Enzymatic function of Fringe

Fringe modulates Notch signaling resulting in the establishment of compartmental boundaries in developing organisms. Fringe is a beta 3N-acetylglucosaminyltransferase (beta 3GlcNAcT) that transfers GlcNAc to O-fucose in epidermal growth factor-like repeats of Notch. Five different Chinese hamster ovary cell glycosylation mutants were used to identify a key aspect of the mechanism of fringe action. Although the beta 3GlcNAcT activity of manic or lunatic fringe is shown to be necessary for inhibition of Jagged1-induced Notch signaling in a coculture assay, it is not sufficient. Fringe fails to inhibit Notch signaling if the disaccharide generated by fringe action, GlcNAc beta 3Fuc, is not elongated. The trisaccharide, Gal beta 4GlcNAc beta 3Fuc, is the minimal O-fucose glycan to support fringe modulation of Notch signaling. Of six beta 4galactosyltransferases (beta 4GalT) in Chinese hamster ovary cells, only beta 4GalT-1 is required to add Gal to GlcNAc beta 3Fuc, identifying beta 4GalT-1 as a new modulator of Notch signaling (Chen, 2001).

O-Fucose has been identified on epidermal growth factor-like (EGF) repeats of Notch, and elongation of O-fucose has been implicated in the modulation of Notch signaling by Fringe. O-Fucose modifications are also predicted to occur on Notch ligands based on the presence of the C2XXGG(S/T)C3 consensus site (where S/T is the modified amino acid) in a number of the EGF repeats of these proteins. Both mammalian and Drosophila Notch ligands are modified with O-fucose glycans, demonstrating that the consensus site is useful for making predictions. The presence of O-fucose on Notch ligands raises the question of whether Fringe, an O-fucose specific ß1,3-N-acetylglucosaminyltransferase, is capable of modifying O-fucose on the ligands. Indeed, O-fucose on mammalian Delta1 and Jagged1 can be elongated with Manic Fringe in vivo, and Drosophila Delta and Serrate are substrates for Drosophila Fringe in vitro. These results raise the interesting possibility that alteration of O-fucose glycans on Notch ligands could play a role in the mechanism of Fringe action on Notch signaling. As an initial step to begin addressing the role of the O-fucose glycans on Notch ligands in Notch signaling, a number of mutations in predicted O-fucose glycosylation sites on Drosophila Serrate have been generated. Interestingly, analysis of these mutants has revealed that O-fucose modifications occur on some EGF repeats not predicted by the C2XXGGS/TC3 consensus site. A revised, broad consensus site, C2X3-5S/TC3 (where X3-5 are any 3-5 amino acid residues), is proposed (Panin, 2002).

Deciphering the Fringe-Mediated Notch Code: Identification of Activating and Inhibiting Sites Allowing Discrimination between Ligands

Fringe proteins (see Drosophila Fringe) are beta3-N-acetylglucosaminyltransferases that modulate Notch activity by modifying O-fucose residues on epidermal growth factor-like (EGF) repeats of Notch. Mammals have three Fringes: Lunatic, Manic, and Radical. While Lunatic and Manic Fringe inhibit Notch1 activation from Jagged1 and enhance activation from Delta-like 1 (see Drosophila Delta), Radical Fringe enhances signaling from both. A mass spectrometry approach was used to determine whether the variable effects of Fringes on Notch1 result from generation of unique glycosylation patterns on Notch1. Lunatic and Manic Fringe were found to modify similar sites on Notch1, while Radical Fringe modified a subset. Fringe modifications at EGF8 and EGF12 enhanced Notch1 binding to and activation from Delta-like 1, while modifications at EGF6 and EGF36 (added by Manic and Lunatic but not Radical) inhibited Notch1 activation from Jagged1. Combined, these results suggest that Fringe modifications 'mark' different regions in the Notch1 extracellular domain for activation or inhibition (Kakuda, 2017).

Invertebrate Fringe homologs

Although Notch signaling is not known to play a role in the formation of segments in Drosophila, it was reasoned that it might do so in other insects such as the grasshopper, where segment boundaries form between cells, not between syncytial nuclei as they do in Drosophila. A single fringe gene has been cloned from the grasshopper Schistocerca. It is not detectably expressed in the forming trunk segments of the embryo until after segment boundaries have formed. It is concluded that fringe is not part of the mechanism that makes segments in Schistocerca. Thereafter it is expressed in a pattern that shows that it is a downstream target of the segmentation machinery and suggests that it may play a role in segment morphogenesis. Like its Drosophila counterpart, Schistocerca fringe is also expressed in the eye, in rings in the legs, and during oogenesis, in follicle cells (Dearden, 2000).

The thoracic and gnathal limbs of S. gregaria develop from the body wall of the embryo between 25% and 60% development. Sgfringe is expressed in each limb prior to the first appearance of joints. Rings of expression appear in the epithelium of each limb from 25% of development. Each ring forms separately, the most basal ring forming first. After the formation of the first two rings the subsequent rings form quickly, apparently by intercalating between the first two rings. The rings of Sgfringe expression appear earliest in the third limb where, by 40% development, seven or eight rings are visible. Most of these rings are located near folds in the limb bud epithelium. At the presumptive femur/tibia joint on the third leg, a joint that forms early and distinctively, expression of Sgfringe quickly becomes restricted to regions proximal to the invagination. This is not the case for all rings; in some cases the ring of Sgfringe expression is distal to the invagination of the epithelium. Sgfringe is also expressed in rings in the epithelia of the antenna, much like the expression of fringe in Drosophila, and in the gnathal appendages. As all of the appendages develop, Sgfringe is weakly expressed in mesodermal cells underlying the limb epithelia (Dearden, 2000).

In Schistocerca the lack of markers makes it difficult to locate Sgfringe expression during the early stages of eye development. However, as the embryonic eye develops, a clear boundary becomes visible between tissue organized into ommatidia and the remaining area of unorganized tissue. Sgfringe is expressed in the unorganized part of the eye with a sharply defined limit of expression along the line where ommatidia become visible. It is assumed that this limit is developmentally equivalent to the morphogenetic furrow in Drosophila. Sgfringe is expressed in the somatic cells of the Schis-tocerca ovary. RNA first appears in the interstitial cells in the germarium, from which all follicle cells are derived. It remains detectable in all follicle cells until at least the end of vitellogenesis. In mature follicle cells Sgfringe mRNA is present near the basal membrane (Dearden, 2000).

Fringe function in fish and frogs

Cell interactions involving Notch signaling are required for the demarcation of tissue boundaries in both invertebrate and vertebrate development. Members of the Fringe gene family encode b-1,3 N-acetyl-glucosaminyltransferases that function to refine the spatial localization of Notch-receptor signaling to tissue boundaries. The isolation and characterization of the zebrafish (Danio rerio) homolog of the lunatic fringe gene (lfng) is described in this study. Zebrafish lfng is generally expressed in equivalent structures to those reported for the homologous chick and mouse genes. These sites include expression along the A-P axis of the neural tube, within the lateral plate mesoderm, in the presomitic mesoderm and the somites and in specific rhombomeres of the hindbrain; however, within these general expression domains species-specific differences in lfng expression exist. In mouse, Lfng is expressed in odd-numbered rhombomeres, whereas in zebrafish, expression occurs in even-numbered rhombomeres. In contrast to reports in both mouse and chicken embryos showing a kinematic cyclical expression of Lfng mRNA in the presomitic paraxial mesoderm, no evidence is found for a cyclic pattern of expression for the zebrafish lfng gene; instead, the zebrafish lfng is expressed in two static stripes within the presomitic mesoderm. Nevertheless, in zebrafish mutants affecting the correct formation of segment boundaries in the hindbrain and somites, lfng expression is aberrant or lost (Prince, 2001).

Two Xenopus homologs of the Drosophila gene fringe, lunatic Fringe (lFng) and radical Fringe (rFng), were identified and the protein product of lFng further characterized. Amino acid identities are 37% between iFng and Drosophila FNG, 33% between rFng and Drosophila FNG and 46% between iFng and rFng. There is no significant sequence similarity among Xenopus and Drosophila sequences before residue 120 of Drosophila FNG, while the C-terminal residues are highly conserved. The messenger RNA of lFng is supplied as a maternal message. Its product is a precursor protein consisting of pre-, pro-, and mature regions. LFng mRNA is detected in animal, marginal and vegetal regions of stage 9 embryos, and is later expressed in the neural tube, in the medial, intermediate and later neurons, and the eyes. The mature lunatic Fringe protein is secreted extracellularly, and it induced mesodermal tissue formation in animal cap assays, including markers Xbra and Xwnt-8, vertebrate homologs of Drosophila T-related gene and optomotor blind and of wingless, respectively. These results indicate that secreted lunatic Fringe can induce mesoderm and reveal that the Fringe proteins are a family of vertebrate signaling molecules (Wu, 1996).

Function of chicken Fringe homologs in neural crest development

Lunatic fringe is a vertebrate homolog of Drosophila fringe, which plays an important role in modulating Notch signaling. This study examines the distribution of chick lunatic fringe at sites of neural crest formation and explores its possible function by ectopic expression. Shortly after neural tube closure, lunatic fringe is expressed in most of the neural tube, with the exception of the dorsal midline containing presumptive neural crest. Thus, there is a fringe/non-fringe border at the site of neural crest production. Expression of excess lunatic fringe in the cranial neural tube and neural crest by retrovirally mediated gene transfer resulted in a significant increase (~60%) in the percentage of cranial neural crest cells 1 day after infection. This effect is mediated by an increase in cell division as assayed by BrdU incorporation. Infected embryos have an up-regulation of Delta-1 in the dorsal neural tube and redistribution of Notch-1 to the lumen of the neural tube, confirming that excess fringe modulates Notch signaling. These findings point to a novel role for lunatic fringe in regulating cell division and/or production of neural crest cells by the neural tube (Nellemann, 2001).

Although Notch signaling is generally thought to affect cell lineage decisions, other data are consistent with this signaling pathway influencing cell proliferation. In the fly wing, activation of Notch in the wing disc results in a strong increase in mitotic activity, and, in chicken, a constitutively active form of Notch-1 causes abnormal proliferation in the retina. In the developing vertebrate nervous system, the Notch signaling pathway has been proposed to maintain a population of dividing, uncommitted precursor cells in the ventricular zone. In the ventricular zone of the neural tube, cells proliferate extensively and give rise to postmitotic neurons and glial cells. One interesting possibility is that signaling between the ventricular zone and adjacent regions determines the proportion of cells that becomes post-mitotic versus those that continue to divide. In the retina, activation of Notch clearly keeps cells cycling. The present data suggest that one potential role for lunatic fringe is to keep cells in a hyperproliferative state. By attenuating Notch signaling, lunatic fringe may place the normally highly proliferative neural crest population into a hyperproliferative state. This is likely to be true not only for neural crest cells but for other cell types as well. For example, the neural tube cell number appears increased in experimental embryos (Nellemann, 2001).

Mammalian Fringe homologs and boundary formation

Fringe, a novel signaling protein, has been shown to induce wing margin formation in Drosophila. Fringe shares significant sequence homology and a predicted secondary structure similarity with bacterial glycosyltransferases7. Thus, fringe may control wing development by altering glycosylation of the cell surface and/or secreted molecules. Three murine fringe genes (lunatic fringe, manic fringe and radical fringe) have been characterized. In several tissues fringe expression boundaries coincide with Notch-dependent patterning centers and with Notch-ligand expression boundaries. Ectopic expression of murine manic fringe or radical fringe in Drosophila results in phenotypes that resemble those seen in Notch mutants (Cohen, 1997).

Function of mammalian Fringe homologs in mesodermal segmentation

Segmentation in vertebrates first arises when the unsegmented paraxial mesoderm subdivides to form paired epithelial spheres called somites. The Notch signalling pathway is important in regulating the formation and anterior-posterior patterning of the vertebrate somite. One component of the Notch signaling pathway in Drosophila is the fringe gene, which encodes a secreted signaling molecule required for activation of Notch during specification of the wing margin. Mice homozygous for a targeted mutation of the lunatic fringe (Lfng) gene, one of the mouse homologs of fringe, have defects in somite formation and anterior-posterior patterning of the somites. Somites in the mutant embryos are irregular in size and shape, and their anterior-posterior patterning is disturbed. Marker analysis reveals that in the presomitic mesoderm of the mutant embryos, sharply demarcated domains of expression of several components of the Notch signaling pathway are replaced by even gradients of gene expression. These results indicate that Lfng encodes an essential component of the Notch signaling pathway during somitogenesis in mice (Zhang, 1998).

The gene lunatic fringe encodes a secreted factor with significant sequence similarity to the Drosophila gene fringe. Fringe has been proposed to function as a boundary-specific signaling molecule in the wing imaginal disc, where it is required to localize signaling activity by the protein Notch to the presumptive wing margin. By targeted disruption in mouse embryos, it has been shown that lunatic fringe is likewise required for boundary formation. lunatic fringe mutants fail to form boundaries between individual somites, the initial segmental units of the vertebrate trunk. In addition, the normal alternating rostral-caudal pattern of the somitic mesoderm is disrupted, suggesting that intersomitic boundary formation and rostral-caudal patterning of somites are mechanistically linked by a process that requires lunatic fringe activity. As a result, the derivatives of the somitic mesoderm, especially the axial skeleton, are severely disorganized in lunatic fringe mutants. Taken together, these results demonstrate an essential function for a vertebrate fringe homolog and suggest a model in which lunatic fringe modulates Notch signaling in the segmental plate to regulate somitogenesis and rostral-caudal patterning of somites simultaneously (Evrard, 1998).

The most obvious segments of the vertebrate embryo are the trunk mesodermal somites, which give rise to the segmented vertebral column and the skeletal muscles of the body. Mechanistic insights into vertebrate somitogenesis have recently been gained from observations of rhythmic expression of the avian hairy-related gene (c-hairy1) in chick presomitic mesoderm (PSM), suggesting the existence of a molecular clock linked to somite segmentation. lunatic Fringe (IFng), a vertebrate homolog of the Drosophila fringe gene, is also expressed rhythmically in PSM. The PSM expression of IFng is observed as coordinated pulses of mRNA resembling the expression of c-hairy1. c-hairy1 and IFng expression in the PSM are coincident, indicating that both genes are responding to the same segmentation clock. The genes have been found to differ in their regulation, however; in contrast to c-hairy1, IFng mRNA oscillations require continued protein synthesis, suggesting that IFng could be acting downstream of c-hairy1 in the clock mechanism. In Drosophila, Fringe has been shown to play a role in modulating Notch-Delta signalling, a pathway that in vertebrates has been implicated in defining somite boundaries. These observations place the segmentation clock upstream of the Notch-Delta pathway during vertebrate somitogenesis (McGrew, 1998).

During somitogenesis, cells are recruited to the caudal presomitic mesoderm (PSM) from the primitive streak (and later the tail bud), while somites separate from the rostral end as epithelial cubes. This is a regular process, one somite forming every 2 hours in the mouse; it can be simulated by clock and wavefront models. The chick basic helix-loop-helix transcription factor encoded by c-hairy1 is expressed in dynamic waves in the PSM, undergoing one cycle for each somite formed. This is compatible with an underlying oscillating molecular clock. Lunatic fringe (L-fng) is likely to be one of the regulators of such a clock. Fringe genes regulate the Notch signaling pathway in boundary formation. Of the known mouse genes, only L-fng is expressed in PSM and it is required for somite segmentation and patterning. L-fng is expressed as dynamic, repetitive and complex waves within the mouse PSM. A wave takes 4 hours to complete one cycle and terminates immediately at, and prior to, somite boundary formation. Consecutive waves are temporally but not spatially overlapping, being initiated in the caudal PSM every 2 hours, and therefore offset by one half-cycle. Waves of expression are not associated with cell movement and do not require cell contact for propagation; they appear to reflect a cell-autonomous clock that is synchronous in all PSM cells (Forsberg, 1998).

The metameric organization of the vertebrate trunk is a characteristic feature of all members of this phylum. The origin of this metamerism can be traced to the division of paraxial mesoderm into individual units, termed somites, during embryonic development. Despite the identification well over 100 years ago of somites as the first overt sign of segmentation in vertebrates, the mechanism(s) underlying somite formation remain poorly understood. Recently, however, several genes have been identified that play prominent roles in orchestrating segmentation, including the novel secreted factor lunatic fringe. To gain further insight into the mechanism by which lunatic fringe controls somite development, a thorough analysis of lunatic fringe expression in the unsegmented paraxial mesoderm of chick embryos was carried out. Lunatic fringe is expressed predominantly in somite -II, where somite I corresponds to the most recently formed somite and somite -I corresponds to the group of cells that will form the next somite. In addition, lunatic fringe is expressed in a highly dynamic manner in the chick segmental plate prior to somite formation and lunatic fringe expression cycles autonomously with a periodicity of somite formation. Moreover, the murine ortholog of lunatic fringe undergoes a similar cycling expression pattern in the presomitic mesoderm of somite stage mouse embryos. The demonstration of a dynamic periodic expression pattern suggests that lunatic fringe may function to integrate notch signaling to a cellular oscillator controlling somite segmentation (Aulehla, 1999).

Somitic segmentation provides the framework on which is established the segmental pattern of the vertebrae, some muscles and the peripheral nervous system. Recent evidence indicates that a molecular oscillator, the 'segmentation clock', operates in the presomitic mesoderm (PSM) to direct periodic expression of c-hairy1 and lunatic fringe (l-fng). The identification and characterization of a second avian hairy-related gene, c-hairy2, is reported, that also cycles in the PSM and whose sequence is closely related to the mammalian HES1 gene, a downstream target of Notch signaling in vertebrates. HES1 mRNA is also expressed in a cyclic fashion in the mouse PSM, similar to that observed for c-hairy1 and c-hairy2 in the chick. In HES1 mutant mouse embryos, the periodic expression of l-fng is maintained, suggesting that HES1 is not a critical component of the oscillator mechanism. In contrast, dynamic HES1 expression is lost in mice mutant for Delta1 that are defective for Notch signaling. In order to investigate the relationship between the dynamic HES1 expression in the PSM and the Notch signaling pathway, HES1 expression was examined in Dll1 homozygous mutant mice in which Notch activation is impaired in the PSM. Homozygous null mutants for the Dll1 gene exhibit strong segmentation defects and a severe down-regulation of l-fng expression. The expression of HES1 at E10.5 was compared in wild type, heterozygous and homozygous null mutants by in situ hybridization. The dynamic expression of HES1 in the PSM is maintained in Dll1+/- embryos as shown by the different expression patterns observed in the PSM. In contrast, all Dll1-/- embryos show the same global downregulation of HES1 expression in the PSM (n=7). This observation suggests that HES1 expression in the PSM is dependent on the Notch signaling pathway and suggests that Notch signaling is required for hairy-like genes cyclic expression in the PSM (Jouve, 2000).

Thus HES1 and l-fng dynamic expression are lost in the PSM of Dll1 mutants, in which Notch signaling is disrupted. Various clock outputs appear, therefore, to be severely downregulated when Notch signaling is disrupted. These observations raise the possibility that in addition to being an output of the segmentation clock as previously proposed, the Notch signaling pathway might also be an important component of the oscillator. Notch activation upon ligand binding involves a proteolytic cleavage liberating the intracytoplasmic domain (NICD), which translocates into the nucleus where together with Su(H)/RBPjk it activates the transcription of genes such as HES1 in vertebrates. The observations in the Dll1-/- mice indicate that HES1 is downstream of the Notch pathway in the PSM. Since c-hairy1 and c-hairy2 share a high similarity in their sequence and in their expression patterns to HES1, they are also likely targets of Notch signaling in the chick PSM. A direct regulation of c-hairy1 and c-hairy2 expression by oscillating Notch activation would explain why c-hairy1 expression is insensitive to cycloheximide, since protein synthesis is not required for transduction of the Notch signal. To achieve oscillations of Notch signaling, the activity of the pathway would need to be modulated by a feedback mechanism. However, known output events resulting from Notch signaling are transcriptional regulation of target genes and the clock is partly independent of protein synthesis. Notch1 and Delta1 are present along the whole presomitic mesoderm and could generate constitutive activation of the pathway in the tissue. The rhythmic modification of this activation could, in principle, be achieved by the periodic expression of l-fng (Jouve, 2000).

Boundary formation plays a central role in differentiating the flanking regions that give rise to discrete tissues and organs during early development. Mechanisms by which a morphological boundary and tissue separation are regulated have been studied by examining chicken somite segmentation as a model system. By transplanting a small group of cells taken from a presumptive border into a non-segmentation site, a novel inductive event has been found where posteriorly juxtaposed cells to the next-forming border instruct the anterior cells to become separated and epithelialized. The molecular mechanisms underlying these interactions was further studied by focusing on Lunatic fringe, a modulator of Notch signaling, which is expressed in the region of the presumptive boundary. By combining DNA in ovo electroporation and embryonic transplantation techniques, a sharp boundary of Lunatic fringe activity has been ectopically made in the unsegmented paraxial mesoderm and a fissure formed at the interface has been observed. In addition, a constitutive active form of Notch mimics this instructive phenomenon. These suggest that the boundary-forming signals emanating from the posterior border cells are mediated by Notch, the action of which is confined to the border region by Lunatic fringe within the area where mRNAs of Notch and its ligand are broadly expressed in the presomitic mesoderm (Sato, 2002).

In the anterior end of the unsegmented paraxial mesoderm (presomitic mesoderm: PSM), an expression boundary of genes exists, including MesP2, a novel mouse gene expressed in the presegmented mesoderm and essential for segmentation initiation. This expression border, which coincides with the next border being formed, is established prior to a morphological change. The segmental patterns of these genes are thought to be regulated by a 'segmentation clock', first demonstrated by wavy and cyclic expression of c-hairy1. Thus, the segmentation clock operates in the continuous young PSM to establish the segmental patterns of gene expression in the anterior PSM, which eventually implements a morphological fissure formation. Both clock and segmentation genes are tightly related to Notch signaling, as revealed mainly by recent knockout and mutant studies: an animal where Notch signaling is (at least in part) deficient displays perturbed patterns of cyclic and segmental expression of genes in PSM, and also shows its consequent malformation of segmented structures later in development. In general, studies using mutants or knockout animals unveil the 'first stage' where the gene of concern is essential during development. However, if a given gene plays a role in the fissure formation as well as at earlier steps of segmentation, it would be difficult to distinguish between them. This may be the reason why the molecular mechanisms underlying the fissure formation have been poorly addressed (Sato, 2002).

A novel inductive event is described that takes place when a segmentation fissure forms. In this event posterior border cells located immediately posterior to the next forming boundary instruct the anterior ones. Molecular mechanisms underlying these events are addressed by focusing on Notch signals where Lunatic fringe (Lfng) is involved. Lfng is a modulator of the Notch receptor with glycosyltransferase activity, and is expressed in a region coinciding with the segmentation border in PSM. By combining DNA in ovo electroporation with embryological manipulations to make an ectopic boundary of a transgene activity in PSM, it was found that Notch signals play major roles in the formation of a fissure. A model is presented in which specific localization of Lfng determines the site of Notch action relevant to the morphological segmentation (Sato, 2002).

A molecular oscillator regulates the pace of vertebrate segmentation. The oscillator (clock) controls cyclic initiation of transcription in the unsegmented presomitic mesoderm (PSM). An evolutionarily conserved 2.3 kb region has been identified in the murine Lunatic fringe (Lfng) promoter that drives periodic expression in the PSM. This region includes conserved blocks required for enhancing and repressing cyclic Lfng transcription, and to prevent continued expression in formed somites. Dynamic expression in the cycling PSM is lost in the total absence of Notch signaling, and Notch signaling acts directly via CBF1/RBP-Jkappa binding sites to regulate Lfng. These results are consistent with a model in which oscillatory Notch signaling underlies the segmentation clock and directly activates and indirectly represses Lfng expression (Morales, 2002).

To search for putative regulatory elements controlling cyclic transcription, it was asked whether the 2.3 kb Lfng cyclic promoter includes evolutionarily conserved cis-acting regulatory sequences. An equivalent promoter fragment was isolated from the human Lfng gene, and whether it also drives cyclic expression of a lacZ reporter gene was tested in transgenic embryos. Like its mouse counterpart, this construct expresses lacZ transcripts dynamically in the PSM, and all cycling stages are represented. Thus, the human Lfng fragment contains a clock response element which can operate in the context of a mouse embryo PSM, indicating that the clock machinery is functionally conserved between human and mouse. Comparing the DNA sequences of the mouse and human Lfng promoter fragments reveals three blocks of high sequence similarity. Block A (109 bp mouse/110 bp human) is located most 5', and shows 89% sequence identity between the two species. Block B (284 bp/292 bp) is 77% identical between the two species. Block C, the most 3' block (200 bp/208 bp), displays 63% sequence identity (Morales, 2002).

Since none of the individual conserved blocks encode an autonomous clock element, it was asked whether such an element might lie between the conserved blocks. Deleting 600 bp between blocks A and B from the 2.3 kb cyclic promoter leads to much reduced expression throughout the PSM. Dynamic expression in the posterior PSM is very weak, although it is still evident in the anterior PSM. Thus, this region (X) may contain a general enhancer of transcription in the PSM, as well as an enhancer required for the cyclic activity in the caudal PSM (Morales, 2002).

Unexpectedly, cyclic Lfng expression is differently regulated within the cycling regions. Regulatory block A is essential for cyclic expression in the anterior PSM, whereas region X is required for cyclic expression in the posterior PSM. It is possible that the boundary between these two zones of cyclic Lfng expression corresponds to the determination wavefront, and that it is regulated by FGF8 signaling. Most likely, FGF8 acts indirectly to modulate target gene specificity of the clock and perhaps, its periodicity (Morales, 2002).

The gene networks controlling Lfng expression continue to evolve as cells enter the differentiating region of the PSM, leading to distinct regulatory requirements in rostral and caudal presomite compartments. Lfng is initially expressed in a somite broad domain and becomes progressively restricted to the rostral compartment. Reporter transcripts suggest that different sequences regulate expression within presomites -II and -I: block B in each rostral compartment, and block A probably in the caudal ones (Morales, 2002).

In summary, these findings offer persuasive evidence that oscillatory Notch signaling underlies the segmentation clock. Notch signaling activates the Lfng promoter directly, is necessary for its expression, and the Lfng promoter integrates positive and negative inputs to generate oscillatory expression. Since Lfng is itself a modulator of Notch signaling and required for segmentation, it is most likely to participate in the clock itself (Morales, 2002).

Somitogenesis requires a segmentation clock and Notch signaling. Lunatic fringe expression in the presomitic mesoderm (PSM) cycles in the posterior PSM, is refined in the segmenting somite to the rostral compartment, and is required for segmentation. Distinct cis-acting regulatory elements have been identified for each aspect of Lfng expression. Fringe clock element 1 (FCE1) represents a conserved 110 bp region that is necessary to direct cyclic Lfng RNA expression in the posterior PSM. Mutational analysis of E boxes within FCE1 indicates a potential interplay of positive and negative transcriptional regulation by cyclically expressed bHLH proteins. A separable Lfng regulatory region directs expression to the prospective rostral aspect of the condensing somite. These independent Lfng regulatory cassettes advance a molecular framework for deciphering somite segmentation (Cole, 2002).

Hes7, a bHLH gene essential for somitogenesis, displays cyclic expression of mRNA in the presomitic mesoderm (PSM). Hes7 protein is also expressed in a dynamic manner, which depends on proteasome-mediated degradation. Spatial comparison reveals that Hes7 and Lunatic fringe (Lfng) transcription occurs in the Hes7 protein-negative domains. Furthermore, Hes7 and Lfng transcription is constitutively up-regulated in the absence of Hes7 protein and down-regulated by stabilization of Hes7 protein. Thus, periodic repression by Hes7 protein is critical for the cyclic transcription of Hes7 and Lfng, and this negative feedback represents a molecular basis for the segmentation clock (Bessho, 2003).

The temporal expression profiles are thought to occur as follows: Hes7 transcription leads to accumulation of Hes7 mRNA and Hes7 protein but is turned off as soon as Hes7 protein is accumulated, but Hes7 mRNA persists for a while. Recent studies have revealed that oscillatory expression of Lfng is controlled at the transcriptional level. Among several elements in the Lfng promoter, the region 2 or the region A contains two E-boxes, which are critical for the cyclic expression. The data indicate that Hes7, which represses transcription via E-boxes, is likely to regulate the cyclic expression of Lfng through the E-boxes in the region 2/A (Bessho, 2003).

Lfng represses its own expression through modulation of the Notch pathway in chick; the negative feedback loop of Lfng is a molecular basis for the segmentation clock. It is likely that Lfng also constitutes a negative feedback loop in mouse, because the expression domains of the lacZ gene, which is knocked into the Lfng allele, become wider in the Lfng-null mutant mouse. However, although Hes7 is required for the dynamic expression of Lfng, Lfng is not required for the dynamic expression of Hes7. Thus, Hes7 is an upstream regulator of Lfng oscillation, and the negative feedback loop of Lfng could be involved in refinement of cyclic expression or keeping accuracy of the 2-h cycle. It was recently reported that Wnt signaling is also involved in the segmentation clock. Interestingly, in hypomorphic mutants for Wnt3a, Lfng oscillation is lost, suggesting the cross-talk between Wnt and Notch signaling. However, the relationship between Hes7 oscillation and Wnt signaling remains to be determined (Bessho, 2003).

To analyze requirements for Notch signalling in patterning the paraxial mesoderm, transgenic mice were generated that express in the paraxial mesoderm a dominant-negative version of Delta1. Transgenic mice with reduced Notch activity in the presomitic mesoderm as indicated by loss of Hes5 expression were viable and displayed defects in somites and vertebrae consistent with known roles of Notch signalling in somite compartmentalization. In addition, these mice showed with variable expressivity and penetrance alterations of vertebral identities resembling homeotic transformations, and subtle changes of Hox gene expression in day 12.5 embryos. Mice that carried only one functional copy of the endogenous Delta1 gene also showed changes of vertebral identities in the lower cervical region, suggesting a previously unnoticed haploinsufficiency for Delta1. Likewise, in mice carrying a null allele of the oscillating Lfng gene, or in transgenic mice expressing Lfng constitutively in the presomitic mesoderm, vertebral identities were changed and numbers of segments in the cervical and thoracic regions were reduced, suggesting anterior shifts of axial identity. Together, these results provide genetic evidence that precisely regulated levels of Notch activity as well as cyclic Lfng activity are critical for positional specification of the anteroposterior body axis in the paraxial mesoderm (Cordes, 2004).

Transformations of vertebral identities, anterior shifts of Hoxb6 expression, and of the position of both fore and hind limb buds were detected in mice lacking Lfng function or expressing Lfng constitutively. An apparent anterior shift of Hoxb6 expression in Lfng mutant embryos could also be expected if fewer segments were generated in the prospective cervical region, whereas the absolute position of the anterior Hoxb6 expression border along the anteroposterior body axis was maintained. Recent models of somite segmentation suggest that the interaction of graded FGF or WNT signals with the segmentation clock generates the periodic somite pattern. Conceptually, increasing the steepness of the gradient or slowing the periodicity of the clock would lead to fewer segments, which in either case would be larger. Thus, if the loss of Lfng would affect the clock (output) and fewer segments would be formed in the prospective cervical region, they should be larger than normal. However, the five cervical segments in Lfng mutant embryos occupied essentially the same space as the anterior five segments in WT embryos, strongly supporting the idea that the rostral Hoxb6 expression border is indeed shifted anteriorly. The positions of the fore and hind limb buds are invariant in WT embryos and correspond to the transition between the cervical and thoracic, and lumbar and sacral regions, respectively. Their anterior shift suggests homeotic transformations throughout the trunk region along the anterior posterior body axis that led to an overall reduction of the number of segments in the trunk (Cordes, 2004).

Different types of oscillations in Notch and Fgf signaling regulate the spatiotemporal periodicity of somitogenesis

Somitogenesis is controlled by cyclic genes such as Notch effectors and by the wave front established by morphogens such as Fgf8, but the precise mechanism of how these factors are coordinated remains to be determined. This study shows that effectors of Notch and Fgf pathways oscillate in different dynamics and that oscillations in Notch signaling generate alternating phase shift, thereby periodically segregating a group of synchronized cells, whereas oscillations in Fgf signaling released these synchronized cells for somitogenesis at the same time. These results suggest that Notch oscillators define the prospective somite region, while Fgf oscillators regulate the pace of segmentation (Niwa, 2011).

Somite formation occurs periodically by segmentation and maturation of a block of cells in the anterior presomitic mesoderm (PSM). It is thought that the pace of segmentation depends on the clock controlled by cyclic genes such as Notch signaling molecules, while the timing of maturation depends on the wave front established by morphogens such as Fgf8. However, Notch signaling oscillations become slower than the pace of segmentation as the oscillations are propagated anteriorly, raising the question of whether such a slowing oscillator regulates the segmentation pace. Furthermore, Fgf signaling seems to sweep back at a steady speed as the PSM grows, raising another question of whether the release from Fgf signaling occurs at different times between the anterior and posterior cells even in the same prospective somites (Niwa, 2011).

In the mouse PSM, Hes7 is expressed in an oscillatory manner and induces oscillatory expression of Lunatic fringe (Lfng), a modulator of Notch signaling. Lfng oscillations in turn lead to cyclic formation of the Notch intracellular domain (NICD), an active form of Notch, which then periodically induces expression of Mesp2, an essential gene for the segmentation and rostro-caudal patterning of each somite. Mesp2 expression depends on NICD and Tbx6 and occurs after the release from Fgf and Wnt signaling in the whole S-1 region, a group of cells that forms a prospective somite. High-resolution in situ hybridization demonstrated that S-1 cells synchronously exhibit nuclear dots of Mesp2 signals, indicating synchronous initiation of Mesp2 transcription in the whole S-1 region. In Lfng-null mice, which have segmentation defects, Mesp2 expression becomes randomized in S-1 cells, displaying a salt-and-pepper pattern. These results suggest that synchronous Mesp2 expression in S-1 cells is important for somite formation. However, how slowing Notch signaling oscillators and steadily regressing Fgf and Wnt signaling regulate periodic and synchronous Mesp2 expression in S-1 cells remains to be determined (Niwa, 2011).

This study found that Notch and Fgf signaling effectors oscillate with different dynamics and that oscillations in Notch signaling periodically segregate a group of synchronized cells, whereas oscillations in Fgf signaling release these synchronized cells for somitogenesis at the same time. These results suggest that Notch oscillators define the prospective somite region, while Fgf oscillators regulate the pace of segmentation, thereby linking the clock and the wave front (Niwa, 2011).

Modifying transcript lengths of cycling mouse segmentation genes

Regular production of somites, precursors of the axial skeleton and attached muscles is controlled by a molecular oscillator, the segmentation clock, which drives cyclic transcription of target genes in the unsegmented presomitic mesoderm (PSM). The clock is based on a negative feedback loop which generates pulses of transcription that oscillate with the same periodicity as somite formation. Mutants in several oscillating genes including the Notch pathway gene Lunatic fringe (Lfng) and the Notch target Hes7, result in defective somitogenesis and disorganised axial skeletons. Both genes encode negative regulators of Notch signalling output, but it is not yet clear if they are just secondary clock targets or if they encode components of a primary, pacemaker oscillator. This study tries to identify components in the primary oscillator by manipulating delays in the feedback circuitry. Recombinant mice were characterized in which Lfng and Hes7 introns are lengthened in order to delay mRNA production. Lengthening the third Hes7 intron by 10 or 20 kb disrupts accurate RNA splicing and inactivates the gene. Lfng expression and activity is normal in mice whose Lfng is lengthened by 10 kb, but no effects on segmentation are evident. These results are discussed in terms of the relative contributions of transcriptional and post-transcriptional delays towards defining the pace of segmentation, and of alternative strategies for manipulating the period of the clock (Stauber, 2012).

Posterior skeletal development and the segmentation clock period are sensitive to Lfng dosage during somitogenesis

The segmental structure of the axial skeleton is formed during somitogenesis. During this process, paired somites bud from the presomitic mesoderm (PSM), in a process regulated by a genetic clock called the segmentation clock. The Notch pathway and the Notch modulator Lunatic fringe (Lfng) play multiple roles during segmentation. Lfng oscillates in the posterior PSM as part of the segmentation clock, but is stably expressed in the anterior PSM during presomite patterning. Mice lacking overt oscillatory Lfng expression in the posterior PSM (LfngFCE) exhibit abnormal anterior development but relatively normal posterior development. This suggests distinct requirements for segmentation clock activity during the formation of the anterior skeleton (primary body formation), compared to the posterior skeleton and tail (secondary body formation). To build on these findings, an allelic series was created that progressively lowers Lfng levels in the PSM. Interestingly, it was found that further reduction of Lfng expression levels in the PSM does not increase disruption of anterior development. However tail development is increasingly compromised as Lfng levels are reduced, suggesting that primary body formation is more sensitive to Lfng dosage than is secondary body formation. Further, while low levels of oscillatory Lfng in the posterior PSM were found to be sufficient to support relatively normal posterior development, the period of the segmentation clock is increased when the amplitude of Lfng oscillations is low. These data support the hypothesis that there are differential requirements for oscillatory Lfng during primary and secondary body formation and that posterior development is less sensitive to overall Lfng levels. Further, they suggest that modulation of the Notch signaling by Lfng affects the clock period during development (Williams, 2014).

Fringe proteins modulate Notch-ligand and interactions to specify signaling states

The Notch signaling pathway consists of multiple types of receptors and ligands, whose interactions can be tuned by Fringe glycosyltransferases. A major challenge is to determine how these components control the specificity and directionality of Notch signaling in developmental contexts. This study analyzed same-cell (cis) Notch-ligand interactions for Notch1 (see Drosophila Notch), Dll1, and Jag1 (see Drosophila Delta), and their dependence on Fringe protein expression in mammalian cells. Dll1 and Jag1 were found to cis-inhibit Notch1, and Fringe proteins modulate these interactions in a way that parallels their effects on trans interactions. Fringe similarly modulated Notch-ligand cis interactions during Drosophila development. Based on these and previously identified interactions, it was shown how the design of the Notch signaling pathway leads to a restricted repertoire of signaling states that promote heterotypic signaling between distinct cell types, providing insight into the design principles of the Notch signaling system, and the specific developmental process of Drosophila dorsal-ventral boundary formation (LeBon, 2014: PubMed).

Function of mammalian Fringe homologs in limb development

The gene Radical fringe positions the apical ectodermal ridge at the dorsoventral boundary of the vertebrate limb. R-fng is homologous to Drosophila fng. Weak but significant similarity of Fng proteins to the Lex1 family of biosynthetic enzymes suggests that R-fng may function as a glycosyltransferase. This homology indicates that R-fng may operate through the recognition, and perhaps modification, of carbohydrate moieties linked to either extracellular proteins or lipids. R-fng is expressed in the dorsal ectoderm and apical ectodermal ridge (AER) prior to the expression of Fgf-8, a gene thought to play a role in the formation of the AER. Abnormal limb phenotypes consisting of AER-like structures are observed in 16% of embryos infected with a replication-competent retroviral vector containing R-fng, suggesting that R-fng misexpression perturbs the normal formation of the AER. While Wnt-7a, expressed in the dorsal ectoderm has no effect on R-fng expression, Engrailed-1 (see Drosophila Engrailed), normally expressed in the ventral ectoderm, strongly represses R-fng and Wnt-7a, suggesting that En-1 regulates dorsal-ventral polarity of the limb and the positioning of the AER. This En-1 role contrasts with the role taken by Engrailed in Drosophila: the regulation of posterior compartment identity. It is suggested that the AER develops at the interface of tissue that is expressing high levels of R-fng adjacent to tissue that does not express R-fng. Because R-fng is not normally expressed on the ventral side of the limb, ectopic expression of R-fng here creates new boundaries that result in additional AERs. Serrate-2 (see Drosophila Serrate) is expressed in the AER from the earliest stages of its formation through at least stage 26. Chicken Notch-1 is also expressed in the AER. Thus R-fng, like its Drosophila counterpart, may act upstream of Notch signaling (Laufer, 1997 and Rodriguez-Esteban, 1997).

Function of mammalian Fringe homologs in tooth development

Teeth develop as epithelial appendages, and their morphogenesis is regulated by epithelial-mesenchymal interactions and conserved signaling pathways common to many developmental processes. A key event during tooth morphogenesis is the transition from bud to cap stage, when the epithelial bud is divided into specific compartments distinguished by morphology as well as gene expression patterns. The enamel knot, a signaling center, forms and regulates the shape and size of the tooth. Mesenchymal signals are necessary for epithelial patterning and for the formation and maintenance of the epithelial compartments. The expression of Notch pathway molecules was studied during the bud-to-cap stage transition of the developing mouse tooth. Lunatic fringe expression is restricted to the epithelium, where it forms a boundary flanking the enamel knot. The Lunatic fringe expression domains overlapped only partly with the expression of Notch1 and Notch2, which are coexpressed with Hes1. The regulation of Lunatic fringe and Hes1 was examined in cultured explants of dental epithelium. The expression of Lunatic fringe and Hes1 depend on mesenchymal signals and both are positively regulated by FGF-10. BMP-4 antagonizes the stimulatory effect of FGF-10 on Lunatic fringe expression but has a synergistic effect with FGF-10 on Hes1 expression. Recombinant Lunatic fringe protein induced Hes1 expression in the dental epithelium, suggesting that Lunatic fringe can act also extracellularly. Lunatic fringe mutant mice do not reveal tooth abnormalities, and no changes were observed in the expression patterns of other Fringe genes. It is concluded that Lunatic fringe may play a role in boundary formation of the enamel knot and that Notch-signaling in the dental epithelium is regulated by mesenchymal FGFs and BMP (Mustonen, 2002).

Fringe and folliculogenesis

Notch genes are expressed in developing mammalian ovarian follicles. Lunatic fringe is an important regulator of Notch signaling. Radical fringe and lunatic fringe are expressed in the granulosa cells of developing follicles. Lunatic fringe null female mice are infertile. Histological analysis of the lunatic fringe-deficient ovary demonstrates aberrant folliculogenesis. Furthermore, oocytes from these mutants do not complete meiotic maturation. This is a novel observation because this is the first report describing a meiotic defect that results from mutations in genes that are expressed in the somatic granulosa cells and not the oocytes. This represents a new role for the Notch signaling pathway and lunatic fringe in mammalian folliculogenesis (Hahn, 2005).

Function of mammalian Fringe homologs in brain segmentation

Three mammalian fringe-related family members have been cloned and characterized: Manic, Radical and Lunatic Fringe. Expression studies in mouse embryos support a conserved role for mammalian Fringe family members in participation in the Notch signaling pathway leading to boundary determination during segmentation. This model is proposed based on unpublished results of V. Panin, V. Papayannopoulos, R. Wilson and K. D. Irvine quoted in Johnston, 1997. Drosophila Fringe works in a positive feedback loop by modulating the signaling between the two Notch ligands at the D-V boundary of the wing imaginal disc. Delta expressed in ventral cells binds and activates Notch in dorsal cells, and Serrate expressed in dorsal cells binds and activates Notch in ventral cells. An important consequence of the boundary mediated signaling is that the initial asymmetric boundary between dorsal and ventral cells is translated by reciprocal activation of Notch at the D-V compartment boundary into symmetrical expression of downstream targets and the expression of wingless as the organizing signaling molecule (Johnston, 1997).

Manic and Lunatic expression domains are suggestive of a role for mammalian Fringe in demarcation of segmental boundaries observed during anterior-posterior patterning of the hindbrain and somites. Both Manic and Lunatic are expressed in anterior-posterior stripes in the hindbrain prior to the formation of morphologically recognized rhombomeric boundaries. In embryos at the 5 somite stage, Manic is expressed in pre-rhombomere (pre-r) 3, while Lunatic is expressed in pre-r3 and pre-r5. In embryos with 12-13 somites, r3 and r5 expression of Manic and Lunatic is observed. The restricted pattern of Manic and Lunatic Fringe to r3 and r5 creates a juxtaposition of Fringe-expressing and non-expressing cells between even and odd numbered rhombomeres, an observation consistent with the demarcation and establishment of boundaries. Of the three mouse Fringe genes, Lunatic is solely expressed in the presomitic mesoderm in a striking pattern that appears to demarcate the formation of intersomitic boundaries and the appearance of separated somites. The expression of Lunatic is dynamic, reflecting the rostral-caudal wave of somitic developmetal progression. In the posterior presomitic mesoderm, Lunatic is expressed in a broad swath of cells. As somitogenesis proceeds, this band of expression becomes narrower, paralleling the cell condensation. Stripes of expression appear to correspond to what will form the posterior half of a somite. The initial broad posterior band of Lunatic expression is found in the midst of Dll1-expressing cells (Dll1 is the mammalian homolog of Drosophila Delta). Lunatic and Dll1 expressions become progressively more refined. Lunatic expression is localized to the posterior half of presumptive somitomere 2, while Dll1 expression is localized to the posterior border of this same somitomere. Dll1 expression also extends caudally throughout the adjoining presomitic mesoderm. Once the somite has formed, transcript levels of Lunatic decline rapidly. Likewise Radical expression is correlated with the rostral-caudal maturation of the spinal cord and the differentiation of neuronal population. In mammalian cells, Drosophila Fringe and the mouse Fringe proteins are subject to posttranslational regulation at the levels of differential secretion and proteolytic processing. When misexpressed in the developing Drosophila wing imaginal disc the mouse Fringe genes exhibit conserved and differential effects on boundary determination. In some cases misexpression of a mammalian Fringe in Drosophila eliminates wing margin tissue and Wingless expression (Johnston, 1997).

Lunatic fringe promotes the lateral inhibition of neurogenesis

Previous studies have identified roles of the modulation of Notch activation by Fringe homologues in boundary formation and in regulating the differentiation of vertebrate thymocytes and Drosophila glial cells. This study investigated the role of Lunatic fringe (Lfng) expression during neurogenesis in the vertebrate neural tube. In the zebrafish hindbrain, Lfng is expressed by progenitors in neurogenic regions and downregulated in cells that have initiated neuronal differentiation. Lfng is required cell autonomously in neural epithelial cells to limit the amount of neurogenesis and to maintain progenitors. By contrast, Lfng is not required for the role of Notch in interneuronal fate choice, which is mediated by Notch1a. The expression of Lfng does not require Notch activity, but rather is regulated downstream of proneural genes that are widely expressed by neural progenitors. These findings suggest that Lfng acts in a feedback loop downstream of proneural genes, which, by promoting Notch activation, maintains the sensitivity of progenitors to lateral inhibition and thus limits further proneural upregulation (Nikolaou, 2009).

Domain-specific control of neurogenesis achieved through patterned regulation of Notch ligand expression

Homeodomain (HD) transcription factors and components of the Notch pathway [Delta1 (Dll1), Jagged1 (Jag1) and the Fringe (Fng) proteins] are expressed in distinct progenitor domains along the dorsoventral (DV) axis of the developing spinal cord. However, the internal relationship between these two regulatory pathways has not been established. This report shows that HD proteins act upstream of Notch signalling. Thus, HD proteins control the spatial distribution of Notch ligands and Fng proteins, whereas perturbation of the Notch pathway does not affect the regional expression of HD proteins. Loss of Dll1 or Jag1 leads to a domain-specific increase of neuronal differentiation but does not affect the establishment of progenitor domain boundaries. Moreover, gain-of-function experiments indicate that the ability of Dll1 and Jag1 to activate Notch is limited to progenitors endogenously expressing the respective ligand. Fng proteins enhance Dll1-activated Notch signalling and block Notch activation mediated by Jag1. This finding, combined with the overlapping expression of Fng with Dll1 but not with Jag1, is likely to explain the domain-specific activity of the Notch ligands. This outcome is opposite to the local regulation of Notch activity in most other systems, including the Drosophila wing, where Fng co-localizes with Jagged/Serrate rather than Dll/Delta, which facilitates Notch signalling at regional boundaries instead of within domains. The regulation of Notch activation in the spinal cord therefore appears to endow specific progenitor populations with a domain-wide autonomy in the control of neurogenesis and prevents any inadequate activation of Notch across progenitor domain boundaries (Marklund, 2010).

HD proteins, which are implicated in pattern formation, as well as several components of the Notch pathway, exhibit specific expression domains along the DV axis of the neural tube, but their internal relationship has not been determined. This study shows that the patterned expression of Notch ligands and Fng genes are controlled by HD transcription factors. Loss of Nkx6.1 led to a ventral expansion of Jag1 expression, accompanied by a reduction of Dll1, Lfng and Mfng expression. Conversely, forced ectopic expression of Nkx6.1 suppressed the expression of Jag1 and induced that of Dll1, Lfng and Mfng. Perturbation of Dbx1 caused ectopic Jag1 expression in the ‘p0 domain’ and a concomitant reduction of Dll1, Lfng and Mfng expression, whereas overexpression of Dbx1 had the opposite effect. By contrast, perturbation of Dll1 and Jag1 did not alter the expression patterns of the Nkx6.1 and Dbx1 proteins, and there was no obvious increase of cell intermingling at progenitor domain boundaries. In conclusion, these findings suggest a mechanism in which HD proteins act upstream of Notch ligands and Fng gene expression, resulting in the establishment of discrete progenitor domains with co-localized expression of Dll1 and Fng, whereas regions expressing Jag1 are devoid of Fng protein expression (Marklund, 2010).

fringe: Biological Overview | Regulation | Developmental Biology | Effects of Mutation | References

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.