| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Departmento de Inmunología y Oncología, Centro Nacional de Biotecnología/Consejo Superior de Investigaciones Científicas, E-28049 Madrid, Spain
Correspondence: Address requests for reprints to: José Luis de la Pompa, Departmento de Inmunología y Oncología, Centro Nacional de Biotecnología/Consejo Superior de Investigaciones Científicas, Darwin 3, Campus de Cantoblanco, E-28049 Madrid, Spain. E-mail: jlpompa{at}cnb.uam.es
| Abstract |
|---|
|
|
|---|
| I. Introduction: Elements of the Notch Signaling Pathway |
|---|
|
|
|---|
Notch is a local signaling mechanism that is evolutionarily conserved throughout the animal kingdom. Mammals have four Notch proteins (Notch 14; Refs. 6, 7, 8, 9, 10 and Fig. 1A
) that are membrane-bound type I receptors (with a single-pass transmembrane domain), harboring a large extracellular domain involved in ligand binding, and a cytoplasmatic domain involved in signal transduction. The extracellular domain contains a variable number of epidermal growth factor (EGF)-like repeats that are critical for binding interactions (11, 12). The EGF-like repeats are followed by three cysteine-rich LIN12/Notch repeats (LNR) that prevent signaling in the absence of the ligand. The Notch intracellular domain (NICD) contains a RAM23 domain (13), six ankyrin/cdc10 repeats involved in protein-protein interactions (14), two nuclear localization signals (N1 and N2), a transcriptional activation domain (TAD) that differs among the four receptors, and a PEST sequence [rich in proline (P), glutamic acid (E), serine (S) and threonine (T)] that negatively regulates protein stability (15).
|
Notch signaling is regulated by posttranslational modification events, such as glycosylation, and by other modifications involving the extracellular domains of both receptors and ligands, such as the extension of sugar residues by the glycosyl transferase fringe (23, 24) (Fig. 1B
). This prevents Notch activation by Jagged but not by Delta ligands (Refs. 25 and 26 ; reviewed in Ref. 27). Notch signaling initiates through ligand-receptor interactions between neighboring cells, leading to two consecutive proteolytic cleavages of the receptor, which ultimately liberate NICD (Fig. 1
, C and D). Thus, after ligand binding, the ubiquitin ligases mind bomb (28, 29) or neuralized (30, 31, 32) interact with the ligand intracellular domain to promote its ubiquitination and internalization (28) (Fig. 1C
). Ligand endocytosis leads to a conformational change in Notch that allows ADAM (metalloprotease and disintegrin) protease TACE (TNF-
-converting-enzyme), to cleave the receptor at a second site (S2) on the extracellular side, near the transmembrane domain (33) (Fig. 1C
); the released extracellular portion of the receptor is then transendocytosed to the ligand-expressing cell (34) (Fig. 1C
). The third cleavage (S3) occurs within the transmembrane domain and is mediated by a
-secretase activity whose key components are presenilin and nicastrin (35) (Fig. 1D
). This final cleavage liberates NICD, which subsequently translocates to the nucleus where it binds via its RAM23 domain to the transcription factor CSL (CBF1 in humans, Suppressor of Hairless in Drosophila, LAG in Caenorhabditis elegans), also called RBPJK in mice. In the absence of Notch activity, CSL proteins bind to promoters of its target genes and recruit histone deacetylases (36) and corepressors (CoR; Fig. 1E
) that inhibit transcription. The corepressor molecules include SMRT/NcoR (37) and SHARP/MINT/SPEN (38). The NICD/CSL interaction converts CSL from a transcriptional repressor into a transcriptional activator by displacing the corepressor complex and recruiting coactivators (Fig. 1E
) such as Mastermind-Like 1 (MAM; Fig. 1E
) (39) and histone acetyltransferase (40). A number of additional proteins modulate Notch signaling, including the RING-domain E3 ubiquitin ligase deltex (41, 42) and the phosphotyrosine-binding domain (PTB)-containing proteins numb and numb-like (43), which act as context-dependent negative or positive Notch regulators (Fig. 1D
).
To date, only a few target genes have been identified; some are Notch-dependent in various tissues, whereas others are tissue-specific. The best-known Notch target genes are members of the basic helix-loop-helix (bHLH) hairy/enhancer of split (Hes) family, the related HRT/Herp (Hes-related repressor protein) transcription factor family (44), the cell cycle regulator p21 (45), the Notch pathway element Notch-regulated ankyrin repeat protein (Nrarp) (46), deltex1, and the pre-T cell receptor-
gene (47). For an in-depth study of the molecular intricacies of Notch signaling elements see Refs. 48, 49, 50, 51 .
Here we will examine the role of Notch in specific developmental processes such as central nervous system (CNS) development, somitogenesis, cardiovascular development, and endocrine development, with attention to the distinct mechanisms by which Notch regulates cell specification and patterning in these tissues. We will then analyze the role of Notch in cancer, both in leukemia and in solid tumors, and describe studies that suggest possible means of therapeutic intervention. As a rule, when we refer to humans, the elements of the pathway are named in capital letters (i.e., NOTCH1 receptor or NOTCH1 gene), and when we refer to experimental models, the elements of the pathway are named in lowercase (Notch1 protein or Notch1 gene).
| II. Notch in Vertebrate CNS Development |
|---|
|
|
|---|
During the past 10 yr, the role of Notch in the differentiation, morphogenesis, and function of the CNS has become increasingly valued. The phenotypic analyses of Notch-targeted mutants in mice and functional manipulation in other vertebrates have greatly benefited from knowledge generated by research in Drosophila neurogenesis (52). Poulson (2) was the first to associate lack of Notch function with an embryonic lethal phenotype in Drosophila; it is caused by failure of the early neurogenic ectoderm to segregate neural and epidermal cell lineages. In homozygous Notch mutant embryos, all cells become neuroblasts, which leads to hypertrophy of the neural tissue at the expense of the epidermis, giving rise to the so-called neurogenic phenotype (2). In vertebrates, Notch is required when the epidermal and neural lineages have already segregated; its inactivation results in a "neurogenic phenotype" represented by premature differentiation of neuronal progenitors, leading to the interpretation that Notch maintains a progenitor state and inhibits differentiation.
A. Notch promotes progenitor diversification and inhibits neuronal differentiation
Notch1 was the first Notch pathway gene to be disrupted by homologous recombination (53, 54). Mutant embryos die at midgestation [embryonic day 11 (E11)] with defective somitogenesis and placentation, although little attention was given to a possible neural phenotype. Subsequently, a detailed analysis of neural development in Notch1 targeted mutants was reported (55). This study examined the expression of pathway components such as Hes1, Hes5, and Delta1 and of early differentiation markers such as NeuroD (or Neurod2), Math4A (or Neurog2), and NSCL-1 (or Nhlh1). Consistent with the view that Notch activity is needed for progenitor maintenance, expression of these markers was increased in mutants. In addition, Hes5 expression was reduced in Notch1 mutants, although Hes1 expression appeared to be unaffected. This result was unexpected in light of the Hes1 mutant phenotype (56) and the extensive literature supporting the idea that Hes1 is a primary Notch/CSL target (reviewed in Ref. 57). Similar results in RBPJk targeted mutants (55) suggested that whereas Hes1 may well be a true Notch target, it is also likely to be activated by other signaling pathways. After the original Notch1 deletion studies, targeted alleles of Notch2 (58, 59), Notch3 (60), and Notch4 (61) were also generated, as well as conditional (floxed) alleles of Notch1 (62). Although Notch3 and Notch4 do not have detectable phenotypes when deleted, Notch2 mutants, similarly to Notch1 mutants, die around E11 (58). In contrast to Notch1 mutants, however, Notch2 mutants do not show alterations in Hes5 expression in the CNS. Notch2 mutants undergo widespread cell death in the CNS starting around E9 (58), but it is not clear whether this phenotype reflects a role for Notch2 in the developing CNS or whether it occurs as a consequence of other embryonic perturbations.
To circumvent the early lethality of Notch1 deletion, several studies have addressed the effect of deleting this receptor in specific brain structures. In one case, conditional Cre-loxP-mediated recombination was used to delete Notch1 from the medial cerebellar primordium (63). Consistent with the traditional model of Notch function in the nervous system, the authors found that Notch1 deletion resulted in up-regulation of the proneural genes Mash1 and Math1 and precocious neuronal differentiation. More recently, conditional deletion of Notch1 in the neural progenitor pool using a nestin-Cre promoter also resulted in precocious neuronal differentiation (64). Deletion of Notch1 in the telencephalon, using the foxg1-Cre line, led to reduced neuron numbers later in development, most likely resulting from precocious neuron differentiation and earlier progenitor pool depletion (65). In support of this finding, the telencephalic deletion of Notch1 led to a reduction in progenitor frequency (assayed as neurospheres) in vitro (66). This result is consistent with the reduced neurosphere frequencies observed after standard deletion of Notch1, RBPJk, PS1 and PS2 (67), or Hes1 and Hes5 (68). In summary, the conditional deletions of Notch1 support the view that Notch signaling inhibits neuron differentiation and maintains the neural progenitor pool.
In addition to the receptor mutations, many Notch ligand mutations have been examined in mice. A few studies have analyzed the effect of deleting Delta1 on neural development (66, 69). One of these studies found that Delta1 mutants had decreased Hes5 expression, consistent with the predicted reduction in Notch activation (66). This finding was previously documented in a comparative study of the role of Notch in somitogenesis (70). In addition, the study by Yun et al. (66) found that Delta1 targeted mutants showed a decrease in radial progenitor markers and an increase in neuronal markers. These findings support the view that Notch signaling inhibits neuron differentiation in the developing CNS. Based on comparisons of the Delta1, Mash1, and other mutants, however, the authors suggest that Notch signaling might also regulate the diversification of the progenitor pool into distinct progenitor subtypes. This function would precede the role of Notch in inhibiting the differentiation of mature cell types (neurons and oligodendrocytes) and could convert a homogeneous proliferative pool into a heterogeneous mixture of stem cells, neuroblasts, and glioblasts. Further evidence that Notch signaling may generate progenitor diversity was obtained by in vitro analysis of Delta1 targeted mutants (69). This report suggested that Notch signaling first specifies glial progenitors and then functions in those cells to promote astrocyte vs. oligodendrocyte fate. Both this study and the work described above indicate that in mice, Notch influences multiple choice points in the neural progenitor lineage.
As discussed previously, the Notch signaling cascade is transduced primarily through the transcriptional regulator CSL (RBPJK in mice), when nuclear translocation of NICD converts CSL from a repressor to an activator. Consistent with the Notch1 mutant phenotype, RBPJk targeted mutants show altered gene expression, suggestive of widespread precocious neuron differentiation (such as decreased Hes5 and increased Delta1 and NeuroD) (55). As in the case of the other Notch pathway genes, conditional deletion of RBPJk in the CNS is likely to be highly informative.
The most widely accepted Notch/CSL targets are the Hes (71) and the HRT gene families (72). Although there are seven Hes genes, not all are clear Notch targets, and studies in the mammalian CNS have focused on Hes1 and Hes5. Hes1 mutant embryos show severe defects in neural development, including lack of cranial neural tube closure and anencephaly (56). Because these animals die perinatally, it is possible to examine alterations in gene expression at late developmental stages. Consistent with the canonical model, precocious neurogenesis in Hes1 mutants was suggested by early expression of Mash1, NSCL1, and neurofilament markers. Studies with double mutant combinations reveal that Hes1 and Hes5 have redundant functions in the CNS, regulating the differentiation of the neural progenitor pool. Hes1; Hes5 double mutants thus have a more severe phenotype than single Hes1 and Hes5 mutant phenotypes combined (68). Hes1 mutants also show increased Hes5 expression, suggesting the existence of a compensatory mechanism between these Notch targets (56). Lastly, although Hes5 mutants showed some precocious neuronal differentiation, these animals were largely normal, suggesting that Hes1 is able to compensate for lack of Hes5 almost completely.
Inactivation of recently identified Notch signaling elements also leads to altered neurogenesis. Targeted mutagenesis of mind bomb1 causes a phenotype relatively similar to that of Notch1 or RBPJk mutants, which result in embryonic lethality at E10.5. There is also a strong neurogenic phenotype in the CNS, with premature neurons undergoing apoptosis soon after differentiation. Aberrant neurogenesis is a direct consequence of lowered Hes1 and Hes5 expression resulting from the inability to generate N1ICD (73).
B. Notch in gliogenesis
In contrast to its "permissive" role in neuronal differentiation, Notch appears to have an instructive role in gliogenesis, directly promoting the differentiation of many glial subtypes. The results of in vivo studies involving mouse (74, 75), zebrafish (76), chick (77, 78), and Xenopus (79) are consistent with an instructive role for Notch signals in gliogenesis, through their conventional bHLH targets. Activation of Notch signaling favors the generation of Müller glia cells at the expense of neurons, whereas reduced Notch signaling induces production of ganglion cells, causing a reduction in the number of Müller glia.
In vertebrates (55, 80) as in Drosophila (81) early neurogenesis, Notch signaling operates among cells belonging to an equivalence groupbecause they express the same set of molecules and are functionally equivalentand controls their commitment to differentiate via a mechanism termed lateral inhibition (82). How a given cell within an equivalence group adopts the neuronal fate or not depends on the Notch ligand expression level. If a progenitor cell expresses more Delta ligand than its neighbors, it will become a neuron and will transmit an inhibitory signal to the Notch-expressing progenitors in contact with it, preventing these progenitors from differentiating prematurely into neurons and from expressing Delta. As a consequence, the cell that produces more ligand forces its neighbors to produce less; ultimately, individual neighboring cells are driven into different developmental pathways (83). Molecular genetic studies show that in this situation, ligand production is regulated by a negative feedback loop (84). This interpretation of the gene expression patterns has been documented extensively by experiments in Xenopus (79, 80), chick (78, 85), zebrafish (86, 87, 88), and mouse (55). Another example of lateral inhibition is the formation of sensory hair cells in the vertebrate inner ear (89), where a single cell within an equivalence domain expresses high Delta levels.
More detailed analyses of Notch function in the CNS has revealed that Notch is likely to regulate progenitor pool diversification and neuronal maturation (90). Emerging data also suggest that Notch signaling has a role in neuronal function in the adult brain (91). It will be of great interest to determine which components of the Notch signaling cascade function in each of these processes. Figure 2
summarizes the roles of Notch in embryonic and adult CNS.
|
| III. Notch in Somitogenesis |
|---|
|
|
|---|
|
Oscillatory expression of other bHLH repressor genes in the PSM was later observed in other vertebrate embryos, although the gene with dynamic expression is not always the same. Hes7 (105) and Her1/7 (106, 107) are thus rhythmically expressed in the mouse and zebrafish embryo, respectively. In Xenopus, homologs of the oscillating zebrafish Her1 gene are expressed in the PSM, but their expression does not oscillate (101). The emerging picture is that Notch signaling undergoes a dramatic spatial and temporal change within the PSM during each segmental cycle. Segmentation appears to rely on two major components; an oscillator, the segmentation clock, sets the periodicity of somite formation, and a dynamic wave-front defines the level at which PSM cells respond to the clock, providing a mechanism that spaces the segment boundaries (108).
How does the segmentation clock work? A negative feedback loop has been proposed as the major mechanism. A study by Hirata et al. (109) showed that the segmentation clock can be mimicked in cultured cells. After serum stimulation, expression of both Hes1 mRNA and protein oscillates with a 2-h cycle. This coincides with the cycle of the segmentation clock in mouse (Fig. 3E
), suggesting that the oscillators in cultured cells and the PSM share the same mechanism. Hes1 protein oscillation is delayed by approximately 15 min relative to Hes1 mRNA oscillation. Both Hes1 mRNA and Hes1 protein have very short half-lives, which enable 2-h cycle oscillatory gene expression. Hes1 protein is degraded by the ubiquitin-proteasome pathway (109). When stabilized by proteasome inhibitors, Hes1 protein constitutively represses its own transcription by binding directly to the Hes1 promoter. Conversely, in the absence of functional Hes1 protein, Hes1 transcription is constitutively up-regulated. The negative feedback loop in which Hes1 protein periodically represses its own transcription is thus the central mechanism for Hes1 oscillation (Fig. 3F
). It is likely that Hes1 protein represses Hes1 mRNA synthesis, but is rapidly degraded, allowing the next cycle of mRNA synthesis (109). Because Hes1 transcription is constitutively up-regulated in the absence of functional Hes1 protein, there appears to be a steady level of transcriptional activators in cultured cells after serum stimulation.
Because a simple negative feedback loop would be insufficient to maintain stable oscillation, another cycling factor might be involved. Such a cycling factor is not yet known for cultured cells, but for the segmentation clock, lunatic fringe (Lfng) in mouse (Fig. 3F
) and chick and deltaC in zebrafish might be implicated in maintaining stable oscillation in the PSM. Dale et al. (110) found that Lfng also establishes a negative feedback loop in chick, and that, in addition to Lfng mRNA, Lfng protein exhibits oscillatory expression in the PSM. Activation of Notch signaling induces Lfng transcription, but Lfng protein inhibits Notch signaling and thereby represses its own transcription. Lfng thus also periodically represses its own expression, like Hes1 and Hes7. Analysis of the regulatory region of Lfng in mouse confirmed that it includes a site that mediates direct stimulation by activated Notch (111) but showed that it contains sites that mediate inhibition by bHLH proteins such as the Hes family members (112). This led to a model in which the inhibitory action of Hes genes plays a key role, and Lfng is assumed to potentiate Delta-Notch signaling (Fig. 3F
), as in Drosophila (25). In contrast to this proposition, Morimoto et al. (113) have shown that Notch1 activity in mice oscillates in the posterior PSM. Somite boundaries would form at the interface between Notch1-activated and Notch1-repressed domains. This interface would be generated by suppression of Notch activity by the bHLH transcription factor Mesp2 through induction of Lfng that, in contrast with its role in the wing margin (25), would inhibit Notch activation. It thus becomes clear that the function of Lfng could vary, depending on cell context and biological system. Another important finding is that Wnt signaling is also involved in the segmentation clock (114). Lfng oscillation is lost in vestigial tail mice (hypomorphic Wnt3a mutants), indicating that the Wnt pathway interacts with the Notch pathway, although the precise mechanism remains to be determined. Although great advances have been made in the past decade in understanding aspects of the somitogenesis process, many aspects, such as the molecular machinery that underlies the segmentation clock, remain to be explained. For a more detailed review on the molecular aspects of somitogenesis see Refs. 108 , 115 , and 116 .
| IV. Notch in Cardiovascular Development and Homeostasis |
|---|
|
|
|---|
Tissue interactions between the myocardium and endocardium in the atrioventricular canal and outflow tract regions lead to epithelial-mesenchyme transition (EMT) of endocardial cells, to participate in cardiac valves and membranous septa formation (119). Cushion formation is followed by the development of more compact myocardium in the periphery of the developing cardiac ventricles.
Vascular development initiates with the differentiation of endothelial precursors, or angioblasts, into endothelial cells (120). These cells assemble to form a primitive vascular plexus of uniformly sized vessels composed entirely of endothelial cells by the process of vasculogenesis. This primitive vascular plexus is then remodeled to form the veins, arteries, and capillaries through the process of angiogenesis.
Different Notch ligands (20, 61) and receptors (6, 10, 121, 122), as well as their downstream effectors and target genes (72), are expressed in the vascular system. Functional studies in zebrafish (123) and mice (61) showed that Notch is a critical regulator of cardiovascular development. Evidence for a crucial function of Notch in vascular development and homeostasis is the finding that the human disease CADASIL (cerebral autosomal dominant arteriopathy with subcortical infarction and leukoencephalopathy), which involves the NOTCH3 gene, causes stroke and vascular dementia (124). In addition, JAGGED1 haploinsufficiency leads to the dominant inherited Alagille syndrome, which among other features is characterized by vascular anomalies (125, 126).
Analysis of mice with targeted mutations in Notch1 or RBPJk revealed that these mutants have severe pericardial distension, indicative of a circulatory defect (53, 54, 98). Likewise, mutations of the Delta1 (99) and Jag1 (127) genes lead to vascular abnormalities and hemorrhage, indicative of an underlying vascular defect. Analysis of the Notch1; Notch4 double mutants provided an insight into Notch function in the vasculature. Whereas Notch4 mutants are viable and fertile, the Notch1; Notch4 double mutants show normal vasculogenesis, but the process of angiogenesis is affected in the embryo proper, the yolk sac, and the placenta (61). This phenotype is more severe than that of Notch1 mutants alone, suggesting that Notch1 and Notch4 have partially redundant roles during embryonic vascular development. From this study, it became clear that Delta4 was the key ligand for Notch1 and Notch4 receptors in the vasculature. In fact, Delta4-targeted mutants and endothelial-specific RBPJk-targeted mutants show a loss of arterial identity and arteriovenous malformations (128, 129), indicating that Notch signaling is required for arterial specification and patterning during development. These studies also suggested a relationship between Notch and EphrinB2/EphB4, another local cell signaling pathway involved in arterial-venous specification (130).
Targeted mutagenesis of potential Notch target genes in the cardiovascular system has helped to delineate the Notch signaling pathway that leads to arterial specification. HRT1; HRT2 double mutants die after E9.5 and display severe angiogenic remodeling defects and massive hemorrhage, probably due to impaired arterial fate determination and maturation. This phenotype is very similar to that of zebrafish gridlock (grl) mutants, that show impaired aorta maturation (131). Grl encodes for the zebrafish homolog of mammalian HRT2. These report identified HRT/CHF1/Hesr/Herp genes as essential Notch effectors in vascular development. The view of Notch as a key regulator of arterial fate is consistent with data indicating that ectopic activation of Notch signaling leads to repression of venous fate (132) and to severe vascular patterning defects (133). More recently, our understanding of how arterial-venous specification occurs has been refined by the study of an endothelial-specific mutant of the COUP-TFII transcription factor. Suppression of Notch signaling activity by COUP-TFII regulates vein identity, which may thus be genetically controlled and not derived by a default pathway (134, 135) (Fig. 4
).
|
In RBPJk-targeted mutants, early cardiogenic differentiation and chamber identity is unaffected, but valve development is severely disrupted, presumably because of defective endocardial maturation and signaling (140). During valvulogenesis in the E9.5 mouse embryo, endocardial cells undergo EMT and form the cardiac cushions that later remodel to give rise to the thin, finely sculpted mature valve leaflets of the E14.5 embryo (141). RBPJk and Notch1 mutants have a collapsed endocardium and lack mesenchymal cushion cells, indicating that endocardial EMT is defective in Notch pathway mutants. Ultrastructural analysis of E9.5 mutant atrioventricular canal endocardia reveals that cells remain in close association, abnormally maintaining adherens junctions, and do not invade the cardiac jelly, despite showing features of activated premigratory endocardial cells. These observations correlate with a specific reduction in the transcription of the snail repressor (snai1) (142) in the atrioventricular canal region. Concomitantly, VE-cadherin expression remains abnormally stabilized in the atrioventricular canal and outflow tract endocardium of the mutants, suggesting that the lack of snail expression prevents down-regulation of VE-cadherin in this tissue, blocking endocardial EMT. These findings demonstrate that Notch activity is required for endocardial EMT (140).
Explant assays with Notch1 and RBPJk mutants demonstrate an impaired endocardial EMT, a phenotype reproduced in wild-type explants cultured with the
-secretase inhibitor N-[N-(3,5-difluorophenacetyl)-l-alanyl]-S-phenylglycine t-butyl ester (DAPT). This finding is supported by Notch inhibition experiments in zebrafish that impair valve development and by gain-of-function experiments in which transient overexpression of N1ICD in the heart leads to formation of hypertrophic atrioventricular valves (140). This result contrasts with recent zebrafish data reporting that restricted N1ICD overexpression in the endocardium of atrioventricular canal inhibits EMT (143). Conditional activation of Notch1 in the cardiac lineage of the mouse using a MesP1-CRE driver impairs atrioventricular myocardial differentiation and ventricular myocardium maturation of MesP1-CRE; N1ICD transgenic mice, but EMT occurs normally (144). This discrepancy may be explained by the fact that N1ICD RNA was ectopically expressed in both endocardium and myocardium (140), which may generate additional myocardial-derived EMT-inductive signals. In the case of the experiments by Beis et al. (143), N1ICD was overexpressed exclusively in the endocardium. Our own data infecting wild-type atrioventricular canal explants with a retrovirus expressing N1ICD are consistent with this possibility, because N1ICD-transduced explants transform more than controls (J. Grego-Bessa, L. Luna-Zurita, and J. L. de la Pompa, unpublished data). The relevance of Notch in human cardiac valve development and homeostasis was demonstrated by the recent finding that NOTCH1 mutations cause aortic valve disease in humans (145). Studies in mice in the same report show that similarly to Notch1, its target genes HRT1/Hey1 and HRT2/Hey2 are expressed in the aortic valve leaflets at E17.5 where they repress Runx2, a regulator of osteoblast cell fate. These results suggest that NOTCH1 mutations cause an early developmental defect in the aortic valve and a later derepression of calcium deposition that causes progressive aortic valve disease (145).
Jag1 and Notch2 are also linked to cardiac development. Mice doubly heterozygous for Jag1 and Notch2 mutations exhibit developmental abnormalities characteristic of Alagille syndrome, such as jaundice, growth retardation, defective bile duct differentiation, and abnormal kidney, eye, and heart development. The cardiac defects include right ventricular hypoplasia, pulmonic valve stenosis, atrial and ventricular septal defect, and dextropositioning of the aorta (59). The Notch targets HRT1/Hey1 and HRT2/Hey2 have also been linked to cardiac development. HRT2-targeted mutant mice have several cardiac anomalies. Almost all HRT2 mutants show growth retardation and die within 10 d of birth. Surviving HRT2 mutants have enlarged atria and ventricles, and echocardiographic analysis revealed abnormal cardiac hemodynamics, including stenosis and regurgitation of the tricuspid valve, mitral valve regurgitation, membranous ventricular septal defect, and secundum atrial septal defect (146). The phenotypic variation reported in HRT2 mutants probably results from the use of different genetic backgrounds and/or functional redundancy between HRT2 and other HRT family members (147). Analysis of HRT2 mutant mice suggests that HRT2 is required for formation of the atrioventricular valves. In addition, mice lacking both HRT1 and HRT2 die during embryogenesis due to severe cardiovascular malformations, including those in the development of the atrioventricular cushions. Few cells undergo EMT in the HRT1; HRT2 mice (148), suggesting that HRT1 and HRT2 function synergistically in the EMT process. In addition, HRT1; HRT2 mice show defects in trabecular myocytes (148). Although the ventricular chamber containing both the compact and trabecular zones forms initially, subsequent apoptosis of trabecular zone myocytes leads to poor trabecular formation. It is thus clear that HRT2 is important for correct cardiomyocyte development. The interaction between HRT1-expressing endocardial and epicardial cells and HRT2-expressing cells in the compact layer of the ventricle is also considered necessary to produce and/or maintain trabecular myocytes (149). Our own data (295) support a role for Notch in the development of ventricular myocardium, as the trabeculation-defective phenotype of standard and endocardial-specific Notch 1 and RBPJk mutants indicates. We propose that Notch mediates an endocardium-myocardium interaction critical for trabeculation and ventricular chamber morphogenesis and identify two distinct Notch-dependent processes: 1) transition of primitive myocardial epithelium to trabecular and compact myocardium (EphrinB2- and NRG1-dependent); and 2) maintenance of a proliferating trabecular cardiomyocyte population (BMP10-dependent) during this transition. The defective ventricular phenotype of Notch mutants is reminiscent of a congenital disorder termed isolated ventricular noncompaction (IVNC) (296), which is characterized by altered myocardial structure. Thus, Notch signaling may be altered in infants with conditions including malseptation, abnormal valve development or conduction system defects, all of which are related to abnormal trabeculation.
Notch activity in the cardiovascular system functions via a mechanism termed lateral induction (83), by which Notch generates contiguous domains of cells that share the same fate, an embryonic field. This signaling mechanism also occurs in flies during wing margin boundary formation (25). Other examples of Notch-mediated lateral induction in vertebrates include induction of proneural domains in the ear (150), formation of the limb bud margin (151), and somite boundary formation (83). In these cases, Notch activation promotes ligand production via a positive feedback loop, so that signaling occurs simultaneously in a developmental field. Loss of Notch signaling leads to down-regulation of ligand expression throughout the embryonic field, indicating the existence of a positive feedback loop.
Figure 5
summarizes the processes in which Notch is involved during cardiogenesis. In a search for novel therapeutic approaches for cardiac disease, and in view of the role of Notch in the maintenance of an uncommitted state (152) and in inhibition of the cardiogenic fate in vitro (139), it will be of great interest to explore the role of Notch in adult cardiac homeostasis.
|
| V. Notch in Endocrine Development: Pancreas, Gut, and Bone Endocrine Cells |
|---|
|
|
|---|
-and ß-endocrine cells (islets of Langerhans) that produce the hormones regulating nutrient homeostasis (insulin and glucagon), and the elaborate branched ductal tree that connects with the intestine (153, 154). Pancreas formation begins early in development (around E9.0 in the mouse) with the formation of two cell buds, ventral and dorsal. These buds express the Pdx1 transcription factor (155, 156) and arise from a specialized endodermal epithelium located in the foregut region that will give rise to the duodenum (157). During pancreatic organogenesis, the two buds undergo a series of morphogenetic, proliferative, and lineage specification events, grouped in three "developmental transitions," to give rise to the mature and functional pancreas (for a review, see Refs. 158 and 159). At E13.5, the developing pancreas is formed by a branched Pdx1-positive epithelial tree, without morphological signs of exocrine differentiation. Endocrine and exocrine precursors are marked by two members of the bHLH transcription factors family, Ngn3 and Ptf1-P48, respectively (160). Specific Notch pathway elements and downstream effectors are expressed in the developing pancreas, suggesting a role for Notch in pancreatic development (161, 162, 163). Loss-of-function of various Notch pathway genes (RBPJk, Delta1, and Hes1) leads to up-regulation of the proendocrine gene Ngn3 (164), a direct Hes1 target (165), and consequent accelerated and increased differentiation of pancreatic endocrine cells (161, 162). Conversely, forced Notch activity in the embryo blocks both endocrine and exocrine pancreas development (166, 167), and in pathological situations it leads to dedifferentiation of exocrine cell types in pancreatic epithelium (168). The physiological relevance of Notch in exocrine pancreas has been shown in the mouse, where Notch is active in committed exocrine progenitors cells and whose ectopic activation in pancreatic bud explants represses acinar cell differentiation (169). The available data indicate that Notch regulates the progressive recruitment of endocrine and exocrine cell types from a common precursor pool in developing pancreas. The inhibitory effect of Notch signaling in exocrine differentiation has been well characterized in zebrafish, where endocrine and exocrine cells arise independently (170). Thus, accelerated exocrine differentiation is observed in zebrafish mind bomb mutants or upon expression of a dominant negative Su (H) version in wild-type embryos (169).
Similarly to the CNS, Notch operates in the developing pancreas by lateral inhibition. This would explain the initially dispersed distribution of endocrine cells within the pancreatic epithelium. Consistent with this idea, Hes1 mutants show increased expression of the Delta 1 and Delta 3 ligands (162). Nevertheless, additional observations suggest that Notch signaling in the pancreas does not function only via lateral inhibition and perhaps respond to additional signals, emanating from the mesenchyme. Fibroblast growth factor 10 (FGF10) produced in pancreatic mesenchyme has been shown to be essential for precursor pool maintenance (171). Transgenic mice overexpressing FGF10 in pancreatic epithelia show pancreatic hyperplasia and a block in exocrine, ductal, and endocrine differentiation (172, 173). In these mice and in contrast to the wild-type situation (161, 162), Notch1 and Notch2 as well as Hes1 are ubiquitously expressed in pancreatic precursor cells, whereas ngn3 expression is abrogated (172, 173). Thus, ectopic FGF10 signaling is capable of maintaining Notch signaling activity in the pancreas. These observations have led to the suggestion that another mechanism different from lateral inhibition and termed "suppressive maintenance" sustains Notch signaling activity in the pancreas and suppress its differentiation (159, 172). The mechanism underlying suppressive maintenance is not well understood, but it may involve FGF10-dependent activation of the Jag 1 and Jag2 Notch ligands in pancreatic epithelium (172). It remains to be seen what is the effect of Notch signaling abrogation on the expression of its ligands.
A further refinement of our view of Notch in pancreatic development has come from the analysis of Hes1-targeted mutant mice. Ptf1 is misexpressed in discrete regions of the primitive stomach and duodenum and throughout the bile duct of Hes1 mutants. Ptf1-expressing cells are reprogrammed to multipotent pancreatic progenitors that differentiate into mature pancreatic exocrine, endocrine, and duct cells. These data demonstrate that Notch is required for proper regional specification of pancreas in developing foregut endoderm through Ptf1 regulation (174). Because Notch is involved in the development and homeostasis of a variety of self-renewing tissues (175, 176), understanding Notch function in pancreatic development will help to design protocols to control ß-cell development in vitro and thus treat diseases such as diabetes, in which the mass of insulin-producing ß-cells is reduced.
B. Gut development
The gastrointestinal tract comprises the small intestine and colon. Small intestine epithelium is organized into finger-like villi and adjacent invaginations, the crypts of Lieberkühn; the colon has a flat epithelial surface with no villi (177). In small intestine, the crypt compartment contains stem cells and progenitors, whereas the villus compartment is made up entirely of differentiated cells. Pluripotent stem cells, which reside at the crypt bottom, give rise to transit amplifying cells, which divide robustly before terminal differentiation. Four lineages can be distinguished in the gastrointestinal tract, i.e., absorptive enterocytes, mucus-secreting goblet cells, hormone-secreting enteroendocrine cells, and lysozyme- and cryptidin-producing Paneth cells (reviewed in175). Enteroendocrine cells can be further subdivided on the basis of the hormones they secrete (178). Intestine homeostasis depends on a small number of evolutionarily conserved pathways, including bone morphogenetic protein (BMP)/TGFß, sonic hedgehog (Shh), wingless (Wnt) and Notch (179).
The analysis of mutant zebrafish or targeted mice deficient for different Notch pathway elements or target genes has implicated Notch in the regulation of the earliest intestinal cell fate decisions. Thus, a recent study describes increases in secretory cells at the cost of absorptive cells in the intestines of zebrafish that are mutant for DeltaD and mind bomb (180). Different Notch pathway elements are expressed in murine crypts (181). The Notch target Hes1 is expressed in crypts (162). Analysis of the developing fetal intestine of Hes1 mutant mice revealed an increase in mucosecreting and enteroendocrine cells at the expense of absorptive enterocytes (162). The crypt progenitor pool in the small intestine seemed unaffected, as judged by an analysis of proliferative activity. Math1 is a target gene of Hes1-mediated repression in several organs, including the intestine (182). Math1 mutant mice die neonatally. Although the crypt-villus architecture was essentially undisturbed in the mutant mice, commitment toward the secretory lineage had entirely halted (182). These results have been interpreted to indicate that Hes1 and Math1 are required to skew the fate of differentiating cells leaving the transit amplifying compartment toward an enterocyte or a secretory phenotype, respectively (162, 182).
Conditional deletion of the RBPJK effector in the epithelium of the small intestine and colon using different CRE-driver lines led to a decrease in Hes1 expression, and Math1 was up-regulated throughout the crypt compartment, whereas it is normally expressed only in secretory cells (183). In addition, the number of goblet cells was increased. In the small intestine, Paneth cells appeared in near-normal numbers at the bottom of the crypts. However, the rapidly dividing transit amplifying compartment, which normally occupies the remainder of the crypt, was entirely replaced by postmitotic goblet cells that expressed Math1 protein but not Hes1. Examination of proliferation by different techniques indicated that basically all epithelial cell division had halted. Essentially identical observations were made in the colon (183). As an explanation for the different severity of the phenotypes observed upon Hes1 deletion and conditional RBPJk deletion in colon, it has been argued that other Hes genes such as Hes6 are affected by conditional RBPJk deletion (183). This phenotype was confirmed using a highly efficient
-secretase inhibitor. Cell proliferation had entirely halted and histological markers revealed that the tissue changes were indistinguishable from those observed after RBPJk deletion (183).
A reciprocal phenotype to that of RBPJk deletion in colon was obtained in transgenic mice overexpressing N1ICD in all cells of the intestinal epithelium, by virtue of the villin-CRE driver (184). N1ICD; villin-CRE transgenic mice have a complete lack of goblet cells in all intestinal tracts. In addition, enteroendocrine cells are markedly reduced, as well as Paneth cells (184). Microscopic examination of earlier developmental stages revealed that already at E18.5, N1ICD expression affects the architecture of the villi, and thus the differentiation of secretory cell lineages along the duodenal-ileal axis and the cranial-to-caudal wave of intestinal differentiation was inhibited (184). Transcriptional analysis revealed a direct correlation between N1ICD expression and elevated levels of Hes1 transcription in the intestinal epithelium of N1ICD transgenic mice, although no other Hes genes were affected. Expression of the Hes1 targets Math-1 (185) and ngn3 (165), involved in secretory cell lineage specification, was reduced in N1ICD; villin-CRE mice. This finding was consistent with the idea that the expression of Math-1 and ngn3 is repressed by Notch activation (184). This intestinal phenotype is reminiscent of Math-1 (lack of goblet and enteroendocrine cells) and opposite to that of Hes1 mutant mice (an excess of secretory cells at the expense of enterocytes). Thus, the gain-of-function phenotype of N1ICD; villin-CRE mice provides direct evidence that Notch signals target Hes1 in the intestine, explaining mechanistically the differentiation defects observed. Overall, these genetic data indicate that Notch-mediated Hes1 expression regulates a binary cell fate decision between adsorptive and secretory cell fates. However, Hes1-deficient mice do not show a change in the proliferative status of the intestinal precursor pool (162), whereas Notch activation profoundly affects the proliferation potential of intestinal progenitors (184), suggesting that other Notch targets may be responsible for the increased in proliferation. The effect of deregulated Notch signaling in colon homeostasis and cancer will be discussed in Section VI.B.
C. Bone development
Bone is a dynamic tissue that is constantly renewed and degraded. Two main types of bone cells are responsible for these processes: bone-forming osteoblasts of mesenchymal origin, and bone-resorbing osteoclasts of hematopoietic origin. Bone formation and resorption are coordinated so that bone remodeling is balanced. When this equilibrium is altered in a way that bone resorption exceeds bone formation, osteoporosis occurs. This disease is prevalent in old age and is characterized by bone loss and a high risk of fractures. Therefore, knowledge about molecular events involved in osteoblast differentiation is crucial.
The in vitro potential of Notch pathway in osteoclastogenesis and osteoblastogenesis has been investigated in several reports. Evidence indicates that Notch down-regulates osteoclastogenesis activation, reduces the surface expression of c-Fms, which is a receptor for macrophage colony-stimulating factor, in osteoclast precursor cells and enhances the expression of osteoprotegerin in stromal cells, which results in the down-regulation of osteoclastogenesis (186). However, controversial results have been obtained with respect to osteoblastic differentiation. Continuous NICD expression inhibits bone morphogenetic protein 2-induced osteoblast differentiation in osteoblast precursor cells (187). In contrast, transient expression of N1ICD in osteoblast precursor cells leads to an enhanced bone mineral deposition (188). Notch1 is expressed in the mesenchymal condensation area and subsequently in the hypertrophic chondrocytes during chondrogenesis (189). Another study shows that Notch1, Delta1, and Jagged1 are expressed in cultured osteoblast precursor cells as well as in differentiating osteoblasts during bone regeneration, and that Notch1 is activated in these cells (190). These results suggest that Notch signaling plays an important role in the commitment of mesenchymal cells to the osteoblastic cell lineage (190). Concomitant expression of Delta1 and Jagged1 during in vivo bone regeneration suggests that there is a functional redundancy between Delta1 and Jagged1 and that these ligands direct osteoprogenitor cells to the differentiated status through identical signaling pathways (190). These data suggest a therapeutic potential for Notch in bone regeneration as well as osteoporosis.
Mammalian bone marrow architecture involves hematopoietic stem cells (HSCs) in close proximity to the endosteal surfaces of bone (191), with more differentiated cells arranged in a graduated fashion as the central longitudinal axis of the bone is approached (192), suggesting a relationship between HSCs and osteoblasts. Osteoblasts produce hematopoietic growth factors (193) and are activated by PTH or the locally produced PTHrP through the PTH/PTHrP receptor (PPR). Experiments involving transgenic mice expressing in osteoblastic cells a constitutively active PPR under the control of the alpha1(I) collagen promoter showed increased numbers of trabeculae and trabecular osteoblastic cells in the long bones of transgenic mice (194). In addition, the number of HSCs was also increased in these mice, but mature cell number was not changed (194). Because this phenotype was reminiscent of the mode of Notch action in hematopoiesis (195) and Jag1 is expressed by marrow stromal cells (196) and by murine osteoblasts (197), the levels of Jag1 protein were analyzed in the bone marrow of transgenic mice. col1-caPPR transgenic mice showed a marked increase in Jag1 expression in osteoblastic cells. The response of hematopoietic stem cells to increased Jag1 expression was measured by N1ICD staining, which was increased in transgenic mice. These data support a model in which PPR activation in the osteoblastic population increases cell number and the overall production of Jag1. This, in turn, may activate Notch on primitive hematopoietic cells, resulting in expansion of the stem cell compartment (194). In another study, these authors demonstrated that PTH treatment led to increased levels of Jagged1 in osteoblastic cells both in vivo and in vitro, in a protein kinase A (PKA)-dependent manner (198). Because Jagged1 is very important in stromal-HSC interactions and PTH regulates HSC expansion through osteoblastic activation, these studies suggested that stimulation of osteoblastic PKA activation may alter the HSC niche. This is of great therapeutic importance because this study suggests that alternative means to stimulate osteoblastic PKA activation may alter the HSC niche. In addition, Jagged1/Notch signaling modulates osteoblastic differentiation (189), and Jagged1 may play a critical role in mediating the effects of PTH on osteoblasts.
| VI. Notch in Cancer |
|---|
|
|
|---|
|
-secretase blockade has antineoplastic effects in Notch-expressing transformed cells in vitro and in xenograft models in vivo (Refs. 211, 212, 213, 214 ; reviewed in Refs. 176 and 209). Notch alone may not be a very efficient oncogene, however, and it must associate with another oncoprotein to cause transformation. Although such partners have not yet been identified in naturally occurring tumors, transformation can be induced in vitro in various cell types by expressing NICD with certain oncoproteins (215, 216). The Notch tumor suppressor function may be a peculiarity of the mouse skin system, or it may also apply in man and include human keratinocytes as well as other human epithelial cell types (217). Available evidence thus suggests that, with the possible exception of some human epidermal malignancies, Notch signaling inhibition is a viable strategy for treatment of certain solid and hematopoietic tumors (176, 209). JAG1 expression in head and neck squamous cell carcinoma (HNSCC) cells triggers Notch activation in neighboring endothelial cells and promotes network formation (218). In xenograft models, HNSCC cells overexpressing JAG1 formed larger tumors with increased vascularization, and JAG1 protein levels were notably higher in HNSCC samples than in normal samples (218). These results offer a causal link between Notch signaling and tumor angiogenesis and describe a novel juxtacrine signaling mechanism from tumors to surrounding vasculature. As we saw in Section IV, Notch is essential for angiogenic remodeling in the embryo and for vascular homeostasis in the adult. This study shows that Notch is also involved in pathological angiogenesis (218), suggesting a possible direction for therapeutic intervention in tumors. In this regard, a recent report has shown that VEGF-induced Delta4 acts as a negative regulator of tumor angiogenesis (219). Thus, tumor-derived VEGF induces Delta4 expression in angiogenic endothelial cells to negatively regulate vascular growth, acting to restrain excessive vascular sprouting and branching and allowing angiogenesis to proceed at a productive rate. Thus, increasing Delta4/Notch activity resulted in decreased vascular density associated with reduced sprouting and branching of the vascular network. On the contrary, the blockade of Delta4/Notch signaling produced enhanced angiogenic sprouting and branching, resulting in a marked increase in tumor vessel density but decreased tumor vessel function (219). These data suggest that an alternative treatment of tumors may be based on the promotion of "nonfunctional" tumor angiogenesis.
A. Notch in hematological tumors
Notch signaling mediates hematopoietic cell fate determination in the embryo (220) and in the adult (221) and is also a critical factor in the maintenance of a pool of self-renewing HSCs (222). Deregulated expression of Notch pathway elements can thus lead to development of hematological malignancies. The prototypical Notch-associated cancer is human acute T cell acute lymphoblastic leukemia/lymphoma (T-ALL), which constitutes approximately 1520% of ALL in children and adults. The NOTCH1 gene was discovered due to its involvement in a chromosome translocation [t (7; 9)] seen in some human T-ALL; this leads to expression of N1ICD in a T cell receptor-ß-regulated manner (Fig. 6A
) (223). Notch1 was later shown to be essential for normal T cell progenitor development (62). Although t (7; 9) is rare (less that 1% of T-ALL; Fig. 6A
), the majority of human T-ALL have gain-of-function mutations in NOTCH1, leading to aberrant increases in downstream signaling (Fig. 6A
) (224), placing the NOTCH pathway at the center of T-ALL pathogenesis. Weng et al. (224) found that more than 50% of human T-ALL without specific (7; 9) chromosome translocation, including tumors from all major molecular oncogenic subtypes, have activating mutations that involve the NOTCH1 extracellular heterodimerization domain and/or the C-terminal PEST destruction box (Fig. 6A
). The detection of mutations in multiple molecular subtypes of T-ALL is the basis for the conclusion that NOTCH1 appears to collaborate with various other proteins that are also deregulated in T-ALL (176). A recent study also identified c-MYC as an important direct target of NOTCH1 in T-ALL and in a critical stage of normal pre-T cell development, showing that c-MYC inhibitors interfere with progrowth effects of activated NOTCH1 and that forced c-MYC expression rescues NOTCH1-dependent T-ALL cell lines from Notch withdrawal (225). Based on these findings, a phase I/II clinical trial was recently begun using a NOTCH pathway inhibitor to treat patients with refractory T-ALL. I