The Notch signaling pathway was initially discovered in Drosophila and Caenorhabditis elegans and was so named because Drosophila carrying a mutant copy of the Notch gene have notched wings. Homology cloning in mammals subsequently identified four Notch receptors (Notch-1, Notch-2, Notch-3, and Notch-4) and five ligands (Jagged-1, Jagged-2, Delta-like-1, Delta-like-3, and Delta-like-4). In humans, mutations in Notch receptors or ligands are associated with several diseases. Most relevant to the field of hepatology is Alagille syndrome (AGS), a pleiotropic developmental disorder characterized by neonatal cholestasis in association with cardiac, skeletal, and ophthalmologic anomalies, and less frequently with renal or vascular deficiencies. AGS exhibits autosomal dominant inheritance and is caused by mutations in JAG11, 2 (Online Mendelian Inheritance in Man #118450).
Neonatal cholestasis in AGS was found to result from bile duct paucity, but the link between Jagged-1 function and the biliary anomaly remained obscure, prompting developmental geneticists to generate mouse models,3 soon followed by zebrafish models,4 deficient in Jag1 expression. JAG1 haploinsufficiency is suggested to cause AGS in humans and it came as a surprise that mice with a genotype mimicking that of patients with AGS did not recapitulate the spectrum of anomalies found in AGS.3 Also, conditional inactivation of Jag1 in hepatic epithelial cells did not induce bile duct paucity.5 However, based on observations that Jagged-1 can bind to Notch-2, Thomas Gridley's team generated mouse mutants which were doubly heterozygous for a Jag1 null allele and a Notch2 hypomorphic allele.6 These mice displayed a phenotype reproducing most AGS features including the bile duct paucity: few bile ducts were present and a small number of biliary cells unable to form ducts were located adjacent to the portal vein. Moreover, the same biliary defects were observed in mice that were homozygous for the Notch2 hypomorphic allele, further suggesting that NOTCH2, like JAG1, is involved in human AGS. This hypothesis was later validated by McDaniel and coworkers, who found NOTCH2 mutations in families with AGS7 (Online Mendelian Inheritance in Man #610205).
This work on Jag1-deficient and Notch2-deficient mice did not solve the issue as to how perturbed Notch signaling leads to bile duct paucity. Indeed, normal bile duct development in the embryo starts with the differentiation of hepatoblasts (liver progenitor cells) into biliary cells (Fig. 1; reviewed in Lemaigre8). The latter form a single cell layer around the branches of the portal vein, called the ductal plate. Then, parts of the ductal plate become bilayered and form tubular structures that give rise to mature ducts, while the remaining nontubular ductal plate is eliminated. This complex process raises questions on the role of the Notch pathway. Does deficient Notch signaling induce bile duct paucity as a result of abnormal cell fate specification of biliary cells, or is bile duct paucity caused by the inability of ductal plate cells to form ducts ? How does Jagged-1 activate Notch signaling during bile duct development ? In this issue of HEPATOLOGY, Geisler and coworkers add new pieces to the puzzle.9 They analyzed biliary development in mice with combined or single targeted disruption of Notch1 and Notch2. They propose that Notch2 is most likely dispensable for biliary cell specification, but that it is required for bile duct formation. A similar conclusion was reached independently by Gridley's team.10
Several hurdles had to be overcome to address the function of Notch-1 and Notch-2. Constitutive Notch1 and Notch2 knockouts display embryonic lethality at a stage preceding that of bile duct development. Therefore, Geisler and coworkers generated conditional mutants with liver-specific inactivation of Notch receptor genes. They mated mice expressing Cre recombinase under control of Albumin gene regulatory sequences with mice harboring loxP-flanked Notch alleles. Combined inactivation of Notch1 and Notch2, or inactivation of Notch2 only, resulted in failure to form normal bile ducts despite the presence of ductal plate cells. Progressively, the biliary system became severely disorganized and appeared as irregular tubular structures, leading to cholestasis. Because this phenotype was rescued in the presence of one wild-type Notch2 allele (Notch1F/F; Notch2F/wt; Alb-Cre) or was absent when only Notch1 is inactivated, Geisler and coworkers concluded that Notch2, but not Notch1, is a key player during bile duct formation.
Beyond the clear demonstration that Notch2 is required for normal biliary morphogenesis, what can be deduced from these data about the mode of action of the Notch signaling pathway? Geisler and coworkers cautiously discuss this issue. The presence of biliary cells in Notch2-deficient newborns suggests that Notch2 is dispensable for biliary lineage commitment from bipotent hepatoblasts, and that the role of Notch2 is at the level of duct formation from the ductal plate. However, as Geisler and coworkers point out, their experiments cannot eliminate the possibility that partial inactivation of Notch2 at early stages of liver development allows residual Notch-2 to drive biliary lineage commitment. The recent work by Lozier and coworkers supports the notion that Jagged-1/Notch-2 signaling is required for morphogenesis rather than for biliary cell specification.6 Indeed, the doubly heterozygous Jag1/Notch2 mice studied earlier by the same group were reinvestigated, and the analysis revealed normal biliary lineage determination, but failure to form patent ducts in the early postnatal period. Again in line with this hypothesis are earlier data from Kodama and coworkers11 who analyzed mice deficient in Hes-1 (hairy and enhancer of split 1) a Notch signaling effector. These mice had apparently normal ductal plates but did not develop tubular structures. Taken together, the data suggest a model whereby Jagged-1 activates a Notch-2/Hes-1 cascade in the ductal plate and so induces duct formation. However, because none of the mouse models analyzed so far proved to have fully inactivated Notch signaling at the time of biliary lineage determination, one cannot firmly conclude on the role of Notch in differentiation. Importantly, in vitro gain-of-function experiments showed that Notch can induce the expression of biliary markers and repress hepatocyte differention.12 Such experiments await in vivo validation.
Considering the role of Notch-2 in duct formation, one may ask how it is activated by Jagged-1. The first concern is to determine which cell types express the ligand and the receptor, an issue which generated much controversy (see summary table in Loomes et al.5). Geisler and coworkers9 addressed this problem by a genetic approach using mouse strains expressing green fluorescent protein or beta-galactosidase under control of Notch1 and Notch2 regulatory sequences (Notch1–green fluorescent protein transgene and LacZ knock-in in Notch2 gene). The results suggest that Notch-2, but not Notch-1, is expressed in developing and mature ducts. This is coherent with earlier immunostainings11 and with a role for Notch-2 in activation of Hes-1, whose expression is restricted to emerging tubular structures in the ductal plate. The expression pattern of Jagged-1 is not well established either, because Jagged-1 was found in periportal mesenchyme, biliary cells, portal endothelium, and hepatic arteries, depending on the detection tool used and developmental stage studied. Determining Jagged and Notch expression patterns in the liver has now become critically important to understand how a single-layered ductal plate gives rise to mature ducts.
Finally, an additional layer of complexity is revealed when comparing the postnatal phenotypes of mice with liver-specific inactivation of Notch2 described in the articles by Geisler et al.9 and Lozier et al.10 Both groups used Albumin-Cre–mediated recombination of loxP-flanked Notch2 alleles. However, although the first group observed high numbers of irregularly organized duct-like structures near the portal tract in the postnatal period, the second group only found a few biliary cells. It is not clear whether the high numbers of biliary cells observed by Geisler and coworkers reflect cell-autonomous consequences of altered Notch signaling or whether the numerous cells should be assimilated to reactive cells. During normal bile duct development, the nontubular parts of the ductal plate involute according to mechanisms that are not fully characterized. The data from Geisler et al. leave open the possibility that Notch-2 signaling may control survival or involution of ductal plate cells. Still, the phenotypic difference between the two studies raises new questions on the mechanisms of Notch-regulated bile duct delopment. Are there stage-dependent Notch signaling effects, what controls Notch signaling intensity, and how is Notch signaling integrated with other pathways such as those initiated by, for example, transforming growth factor-beta,13 Wnt/beta-catenin,14, 15 fibroblast growth factors,16 or bone morphogenic proteins16, 17? The variable expressivity of AGS may depend on subtle interindividual variations in the control exerted on and by the Jagged-1/Notch-2 cascade. Understanding the fine-tuning of this cascade should shed light on the still obscure mechanisms underlying AGS. Obviously, exciting challenges are ahead of us.