Haploinsufficiency of the Mouse Forkhead Box f1 Gene Causes Defects in Gall Bladder Development*

The forkhead box f1 (Foxf1) transcription factor is expressed in the visceral (splanchnic) mesoderm, which is involved in mesenchymal-epithelial signaling required for development of organs derived from foregut endoderm such as lung, liver, gall bladder, and pancreas. Our previous studies demonstrated that haploinsufficiency of the Foxf1 gene caused pulmonary abnormalities with perinatal lethality from lung hemorrhage in a subset ofFoxf1+/− newborn mice. During mouse embryonic development, the liver and biliary primordium emerges from the foregut endoderm, invades the septum transversum mesenchyme, and receives inductive signaling originating from both the septum transversum and cardiac mesenchyme. In this study, we show that Foxf1 is expressed in embryonic septum transversum and gall bladder mesenchyme.Foxf1+/− gall bladders were significantly smaller and had severe structural abnormalities characterized by a deficient external smooth muscle cell layer, reduction in mesenchymal cell number, and in some cases, lack of a discernible biliary epithelial cell layer. This Foxf1+/− phenotype correlates with decreased expression of vascular cell adhesion molecule-1 (VCAM-1), α5 integrin, platelet-derived growth factor receptor α (PDGFRα) and hepatocyte growth factor (HGF) genes, all of which are critical for cell adhesion, migration, and mesenchymal cell differentiation.

At 9-days postcoitum (dpc) 1 during mouse embryogenesis, the liver primordium emerges from the foregut endoderm, invades the septum transversum mesenchyme, and receives bone morphogenetic protein 4 (Bmp-4) and fibroblast growth factor 2 (Fgf2) signaling originating from the septum transversum and cardiac mesenchyme, respectively (1,2). This hepatic specification is associated with expression of the Foxa2 (HNF-3␤) and Gata4 transcription factors (3). Liver morphogenesis involves a proliferative expansion period and development of the bipotential hepatoblasts that begin to differentiate into hepatocytes and bile duct epithelial cells of the intrahepatic bile ducts (IHBD) at 13.5 dpc (3,4). Interestingly, targeted disruption of either the homeodomain Hex gene or the Hgf gene allows normal development of the mouse hepatic diverticulum, but these cells fail to migrate into the septum transversum and undergo liver morphogenesis (5)(6)(7).
In the adult liver, bile is synthesized in hepatocytes from cholesterol, secreted into the bile canaliculi and transported through the intrahepatic and extrahepatic biliary system to the gall bladder, where it is stored for secretion into the digestive tract to emulsify lipids (8). The gall bladder and extrahepatic bile ducts (EHBD) develop from the caudal portion of the liver primordium at 10 dpc of mouse embryogenesis (9). However, little is known regarding visceral (splanchnic) mesoderm transcription factors that regulate expression of genes involved in mesenchymal-epithelial induction of gall bladder development from the liver primordium. Visceral mesenchymal expression of the homeodomain Hlx gene is required for the proliferative expansion required for mouse liver and intestine morphogenesis, but it is not essential for normal specification of the liver primordium (10).
The forkhead box (Fox) proteins are an extensive family of transcription factors that share homology in the winged helix/ fork head DNA binding domain (11) and play important roles in regulating expression of genes involved in cellular proliferation, differentiation, and transformation, and in metabolic homeostasis (3,4,(12)(13)(14)(15)(16)(17)(18)(19). Expression of Foxf1 is restricted to the visceral (splanchnic) mesoderm, which provides mesenchymalepithelial signaling for gut-derived organs such as liver, gall bladder, lung, stomach, pancreas, and intestine (20,21). The Foxf1 gene is required for mesodermal differentiation and cell adhesion and its disruption results in embryonic lethality at midgestation (22). Interestingly, haploinsufficiency of the Foxf1 gene in heterozygous mice causes perinatal lethality from pulmonary hemorrhage with severe defects in formation of alveolar sacs and capillaries and fusion of the lung lobes (23)(24)(25). Lung hemorrhage was observed in a subset of newborn Foxf1ϩ/Ϫ mice that displayed an 80% reduction in pulmonary Foxf1 levels. This pulmonary defect is associated with diminished expression of Bmp-4, vascular endothelial growth factor (VEGF) receptor type 2 (Flk-1), and the transcription factors of the brachyury T-box family (Tbx2-Tbx5) and lung Kruppel-like factor (23).
In this study, we show that Foxf1 is expressed in mouse embryonic septum transversum and that Foxf1 haploinsuffi-ciency results in abnormal development of the gall bladders. Foxf1ϩ/Ϫ gall bladders exhibited malformation of the external smooth muscle cell layer, reduction in mesenchymal cell number, and in some cases, lack of a discernible biliary epithelial cell layer. Abnormalities in gall bladder formation are associated with diminished expression of vascular cell adhesion molecule 1 (VCAM-1), ␣ 5 integrin, platelet-derived growth factor receptor ␣ (PDGFR␣), and HGF. Normal development of the liver and intrahepatic bile ducts was found in Foxf1ϩ/Ϫ mice, and this was associated with compensatory increase in hepatic Foxf1 levels.

MATERIALS AND METHODS
Foxf1ϩ/Ϫ Mice-We previously described the generation of heterozygous (ϩ/Ϫ) Foxf1 mice in which the Foxf1 allele was disrupted by an in-frame insertion of a nuclear-localizing ␤-galactosidase (␤-gal) gene, and thus staining for ␤-gal enzyme activity allowed identification of Foxf1-expressing cells (23). Foxf1ϩ/Ϫ mice used in the gall bladder analysis were bred for three or four generations into the Black Swiss mouse genetic background. Both male and female Foxf1ϩ/Ϫ mice exhibited similar defects in gall bladder formation.
Immunohistochemical and ␤-Galactosidase Enzyme Staining-Paraffin sections were stained with hematoxylin and eosin for morphological examination or used for immunohistochemistry as described previously (23,26). Rat monoclonal PDGF receptor ␣-chain antibody (Ab; clone APA5; BD PharMingen, San Diego, CA) was used at 1:50 dilution, and immunoreactivity was visualized using biotinylated anti-rat Ab, avidin-alkaline phosphatase complex, and BCIP/NBT substrate (all from Vector Laboratories, Burlingame, CA). Tissue was stained with mouse monoclonal ␣-smooth muscle actin Ab (clone1A4; Sigma), and immunoreactivity was detected using an anti-mouse Ig protein conjugated to alkaline phosphatase with BCIP/NBT substrate as described (26). To stain for biliary epithelial cells, we used mouse monoclonal cytokeratin Ab (clone PCK-26; Sigma), which was detected by antimouse Ig protein conjugated with TRITC (Dako Corp, Carpinteria, CA), and endothelial cells were visualized using fluorescein isothiocyanateconjugated isolectin B4 from Griffonia Simplicifolia (GS; Vector Laboratories) as previously described (23). To develop a Foxf1 Ab, rabbits were immunized with mouse Foxf1 RYHSQSPSMCDRKE peptide (Foxf1 amino acids 327-340) conjugated to keyhole limpet hemocyanin (KLH) protein using Sigma Genosys custom peptide antisera service. The Foxf1 peptide Ab was affinity-purified using the same peptide coupled to N-hydroxysuccinimide-activated Sepharose 4 Fast Flow beads (Amersham Biosciences, Inc.) and eluted from the column as described previously (27). For immunohistochemical staining with the Foxf1 peptide Ab, paraffin was removed from the tissue section, was rehydrated, and was then incubated for 15 min at room temperature with phosphate-buffered saline containing 20 g/ml of proteinase K (Roche Molecular Biochemicals, Indianapolis, IN). The Foxf1 peptide Ab was used at 1:100 dilution and visualized using biotinylated antirabbit Ab (BD PharMingen), avidin-alkaline phosphatase complex, and BCIP/NBT substrate (Vector Laboratory). To determine Foxf1-expressing cells, Foxf1ϩ/Ϫ embryos or liver tissue containing gall bladder (15 dpc and adult) were stained for ␤-gal activity with 5-bromo-4-chloro-3indolyl ␤-D-galactopyranoside substrate and paraffin-embedded and sectioned as described previously (23,26). Nuclear fast red (Vector Laboratories) was used as a counterstain.
RNase protection assays was used to determine Foxf1 expression levels in embryonic liver and the ImageQuant program (Amersham Biosciences, Inc.) was used for quantification of phosphorimager scans as described previously (18,23,26). Expression levels were normalized to cyclophilin mRNA levels and expressed as the mean-fold wild type levels Ϯ S.D.
Western Blot Analysis-Total protein extracts were prepared from gall bladder for Western blot analysis (23, 26) using the following antibodies for immune detection: rat monoclonal PDGFR␣ (clone APA5; BD PharMingen; 1:500 dilution), mouse monoclonal ␣-smooth muscle actin (clone1A4; Sigma; 1:1000 dilution), mouse monoclonal ␤-actin (clone AC-15; Sigma; 1:3000 dilution), and rabbit polyclonal CDK2 (BD PharMingen; 1:1000 dilution). Detection of the immune complex was accomplished by using secondary Ab directly conjugated with horseradish peroxidase followed by chemiluminescence (ECLϩplus, Amersham Biosciences, Inc.). Quantitation of expression levels was determined with Tiff files from scanned films by using the BioMax 1D program. The CDK2 expression signal was used for normalization control between different protein samples.

Foxf1ϩ/Ϫ Mice Exhibit a Wide Range of Abnormalities in
Gall Bladder Structure-Foxf1ϩ/Ϫ mice either lacked a discernible gall bladder or possessed significantly smaller gall bladders with severe structural abnormalities. Examination of 25 adult Foxf1ϩ/Ϫ mice revealed that ϳ49% of adult Foxf1ϩ/Ϫ mice lacked a discernible gall bladder (Fig. 1, A and B), another 49% of the heterozygous mice possessed a rudimentary gall bladder, and 2% exhibited normal gall bladder formation. Histological examination of the rudimentary Foxf1ϩ/Ϫ gall bladders (Fig. 1, D-F) demonstrated that they consisted of either a constricted lumen with few convolutions (Fig. 1D) or a collection of extrahepatic bile ducts (Fig. 1, E and F) compared with wild type gall bladder (Fig. 1C).
Because the targeted Foxf1 allele possesses an in-frame insertion of the nuclear-localizing ␤-gal gene, staining for ␤-gal enzyme activity allows the identification of Foxf1-expressing cells (23,26). In 9.5-dpc Foxf1ϩ/Ϫ embryos, ␤-gal staining was found in the septum transversum mesoderm of the hepatic primordium (HP) and visceral mesenchyme juxtaposed to foregut (Fg) endoderm (Fig. 2, B and C), suggesting that Foxf1 regulates mesenchymal genes involved in liver and gall bladder development. In 15-dpc Foxf1ϩ/Ϫ mouse embryos, ␤-gal staining was observed in the gall bladder mesenchyme surrounding the biliary epithelial cell layer and in mesenchymal cells of the liver, but Foxf1 expression was not detected in the vessel mesenchyme (Fig. 2G). In adult Foxf1ϩ/Ϫ mice, ␤-gal staining was found in mesenchymal cells of the gall bladder (Fig. 2I), extrahepatic bile duct (Fig. 2J), and hepatic sinusoids (Fig. 2K), but no Foxf1 expression was detected in either heterozygous (Fig.  2K) or wild type (data not shown) IHBD.
Although embryonic 13-, 15-, and 18-dpc Foxf1ϩ/Ϫ mice displayed variation in severity of their gall bladder defects, all of them exhibited significant reduction in the mesenchymal cell layer (Fig. 2, E, G, H, M, and N) compared with that of wild type gall bladder (Fig. 2, D, F, and L). The entire mesenchymal layer displayed Foxf1-dependent ␤Ϫgal enzyme staining in 15-dpc Foxf1ϩ/Ϫ gall bladders (Fig. 2, G and H). We identified a subset of Foxf1ϩ/Ϫ mouse gall bladders that lacked the biliary epithelial cell layer (Fig. 2, H and N) as evidenced by undetectable immunohistochemical staining with cytokeratin (Fig. 3C, CK) compared with either wild type (Fig. 3A) or less severely affected Foxf1ϩ/Ϫ gall bladder (Fig. 3B). The large vessels adjacent to the Foxf1ϩ/Ϫ gall bladder exhibited normal development of the endothelial cells as evidenced by staining with fluorescein isothiocyanate-conjugated isolectin B4 (Fig. 3, A-C, GS), a finding consistent with the lack of Foxf1 expression in large vessels (Fig. 2, G-H). Furthermore, RT-PCR expression analysis of three discernible adult Foxf1ϩ/Ϫ gall bladders revealed a 90% decrease in expression of the HNF6, HNF3␤, and Hex transcription factors, suggesting that they lacked a normal epithelial cell layer (Fig. 4A). These results suggest that Foxf1ϩ/Ϫ mice exhibit defects in the development of epithelial cells in the gall bladder, possibly due to reduced mesenchymal signaling.
Even in the severely affected Foxf1ϩ/Ϫ gall bladders, the heterozygous liver architecture (data not shown) as well as the cytokeratin staining pattern (Fig. 3, E-F) and structure of intrahepatic bile ducts (Fig. 3H, IHBD) was indistinguishable from that of wild type liver (Fig. 3, D and G). Furthermore, Foxf1ϩ/Ϫ mice displayed wild type liver function as determined by normal serum levels of liver aminotransferases, alkaline phosphatase enzymes, glucose, bilirubin, and albumin (Table I). Interestingly, RNase protection assays demonstrated that 18-dpc Foxf1ϩ/Ϫ liver displays a 3-fold compensatory increase in Foxf1 mRNA levels compared with wild type livers (Fig. 3I).
Defects in Foxf1ϩ/Ϫ Gall Bladder Mesenchyme Are Associated with Significant Decreases in Foxf1, VCAM-1, ␣ 5 integrin, and HGF Expression-RT-PCR analysis of Foxf1ϩ/Ϫ gall bladder RNA showed that defective gall bladder formation was associated with a 70% reduction in Foxf1 levels (Fig. 4, A and  B). This finding is consistent with Foxf1ϩ/Ϫ pulmonary defects, which were also associated with 80% reduction in Foxf1 mRNA levels (23). Because expression of the cell adhesion ␣ 5

FIG. 2. Foxf1؉/؊ embryonic gall bladder exhibit reduction in mesenchymal cell layer.
A-C, Foxf1 is expressed in septum transversum. Wild type (ϩ/ϩ) or Foxf1ϩ/Ϫ (ϩ/Ϫ) mouse embryos (9.5 dpc) were stained for ␤-gal enzyme activity (blue), and tissue sections were counterstained with nuclear fast red (red). Foxf1-dependent ␤-gal staining is observed in mesenchyme adjacent to the foregut (Fg) and in the septum transversum of the hepatic primordium (HP). Box in B shows area magnified for C. D and E, sections of 13-dpc wild type and Foxf1ϩ/Ϫ gall bladder (GB) were stained with nuclear fast red. F-K, mesenchyme of 15 dpc or adult Foxf1ϩ/Ϫ gall bladders stained for ␤-gal enzyme activity. Foxf1 is expressed in mesenchyme of 15 dpc (G and H) and adult gall bladder (I) as well as mesenchymal cells around extrahepatic bile duct (EHBD, J). Foxf1 expression is absent from mesenchymal layers surrounding intrahepatic bile ducts (IHBD, K) adjacent to the portal vein (PV). L-N, sections of wild type and Foxf1ϩ/Ϫ 18-dpc gall bladders were stained with hematoxylin and eosin. Magnification: A and B, ϫ50; C, ϫ150; D and E, ϫ300; F-H and L-N, ϫ100; I and J, ϫ200; K, ϫ400. integrin and VCAM-1 genes is essential for mesoderm formation (22,30), we examined their levels in Foxf1ϩ/Ϫ gall bladders. RT-PCR analysis showed a drastic reduction in VCAM-1 and ␣ 5 integrin expression in Foxf1ϩ/Ϫ gall bladders, suggesting that their diminished levels contribute to the observed mesenchymal defect (Fig. 4, A and B). Furthermore, reduced expression of HGF was found in Foxf1ϩ/Ϫ embryonic and adult gall bladders (Fig. 4, A and B). Because HGF plays a critical role in liver and bile duct morphogenesis (7,9,10,31), the decline in HGF expression may explain defective formation of Foxf1ϩ/Ϫ gall bladders.
We generated an affinity-purified Foxf1 peptide antibody to determine the Foxf1 expression pattern in gall bladder using immunohistochemical staining. Western blot analysis demonstrated that the Foxf1 antibody detects an appropriate 39-kDa protein using protein extracts prepared from lung, a tissue that expresses high levels of Foxf1 (Fig. 5A). Immunohistochemical staining of 18-dpc wild type gall bladder with this affinitypurified Foxf1 peptide antibody demonstrated that mesenchymal Foxf1 staining gradually decreased toward the smooth muscle cell layer (Fig. 6A), and its expression continued in the submucosa and mucosa mesenchyme of the adult gall bladder (Fig. 6C). Interestingly, this Foxf1 protein expression pattern was significantly diminished in embryonic and adult Foxf1ϩ/Ϫ gall bladders (Fig. 6, B and D), confirming the reduction in Foxf1 mRNA levels.

Diminished Number of Smooth Muscle Cells in Foxf1ϩ/Ϫ Gall Bladder Is Associated with Reduced Expression of
PDGFR␣-Immunohistochemical staining of Foxf1ϩ/Ϫ gall bladder with an ␣SM antibody revealed a defective formation of the smooth muscle cell layer (Fig. 6, F and H) compared with that of wild type gall bladder (Fig. 6, E and G). Western blot analysis with protein extracts from Foxf1ϩ/Ϫ and wild type gall bladders supported this diminished expression in heterozygous ␣SM protein levels, whereas ␤Ϫactin protein levels remained unchanged (Fig. 5B). Because PDGFR␣ is critical for differentiation of smooth muscle cell (27), we examined its expression in Foxf1ϩ/Ϫ gall bladders. In wild type gall bladder, high expression levels of PDGFR␣ protein were found in the mesenchymal cells juxtaposed to the epithelial cell layer (Fig.  6, I and K). In contrast, PDGFR␣ immunohistochemical staining was barely detectable in Foxf1ϩ/Ϫ gall bladders (Fig. 6, J and L). Western blot analysis revealed a 95% reduction in PDGFR␣ protein expression compared with wild type gall bladder (Fig. 5B), suggesting that its diminished levels contributes to defective formation of smooth muscle cells in Foxf1ϩ/Ϫ gall bladder.

DISCUSSION
The Foxf1 protein is a potent transcriptional activator and its expression is restricted to the visceral (splanchnic) mesoderm, which expresses genes involved in mesenchymal-epithelial signaling required for development of foregut-derived organs (20,21). Our previous studies demonstrate that haploinsufficiency of the Foxf1 gene caused lung hemorrhage and perinatal lethality in a subset of Foxf1ϩ/Ϫ newborn mice that exhibited an 80% reduction in wild type pulmonary Foxf1 levels (23). Abnormalities in pulmonary alveolarization and vascularization and increased apoptosis of mesenchymal cells were associated with the severe Foxf1ϩ/Ϫ phenotype (23). In this study, we find that Foxf1 haploinsufficiency also resulted in defective formation of the gall bladder, which was associated with a 70% reduction in wild type Foxf1 mRNA levels. Similar to many heterozygous phenotypes, Foxf1ϩ/Ϫ embryos show differences in the severity of the gall bladder defect, and its severe phenotype does not exhibit total penetrance. The Foxf1ϩ/Ϫ mice either lacked an appreciable gall bladder or possessed a smaller rudimentary structure with severe abnormalities. All of the Foxf1ϩ/Ϫ gall bladders displayed a significant reduction in cell numbers within the mesenchymal layers, paucity of smooth muscle cells, and in some cases, lack of a discernible biliary epithelial cell layer. These gall bladder defects are associated with reduced expression of cell adhesion molecules VCAM-1 and ␣ 5 integrin as well as diminished levels of signal transduction PDGFR␣ and HGF proteins. Collectively, our data demonstrate that Foxf1 regulates expression of mesenchymal genes required for proper gall bladder development and function.
We used Foxf1ϩ/Ϫ embryos with a ␤Ϫgal knock-in gene to demonstrate that Foxf1 is expressed in septum transversum mesenchyme during formation of the hepatic primordium and induction of gall bladder development. Although Foxf1 expression continues in the mesenchymal cells of the liver sinusoids, we found no defects in hepatic architecture, and serum analysis demonstrated that liver function was normal in Foxf1ϩ/Ϫ mice. Interestingly, unlike the gall bladder, we found that Foxf1ϩ/Ϫ livers exhibited an increase in Foxf1 mRNA, suggesting the possibility that its compensatory increase prevents defects in liver development. Moreover, Foxf1 is not expressed in the mesenchyme of the IHBD, and we found that Foxf1ϩ/Ϫ liver exhibited no defects in IHBD.
In vivo transplantation studies indicate that bile ducts differentiate abundantly when fetal liver tissue was placed adjacent to mesenchyme (9), suggesting that bile duct development requires mesenchymal signaling. Moreover, the mesenchymederived HGF has been shown to induce lumen formation in cultures of epithelial cell lines and to promote cyst maturation and proliferation of gall bladder epithelial cells (9,31). Our data show that HGF expression is significantly reduced in Foxf1ϩ/Ϫ gall bladders and may therefore contribute to the mesenchymal and epithelial cell defects found in Foxf1ϩ/Ϫ gall bladder. Our finding that the HGF promoter region contains potential binding sites for the Foxf1 transcription factor (12) suggests the possibility that Foxf1 directly regulates transcription of the HGF gene. Moreover, the most severely affected Foxf1ϩ/Ϫ gall bladders lack a discernible biliary epithelial cell layer as evidenced by lack of expression of cytokeratin and the transcription factors Hex, HNF-6, and HNF-3␤. Because epithelial cells secrete growth factors that impinge upon development of the mesenchymal cells, it is interesting to note that Foxf1ϩ/Ϫ gall bladder lacking an biliary epithelial cell layer possess more severe defects in the mesenchymal cells (Figs. 2H and 3C).
We also observed that Foxf1ϩ/Ϫ gall bladders show significant reduction in expression of ␣-smooth muscle actin, indicating that they lack a discernible external smooth muscle cell layer. Although Foxf1ϩ/Ϫ gall bladders possess fewer mesenchymal cells, we observed almost undetectable levels of PDGFR␣, ␣ 5 integrin, and VCAM-1 in developing Foxf1ϩ/Ϫ gall bladders (Figs. 4B and 5B), which cannot be explained solely by the reduction of mesenchymal cells. Our data suggest that reduced Foxf1 levels in heterozygous gall bladders are associated with diminished expression of these mesenchymal receptor and cell adhesion proteins. Mice containing a homozygous mutation for the PDGFR␣ gene exhibit disrupted formation and differentiation of smooth muscle cells in the developing intestinal tract (32). Taken together, these results suggest that diminished mesenchyme expression of PDGFR␣ may contribute to the smooth muscle defect in Foxf1ϩ/Ϫ gall bladders. Because HGF signaling has been implicated in smooth muscle cell migration (33), diminished HGF levels may also contribute to the reduced smooth muscle cell layer in Foxf1ϩ/Ϫ gall bladder. This defective cell migration may be further exacerbated because of the diminished expression of cell adhesion molecules VCAM-1 and ␣ 5 integrin. Moreover, because proper expression of ␣ 5 integrin and VCAM-1 is essential for mesoderm formation (22,30), its decreased levels in Foxf1ϩ/Ϫ gall bladders provide a potential explanation for the observed mesenchyme defects.
In summary, we show that haploinsufficiency of the Foxf1 gene causes defects in gall bladder development characterized by abnormalities in the mesenchymal, smooth muscle, and biliary epithelial cell layers. This gall bladder phenotype is associated with decreased expression of VCAM-1, ␣ 5 integrin, PDGFR␣, and HGF genes, which are critical for cell adhesion, migration, and mesenchymal cell differentiation. Collectively, our data suggest that wild type levels of Foxf1 are necessary to regulate expression of mesenchymal genes, whose expression is essential for proper gall bladder development. FIG. 5. Defects in Foxf1؉/؊ gall bladder are associated with decrease in ␣SM actin and PDGFR␣ levels. A, specificity of Foxf1 peptide antibody. One hundred micrograms of total lung protein extract was used for Western blot analysis with affinity-purified Foxf1 peptide antibody. Indicated on the blot is the position of the 39-kDa Foxf1 protein and the nonspecific 50-kDa protein (NS) that is found with the secondary antibody control. B, Foxf1ϩ/Ϫ gall bladder exhibit diminished ␣SM and PDGFR␣ protein levels. Western blot analysis of adult gall bladder protein extracts with ␣SM and PDGFR␣ antibodies. CDK2 and ␤Ϫactin Western blots were used as controls. Protein levels were normalized to CDK2, and the numbers below each panel represent the average of expression levels in Foxf1ϩ/Ϫ gall bladder with respect to the wild type gall bladder.