Potential conflict of interest: Nothing to report.
The Notch signaling pathway is involved in determination of cell fate and control of cell proliferation in multiple organ systems. Jag1 encodes a ligand in the Notch pathway and has been identified as the disease-causing gene for the developmental disorder Alagille syndrome. Evidence from the study of human disease and mouse models has implicated Jag1 as having an important role in the development of bile ducts. We have derived a conditional knockout allele (Jag1loxP) to study the role of Jag1 and Notch signaling in liver and bile duct development. We crossed Jag1loxP mice with a transgenic line carrying Cre recombinase under the control of the albumin promoter and α-fetoprotein enhancer to ablate Jag1 in hepatoblasts. The liver-specific Jag1 conditional knockout mice showed normal bile duct development. To further decrease Notch pathway function, we crossed the Jag1 conditional knockout mice with mice carrying the hypomorphic Notch2 allele, and bile duct anatomy remained normal. When Jag1 conditional mice were crossed with mice carrying the Jag1 null allele, the adult progeny exhibited striking bile duct proliferation. Conclusion: These results indicate that Notch signaling in the liver is sensitive to Jag1 gene dosage and suggest a role for the Notch pathway in postnatal growth and morphogenesis of bile ducts. (HEPATOLOGY 2007.)
The Notch signaling pathway is an intercellular signaling mechanism that is evolutionarily conserved across species from Drosophila to humans. Notch signaling is essential in embryonic development and decisions on cell fate in multiple organ systems. In mammals, the major components of the Notch pathway are 5 ligands (Jagged1, Jagged2, Delta-like-1 [Dll1], Dll3, and Dll4) that signal 4 transmembrane receptors (Notch1, Notch2, Notch3, Notch4).1 Mutations in several of the Notch pathway genes have been associated with human disease. For example, DLL3 has been identified as the disease-causing gene for the skeletal disorder spondylocostal dysostosis, and NOTCH3 mutations have been found in patients with CADASIL, an adult-onset stroke and dementia syndrome.2, 3 Mutations in JAGGED1 (JAG1) cause the multisystem developmental disorder Alagille syndrome (AGS; OMIM #118450).4, 5
Alagille syndrome is an autosomal dominant disorder characterized by cholestasis and paucity of bile ducts and associated with congenital heart defects, vertebral anomalies, and ocular abnormalities. A subset of patients also exhibits renal disease or vascular anomalies. In a large study, the prevalence of bile duct paucity on liver biopsy increased from 60% in infants less than 6 months of age to more than 95% in patients older than 6 months.6 In addition, many liver biopsies of AGS patients obtained during infancy show bile duct proliferation. These findings suggest that embryonic bile duct development occurs normally in AGS and that Notch signaling may be important for postnatal bile duct growth and remodeling. Currently, mutations in JAG1 can be found in more than 90% of patients with a definitive clinical diagnosis of AGS.7 No clear genotype-phenotype correlation has been identified; patients with gene deletions or truncating or missense mutations all have a similar spectrum of disease severity.8
Several mouse mutants in the Notch pathway have shed light on the role of Jagged1 and Notch signaling during development. Mice homozygous for a null mutation in Jag1 (Jag1dDSL) die on embryonic day (E) 10.5 because of hemorrhage and severe defects in vascular remodeling in the yolk sac and cranial mesenchyme.9 Mice heterozygous for the Jag1 mutation exhibit only minor eye anomalies, with no detectable abnormalities of the liver, heart, and other organs. In contrast, mice with a compound heterozygous genotype for the Jag1dDSL mutation and a hypomorphic mutation in Notch2 (Notch2del1) exhibit many features of AGS, including jaundice, growth failure, bile duct paucity, pulmonary artery hypoplasia, and glomerular abnormalities.10 In this mouse model, bile duct paucity is present at birth, unlike in the human disease, in which paucity evolves postnatally. The phenotype of this double heterozygous mouse model provides evidence of a genetic interaction between Jag1 and Notch2 during development.
Intrahepatic bile duct development begins at about 7-8 weeks' gestation in the human liver and on E14.5 in the mouse with the formation of the ductal plate.11 The hepatoblasts adjacent to the portal vein become strongly reactive for cytokeratin 19 and develop into a continuous layer of cuboidal biliary cells.12 Subsequently, this layer of cells duplicates into a double layer that surrounds the portal tract. During week 12 of gestation, distinct tubular spaces are created through a process of remodeling and apoptosis. These tubules migrate centrally into the portal tracts and are incorporated into the portal mesenchyme. This complicated process is still incompletely understood but involves signaling from several genetic pathways, including the Notch pathway, and interactions with extracellular matrix proteins. Postnatally, liver growth is thought to occur from the central regions to the periphery, with an increase in the number of lobules and branching and elongation of existing bile ducts.13
There is controversy in the literature about the expression pattern of Jag1 in the liver and bile ducts. JAG1 is expressed in blood vessel endothelium in both the portal vein and hepatic artery branches in human embryonic liver.14–16 Similarly, the JAGGED1 protein is in the blood vessels of embryonic and adult human livers.17, 18 However, expression of JAG1 mRNA16 and JAGGED1 protein17, 19 was also documented in the ductal plate in developing human liver. In studies of diseased adult human liver, JAGGED1 protein expression has been reported in bile ducts and in proliferating bile ductules.18 Other reports have suggested that Jagged1 protein expression may be localized primarily in the endothelium in close proximity to Notch2-expressing ductal plate epithelium in mouse liver.20, 21
Due to early embryonic lethality, the Jag1 knockout mouse is not a useful model to investigate the role of Jag1 in bile duct development. Through the use of the Cre/lox conditional gene targeting system, we generated a conditional knockout allele of Jag1 (Jag1loxP) in order to derive a model that would be useful for studying the role of Jag1 and Notch signaling in the development of multiple organ systems. By employing a hepatoblast-specific transgenic Cre line, we ablated Jag1 function in the embryonic liver. We found that mice that did not express Jag1 in the ductal plate were still able to form normal intrahepatic bile ducts at all stages throughout development. However, mice heterozygous at the Jag1 locus for both the null and conditional alleles exhibit increased ductular proliferation, indicating a role for the Notch pathway in postnatal growth and remodeling of the bile ducts.
AGS, Alagille syndrome.
Materials and Methods
Derivation of Jag1 Conditional Knockout Mice.
A BAC clone containing the mouse Jag1 gene was identified by screening a 129/Sv strain BAC library (Research Genetics) using PCR primers specific for the DSL region of Jag1. A 6.5-kb EcoRI/SmaI fragment was subcloned, and a Neo/TK cassette flanked by loxP sites was inserted 5′ to exon 4 by standard techniques. Another loxP site and an EcoRI site were inserted 3′ to exon 5. Finally, a diphtheria toxin cassette was inserted 5′ to the construct to select against random insertion events (Fig. 1A). This targeting construct was electroporated into R1 mouse embryonic stem cells.22 Stably transfected cells were isolated after selection in 350 μg/ml G418 (Gibco), and 288 clones were screened for the desired homologous recombination event. Initial screening was accomplished by PCR. Primers were designed to amplify a segment spanning the region of homologous recombination. PCR products were transferred to a membrane that was probed with a γP32 ATP end-labeled loxP oligonucleotide to identify correctly targeted clones. Candidate clones were screened by Southern blot analysis of genomic DNA digested with EcoRI, using a 455-bp PCR product as an external probe (Fig. 1A,B). The Neo/TK cassette was removed from correctly targeted clones by partial Cre-mediated recombination in vitro. Two clones were expanded and transfected transiently with 2 μg of Cre-expression vector.23 Two days post-transfection, cells were treated with 2 μM gancyclovir to select for the cells where the Neo/TK cassette had been deleted. Ninety-six individual gancyclovir-resistant clones were analyzed by PCR, and approximately 20% of the clones were identified as having undergone partial recombination to generate the Jag1loxP allele. Two of the Jag1loxP ES-cell clones were injected into C57BL/6J mouse blastocysts, which were then transferred to pseudopregnant NMRI females, and chimeric offspring were identified by the presence of agouti hair. Chimeric males were mated to C57BL/6J females to obtain ES-derived offspring that were analyzed by PCR analysis of tail DNA to identify the Jag1loxP/+ heterozygote mice.
Animal Breeding and Genotyping.
As detailed in Table 1, the genetic strains used in these experiments were Jag1loxP, Jag1dDSL,9Notch2del1,24 and the Alfp-Cre transgenic line.25 The Jag1loxP and Alfp-Cre mice were maintained on a C57Bl6/SvEv background. The Jag1dDSL and Notch2del1 mice were maintained on a C57Bl/6J background (backcrossed > 10 generations for Jag1 and > 5 generations for Notch2). Genotyping for all mice was performed by PCR analysis using genomic DNA isolated from the tail tip of weanlings. Genetic crosses were carried out as described in Table 1. All procedures involving mice were conducted in accordance with federal guidelines and approved Institutional Animal Care and Use Committee protocols. All animals received humane care according to the criteria outlined in the “Guidelines for the Care and Use of Laboratory Animals.”
Table 1. Summary of Mouse Strains and Genetic Crosses
Embryos and liver tissue were collected and fixed in 4% paraformaldehyde for 24 hours. Fixed tissues were embedded in paraffin following standard protocols, and 5-μm sections were cut. For initial analysis, sections were stained with hematoxylin and eosin following standard protocols. The extent of hepatic fibrosis was assessed in sections stained with Sirius red.
Paraffin tissue sections were dewaxed and rehydrated. Antigen retrieval was performed by immersion in 0.01 M citric acid at pH 6.0 and heating for 15 minutes. The tissue was then treated with proteinase K for 5 minutes at room temperature. Slides were stained with Polyclonal Rabbit Anti-Bovine Cytokeratin (Code Z0622; DakoCytomation, Carpinteria, CA) at a dilution of 1:100 for 60 minutes. The slides were then washed and incubated with Peroxidase Labelled Polymer (Dako). The tissue was then stained with DAB solution (Dako) and counterstained in 50% hematoxylin. The slides were viewed on a Nikon Eclipse E600 microscope equipped with a digital SPOT camera.
Derivation of Liver-Specific Jag1 Knockout Mice
We used the Cre-loxP conditional gene targeting system to derive a Jag1 conditional knockout allele. This targeting strategy involves flanking Jag1 exons 4 and 5 with loxP sites (Fig. 1A). Exon 4 encodes the DSL region, which is conserved across species and is known to be crucial for ligand-receptor interactions. Deletion of exon 4 (255 bases) and exon 5 (61 bases) is predicted to result in a frameshift mutation and premature truncation of the protein after 192 (of 1,218) amino acids. To target the deletion of Jag1 in the hepatoblasts, we chose the Alfp-Cre transgenic line, which expresses the Cre recombinase under the control of the albumin promoter and the α-fetoprotein enhancer. In Alfp-Cre mice, Cre recombinase expression has been shown to be restricted to hepatoblasts, with complete deletion of the target gene by E15.5.26 We crossed Jag1loxP/loxP mice with mice carrying the Alfp-Cre transgene to obtain Jag1loxP/+;Tg Alfp-Cre animals. These mice were then crossed with Jag1loxP/loxP mice to produce litters with the genotypes Jag1loxP/loxP, Jag1loxP/+, Jag1loxP/+;Tg Alfp-Cre, and Jag1loxP/loxP;Tg Alfp-Cre (see Table 1).
We analyzed liver cDNA from mutant and control livers to confirm that the correct Jag1 deletion had occurred. PCR primers were designed to amplify exons 3 to 6 in wild-type cDNA, resulting in a 675-bp band. In Jag1loxP/loxP liver, only the wild-type band was present (Fig. 1C). As predicted, in Jag1loxP/+;Tg Alfp-Cre and Jag1loxP/loxP;Tg Alfp-Cre liver cDNA, the smaller 359-bp band was also present. Sequencing of this band revealed that indeed Jag1 exons 4 and 5 were deleted in Jag1loxP; Tg Alfp-Cre mouse livers (data not shown). The deleted band was more abundant in the homozygous Jag1loxP/loxP;Tg Alfp-Cre livers. The residual wild-type band in the mutant livers is likely to have been a result of Jag1 expression in vascular endothelium, which would not be targeted by the Alfp-Cre line.
Bile Ducts Develop Normally in Liver-Specific Jag1 Conditional Knockout Mouse Livers.
We examined Jag1 conditional mice (Jag1loxP/loxP;Tg Alfp-Cre) at various stages throughout development including embryonic (E14.5 and E16.5, n = 6), newborn (P0-P7, n = 6), and adult (6-52 weeks, n = 13) development. Age-matched Jag1loxP/loxP and Jag1loxP/+;Tg Alfp-Cre littermates were used as controls. The conditional knockout offspring were produced at the expected Mendelian frequency, and appeared healthy and viable at all stages. The mice were not visibly jaundiced, and their growth and development were normal. At dissection, the gross appearance of the liver was also normal. The liver structure was assessed by observation of hematoxylin and eosin (H&E)–stained sections, and a cytokeratin antibody stain was used to identify bile ducts. Mutant embryonic and neonatal livers showed normal formation of the ductal plate around portal vein branches, and appropriate ductal remodeling (data not shown). The vascular structures were also normal in appearance. Adult animals at all stages examined, from 6 weeks to 1 year of age, showed normal mature bile ducts in portal tracts, with an average of 1 to 2 bile ducts per portal tract. Bile ducts were well formed, surrounded by minimal connective tissue, and incorporated into the center of the portal tracts (Fig. 2). The blood vessels were also unremarkable in the adult animals.
The finding of normal development of bile ducts in Jag1 conditional knockout mice suggests that residual Jag1 expression in the vascular endothelium may be sufficient to allow Notch signaling to occur during bile duct development. We hypothesized that crossing the Jag1 conditional knockout mice with other Notch-pathway mutant mice (see Table 1) would alter Notch signaling. Thus, we made additional genetic crosses to further decrease Notch pathway gene expression, and then we examined the effects of these crosses on the liver phenotype.
Normal Bile Duct Development in Jag1 Conditional/Notch2 Mouse Livers.
Mice heterozygous for a null mutation in Jag1 and a hypomorphic mutation in Notch2 had a paucity of bile ducts and other features of Alagille syndrome.10 Thus, we hypothesized that crossing the liver-specific conditional knockout mice with those carrying the Notch2del1 mutation would result in progeny with altered liver development. We performed a 2-step genetic cross as detailed in Table 1. We collected mouse livers from Jag1 conditional/Notch2 (Jag1loxP/loxP;Tg Alfp-Cre;Notch2del1/+) adult mice ranging in age from 6 to 10 weeks (n = 8). Liver samples were stained with H&E for general tissue morphology and with cytokeratin antibody to identify bile ducts. At all stages examined, portal tracts appeared normal in both mutant and control livers, with appropriate ratios of portal veins, arteries, and bile ducts (data not shown).
Jag1loxP/dDSL;Tg Alfp-Cre Mouse Livers Demonstrate Bile Duct Proliferation.
In light of the normal development of bile ducts in Jag1 conditional Notch2del1 livers, it is likely that residual Jag1 expression in the endothelium is sufficient for Notch signaling during bile duct development. Therefore, we decided to further decrease Jag1 expression by crossing the Jag1 conditional and the null mice in order to generate the Jag1loxP/dDSL;Tg Alfp-Cre (Jag1 conditional/null) allele. These mice are heterozygous for the Jag1 null allele in every cell, and the remaining Jag1 allele has been ablated in hepatoblasts and their derivatives. We examined Jag1 conditional/null livers of adult mice 8 through 16 weeks of age (n = 7). Controls included littermates with the genotypes Jag1loxP/dDSL, Jag1loxP/+, and Jag1loxP/+;Tg Alfp-Cre (n = 8). We also included several age-matched Jag1dDSL/+ mice from other crosses as additional controls (n = 3). The Jag1 conditional/null mice were viable and healthy, with normal growth and development. Their livers and portal venous system appeared normal on gross examination. However, microscopic examination showed striking biliary abnormalities in 3 of the 7 mutant livers. Two mutant livers examined at 8 weeks of age revealed different phenotypes. With H&E staining, 1 Jag1 conditional/null liver showed an increased number of dilated bile ducts surrounding expanded portal tracts (Fig. 3A), whereas the liver of a littermate with the same genotype illustrated normal histology (Fig. 3B). At both 8 and 16 weeks of age, cytokeratin staining showed irregular, dilated ducts surrounding the periphery of many portal tracts (Fig. 3C,D,H). Irregular ducts extended into the liver lobule in some areas (Fig. 3C,D, arrow). In control livers, bile ducts appeared mature and fully remodeled (data not shown). At 12 weeks of age, bile ducts proliferating along the periphery of many portal tracts could be identified in mutant livers (Fig. 3E,F), unlike the normal bile ducts in the control livers. On higher magnification the ducts appeared small and slitlike (Fig. 3F), unlike the large irregular, dilated ducts seen in a liver sample from an 8-week-old mouse. On H&E staining, bile stasis was identified within the lumina of several proliferating ducts (Fig. 3G). However, there was no evidence of occlusion of the common bile duct (data not shown). Staining for collagen with Sirius red demonstrated slight expansion of the portal tracts and distortion of the normal architecture in the mutant liver compared with those in the control (Fig. 4A,B), but did not show evidence of established fibrosis or cirrhosis.
Because the proliferating ducts at the periphery of the portal tracts (Fig. 3A,C) resembled the abnormally formed ducts in ductal plate malformation found in several human liver diseases, we examined liver samples from newborn and 4-week-old mice to ascertain whether the phenotype was apparent early in life. Liver samples from newborn Jag1 conditional/null mice (n = 4) and littermate controls (n = 3) were indistinguishable by both H&E and cytokeratin staining (data not shown). Both the mutant and control livers showed multiple bile duct structures surrounding each portal tract, with relatively incomplete remodeling, which is typical of mouse livers in newborn mice. However, the liver sample from a 4-week-old Jag1 conditional/null mouse showed diffuse bile duct proliferation around both large and small portal tracts (Fig. 5C -F). The control liver showed normal histology (Fig. 5A,B).
In this study, we investigated the role of Jag1 during the development, growth, and maintenance of bile ducts. Because knockout of Jag1 is lethal to mouse embryos, the homozygous Jag1 null mouse model was inappropriate for studies of liver and bile duct development. We therefore derived a Jag1 conditional knockout allele, and targeted the ablation of Jag1 function to the developing liver by using the Alfp-Cre transgenic line. Our hypothesis was that if Jag1 expression were deleted in hepatoblasts, the bile ducts would not develop normally and bile duct paucity would result. In fact, the Jag1 conditional knockout mice displayed a normal liver phenotype. These findings demonstrate that Jag1 expression is not absolutely required in hepatoblasts or in the ductal plate for bile ducts to develop normally. It is well established in the literature that Jag1 is expressed in endothelial cells, so it is certainly possible that this expression could compensate for the lack of Jag1 in hepatoblasts. A further reduction of Notch signaling by crossing with the Notch2 hypomorphic allele to generate Jag1 conditional Notch2 mice did not result in any hepatic or biliary abnormalities. Only when we decreased Jag1 expression by crossing mice carrying the conditional and null alleles in order to generate Jag1 conditional null mice did we observe a mutant phenotype of increased proliferation of bile ducts in the adult mice, with a penetrance of 50%. Although unexpected, this phenotype of bile duct proliferation sheds new light on the role of Jag1 in the postnatal growth and branching morphogenesis of bile ducts.
Many patients with Alagille syndrome have normal bile duct ratios or bile duct proliferation at birth, with bile duct paucity evolving over the first 6 months of life.6 The hypothesis that the Jag1 defect in Alagille syndrome leads to impaired postnatal bile duct growth and branching morphogenesis is supported by the finding in at least 1 patient of normal bile ducts near the liver hilum, with paucity in the peripheral portal tracts.27 In our genetic model, the mice develop bile ducts postnatally and do not progress to having a paucity of bile ducts over the first 4 months of life, even though the livers have grown to adult size. This difference highlights that this developmental process is not identical between mice and humans because of multiple factors including genetic background and the shorter maturation period of the mouse.
In addition to its role in early development and determination of cell fate, the Notch signaling pathway is known to be involved in the control of cell growth and proliferation. Inducible inactivation of Notch1 in mouse liver causes nodular regenerative hyperplasia.28 The striking cell proliferation observed in these mice, demonstrated by BrdU uptake, is restricted to hepatocytes, and bile ducts are reported to be normal. Similarly, inhibition of Notch signaling in the prostate results in increased proliferation of the duct epithelial cells.29 These models suggest that the phenotype of bile duct proliferation observed in the Jag1 conditional/null mice might be a result of decreased overall Notch signaling. In crossing the Jag1 conditional and null alleles, Jag1 signaling may have been reduced below a crucial threshold. An alternative hypothesis is that these mice lack Jag1 in all liver parenchymal cells and are haploinsufficient for Jag1 function in all other cells. Jag1 expression may be required in a specific cell type at a specific time in order to signal to a particular Notch receptor. Prior studies have shown that Jag1 is expressed in liver parenchymal cells as well as in other cell types such as blood vessels and mesenchymal cells (see Supplementary Table 1 for a summary of known expression data for Notch pathway genes in the liver (supplementary material for this article can be found at http://interscience.wiley.com/jpages/0270-9139/suppmat/index.html). It has also been established that all 4 Notch receptors are expressed in developing and adult livers in a number of cell types that may be adjacent to Jag1-expressing cells. Further genetic and in vitro studies will be necessary to elucidate the specific ligand-receptor interactions that are disrupted in this mouse model.