BIOLOGICAL OVERVIEW

Notch and its ligands are modified by a protein O-fucosyltransferase (O-fut1, also known as Neurotic or Ofut1) that attaches fucose to a serine or threonine within EGF domains. By using RNAi to decrease O-fut1 expression in Drosophila, it has been demonstrated that O-linked fucose is positively required for Notch signaling, including both Fringe-dependent and Fringe-independent processes. The requirement for O-fut1 is cell autonomous, in the signal-receiving cell, and upstream of Notch activation. Therefore, O-fut1 activity is required for the cell's ability to receive ligand signals, and would thus be consistent with the hypothesis that the key substrate of O-fut1 is Notch. The transcription of O-fut1 is developmentally regulated, and surprisingly, overexpression of O-fut1 inhibits Notch signaling. Together, these results indicate that O-fut1 is a core component of the Notch pathway, one that is required for the activation of Notch by its ligands, and whose regulation may contribute to the pattern of Notch activation during development (Okajima, 2002).

A mutation has been isolated in the gene encoding O-fucosyltransferase, and analysis of the mutant phenotype confirms the RNAi studies and reveals an unprecedented example of an absolute requirement of a protein glycosylation event for a ligand-receptor interaction. A novel maternal neurogenic gene, neurotic, is essential for Notch signalling. neurotic functions in a cell-autonomous manner, and genetic epistasis tests reveal that Neurotic is required for the activity of the full-length but not an activated form of Notch. neurotic has been shown to be required for Fringe activity. fringe encodes a fucose-specific ß1, 3 N-acetylglucosaminyltransferase that modulates Notch receptor activity. Neurotic is essential for the physical interaction of Notch with its ligand Delta, and for the ability of Fringe to modulate this interaction in Drosophila cultured cells. These results suggest that O-fucosylation catalysed by Neurotic is also involved in the Fringe-independent activities of Notch and may provide a novel on-off mechanism that regulates ligand-receptor interactions (Sasamura, 2003).

Two types of Notch ligands, Delta and Serrate/Jagged, have been identified and shown to possess redundant and distinct functions. Specific functions of each of these two ligands are modified by Fringe (Fng), which adds N-acetylglucosamine (GlcNAc) to O-fucose residues on the EGF repeats of Notch. Fng is involved in establishing a boundary that segregates two cell groups, such as dorsal and ventral (D/V) compartments of the wing disc or leg segments of the leg disc in Drosophila. This GlcNAc modification had been thought to make Notch more sensitive to Delta and less sensitive to Serrate. An additional modification of the GlcNAc by a galactosyltransferase has been shown to be required for the inhibition of Jagged1. The O-fucose glycans are found at serine or threonine residues of consensus sequence CXXGGS/TC (X is any amino acid) that is between the second and third conserved cysteines of the EGF repeats. In mammals, the O-fucosyltransferase that catalyses this fucosylation has been purified and cloned. The functions of a Drosophila O-fucosyltransferase (O-fut1) have been examined using RNA interference (RNAi) and O-fut1 has been implicated in Notch signalling (Okajima, 2002). A knock out of the mouse Pofut1 gene that encodes mouse O-FucT-1 has been reported (Shi, 2003); Pofut1^–/– mice have a phenotype similar to those of the embryos lacking downstream effectors of all Notch signalling pathways. However, the molecular mechanism underlying the function of O-fucose was not determined (Sasamura, 2003).

Deficiency of essential genes for Notch signalling causes overproduction of neurons in the Drosophila embryo: the 'neurogenic' phenotype. Some essential components of Notch signalling are associated with the neurogenic phenotype when both the zygotic genes and maternally contributed mRNA are abolished. These genes are referred to as 'maternal neurogenic genes'. Several genes relatively recently identified as components of Notch signalling, such as Suppressor of Hairless, Kuzbanian, Presenilin and Nicastrin belong to this class of genes. Because of the difficulty to screen for maternal phenotypes, not all maternal neurogenic genes have yet been identified. A novel maternal neurogenic gene, neurotic (nti O-fut1 -- FlyBase), encodes the O-fucosyltransferase. The activity of nti is also needed in wing margin formation, indicating that nti is an essential component of Notch signalling. Epistatic analysis showed that nti is essential for full-length Notch and Fng function but not for Notch^ICD. Nti modification of Notch is essential for the physical interaction between Notch and Delta in Drosophila cultured cells. These results establish Neurotic/O-fut1 as a moderator of Notch-ligand interactions, this moderation has both Fng-dependent and Fng-independent functions (Sasamura, 2003).

Notch signalling is involved in two major classes of cell-cell signalling: lateral inhibition and inductive signalling. Neuroblast segregation in early embryogenesis is a well known example of lateral inhibition, and failure of the Notch signalling pathway during this process leads to an hypertrophy of neuroblasts, known as the neurogenic phenotype. However, induction of several genes, such as wg, along the dorsal/ventral compartmental boundary of the wing disc depends on inductive Notch signalling. In both classes of Notch signalling events, the Notch and nti mutant phenotypes are identical, indicating that Nti is essential for Notch signalling. This finding differs from the recent report using nti RNAi (Okajima, 2002), since apparent embryonic neurogenic phenotypes were not noted. This discrepancy is probably due to the partial knockdown effect generated by RNAi (Sasamura, 2003).

There are a considerable number of genes, including the ligands of the Notch receptor, that encode putative proteins that contain EGF motifs with consensus sequences for O-fucosylation in the Drosophila genome. However, mutant phenotypes of nti are strikingly similar to those of the Notch receptor. In fact, in addition to the wing phenotype, nti somatic clones exhibit mutant phenotypes that are characteristic of Notch loss-of-function mutations. These include defects on the notum bristles, rough eye and the leg segment fusion. These phenotypes were also observed in the knockdown study of nti (Okajima, 2002).

These results indicate that the primary target of O-fucosylation of Nti is Notch, while there may be unrevealed signalling pathways in which nti function is required. There is another putative O-fucosyltransferase in Drosophila and mammals (Roos, 2002). Thus, it is tempting to speculate that each O-fucosyltransferase has distinct target(s) that are involved in specific cell signalling pathways (Sasamura, 2003).

Since O-fucose on Notch has been shown to act as a molecular scaffold for GlcNAc that is elongated by Fng, one would expect that the phenotypes of O-fucosyltransferase mutant might be the same as those of fng. Unexpectedly, however, the nti and fng mutant phenotypes are quite different. Strikingly, the embryonic neurogenic phenotype that is evident in nti mutant, and is an indication of its essential role in Notch signalling, is not evident in fng mutants. Furthermore, it is thought that Fng does not have a significant role in lateral inhibition, while it is involved in the generation of the cell boundary between cells expressing Fng and cells not expressing Fng. Additionally, an in vitro binding assay revealed that Nti is essential for binding between Notch and Delta. Based on the previous findings and the present results, it is proposed that O-glycosylation of Notch EGF repeats has two distinct roles for binding to Delta. (1) O-fucosylation catalysed by Nti is an absolute requirement for binding between Notch and the ligand, and this binding is sufficient to accomplish lateral inhibition. For this function, no additional glycosylation to O-fucose residue is required. This idea is also supported by the observation that in the tissues and organisms that do not express fng, Delta is competent to activate the Notch receptor. In this respect, it is worth noting that in the C. elegans genome there is a highly conserved nti, while a fng homolog is not found. (2) Addition of GlcNAc to the O-fucose residue by Fng enhances the interactions between Notch and Delta, modulating the receptor-ligand interactions. In fact, the expression of fng shows a high degree of regional specificity, and the boundary of the cells expressing and not expressing Fng often defines the border of distinct tissue structures. Thus, the region-specific expression of fng allows modulation of Notch signalling, resulting in generation of complex structure of organs. As expected from the second function of Nti, its function is essential for Fng-dependent modulation of Notch signalling as well as Fng-independent function. In the wing disc, nti is epistatic to fng, and fng requires nti to induce Wg at the dorsal and ventral compartment boundary. Additionally, in the in vitro binding assay, Fng depends on Nti to enhance the binding between Notch and Delta. These lines of evidence indicate that Nti is involved in Fng-dependent modulation of Notch signalling, which is consistent with an O-glycan structure of the Notch EGF repeats (Sasamura, 2003).

Given that nti has novel functions, what is the significance of Nti in Notch signalling at the level of cells and the organism? A model is proposed for nti function during development. This model involves a novel receptor on-off mechanism. In the early embryo, nti mRNA is detected uniformly. However, Okajima (2002) has reported that expression of nti/O-fut1 was spatially regulated during the late stages of embryogenesis and the imaginal discs. Therefore, O-fucosylation of Notch EGF repeats is probably regulated temporally and spatially, and may provide a novel mechanism to regulate Notch signalling. For example, in cells expressing nti at higher level than their neighbors, Notch may become more potent to receive a signal. Removal of O-fucose from the Notch EGF repeats may be part of this regulation (Sasamura, 2003).

In this work, no effect of nti on Notch-Serrate binding was demonstrated, because, as reported before, Notch-Serrate-AP binding can not be detected by this binding assay. However, phenotypes observed in the somatic clones of nti generated in the wing disc strongly suggests that Nti affects the Notch-Serrate binding. For example, in the nti clones near the dorsal and ventral boundary from the ventral side, induction of Wg is impaired, indicating that nti clones fail to receive the Serrate signal. Thus, it is proposed that the requirement of nti for Notch-ligand binding is not restricted to Delta. However, the neurogenic phenotype of nti can be wholly explained by the effect on Notch-Delta binding, since Delta is the only the ligand of Notch involved in neuroblast segregation during early embryogenesis (Sasamura, 2003).

In the Notch-Delta binding assay, co-expression of Nti and Fng with Notch increases the Notch-Delta binding over that of Fng with Notch, whereas co-expression of a wild-type Nti with Notch does not increase this binding. It is speculated that the amount of Nti endogenously expressed in S2 cells saturates the ability of fucose alone to promote the ligand binding. Indeed, it has been shown that S2 cells endogenously expressed Nti/O-fut1 (Okajima, 2002). However it does not imply that addition of O-fucose to Notch by Nti is also saturated under these conditions. Namely, the increase of O-fucosyltransferase activity, which does not lead to elevation of Delta binding per se, could increase the Notch-Delta binding after elongation of the O-fucose residue by Fng. These results suggest that the linear range by which fucose-GlcNAc moiety can elevate the ligand binding is wider than that by which fucose alone can. Multiple sites that are possibly O-fucosylated were found in the EGF repeats of Notch (Okajima, 2002). Thus, it is possible that some of these potential O-fucosylation sites may influence the Delta binding only after modification by Fng. It was also noticed that Nti-myc probably has a dominant-negative function. This dominant-negative activity of Nti-myc is apparently restored by co-expression with Fng; co-expression with Fng makes the level of ligand binding well over that by Fng expression alone. This is in contrast with the fact that the decreased Notch-Delta binding associated with RNAi of nti, which is supposed to merely reduce the endogenous Nti protein, is hardly restored by co-expression with Fng. Therefore, it is speculated that the Myc-tag does not simply disrupt the enzymatic activity of Nti-myc. Some kind of competition among Nti-myc, endogenous Nti, Fng and Notch may be involved in this dominant-negative effect. In support of this model, physical interaction between Notch and Fng has been reported. This dominant-negative effect may also account for the reduction of the Notch-Delta binding associated with Nti-G3-myc. It is speculated that Ntimyc and Nti-G3-myc, both of which carry the same Myc-tag, have similar dominant negative effects, which do not depend on the enzymatic activity of O-fucosyltransferase and which are restored by co-expressing Fng. The slight reduction of the Notch-Delta binding associated with Nti-G3-myc and Fng over that of Fng alone may also be caused by this dominant-negative effect of Nti-G3-myc (Sasamura, 2003).

The importance of glycoconjugates in cell adhesion and cell-cell communication has been affirmed for many years, since nearly all cells are covered with numerous carbohydrate-rich molecules. Gene disruption of a number of glycosyltransferases has been shown to resulted in embryonic lethality in mice, underscoring their importance in development of multicellular organisms. However, most of the previous reports are only descriptive of the phenotypes and the molecular mechanisms are still elusive. This study presents one of the very few examples of molecular explanation for such mechanisms of action. Furthermore, it is an unprecedented example of an absolute requirement of a protein glycosylation event for a ligand-receptor interaction. The simplest interpretation of these phenomena would be that association of Delta to its receptor involves lectin-like protein-carbohydrate interaction. This hypothesis is consistent with the result that O-fucosylation induces little conformational change (Kao, 1999) in EGF repeat of factor VII (Sasamura, 2003).

The first protein shown to be O-fucosylated was urinary type plasminogen activator (uPA), followed by tissue-type plasminogen activators and several clotting factors. It has been shown that O-fucosylation of uPA is essential for activation of the uPA receptor that has diverse functions including plasminogen activation, although uPA receptor-deficient mice survive to adulthood with no overt phenotypic abnormalities and are fertile. Another O-fucosylated protein is Cripto, a member of the EGF-CFC family. Cripto is thought to act as an essential co-factor for Nodal signalling. Amino acid substitution of threonine to alanine, which prevents O-fucosylation reduces the signalling activity of Nodal, suggesting that O-fucosylation is important for Cripto function. Until now, more than ten proteins have been found to carry the consensus sequence of O-fucosylation site. In this paper, it has been shown that O-fucosylation is mandatory for Notch signalling. Although the O-fucosyltransferase presented in this study has remarkable functional specificity to the Notch receptor, there seems to be plurality of such enzymes. Therefore, the regulation of protein activity or function by O-fucosylation could be a general mechanism for many signal transduction systems (Sasamura, 2003 and references therein).

Proteins that bind to sugars without modifying them are classified as lectins. If the action of O-fucose glycans involves lectin-based recognition, it could be either that the Notch ligands themselves are lectins, or that there exist other Notch signaling cofactors that function as lectins. The hypothesis that lectin-based recognition is essential for Notch signaling suggests an explanation as to why the extracellular domain of Notch contains so many EGF repeats. Lectins typically have only low affinity for individual sugars, and high avidity lectin-sugar binding is achieved by multivalent interactions of lectins with sugars. Thus, it is proposed that an array of EGF repeats acts as a scaffold for the display of O-fucose glycans, and thus allows for high avidity binding to multivalent lectins. This hypothesis further implies that quantitative differences in the binding of lectins to EGF repeats may be achieved by varying the fucosylation status of the Notch receptor or its ligands (Okajima, 2002).

The lectin hypothesis also suggests explanations for how both the O-fucose monosaccharide and the more elongated O-fucose glycans that form in the presence of Fringe can influence Notch signaling. Because different lectins can recognize distinct glycans, the existence is postulated of O-fucose binding lectins that are required for all Notch signaling, as well as distinct lectins that bind Fringe-dependent glycans, such as the Gal-β1,4-GlcNAc-β1,3-Fuc trisaccharide that has been implicated in the inhibition of Jagged1 signaling (Chen, 2001). In this model, the general positive requirement for the O-fucose monosaccharide in Notch signaling and the localized positive effect of Fng-dependent glycans in certain tissues could be reconciled if only a subset of O-fucosylated EGF repeats are actually substrates for Fng. Alternatively, it could be that a positively required lectin binds O-fucose regardless of whether it is the terminal or an internal sugar, but its affinity is altered. In contrast, since the negative influence of O-fut1 overexpression on Notch signaling is suppressed by co-expression with Fng, it may involve a lectin that specifically requires a terminal O-fucose. The identification of lectins that bind to O-fucose glycans would make it possible to test and extend these hypotheses, and thus constitutes an important goal for future research (Okajima, 2002).

GENE STRUCTURE

cDNA clone length - 1332 bp

Bases in 5' UTR - 31

Exons - 2

Bases in 3' UTR - 94

PROTEIN STRUCTURE

Amino Acids - 402

Structural Domains

Blast searches of Drosophila genomic and cDNA sequences identify a single closely related homolog (40% identity) of the human GDP-fucose protein O-fucosyltransferase 1. This Drosophila gene (CG12366) is symbolized as O-fut1, and the protein product it encodes as O-fut1. Like most glycosyltransferases, the gene product includes a predicted amino-terminal type II transmembrane domain. Typically, deletion of the transmembrane domain of a glycosyltransferase does not affect its activity, but prevents its retention in the Golgi, resulting in the secretion of a functional enzyme (Okajima, 2002).

Recently a human O-fucosyltransferase has been identified by Wang (1998). The Drosophila genome was searched using partial sequence data and two putative candidates for O-fucosyltransferase: CG12366 and CG14789, called O-FUT1 and O-FUT2, were identified. Both of these predicted genes are active as can be deduced from the existence of similar expressed sequence tag sequences in the data bases (Roos, 2002).

date revised: 20 September 2003

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.