FOXF1 Transcription Factor Promotes Lung Morphogenesis by Inducing Cellular Proliferation in Fetal Lung Mesenchyme

Abstract

Organogenesis is regulated by mesenchymal-epithelial signaling events that induce expression of cell-type specific transcription factors critical for cellular proliferation, differentiation and appropriate tissue patterning. While mesenchymal transcription factors play a key role in mesenchymal-epithelial interactions, transcriptional networks in septum transversum and splanchnic mesenchyme remain poorly characterized. Forkhead Box F1 (FOXF1) transcription factor is expressed in mesenchymal cell lineages; however, its role in organogenesis remains uncharacterized due to early embryonic lethality of Foxf1^−/− mice. In the present study, we generated mesenchyme-specific Foxf1 knockout mice (Dermol-Cre Foxf1^−/−) and demonstrated that FOXF1 is required for development of respiratory, cardiovascular and gastrointestinal organ systems. Deletion of Foxf1 from mesenchyme caused embryonic lethality in the middle of gestation due to multiple developmental defects in the heart, lung, liver and esophagus. Deletion of Foxf1 inhibited mesenchyme proliferation and delayed branching lung morphogenesis. Gene expression profiling of micro-dissected distal lung mesenchyme and ChIP sequencing of fetal lung tissue identified multiple target genes activated by FOXF1, including Wnt2, Wnt11, Wnt5A and Hoxb7. FOXF1 decreased expression of the Wnt inhibitor Wif1 through direct transcriptional repression. Furthermore, using a global Foxf1 knockout mouse line (Foxf1^−/−) we demonstrated that FOXF1-deficiency disrupts the formation of the lung bud in foregut tissue explants. Finally, deletion of Foxf1 from smooth muscle cell lineage (smMHC-Cre Foxf1^−/−) caused hyper-extension of esophagus and trachea, loss of tracheal and esophageal muscle, mispatterning of esophageal epithelium and decreased proliferation of smooth muscle cells. Altogether, FOXF1 promotes lung morphogenesis by regulating mesenchymal-epithelial signaling and stimulating cellular proliferation in fetal lung mesenchyme.

INTRODUCTION

Lung development in the mouse begins at 9.5 days post coitum (E9.5) when the foregut endoderm invades the splanchnic mesenchyme and undergoes dichotomous branching (Cardoso, 2001; Morrisey and Hogan, 2010; Warburton et al., 1999; Whitsett et al., 2004). Lung morphogenesis depends on mesenchymal-epithelial crosstalk between splanchnic mesenchyme and endoderm-derived epithelial cells, which is mediated by SHH, WNTs, FGFs, TGF-β, BMP4 and NOTCH (Bolte et al., 2018; Hogan and Yingling, 1998; Morrisey and Hogan, 2010). These signaling molecules regulate proximal-distal patterning, cellular proliferation and differentiation of various respiratory cell types by inducing expression of cell-type specific transcription factors. The developing respiratory epithelium produces SHH, which acts via PTCH receptor and the GLI family of zinc finger transcription factors to promote mesenchyme survival and stimulate its differentiation into multiple stromal cell lineages, including fibroblasts, pericytes and smooth muscle cells (Weaver et al., 2003). Conditional inactivation of Shh in respiratory epithelium or deletion of either Gli2 or Gli3 results in lung hypoplasia, diminished epithelial branching and loss of mesenchyme and smooth muscle cells (Grindley et al., 1997; Miller et al., 2004; Motoyama et al., 1998). Compound mutations in Glil/2 cause severe lung hypoplasia, whereas Gli2/3 null mutants completely fail to specify the respiratory lineage (Grindley et al., 1997; Motoyama et al., 1998; Rankin et al., 2016). While these studies implicate SHH/PTCH/GLI signaling in regulation of lung development, downstream targets of the SHH signaling pathway in lung mesenchyme remain poorly characterized.

FOXF1 (previously known as HFH-8 or Freac-1) is a transcription factor from the Forkhead Box (FOX) family, which is expressed in septum transversum and splanchnic mesenchyme prior to the initiation of lung development (Mahlapuu et al., 2001b; Peterson et al., 1997). Later in development, FOXF1 is expressed in lung mesenchyme, endothelial cells and airway smooth muscle cells (Cai et al., 2016; Kalin et al., 2008; Kalinichenko et al., 2001b; Mahlapuu et al., 1998; Ren et al., 2014). FOXF1 induces migration of lung mesenchymal cells in vitro by activating transcription of Integrin β3 and Notch-2 (Kalinichenko et al., 2004; Malin et al., 2007). Fox/Γ^A mice exhibit an embryonic lethal phenotype due to lack of vasculature in the yolk sac and allantois (Mahlapuu et al., 2001b). Foxf1 haploinsufficiency (Foxf1^+/-) causes lung hypoplasia, decreased formation of pulmonary capillaries and increased mortality after birth (Kalinichenko et al., 2001a; Kalinichenko et al., 2002a; Kalinichenko et al., 2002b; Lim et al., 2002; Mahlapuu et al., 2001a; Ormestad et al., 2006). Inactivating mutations in the FOXF1 gene are often found in patients with Alveolar Capillary Dysplasia with Misalignment of Pulmonary Veins (ACD/ MPV) (Sen et al., 2013; Szafranski et al., 2013), a fatal congenital disorder of newborns and infants associated with lung hypoplasia and the loss of peripheral pulmonary capillaries. Mouse genetic studies demonstrated that FOXF1 acts downstream of GLI to promote SHH signaling during cardiac, intestinal and craniofacial development (Hoffmann et al., 2014; Mahlapuu et al., 2001b; Xu et al., 2016). Shhmouse embryos exhibit decreased Foxf1 expression in mesenchyme (Mahlapuu et al., 2001a). SHH/GLI signaling induces Foxf1 expression through GLI-mediated transcriptional activation of the Foxf1 promoter (Kim et al., 2005; Madison et al., 2009; Szafranski et al., 2013). Mutations in GLI-binding sites in the human FOXF1 promoter cause ACD/MPV in spite of the preservation of normal FOXF1 coding sequences (Szafranski et al., 2013), emphasizing the importance of the SHH/GLI signaling pathway in regulation of FOXF1 gene expression. While previous studies mostly focused on the role of FOXF1 in endothelial cells during lung development, FOXF1 functions in lung mesenchyme remain uncharacterized.

In the present study, we demonstrated that Foxf1 expression in mesenchymal cells is required for development of respiratory, cardiovascular and gastrointestinal organ systems as deletion of Foxf1 from Dermol-expressing mesenchyme (Dermol-Cre Foxf1^−/−) causes embryonic lethality and a variety of developmental defects. Furthermore, deletion of Foxf1 inhibits mesenchyme proliferation and delays branching lung morphogenesis. FOXF1 promotes mesenchymal-epithelial signaling and stimulates cellular proliferation by inducing expression of Wnt2 and Hoxb7 and suppressing the Wnt-inhibitor Wifi in developing lung mesenchyme. Deletion of Foxf1 from smooth muscle cells (smMHC-Cre Foxf1^−/−) causes perinatal lethality due to enlargement of the esophagus and trachea, thinning of muscle layers and decreased proliferation of smooth muscle cells. Altogether, our data indicate that FOXF1 promotes lung morphogenesis by regulating mesenchymal-epithelial signaling and stimulating cellular proliferation in mesenchyme-derived cells.

METHODS

Mouse strains

The generation of Foxf1^flox/flox(Foxf1^fl/fl) mice was described previously (Ren et al., 2014). Foxf1-floxed allele contains LoxP sequences flanking the DNA binding domain (exon 1) (Cai et al., 2016; Ren et al., 2014). Foxf1^fl/fl mice were bred with Dermol-Cre^tg/- mice (Jackson Laboratory, (Yu et al., 2003)) to generate Dermol-Cre^tg/- Foxf1^fl/fl double transgenic mice. Foxf1^fl/fl and Dermol-Cre^tg/- Foxf1^fl/+ littermates were used as controls. Generation of Foxf1^+/− and smMHC-Cre^tg/- Foxf1^fl/+ mouse lines was described previously (Hoggatt et al., 2013; Kalinichenko et al., 2001a; Sen et al., 2014). Foxf1^fl/fl and Foxf1^+/− mice were generated as C57Bl/6 x 129/SVEV hybrids (Ren et al., 2014) and were bred into C57Bl/6 background for 10 generations. mTmG mice (Ren et al., 2014) were purchased from the Jackson Laboratory. Animal studies were reviewed and approved by the Animal Care and Use Committee of Cincinnati Children’s Research Foundation.

Immunohistochemical staining

Embryos were harvested, fixed overnight with 4% buffered paraformaldehyde and then embedded into paraffin blocks. Five μm paraffin sections were used for staining with hematoxylin and eosin (H&E), immunohistochemistry and immunofluorescent co-staining as described previously (Ren et al., 2013; Wang et al., 2010; Wang et al., 2003). The following antibodies were used for immunostaining: FOXF1 (Ren et al., 2014); αSMA (A5228, Sigma); γSMA (LMAB-B4, Seven Hills Bioreagents); FLK1 (2479, Cell signaling); FOXM1 (Cheng et al., 2014; Weiler et al., 2017); BrdU (G3G4, DSHB); Ki-67 (clone Tec-3, Dako); PH3 (sc8656r, Santa Cruz Biotechnology); Cleaved Caspase 3 (MAB835, R&D Systems); SOX2 (WRAB-1236, Seven Hills Bioreagents); SOX9 (AB5535, Millipore); NKX2.1 (WRAB-1231, Seven Hills Bioreagents); E-Cadherin (MAB7481, R&D Systems; and 4065, Cell Signaling); CCSP (sc9772, Santa Cruz Biotechnology); β-tubulin (MU 178-UC, BioGenex); p63 (sc71827, Santa Cruz Biotechnology); Cytokeratin 13 (ab92551, Abcam); Cytokeratin 14 (ab7800, Abcam); and activated β-catenin (phospho-Ser552) (5651, Cell Signaling). Antibody-antigen complexes were detected using biotinylated secondary antibody followed by avidin-horseradish peroxidase (HRP) complex and DAB substrate (all from Vector Lab) as described (Balli et al., 2012; Ren et al., 2010; Wang et al., 2012; Wang et al., 2014). Sections were counterstained with nuclear fast red (Vector Labs, Burlingame, CA). BrdU incorporation was performed according to the manufacturer’s protocol as previously described (Ren et al., 2014; Sengupta et al., 2013). BrdU was injected intraperitoneally into pregnant females 2 hours before embryo harvest. For colocalization experiments, secondary antibodies conjugated with Alexa Fluor 488, Alexa Fluor 594 or Alexa Fluor 647 (Invitrogen) were used as previously described (Bolte et al., 2011; Bolte et al., 2012; Ustiyan et al., 2012; Ustiyan et al., 2016). Slides were counterstained with DAPI (Vector Laboratory). Images were obtained using a Zeiss Axioplan 2 microscope with AxioCam MRc5 and AxioCam MRm digital cameras and Axio Vision 4.8 Software (Carl Zeiss). The number of αSMA⁺Foxf1⁺ cells and the percentage of the airway epithelium circumference covered by αSMA⁺cells were calculated as described (Hines et al., 2013).

Ex vivo culture of foregut and lung explants

Mouse foregut regions (E8.5) or whole lungs (E11.5) were dissected from embryos in Hank’s balanced salt solution, explanted onto 8 μm pore Whatman Nucleopore Track-Etch Membranes (Millipore) and cultured in DMEM (Gibco) with 5% fetal bovine serum (FBS, Sigma) for lung explant culture or 20% FBS for foregut explant culture (Havrilak et al., 2017; Havrilak and Shannon, 2015). Explants were cultured at 37°C in a 5% CO₂ incubator for 1-3 days. Whole-mount immunostaining was performed as previously described (Havrilak and Shannon, 2015). Images were captured on a Nikon A1Rsi inverted laser confocal microscope and composed using Bitplane Imaris software.

Quantitative real-time RT-PCR (qRT-PCR)

Lungs were dissected from embryos or newborn mice and homogenized in STAT60. Total lung RNA was isolated as described previously (Milewski et al., 2017a; Milewski et al., 2017b; Sun et al., 2017) and analyzed by qRT-PCR using a StepOnePlus Real-Time PCR system (Applied Biosystems, Foster City, CA) (Bolte et al., 2015; Pradhan et al., 2016; Xia et al., 2015). TaqMan Gene Expression Master Mix (Applied Biosystems) was used to amplify samples in combination with inventoried TaqMan gene expression assays for the gene of interest (Suppl. Table S1). Reactions were analyzed in triplicate and expression levels were normalized to β-actin mRNA.

ChIPseq and RNAseq

Analysis of FOXF1 binding sites for genes of interest was performed using a published ChIPseq data set generated from E18.5 mouse lungs (Accession number GSE77951) (Dharmadhikari et al., 2016). Data analysis was performed using the Biowardrobe platform as previously described (Bolte et al., 2017). RNAseq analysis was performed using distal lung mesenchyme micro-dissected from control and Dermol-Cre Foxf1^−/− mouse embryos as previously described (Bolte et al., 2017; Cai et al., 2016; Sen et al., 2014). DESeq package was used to determine and analyze differences in gene expression. Differentially expressed genes were used to build heat map with JMP Genomics 6.0 software. RNAseq data is available as GEO accession GSE78184.

Statistical analysis

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).

RESULTS

Deletion of Foxf1 from mesenchymal cells causes embryonic lethality and defects in multiple organ systems

Immunohistochemistry was performed to examine FOXF1 expression in the developing mouse lung. At E11.5, FOXF1 staining was detected in endothelial cells of pulmonary blood vessels and pulmonary mesenchyme adjacent to respiratory epithelial tubules (Fig. 1A), a finding consistent with published studies (Kalinichenko et al., 2003; Ren et al., 2014). FOXF1 was also expressed in visceral smooth muscle and mesenchyme surrounding esophagus, trachea, bronchi, stomach and intestine (Fig. 1A, Suppl. Fig. S1A-C). FOXF1 was not detected in vascular smooth, cardiac or skeletal muscles (Suppl. Fig. S1D-F). To examine requirements for FOXF1 in mesenchyme and its derivatives, we crossed Foxf1^fl/fl mice with mice containing Dermol-Cre, a mesenchyme-specific Cre driver (Yu et al., 2003). In Dermol-Cre Foxf1^−/− embryos (abbreviated as Dermol-Foxf1^−/−), the exon 1, encoding the FOXF1 DNA-binding domain, was deleted (Fig. 1B). The presence of Dermol-Cre and Foxf1-floxed alleles were confirmed by PCR (Fig. 1C). Dermol-Foxf1^−/− embryos exhibited severe growth retardation and died in utero after E16.5 (Fig. 1D). Genotype analysis of embryos from crosses between Dermol-Cre Foxf1^fl/+ males and Foxf1^fl/fl females revealed a gradual reduction in percentages of Dermol-Foxf1^−/− embryos between E13.5 and E16.5 (Table 1). No Dermol-Foxf1^−/− embryos were recovered after E16.5.

Figure 1. Deletion of Foxf1 from mesenchyme causes growth retardation and heart ventricular hypoplasia.

A, Foxf1 is expressed in lung mesenchyme and endothelial cells during embryonic development. Paraffin lung sections from wild type mouse embryos (E11.5, E13.5 and E16.5) were stained with FOXF1 antibodies (dark brown) and counterstained with nuclear fast red (red). B-C, Schematic drawing illustrates deletion of Foxf1-floxed allele from mesenchymal cells. Foxf1^fl/fl mice were bred with Dermo1-Cre^tg/- mice to generate Dermol-Cre^tg/- /Foxm1^fl/fl double transgenic mice (Dermo1-Foxf1^−/−). PCR shows deletion of Foxf1 and presence of Cre transgene. D, Foxf1 deletion from mesenchyme causes growth retardation. Representative photographs of E16.5 Dermo1-Foxf1^−/− and Dermo1-Foxf1^+/− embryos are shown. E-H, Hematoxylin and eosin (H&E) staining shows hypoplasia of the lung (E), liver (E), esophagus (F, top panels), lack of blood islands in liver tissue (F, bottom panels), interventricular septum defect (G) and myocardial hypoplasia (G-H) in E16.5 Dermo1-Foxf1^−/− embryos. Thickness of the compact layer of ventricular myocardium (H) was quantitated using 10 random x400 microscope fields (n = 3 embryos in each group; * indicates p < 0.05). Abbreviations: Ep, epithelium; Me, mesenchyme; V, vein; Lu, lung; Li, liver; St, stomach; RV, right ventricle; LV, left ventricle; RA, right atrium; LA, left atrium; IVS, interventricular septum. Magnification: A (top panels) and G, x50; A (bottom panels), F and H, x400; E (top panels), x32; E (bottom panels), x16.

Table 1.

Genotypes of embryos from crosses between Dermo1-Cre Foxf1^fl+ males and Foxf1^fl/fl females were identified using PCR analysis of genomic DNA.

	Embryos	Foxf1fl/+	Foxf1fl/fl	Dermo1-Foxf1fl+	Dermo1-Foxf1−/−
E9.5-10.5	34	8 (23.5%)	9 (26.5%)	9 (26.5%)	8 (23.5%)
E11.5-12.5	70	14 (20.0%)	16 (23.0%)	21 (30.0%)	19 (27.0%)
E13.5-14.5	70	13 (18.6%)	21 (30.0%)	22 (31.4%)	14 (20.0%)
E15.5-16.5	23	9 (39.1%)	4 (17.4%)	7 (30.4%)	3 (13.0%)

Histological studies were carried out to identify the cause of growth retardation and lethality in Dermol-Foxf1^−/− embryos. At E16.5, the liver, lung and esophagus in Dermol-Foxf1^−/− embryos were smaller compared to control littermates (Fig. 1E-F). The lack of blood islands in the liver and the pale appearance of Dermol-Foxf1^−/− embryos were consistent with hematopoietic deficiency (Fig. 1D and Fig. 1F). Dermol-Foxf1^−/− E16.5 embryos exhibited heart abnormalities, including ventricular hypoplasia, inter-ventricular septal defect, thinning of the compact layer of the myocardium and atrial chamber malformations (Fig. 1G-H and Suppl. Fig. S2A). The trabecular layer of the myocardium was unaltered in Dermol-Foxf1^−/− mutants (Fig. 1G-H). Earlier time-points were examined to determine timing of the organ abnormalities. At E12.5-E13.5, there were no differences in body size between Dermol-Foxf1^−/− and control embryos (Fig. 2A). However, Dermol-Foxf1^−/− livers and lungs were smaller (Fig. 2C-D), and heart defects were already present (Fig. 2E). Distal parts of Dermol-Foxf1^−/− atrial chambers were collapsed (Suppl. Fig. S3). FOXF1 staining was reduced in septum transversum and venous pole mesenchyme, and the dorsal mesenchymal protrusion was absent in Dermol-Foxf1^−/− embryos (Suppl. Fig. S4). Heart defects in Dermol-Foxf1^−/− embryos were consistent with venous pole mesenchymal defects during heart morphogenesis (Snarr et al., 2007). There were no histological abnormalities in the yolk sac of Dermol-Foxf1^−/− embryos as shown by H&E staining and immunostaining for the endothelial marker FLK1 (Fig. 2B). At E13.5, developing limbs in Dermol-Foxf1^−/− embryos had all 5 digits (Suppl. Fig. S2B). Altogether, deletion of Foxf1 from mesenchymal cells causes growth retardation and embryonic lethality due to defects in multiple organ systems.

Figure 2. Developmental abnormalities in Dermo1-Foxf1^−/− embryos at E12.5-E13.5.

A, Representative photographs of Dermo1-Foxf1^−/− and control Foxf1^fl/fl embryos show no differences in body sizes at E12.5. B, H&E and immunostaining for FLK1 show no vascular abnormalities in yolk sacs of E13.5 Dermo1-Foxf1^−/− embryos. FLK1-stained slides (dark brown) were counterstained with nuclear fast red (red nuclei). C-E, H&E staining of E10.5-E13.5 embryos show lung and liver hypoplasia (C-D) and heart defects (E) after deletion of Foxf1. Abbreviations: lu, lung; li, liver; in, intestine; he, heart; A, heart atrium; V, heart ventricle; t.v., third brain ventricle; f.v., fourth brain ventricle. Magnification: B (left panels), x200; B (right panels), x400; C, x16; D and E (middle and right panels), x50; E (left panels), x100.

Deletion of Foxf1 causes lung hypoplasia

Next, we focused on lung development in Dermol-Foxf1^−/− embryos. At E11.5, Dermol-Foxf1^−/− lungs were hypoplastic and contained less branching points at the time of harvest (Fig. 3 A and Suppl. Fig. S5). Isolated Dermol-Foxf1^−/− lungs were capable of branching ex vivo but remained hypoplastic during the tissue culture period (Fig. 3B). Immunohistochemistry was carried out to examine cell-specific markers in Dermol-Foxf1^−/− lungs. SOX2 and SOX9 were present in proximal and distal epithelial progenitors (Fig. 3C and Suppl. Fig. S6) and there were no changes in Sox2 and Sox9 mRNAs (Fig. 4A). At E16.5, the terminal tips of airways were not dilated in Dermol-Foxf1^−/− mutants and lung histology was consistent with the pseudoglandular stage of lung development (Suppl. Fig. S6).

Figure 3. Lung hypoplasia in Dermo1-Foxf1^−/− embryos.

A, Representative photographs of E11.5 and E13.5 mouse lungs show lung hypoplasia in Dermo1-Foxf1^−/− embryos. B, E11.5 lungs from Dermo1-Foxf1^−/− and control Foxf1^fl/fl embryos were cultured ex vivo for three days. Dermo1-Foxf1^−/− lungs were capable of branching but remained hypoplastic. C, Paraffin lung sections from E13.5 embryos were used for H&E staining or immunostained with antibodies against SOX2 and SOX9 (dark brown). Slides were counterstained with nuclear fast red (red). Magnification in panels C is x50.

Figure 4. Foxf1 deletion from mesenchyme inhibits smooth muscle cell lineage during lung development.

A, E11.5 lungs from Dermo1-Foxf1^−/− and control embryos were used for qRT-PCR. mRNAs were analyzed in triplicate and expression levels were normalized to β-actin. Data are presented as means ± SD (* indicates p < 0.05). qRT-PCR shows decreased expression of Foxf1 and several smooth muscle specific genes (γSMA, smMHC and Myocardin) in E11.5 Dermo1-Foxf1^−/− lungs. B, Immunostaining shows decreased expression of FOXF1, αSMA and γSMA in E13.5 Dermo1-Foxf1^−/− lungs (dark brown). Slides were counterstained with nuclear fast red (red). C, Co-localization studies were performed using E13.5 lung sections and antibodies against FOXF1 and αSMA. Loss of peribronchial αSMA-positive cells was observed in Dermo1-Foxf1^−/− lungs. Decreased αSMA staining was observed in areas with the highest deletion of Foxf1 (white arrowheads). Magnification: B panels, x200; C panels, x400.

FOXF1 staining and Foxf1 mRNA were decreased in Dermol-Foxf1^−/− lungs (Fig. 4A-B), a finding consistent with efficient deletion of Foxf1 by the Dermol-Cre transgene. Expression of smooth muscle markers αSMA, γSMA and smMHC was decreased in Dermol-Foxf1^−/− lungs as shown by immunostaining and qRT-PCR (Fig. 4A-B). Loss of αSMA occurred in mesenchymal regions with the highest Foxf1 deletion efficiency (Fig. 4C). Loss of αSMA was also evident in the trachea and bronchi of Dermol-Foxf1^−/− embryos (Suppl. Fig. S7A and S8). Deletion of Foxf1 from mesenchyme significantly decreased the number of αSMA⁺FOXF1⁺ cells (Suppl. Fig. S7B) and reduced the percentage of the airway epithelium circumference covered by αSMA⁺ cells (Suppl. Fig. S7C). Vascular smooth muscle was not affected in Dermol-Foxf1^−/− mutants (Suppl. Fig. S7D).

Foxf1 deletion from mesenchyme led to mispatterning of tracheal epithelium and cartilage. While the tracheal epithelium of Dermol-Foxf1^−/− embryos expressed E-cadherin, NKX2.1 and SOX2, the number of p63-positive basal cells was lower (Suppl. Fig. S8). SOX9 staining was reduced in the developing cartilage of Dermol-Foxf1^−/− embryos (Suppl. Fig. S8). There were no differences in expression of endothelial markers FLK1 and PECAM-1 in Dermol-Foxf1^−/− lungs (Fig. 4A and Suppl. Fig. S9). mRNAs of Foxf2, Gli1, Gli2, Ptchl, Vegfa and Pdgfra were unchanged after deletion of Foxf1 (Fig. 4A). Altogether, deletion of Foxf1 from mesenchyme decreases smooth muscle cell markers and causes mispatterning in tracheal epithelium and mesenchyme.

Deletion of Foxf1 decreases mesenchyme proliferation

To measure cellular proliferation in Dermol-Foxf1^−/− lungs, we used BrdU incorporation to label cells undergoing DNA replication. The number of BrdU-positive mesenchymal cells was significantly decreased in Dermol-Foxf1^−/−embryos (Fig. 5A-B), a finding consistent with reduced proliferation rates in FOXF1-deficient endothelial cells during embryogenesis (Ren et al., 2014) and lung regeneration (Bolte et al., 2017). There was no difference in cell apoptosis in Dermol-Foxf1^−/− lungs as shown by immunostaining for activated (cleaved) Caspase 3 (Suppl. Fig. S9). Thus, deletion of Foxf1 reduces mesenchyme proliferation but does not influence cell survival.

Figure 5. Foxf1 deletion decreases mesenchyme proliferation.

A, Immunostaining of paraffin sections from E11.5 embryos show decreased incorporation of BrdU into mesenchyme of Dermo1-Foxf1^−/− lungs and esophagus (dark brown). Slides were counterstained with nuclear fast red (red). B, The number of BrdU-positive cells was decreased in Dermo1-Foxf1^−/− lung and esophagus. Ten random x400 microscope fields were counted (n = 5 embryos in each group; * indicates p < 0.05). Magnification: top panels, x100; remaining panels, x400. C, mT/mG transgene was used to show a mosaic pattern of Dermo1-Cre-mediated recombination in mesenchyme surrounding foregut of E9.5 embryos. Images show GFP and tdTomato fluorescence in Dermo1-Cre mTmG embryos. Scale bars are 50μm (left panel) and 150 μm (right panel). D, Global deletion of Foxf1 inhibits lung budding ex vivo. Foreguts with surrounding mesenchyme were micro-dissected from E8.5 Foxf1^−/− and control WT embryos and cultured ex vivo for two days. Lung and thyroid buds were visualized by immunostaining for NKX2.1 (TTF1, green) and E-Cadherin (purple). Scale bars are 100 μm. Abbreviations: Fg, foregut; Sm, splanchnic mesenchyme; th, thyroid; lu, lung.

Deletion of Foxf1 disrupts lung budding

Next, we examined whether FOXF1 regulates the formation of lung buds, which normally occurs at E9.5 in the mouse (Morrisey and Hogan, 2010; Rankin et al., 2016). Dermol-Cre driven recombination was ineffective to delete Foxf1 from splanchnic mesenchyme at E9.5 as shown by GFP fluorescence in Dermol-Cre mTmG embryos (Fig. 5C). Therefore, we used a global Foxf1 knockout mouse line (Foxf1^−/−) to examine the formation of lung bud. Since Foxf1^−/− mice die in utero after E8.5, which is approximately one day earlier than the initiation of the lung bud (Mahlapuu et al., 2001b), we isolated foregut regions from Foxf1^−/− and WT E8.5 embryos and cultured them ex vivo. NKX2.1 staining was used to visualize lung and thyroid buds formed in the foregut. While lung buds were present in Foxf1^−/− foregut explants, they were smaller compared to WT (Fig. 5D). There were no differences in sizes of thyroid buds (Fig. 5D). Thus, Foxf1 expression in mesenchyme is critical for lung budding ex vivo.

FOXF1 regulates mesenchymal genes critical for mesenchymal-epithelial signaling and cell proliferation

To identify FOXF1 target genes in lung mesenchyme, we used Dermol-Foxf1^−/− and control Foxf1^fl/fl embryos to micro-dissect mesenchyme from E10.5 distal lung tips and perform RNAseq analysis. Significant changes in expression of 1714 genes were found in lung mesenchyme from Dermol-Foxf1^−/− embryos compared to Foxf1^fl/fl embryos (Fig. 6A). These include 721 genes in which expression was significantly decreased (> 1.5-fold, p < 0.05) and 993 genes with increased expression (> 1.5-fold, p < 0.05) (Table 2 and GEO accession GSE78184). Differentially expressed genes in Dermol-Foxf1^−/− mesenchyme were functionally classified according to Gene Ontology. Functional categories of “mesenchymal cell development“, “smooth muscle contraction”, “fibroblast proliferation” and “canonical Wnt signaling pathway” were significantly enriched in the subset of down-regulated genes (Suppl. Fig. S10). Expression levels of selected genes were confirmed by RT-PCR (Fig. 6B). Foxf1 mRNA was decreased by 75% in Dermol-Foxfl^−/− lung mesenchyme compared to Foxf1^fl/fl controls (Fig. 6B). Consistent with diminished cellular proliferation (Fig. 5A-B), mRNAs of Igf1, Egfr and Hoxb7 were reduced in Dermol-Foxf1^−/− mesenchyme (Table 2 and Fig. 6A-B). Interestingly, expression of genes regulating canonical Wnt signaling pathway was significantly decreased in Dermol- Foxf1^−/− lung mesenchyme (Table 2 and Fig. 6A-B). These include Wnt ligands Wnt2, Wnt5a and Wnt11, as well as known regulators of the canonical Wnt signaling pathway, including Snai2, Siah2, Zfp 703, Biccl, Nog, Rspol, Is11, Trpm4 and Nkd2 (DiMeo et al., 2009; Morrisey and Hogan, 2010; Peng et al., 2013). In contrast, expression of the Wnt inhibitor Wif1 was significantly increased in Dermol-Foxf1^−/− lungs (Table 2), a finding confirmed by qRT-PCR (Fig. 6B). Cross-reference of the RNAseq data with previously published FOXF1 ChIPseq of mouse fetal lung tissue (Dharmadhikari et al., 2016) demonstrated that FOXF1 directly bound to gene promoters and introns of Wnt2, Wnt11, Wnt5a, Wif1, Hoxb7, Igf1 and Egfr (Fig. 6C, Suppl. Fig. S11 and Suppl. Table S2), implicating FOXF1 in regulation of these genes. Using antibodies against the activated form of β-catenin, we found that activated β-catenin was reduced in respiratory epithelium and mesenchyme of Dermol-Foxf1^−/− lungs (Suppl. Fig. S12), a finding consistent with reduced Wnt/β-catenin signaling. Thus, FOXF1 regulates multiple mesenchymal genes critical for lung morphogenesis, canonical Wnt signaling and mesenchyme proliferation.

Figure 6. FOXF1 regulates expression of Hoxb7, Wnt2 and Wif1 in lung mesenchyme.

A, RNAseq heat-map shows gene expression in distal lung mesenchyme of Dermo1-Foxf1^−/− and control Foxf1^fl/fl embryos. RNA was extracted from micro-dissected mesenchyme of distal lung tips of E10.5 Dermo1-Foxf1^−/− and control embryos (n = 3 embryos were used in each group). B, qRT-PCR of E10.5 lung mesenchyme shows decreased expression of Foxf1, Hoxb7 and Wnt2 and increased Wif1 mRNA in Dermo1-Foxf1^−/− embryos. mRNA was analyzed in triplicate and expression levels were normalized to β-actin mRNA (n = 3 embryos in each group; * indicates p < 0.05). C, ChIPseq of mouse embryonic lungs shows FOXF1 binding to DNA regions in Hoxb7, Wnt2 and Wif1 genes. Statistically significant binding is indicated by arrows.

Table 2.

RNAseq of distal tip lung mesenchyme from E10.5 embryos.

Gene	Folds	p-value	Gene name
Wnt signaling pathway
Wnt2	−2.17	1.07E-14	Wnt family member 2
Wnt 11	−2.14	2.67E-07	Wnt family member 11
Nog	−1.94	0.0062	Noggin
Trpm4	−1.92	7.64E-06	Transient receptor potential cation channel, subfamily M, member 4
Nkd2	−1.90	0.00026	Naked cuticle homolog 2
Biccl	−1.83	1.55E-13	BicC family RNA binding protein 1
Isl1	−1.82	9.96E-06	ISL LIM homeobox 1
Snai2	−1.79	0	Snail family zinc finger 2
Wnt5a	−1.77	0.0006	Wnt family member 5A
Siah2	−1.71	0.0043	Siah E3 ubiquitin protein ligase 2
Zfp703	−1.65	0.0006	Zinc finger protein 703
Rspo1	−1.59	4.44E-11	R-spondin 1
Wif1	4.87	0	Wnt inhibitory factor 1
Regulation of smooth muscle proliferation and differentiation
Cdh13	−10.7	0	Cadherin13
Hoxb7	−10.47	2.72E-10	Homeobox B7
Trpc3	−4.62	5.77E-15	Transient receptor potential cation channel subfamily C member 3
Calcb	−3.98	0.00028	Calcitonin related polypeptide beta
Cacna1c	−3.81	0	Calcium voltage-gated channel subunit alpha 1C
Kcnb2	−2.95	1.75E-11	Potassium voltage-gated channel subfamily B member 2
Twist2	−2.52	1.18E-08	Twist family bHLH transcription factor 2
Actg2	−2.27	0.0024	Gamma2 actin
Hand2	−2.24	0.0067	Heart and neural crest derivatives expressed 2
Edn3	−2.07	2.8E-6	Endothelin 3
Egfr	−2.05	0	Epidermal growth factor receptor
Gucy1a3	−1.84	0	Guanylate cyclase 1 soluble subunit alpha
Chrm3	−1.77	0.011	Cholinergic receptor muscarinic 3
Igf1	−1.7	0	Insulin like growth factor 1
Klf4	−1.68	1.77E-05	Kruppel like factor 4
Cald1	−1.63	0	Caldesmon 1
Cardiovascular system development
Bmp4	1.86	0	Bone morphogenic protein 4
Tnnt2	2.12	1.11E-16	Troponin T2, cardiac type
Tnni3	2.9	2.31E-05	Troponin I3, cardiac type
Tnni1	3.46	0	Troponin I1, slow skeletal; cardiac during embryonic development
Tnnc1	5.09	7.24E-13	Troponin C1, slow skeletal and cardiac type
Csrp3	13.26	0	Cysteine and glycine rich protein 3

Deletion of Foxf1 from smooth muscle cells causes tracheal and esophageal abnormalities

To examine the requirements for FOXF1 in smooth muscle cell lineage, we used the smMHC-Cre transgene (Fig. 7A), which is active after E13.5 (Xin et al., 2002). FOXF1 staining was decreased in esophageal smooth muscle of E17.5 smMHC-Cre Foxf1^−/− embryos, consistent with efficient Foxf1 deletion from smooth muscle cells (Fig. 7B-C). smMHC-Cre Foxf1^−/− mice died immediately after birth due to postnatal hemorrhage, causing accumulation of blood inside airways (Fig. 7E). smMHC-Cre Foxf1^−/− newborn mice exhibited hyperextension of esophagus, trachea and bronchi (Fig. 7D-E), which was associated with thinning of smooth muscle layers and loss of smooth muscle markers αSMA and γSMA (Fig. 7F-G). Decreased mRNAs of αSMA, γSMA and smMHC were found in micro-dissected smMHC-Cre Foxf1^−/− esophagi, consistent with the loss of smooth muscle cells (Fig. 7I). Proliferation of smooth muscle cells was reduced in smMHC-Cre Foxf1^−/− mice as shown by immunostaining for proliferation markers Ki-67, phospho-histone H3 (PH3) and FOXM1 (Fig. 7F and Fig. 7H).

Figure 7. Deletion of Foxf1 from smooth muscle cells causes tracheal and esophageal abnormalities.

A, Schematic drawing shows deletion of Foxf1-floxed allele from smooth muscle cells. Foxf1^fl/fl mice were bred with smMHC-Cre^tg/- mice to generate smMHC-Cre^tg/- Foxf1^fl/fl double transgenic mice (smMHC-Foxf1^−/−; abbreviated as smFoxf1^−/−). B, Immunostaining for FOXF1 was performed using paraffin sections from E17.5 smMHC-Foxf1^−/− and control Foxf1^fl/fl embryos. FOXF1 staining (dark brown) was decreased in esophageal smooth muscle of smMHC-Foxf1^−/− embryos. Slides were counterstained with nuclear fast red (red). C, The number of FOXF1 -positive cells was decreased in esophageal smooth muscle of smMHC-Foxf1^−/− embryos. Ten random x400 microscope fields were used for quantification (n = 3 mice per group; * indicates p < 0.05). D-E, H&E staining shows thinning and hyper-extension of esophagus and trachea in newborn smMHC-Foxf1^−/− mice. F, Immunostaining shows reduced smooth muscle cell proliferation in smMHC-Foxf1^−/− newborn mice. Immunostaining was performed using antibodies against αSMA, γSMA, Ki-67, PH3 and FOXM1. G, Thickness of esophageal smooth muscle and mesenchymal layers was decreased smMHC-Foxf1^−/− newborn mice compared to control Foxf1^fl/fl mice (n=3 mice per group). H, Percentage of proliferating cells was reduced in esophageal smooth muscle of smMHC-Foxf1^−/− newborn mice. Cells were counted using 10 random x400 microscope fields (n = 3 mice per group). I, qRT-PCR of RNA isolated from esophagus of smMHC-Foxf1^−/− newborn mice shows decreased expression of Foxf1, αSMA, γSMA and smMHC. Abbreviations: Eso, esophagus; Br, bronchus; Tr, trachea. Magnification: B (left panels), D (right panels) and F, x400; E and D (left panels), x100; B (right panels), x800.

Esophageal epithelium was hypoplastic in smMHC-Cre Foxf1^−/− mice (Supp. Fig. S13). While CK13-positive suprabasal cells were present in smMHC-Cre Foxf1^−/− tracheas, the basal cell layer had diminished expression of SOX2, p63 and CK14 (Supp. Fig. S13). CCSP-positive Clara and β-tubulin-positive ciliated cells were present in proximal and distal airways of both smMHC-Cre Foxf1^−/− and control mice (Suppl. Fig. S14). αSMA staining was reduced in proximal bronchioles of smMHC-Cre Foxf1^−/− lungs (Suppl. Fig. S14), a finding consistent with the loss of peribronchial smooth muscle cells. The number of proliferative cells was significantly reduced in peribronchial smooth muscle of smMHC-Cre Foxf1^−/− mice compared to controls (Suppl. Fig. S15). There were no differences in the thickness of smooth muscle layers surrounding intestine, and vascular smooth muscle was normal in smMHC-Cre Foxf1^−/− mice (Suppl. Fig. S16). Taken together, deletion of Foxf1 decreases cell proliferation and disrupts proper development of smooth muscle in esophagus, trachea and bronchi.

DISCUSSION

Published studies demonstrate that FOXF1 is expressed in extraembryonic mesoderm, the second heart field and lateral mesoderm (Mahlapuu et al., 2001b; Peterson et al., 1997). Later in development, FOXF1 is detected in septum transversum, venous pole mesenchyme and splanchnic mesoderm, which are critical for mesenchymal-epithelial signaling during formation of the heart, lung, liver, esophagus, stomach and intestine (Kalinichenko et al., 2002a). Consistent with this expression pattern, we found that FOXF1 is critical for proper development of respiratory, cardiovascular and gastrointestinal organ systems as Dermol-Cre-mediated deletion of Foxf1 from mesenchyme caused multiple developmental defects in the heart, lung, liver and esophagus. The presence of atrioventricular septal defects in Dermol-Foxf1^−/− hearts is consistent with published studies that reported similar findings in hearts of Foxf1^+/− Foxf2^+/− embryos (Hoffmann et al., 2014), implicating FOXF1 in mesenchymal signaling during cardiac septation. Compared to Foxf1^+/− Foxf2^+/− embryos, heart defects in Dermol-Foxf1^−/− embryos were more severe and included ventricular hypoplasia, thinning of myocardium and atrial chamber malformations, the latter of which could be a result of abnormal venous pole mesenchymal development. In addition to heart defects, Dermol-Foxf1^−/− embryos had small livers that lacked blood islands. This can result in hematopoietic deficiency and contribute to embryonic lethality in Dermol-Foxf1^−/− embryos. Since FOXF1 is expressed in septum transversum mesenchyme prior to liver development (Kalinichenko et al., 2002a), it is possible that mesenchymal Foxf1 deletion inhibits proper liver morphogenesis by disrupting signaling between the mesenchyme and endoderm-derived hepatic progenitors. In contrast to Foxf1^−/− and Tie2-Cre Foxf1^−/− embryos that exhibited vascular defects in the yolk sac (Mahlapuu et al., 2001b; Ren et al., 2014), there were no abnormalities in the yolk sac of Dermol-Foxf1^−/− embryos, suggesting that the Dermol-Cre transgene is inefficient in targeting the yolk sac vasculature. Our results also suggest that combined heart/liver insufficiency causes growth retardation and lethality in Dermol-Foxf1^−/− embryos.

Published studies demonstrate that haploinsufficiency of Foxf1 in mice (Foxf1^+/−) causes alveolar capillary dysplasia, fusion of the lung lobes and various developmental defects in mesenchyme of the gallbladder, esophagus, intestine and trachea (Kalinichenko et al., 2001a; Kalinichenko et al., 2002a; Lim et al., 2002; Mahlapuu et al., 2001a; Ormestad et al., 2006). Interestingly, inactivating mutations in FOXF1 gene were found in patients with Alveolar Capillary Dysplasia with Misalignment of Pulmonary Veins (ACD/MPV), a fatal congenital disorder which is often associated with lung hypoplasia, liver abnormalities and congenital heart defects, including AVS (Bishop et al., 2011). The lung, heart and liver defects in ACD/MPV patients are consistent with mouse genetic studies described in this manuscript, suggesting that FOXF1 deficiency in mesenchymal cells contributes to pathogenesis of ACD/MPV. While our study did not focus on FOXF1 requirements in the heart and liver, our results implicate FOXF1-expressing mesenchymal cells in development of these organs.

Lung morphogenesis requires crosstalk between respiratory epithelium and pulmonary mesenchyme mediated by SHH, WNTs, FGFs and BMPs (Bolte et al., 2018; Hogan and Yingling, 1998; Morrisey and Hogan, 2010). SHH is secreted by respiratory epithelial cells and acts via mesenchymal PTCH receptor and GLI transcription factors to induce FOXF1 through direct binding of GLI to the Foxf1 promoter (Madison et al., 2009; Szafranski et al., 2013). In the present study, we found that Dermol-Foxf1^−/− embryos exhibited severe lung hypoplasia, a phenotype similar to Gli2^+/−;Gli3^+/− mutants and embryos with epithelial-specific inactivation of Shh (Grindley et al., 1997; Miller et al., 2004; Motoyama et al., 1998). Therefore, FOXF1 could be an important downstream mediator of SHH signaling in lung mesenchyme. Furthermore, deletion of Foxf1 from mesenchyme disrupted epithelial development in the trachea and esophagus, which is consistent with previous studies demonstrating a critical role of mesenchymal-epithelial signaling in epithelial differentiation (Hines et al., 2013; Lee et al., 2017). Our data also suggest that FOXF1 deficiency causes lung hypoplasia, at least in part by inhibiting cell proliferation in fetal lung mesenchyme, leading to diminished expansion of mesenchyme-derived tissues. These data are consistent with recent reports showing decreased cellular proliferation in FOXF1-deficient endothelial cells (Bolte et al., 2017; Ren et al., 2014) and rhabdomyosarcomas (Milewski et al., 2017b). Altogether, our findings in Foxf1^−/− foregut explants and Dermol-Foxf1^−/− embryos indicate that FOXF1 expression in mesenchymal cells is critical for mesenchyme proliferation, as well as lung budding and branching.

In addition to proliferation defects, FOXF1-deficient mesenchyme had diminished expression of several WNT ligands and downstream targets of canonical WNT signaling pathway. Activated β-catenin was reduced in lung epithelium and mesenchyme of Dermol-Foxf1^−/− embryos, consistent with diminished WNT/β-catenin signaling. The lack of mesenchymal WNT ligands could lead to proliferation defects in respiratory epithelium and smooth muscle cells, contributing to lung hypoplasia in Dermol-Foxf1^−/− embryos. It is possible that FOXF1 directly induces transcription of Wnt2, Wnt11 and Wnt5A in lung mesenchyme since Foxf1-binding sites were detected by ChIPseq in promoters and enhancers of these genes. Canonical WNT signaling is required for specification of the lung bud (Herriges and Morrisey, 2014). Mice lacking Wnt2/2b or β-catenin fail to generate a lung field, whereas over-activation of the canonical WNT pathway expands the specified lung field into the anterior foregut (Goss et al., 2009). Therefore, reduced expression of Wnt2 and other mesenchymal Wnt ligands can disrupt lung budding and contribute to lung hypoplasia in Dermol-Foxf1^−/−embryos.

In summary, FOXF1 directly activates expression of genes critical for WNT signaling, mesenchyme proliferation and formation of the lung bud. FOXF1 acts downstream of SHH signaling and is required for development of multiple organ systems. Targeting FOXF1 or its effectors could be beneficial for treatment of human congenital disorders, including ACD/MPV.

Supplementary Material

Supplemental Figure S1. FOXF1 is expressed in mesenchyme and smooth muscle cells.

FOXF1 staining was observed in mesenchyme and smooth muscle cells surrounding esophagus and trachea (A), stomach (B) and intestine (C) of E13.5 wild type embryos. FOXF1 was not expressed in smooth muscle cells surrounding artery (D). There was no FOXF1 staining in myocardium (E) or skeletal muscle (F). Abbreviations: Eso, esophagus; Tr, trachea; St, stomach; In, intestine; Ar, artery; LV, left ventricle; RV, right ventricle; Sk, skeletal muscle. Magnification: A, C, D and F, x200; B, x100; E, x50.

Supplemental Figure S2. Deletion of Foxf1 from mesenchyme causes heart defects.

A, H&E staining shows intraventricular septum defect (arrow) and lack of blood in atriums of E13.5 Dermol-Foxf1^−/− hearts. B, Limbs in E13.5 Dermol-Foxf1^−/− embryos had all 5 digits (indicated by numbers). Abbreviations: eso, esophagus; tr, trachea; a, artery; LV, left ventricle; RV, right ventricle; RA, right atrium; LA, left atrium. Magnification: A, x20; B, x50.

Supplemental Figure S3. Atrial abnormalities in Dermol-Foxf1^−/− hearts.

H&E was used to stain consecutive paraffin sections of E12.5 Dermol-Foxf1^−/− and control Foxf1^fl/fl hearts. Sections are shown from the top to bottom and from the left to right. Distal parts of Dermol-Foxf1^−/− atrial chambers are collapsed (arrows). Magnification is x50.

Supplemental Figure S4. FOXF1 expression in septum transversum and venous pole mesenchyme.

FOXF1 staining (dark brown) is observed in venous pole mesenchyme (VPM), dorsal mesenchymal protrusion (DMP) and septum transversum mesenchyme (STM) of E12.5 embryos. Slides were counterstained with nuclear fast red (red). High magnification images (boxes) are shown in middle and bottom panels. DMP is absent in Dermol-Foxf1^−/− embryos. FOXF1 staining is decreased in VMP and STM of Dermol-Foxf1^−/− embryos. Abbreviations: A, atrium; V, ventricle; Li, liver. Magnification: top panels, x50; middle and bottom panels, x400.

Supplemental Figure S5. Lung hypoplasia in Dermol-Foxf1^−/− embryos.

Whole lungs were dissected from E11.5 Dermol-Foxf1^−/− and control Foxf1^fl/fl embryos. Photographs show lung hypoplasia in Dermol-Foxf1^−/− embryos (n = 3 embryos in each group).

Supplemental Figure S6. Lung histology in Dermol-Foxf1^−/− embryos.

Representative H&E images of Dermol-Foxf1^−/− and control Dermol-Foxf1^+/− lungs show no differences in histology at E16.5 (top and middle panels). Immunostaining for SOX2 (red) and SOX9 (green) was used to visualize proximal and distal airways (bottom panels). Magnification: top panels, x200; middle and bottom panels, x400.

Supplemental Figure S7. Foxf1 deletion from mesenchyme reduces the number of peribronchial smooth muscle cells.

A, Immunostaining shows decreased expression of αSMA in trachea and bronchi of Dermol-Foxf1^−/− embryos. Co-localization studies were performed using E13.5 lung sections and antibodies against FOXF1, SOX9 and αSMA. Slides were counterstained with DAPI. B-C, The number of αSMA⁺FOXF1⁺ cells and the percentage of the airway epithelium circumference covered by αSMA⁺ cells were calculated using ten x200 microscope fields (n = 3 embryos in each group; * indicates p < 0.05). D, Immunostaining shows normal expression of αSMA in pulmonary arteries (arrows). Slides were counterstained with nuclear fast red. Abbreviations: Tr, trachea; Br, bronchi; Ar, artery. Magnification: top panels in A, x400; bottom panels in A, x300; panels in D, x400.

Supplemental Figure S8. Decreased expression of p63 and SOX9 in trachea of Dermol-Foxf1^−/− embryos.

Co-localization studies were performed using E16.5 paraffin sections and antibodies against FOXF1, αSMA, SOX9, SOX2, E-Cadherin, p63, Ki-67 and NKX2.1. Loss of peritracheal αSMA-positive cells was observed in Dermol-Foxf1^−/− embryos. p63 and SOX9 were reduced in trachea (Tr) of Dermol-Foxf1^−/− E16.5 embryos compared to Dermol-Foxf1^+/− controls. Magnification: top panels in each staining, x300; remaining panels, x1000.

Supplemental Figure S9. Deletion of Foxf1 from mesenchyme does not influence cell survival.

Immunostaining was performed using paraffin sections from El3.5 Dermol-Foxf1^−/− and control embryos. FLK1-positive blood vessels were observed in Dermol-Foxfl^−/− and control lungs. Apoptosis was not observed in lungs or esophagi of Dermol-Foxf1^−/− or control embryos. Inserts show rare apoptotic cells (positive for Cleaved caspase-3) in ganglions adjacent to the spinal cord. Magnification is x400 for panels and x800 for inserts.

Supplemental Figure S10. Biological processes influenced in Dermol-Foxf1^−/− lungs are identified using RNA-seq analysis.

RNA-seq was performed using distal tip lung mesenchyme micro-dissected from E10.5 Dermol-Foxf1^−/− and control Foxf1^fl/fl embryos (n=3 embryos in each group). Biological processes influenced by Dermol-Cre-mediated Foxf1 deletion were identified using ToppGene Suite (http://toppgene.cchmc.org). Statistical significance of each bioprocess is shown using negative log2 transformation of P value.

Supplemental Figure S11. FOXF1 binds DNA regulatory regions in Egfr, Igf1, Wnt5a and Wnt11 genes.

ChIPseq shows Foxf1-binding sites (arrows) in Egfr, Igf1, Wnt5a and Wnt11 genes. Whole E18.5 lungs were used for ChIPseq. Data were aligned using the Biowardrobe platform.

Supplemental Figure S12. Deletion of Foxf1 from mesenchyme reduces nuclear localization of activated β-catenin.

Immunostaining was performed using antibody to the activated form of β-catenin (phospho-Ser552) and paraffin sections from E13.5 (A) and E16.5 (B) Dermol-Foxf1^−/− embryos. Dermol-Foxf1^+/− littermates were used as controls. Arrowheads show epithelial cells containing nuclear β-catenin. Magnification: top panels in A, x200; bottom panels in A, x800; panels in B, x400; inserts in B, x1000.

Supplemental Figure S13. Foxf1 deletion from smooth muscle cells reduces p63, CK14 and SOX2 in esophageal epithelium.

Immunostaining shows decreased expression of p63, SOX2 and CK14 in esophagus (Eso) of smMHC-Cre Foxf1^−/− newborn mice (abbreviated as smFoxf1^−/−). smMHC-Cre Foxf1^+/− littermates (abbreviated as smFoxf1^+/−) were used for comparison. Slides were counterstained with nuclear fast red. Magnification is x1000.

Supplemental Figure S14. Expression of αSMA and epithelial markers in smMHC-Cre Foxf1^−/− lungs.

Immunostaining was performed using lung paraffin sections from smMHC-Cre Foxf1^−/− newborn mice (abbreviated as smFoxf1^−/−). smMHC-Cre Foxf1^+/− littermates (abbreviated as smFoxf1^+/−) were used as controls. Proximal (top panels) and distal airways (bottom panels) are shown for comparison. Slides were counterstained with nuclear fast red. Abbreviations: Br, bronchi; Ar, artery. Magnification is x400.

Supplemental Figure S15. Decreased number of proliferative peribronchial smooth muscle cells in smMHC-Cre Foxf1^−/− mice.

A, Immunostaining shows decreased Ki-67 staining in peribronchial smooth muscle of smMHC-Cre Foxf1^−/− newborn mice (abbreviated as smFoxf1^−/−). smMFIC-Cre Foxf1^+/− littermates (abbreviated as smFoxf1^+/−) were used as controls. Slides were counterstained with nuclear fast red. Magnification is x400. B, The number of Ki-67⁺ smooth muscle cells was counted using ten x400 microscope fields (n = 3 embryos in each group; * indicates p < 0.05). Abbreviations: Br, bronchi.

Supplemental Figure S16. Vascular and intestinal smooth muscles are not affected in smMHC-Cre Foxf1^−/− mice.

Paraffin sections from smMFIC-Cre Foxf1^−/− newborn mice (abbreviated as smFoxf1^−/−) were used for H&E staining and immunostaining for αSMA and γSMA, Abbreviations: A, artery; In, intestine. Magnification: left panels, x200; middle and right panels, x400.

Highlights:

• -FOXF1 is required for embryonic development of respiratory, cardiovascular and gastrointestinal organ systems.
• -FOXF1 expression in mesenchymal cells is critical for lung budding and branching.
• -FOXF1 stimulates cellular proliferation in fetal lung mesenchyme.
• -Deletion of Foxf1 from mesenchyme causes the loss of visceral smooth muscle cells and disrupts epithelial development in the trachea and esophagus.
• -FOXF1 induces canonical WNT signaling in respiratory epithelium and lung mesenchyme.