed entry into mitosis in Foxm1-deficient smooth muscle cells. These results are consistent with previous studies that reported a direct transcriptional activation of these genes by Foxm1 in cultured mouse embryonic fibroblasts (MEFs) and U2OS human osteosarcoma cell line (Wang et al., 2005a).
Our studies demonstrated that several genes critical for G1/S transition were decreased in Foxm1-deficient smooth muscle cells in vivo and in vitro. These include JNK1, c-Myc transcription factor and cyclins A2, D1 and E2 that activate Cdk proteins during G1/S transition. Cdk2 complexes with either Cyclin E or Cyclin A cooperate with Cyclin D-Cdk4/6 to phosphorylate the Retinoblastoma (RB) protein, which releases bound E2F transcription factor and allows it to stimulate expression of genes required for DNA replication (Harbour and Dean, 2000; Ishida et al., 2001). Therefore, defects in cell proliferation in smFoxm1−/− smooth muscle cells may be a direct consequence of reduced Cdk2 and Cdk4 activity causing diminished DNA replication. Our results are consistent with previously published data demonstrating that transgenic over-expression of Foxm1 in Rosa26-Foxm1 transgenic mice is associated with premature DNA replication in pulmonary smooth muscle cells following Butylated Hydroxytoluene (BHT) lung injury (Kalinichenko et al., 2003).
In our previous studies, we generated mice with global deletion of the Foxm1 gene (homozygous for Foxm1 null allele or Foxm1−/− mice) and demonstrated that these mice displayed an embryonic lethal phenotype between E13.5 and E16.5 (Krupczak-Hollis et al., 2004; Kim et al., 2005a; Ramakrishna et al., 2007). Foxm1−/− embryos displayed multiple abnormalities in development of embryonic heart, liver, lung and blood vessels as well as differentiation defects in capillary endothelial cells (Kim et al., 2005a) and bile duct epithelial cells (Krupczak-Hollis et al., 2004). However, since Foxm1 is required for normal development of distinct Foxm1−/− cell lineages, these studies did not clarify specific requirements for Foxm1 activity in smooth muscle cells. Furthermore, severe ventricular hypoplasia in Foxm1−/− embryos may cause a diminished circulatory output and altered signaling, causing secondary defects in developing blood vessels. In this study, we demonstrated that Foxm1-deletion from smFoxm1−/− smooth muscle cells is sufficient to disrupt vascular development in the absence of structural heart abnormalities. Therefore, Foxm1 promotes embryonic vascular development by directly influencing proliferation of developing smooth muscle cells. Interestingly, although increased apoptosis was observed in smFoxm1−/− newborn mice, the apoptosis was not significantly increased in smFoxm1−/− embryos prior to birth. These results raise a possibility that the apoptotic phenotype in smFoxm1−/− mice was an indirect consequence of Foxm1 deficiency. Furthermore, a specific Foxm1 depletion by siRNA did not cause apoptosis in cultured smooth muscle cells, providing further support for this concept.
In contrast to structural defects in smFoxm1−/− blood vessels, no structural abnormalities were found in the developing airways, stomach or intestine of the smFoxm1−/− embryos. These results are surprising and suggest different requirements in Foxm1 function in different populations of developing smooth muscle cells. Published studies demonstrated that vascular and visceral smooth muscle cells (intestinal, esophageal and peribronchial muscle) use distinct transcription factor complexes to activate smooth muscle-specific gene promoters. Differential deletion of the serum response factor (SRF) binding sites (CArG boxes) revealed fundamental differences in importance of particular SRF-binding regions between smooth muscle subtypes (Manabe and Owens, 2001). It was also demonstrated that promoter of CRP-1 gene contains a CArG box-dependent enhancer, which directs expression in arterial but not venous or visceral smooth muscle cells (Lilly et al., 2001). Expression of GATA-5 and Foxf1 transcription factors was found to be restricted to visceral smooth muscle cells with little or no expression in vascular smooth muscle tissues (Morrisey et al., 1997; Kalinichenko et al., 2001). Selective smooth muscle defects in smFoxm1−/− embryos indicate that Foxm1 may regulate distinct genes in different subsets of smooth muscle cells. However, molecular mechanisms underlying this selectivity remain unknown. Alternatively, it is possible that Foxm1-deletion does not efficiently occur in all smooth muscle tissues during embryogenesis. While Cre recombinase protein was detected in inner circumferential muscle layers of the developing stomach and intestine, no Cre staining was observed in outer longitudinal layers of smFoxm1−/− intestine until E18.5. In contrast, smFoxm1−/− esophagi contained Cre protein in both inner and outer smooth muscle layers. Interestingly, approximately 13% of smFoxm1−/− mice survived after birth and developed a megacolon by 4 weeks of age (data not shown). Since a histological structure of the intestine in these adult smFoxm1−/− mice was normal (data not shown), the megacolon may be a result of diminished muscular tone and/or intestinal contractility in these Foxm1 mutant mice.
In summary, smFoxm1−/− mice died immediately after birth due to severe pulmonary hemorrhage, structural defects in arterial wall, and esophageal abnormalities. Although Foxm1 deletion did not influence differentiation of smooth muscle cells, decreased cellular proliferation was observed in muscle layers of embryonic blood vessels and esophagus. Depletion of Foxm1 in vitro and in vivo was associated with reduced expression of genes required for cell cycle progression. The identification of critical regulators of myocyte proliferation, such as Foxm1, may provide novel strategies for diagnosis and treatment of congenital vascular and esophageal abnormalities.
We thank Dr. J. Whitsett for critically reviewing the manuscript. This work was supported by grants from National Institute of Health (HL 84151-01) and March of Dimes Birth Defects Foundation (6-FY2005-325).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.