In Drosophila melanogaster the Enhancer of split-Complex [E(spl)-C] consists of seven highly related genes encoding basic helix-loop-helix (bHLH) repressors, intermingled with four genes that belong to the Bearded (Brd) family. Both gene classes are targets of the Notch signalling pathway. The Achaete-Scute-Complex [AS-C] comprises four genes encoding bHLH activators. Focussing on Diptera and the Hymenoptera Apis mellifera, the question arose how these complexes evolved with regard to gene number in the evolution of insects. In Drosophilids, both gene complexes are highly conserved, spanning roughly 40 million years of evolution. However, in species more diverged, like Anopheles or Apis, dramatic differences are found. Here, the E(spl)-C consists of one bHLH (mß) and one Brd family member (malpha) in a head to head arrangement. Interestingly in Apis but not in Anopheles, there are two more E(spl) bHLH like genes within 250 kb, which may reflect duplication events in the honeybee that occurred independently of those in Diptera. The AS-C may have arisen from a single sc/l'sc like gene which is well conserved in Apis and Anopheles and a second ase like gene that is highly diverged, however, located within 50 kb. Thus, E(spl)-C and AS-C presumably evolved by gene duplication to the current complex composition in Drosophilids in order to govern the accurate expression patterns typical for these highly evolved insects. The ancestral ur-complexes, however, consisted most likely of just two genes: (1) E(spl)-C contains one bHLH member of mß type and one Brd family member of malpha type, and (2) AS-C contains one sc/l'sc and a highly diverged ase like gene (Schlatter, 2005).

In total, 12 genes in D. melanogaster are known to encode Hairy/E(spl)-like proteins, i.e. bHLH proteins that also have the orange domain and a WRPW-type Gro-binding motif. Apart from the seven E(spl) bHLH proteins, these include Hairy, Deadpan, Side, Hey and Her. Moreover, there is similarity to Stich1/Sticky, which has a bHLH and an orange domain but not the typical Gro-binding motif. Since the number of E(spl) bHLH genes is not conserved in honeybee and mosquito, it was interesting to ask whether all the other genes were present. The Ensembl database was searched with the respective D. melanogaster protein sequences: orthologs were found of all genes except Her in both species. However, most of the predictions are incomplete. It is known from D. melanogaster that these genes contain introns, which complicates the search for potential coding sequences within genomic DNA. Thus, the protein sequence predictions are uncertain. With the sole exception of Dpn, all the proteins are better conserved between Drosophila and Anopheles than between Drosophila and Apis, confirming the evolutionary relationship. The best conserved proteins are Hey and Hairy. The Hey orthologs are 76% identical between Drosophila and Anopheles and 66% between Drosophila and Apis and the Hairy orthologs between 72% and 65%, respectively. Less conservation is found for Side, Dpn and Stich1 (62%/57% Side identity, 57%/59% Dpn identity and 60%/57% Stich1 identity), comparing fly with mosquito and honeybee, respectively. All proteins share the bHLH and orange domains. The WRPW motif of Hairy, Dpn and Side as well as the YRPW motif of Hey is present in the orthologs (Schlatter, 2005).

Extensive genome analyses in recent years revealed that there are not many examples of large gene complexes that are widely conserved. Prominent examples are the HOX (homeobox) complexes, which contain homeotic genes in Drosophila. HOX complexes are well conserved in metazoans despite some variations in gene number. HOX-genes encode regulatory proteins with specific individual functions and mutations affect different aspects of the body plan. Not surprisingly, it is almost only the homeodomain, which serves as sequence-specific DNA binding motif, that is conserved amongst different species. In contrast, similarity among bHLH proteins encoded by the E(spl)-C extends over the entire length, even within the same species, indicating rather recent duplication events. The D. melanogaster proteins M8/M5 and Mß/M3 are most similar with over 70% identity, whereas Mdelta is the most diverged. However, Mdelta still shares at least 50% identity with other E(spl) bHLH protein members. More interesting is the analysis of the overall similarity among these proteins. Here, any one of the proteins is compared with the other six and the result is averaged. Clearly, Mß (73%/64%, similarity/identity) closely followed by Mgamma (72%/63%) is most similar to all others, whereas Mdelta (66%/55%) shows the lowest values. One interpretation might be that the different bHLH genes evolved by duplication out of mß or mgamma. Remarkably, these two bHLH proteins in addition to M3 are the best conserved in the three Drosophila species. It is postulated that these are the most ancient proteins with the most general function and, therefore, the highest selection pressure. This hypothesis is supported by the finding that mß has the most general expression pattern from which the others can be derived by a decrease of gene activity. The conspicuous conservation of M3 might hint to an important function during egg development as this gene is also expressed maternally. The high degree of conservation of all E(spl) bHLH orthologous proteins in Drosophilids, which is clearly higher than the similarity within this protein family in D. melanogaster, indicates specific and non-redundant roles during development. Some of these functions have been identified in the past. It is conceivable that regulatory sequences were not duplicated or evolved more rapidly so that now highly dynamic expression patterns of these genes are found (Schlatter, 2005).

mß appears to be the ancestral bHLH gene of the E(spl)-C in Drosophilids based on its great similarity with all the other bHLH proteins. This assumption is strongly supported by the sequence conservation of the E(spl) bHLH proteins in A. gambiae and A. mellifera. The single E(spl) bHLH protein encoded by the mosquito genome has the highest identity to Mß. The genome of honeybee contains three prospective genes that encode proteins most highly related to E(spl) D.m.Mß and D.m.Mgamma. All three are clustered within a single sequence contig, albeit they span a large segment of about 250 kb, whereas the whole E(spl)-C in D. melanogaster comprises roughly 50 kb. Despite the fact that two of these genes possess introns just within the bHLH domain and at positions close to the ones found in the D. melanogaster genes dpn, hairy or Her , the amino acid sequence similarity classifies them clearly as E(spl) bHLH proteins. A comparison of Anopheles and Apis proteins reveals, that the presumptive Mß homologs have highest similarity (83%) and identity (76%), whereas the protein that classified as A.m.Mgamma is just 70% similar and 66% identical to A.g.Mß (Schlatter, 2005).

In Drosophilids, malpha is located close to mß and is transcribed in the opposite direction (head to head). This arrangement is likewise found in Anopheles and Apis. Notably, A.m.malphais next to A.m.mß, whereas the two Apis A.m.mgamma and A.m.mß' genes are much further apart. This arrangement is thought to be very ancient. In the beetle Tribolium, which on the tree of evolution is found even more deeply rooted (~300 Myr to Dipterans), two similar genes coding for Mß-like proteins (~65% and ~67% identity to D.m.Mß) are found and one is within ~18 kb of a gene coding for an Malpha-like protein (~52% identity to D.m.Malpha). It is postulated that the ur-complex consists of these two ancestral genes, malpha and mß. It is intriguing that they belong to the two different classes of Notch-responsive genes in the E(spl)-C, the bHLH and the Brd-class. In the fly, malpha and bHLH genes are similarly expressed. It is not unlikely that they share common regulatory elements that could explain their co-segregation in the process of evolution (Schlatter, 2005).

In conclusion this study found that both E(spl)-C and AS-C expanded rather recently because they are only present in their current complex structures in Drosophilids. In Apis and in Anopheles, very similar arrangements are found indicative of an ancient ur-complex. The E(spl)-C seems to have evolved from two genes, one HES-like and one Brd-like that are arranged in a head to head orientation. Both types of genes are responsive to Notch signalling in Drosophila. The data suggest that the most ancient genes are E(spl) bHLH mß and E(spl) malpha from which the other E(spl)-C genes derived by duplication and subsequent change. Moreover, an E(spl) ur-complex is likewise detected in Tribolium castaneum that belongs to the order Coleoptera. In Drosophila the complex also gained unrelated genes like m1 and gro. The latter is highly conserved, however, located at different genomic positions. Whereas in Anopheles the ur-complex seems to exist in its original form, two additional mß-like bHLH genes are found in the Apis genome that possess introns. These introns are at similar positions as the introns of two other HES-like genes, dpn and h which themselves are highly conserved in the three insect species, arguing for a common evolutionary history. Presumably, the introns are evolutionarily ancient because they are also found in the C. elegans E(spl)/h like gene lin-22. The AS-C seems to originate from a single sc/l'sc like bHLH gene and a second largely diverged bHLH gene that shares similarity with Drosophila ase. The high degree of variation in the latter makes it difficult to conclusively decide on the original arrangement of this gene complex (Schlatter, 2005).

Extraordinary parallels exist between the fly neurogenic pathway and that of vertebrates. Hairy enhancer of split (HES-1), the mouse homolog of E(spl), has been isolated. HES-1 can be activated by the intracellular domain of murine Notch (mNotch). When combined with the human analog of Su(H), it can block myoD-induced myogenesis. Activated forms of mNotch associate with KBF2/RBP-J Kappa, the human analog of Su(H). This complex acts as a transcriptional activator through the KBF2-binding sites of the HES-1 promoter (Jarreault, 1995).

HES-1 acts as a negative regulator and binds preferentially to the Nbox, instead of the Ebox, thus exhibiting the same activity and promoter specificity as the fly homolog (Takakayashi, 1994). Targeted disruption of HES-1 leads to up-regulation of neural helix-loop-helix factors, premature neurogenesis and severe neural tube defects (Ishibashi, 1995). The conserved function of vertebrate Notch and Su(H) and of Hairy enhancer of split, as well as the conservation of promoter elements, indicates that from the fly to man the Notch pathway uses the same elements. It remains to be seen, however, whether the pathways are identical. There are hints that portions of the Notch receptor can accompany the human analog of Su(H) into the nucleus. This may not occur in the fly.

Transcription of a zebrafish gene of the hairy-Enhancer of split family delineates the midbrain anlage in the neural plate. her5 encodes a basic helix-loop-helix protein with all features characteristic of the Drosophila hairy-E(spl) family. her5 is expressed in a band of cells within the neural anlage from about 90% epiboly on to at least 39 hours postfertilization. After completion of brain morphogenesis, her5-expressing cells are located in the caudal region of the midbrain, at the boundary with the rhombencephalon. The her5 expression domain corresponds to the midbrain primordium, including both the tectum and the tegmentum, in the neural plate. Four other genes, pax(zf-b) and three engrailed family genes (eng3, eng2 and eng1) are expressed in the same mesencephalic-rhombencephalic territory, but her5 is more exclusively transcribed in the midbrain primordium. Does her5 expression contribute to regionalization or is the activity necessary for partitioning the neural plate into different territories? So far there is no answer to this question (Müller, 1996).

her4 encodes a zebrafish bHLH protein of the Hairy-E(spl) family. The gene is transcribed in a complex pattern in the developing nervous system and in the hypoblast. During early neurogenesis, her4 expression domains include the regions of the neural plate from which primary neurons arise, suggesting that the gene is involved in directing their development. Indeed, misexpression of specific her4 variants leads to a reduction in the number of primary neurons formed. The amino-terminal region of her4, including the basic domain, and the region between the putative helix IV and the carboxy-terminal tetrapeptide WRPW are essential for this effect, since her4 variants lacking either of these regions are non-functional. However, the carboxy-terminal WRPW itself is dispensable. The interrelationships between deltaD, deltaA, notch1, her4 and neurogenin1 have been examined by means of RNA injections. her4 is involved in a regulatory feedback loop that modulates the activity of the proneural gene neurogenin, and as a consequence, the activity of deltaA and deltaD as well. Activation of notch1 leads to strong activation of her4, to suppression of neurogenin transcription and, ultimately, to a reduction in the number of primary neurons. These results suggest that her4 acts as a target of Notch-mediated signals that regulate primary neurogenesis (Takke, 1999).

The main conclusion of this study is that her4 encodes a zebrafish homolog of the Drosophila E(Spl)-C proteins, which acts as a target of Notch to suppress primary neurogenesis. This conclusion is based on structural and functional considerations. HER4 shows considerable sequence identity in the bHLH domain, the region that binds DNA and is involved in target recognition. Furthermore it also exhibits the other characteristics of the Hairy-E(Spl) protein family, such as the carboxy-terminal tetrapeptide WRPW. her4 is expressed in the neural plate region in which the primary neurons are formed. In Drosophila, the E(Spl) proteins suppress the activity of the proneural genes. Misexpression of her4 suppresses ngn1 and the development of primary neurons. However, there are two major differences between HER4 and the E(Spl) proteins. In Drosophila the latter require Groucho for their function in neurogenesis whereas the data presented here do not provide evidence for interactions between HER4 and zebrafish Groucho2. Injections of groucho2 RNA do not have any apparent effect on the development of islet-1 cells, and co-injection of her4 and groucho2 RNA does not enhance the neural suppression mediated by her4 alone. In the association between Groucho and E(Spl)-C proteins in Drosophila, the WRPW domain plays an essential role in segmentation and neurogenesis; thus removal or mutation of this domain renders the protein non-functional. In contrast, the WRPW of HER4 is apparently not needed to suppress primary neurogenesis in zebrafish. It is important to note in this context that there is at least one precedent for the present result: it has been reported that the WRPW of Hairy is not required to suppress Scute in the sex determination pathway in Drosophila. The other important difference between the Drosophila E(Spl)-C proteins and HER4 concerns their amino-terminal regions, including the basic domain, which has been shown to bind a specific DNA sequence called the N-box. In misexpression experiments using the Gal4-UAS system in Drosophila, deletion of either the basic domain or both the basic and the HLH domains does not seriously affect the ability of E(Spl) proteins to suppress neural development. However, the amino-terminal domain of HER4 seems to be essential for activity, as its deletion resulted in non-functional proteins. Therefore, DNA binding of the HER4 protein might play a more important role during neurogenesis in the zebrafish than in the case of the E(SPL) proteins in Drosophila (Takke, 1999a and references).

In both Xenopus and zebrafish, differentiation of primary neurons has been shown to be perturbed following misexpression of Notch and Delta homologs. The results of these experiments, as well as those presented here, strongly support the idea that primary neurons are selected from equivalence groups within the neural plate, and that this process is mediated by lateral inhibition. her4 is one of the target genes of the Notch signaling cascade in the zebrafish. Transcription of her4 is activated by the constitutively active Notch1 variant encoded by nic, and misexpression of her4 leads to a reduction in the number of islet-1 positive cells. Similar observations have been made concerning ESR-1, a Xenopus homolog of the E(SPL) proteins. However, the effects of misexpression of her4 on the islet-1 cells are less severe than those caused by nic. It is suggested that, in analogy to Drosophila, this difference is due to the concomitant activation by Notch of additional bHLH genes involved in regulation of primary neurogenesis. The reduction in the number of islet-1 positive cells following misexpression of her4 might be due either to direct inhibition of proneural gene transcription, or to an effect on the target genes of the proneural proteins. Transcription of zebrafish ngn1 is modulated by misexpression of Delta variants that are assumed to activate or repress Notch. Inhibition of ngn1 transcription correlates with transcriptional activation of her4, thus suggesting a causal relationship between the two events (Takke, 1999a and references).

The existence of a feedback loop between Notch and Delta in zebrafish that is organized similar to one in Drosophila is strongly supported by the data. The activation of Notch has two detectable effects: (1) it leads to an increase in transcription of her4, resulting in repression of proneural activity and primary neural fate and (2) a feedback loop. Whether proneural activity is directly repressed by transcriptional suppression, and/or indirectly by posttranslational modifications, has to be analyzed in further detail. In any case, the second effect, a feedback loop, has been established; it leads to a reduction in the concentration of delta RNA in the cells in which Notch has been activated. This in turn reduces the intensity of Notch activation in the neighboring cells, and allows them to differentiate as primary neurons (Takke, 1999a).

Within the neural plate, the number of islet-1-positive cells is reduced following ngn1 RNA injection, while cells of this type develop ectopically outside the neural plate of the same animals. A similar behavior has been observed with XASH-3 in Xenopus. However, misexpression of Xenopus neurogenin leads in Xenopus to ectopic development of primary neurons in the neural plate. These apparently paradoxical effects of ngn1 can be understood when one considers that it activates transcription of deltaA, deltaD and her4 (i.e. the effectors responsible for lateral inhibition) only within the limits of the neural plate. This may explain the reduction in the number of islet-1 cells observed within the neural plate following injection of ngn1 RNA. ngn1 cannot activate deltaA, deltaD and her4; consequently, lateral inhibition outside the neural plate is not activated, thus permitting ectopic development of islet-1-positive cells in the non-neural ectoderm. The observation that the co-injection of her4 suppresses ngn1-mediated development of ectopic islet-1 cells supports this hypothesis. In a Drosophila embryo that overexpresses proneural genes the situation is remarkably similar to that of ngn1 misexpression in the zebrafish. Gal4-mediated overexpression of lethal of scute leads to ectopic development of neurons only within the amnioserosa; this effect is suppressed by the concomitant activation of Notch in the amnioserosa. However, selection of neural and epidermal progenitor cells takes place normally in the neuroectoderm -- in spite of the presence of large amounts of proneural gene products. The proneural gene products seem to activate lateral inhibition strongly, since a reduction in the complement of copies of the Notch+ gene from 2 to 1, leads to strong neurogenic phenotypes (Takke, 1999a and references).

Based on these findings, it is suggested that, like the E(Spl)-C genes in Drosophila, her4 constitutes the last link in the Notch signaling cascade in zebrafish neurogenesis. There are several other her genes in the zebrafish. Their complex patterns of expression during embryogenesis suggest their involvement in other processes besides differentiation of islet-1-positive cells. Since other genes encoding bHLH proteins of the same family are known to be expressed in other regions of the body, e.g. her1 and her5 in the presomitic mesoderm and in the midbrain anlage, respectively, they may represent different endpoints of this signaling pathway and regulate cell fate decisions other than those that result in the appearance of the primary neurons (Takke, 1999a and references).

During vertebrate embryonic development, the paraxial mesoderm becomes subdivided into metameric units known as somites. In the zebrafish embryo, genes encoding homologs of the proteins of the Drosophila Notch signaling pathway are expressed in the presomitic mesoderm and expression is maintained in a segmental pattern during somitogenesis. This expression pattern suggests a role for these genes during somite development. Various zebrafish genes of this group were misexpressed by injecting mRNA into early embryos. RNA encoding a constitutively active form of NOTCH1a (notch1a-intra) and a truncated variant of deltaD [deltaD(^Pst)], as well as transcripts of deltaC and deltaD, the hairy-E(spl) homologs her1 and her4, and groucho2 were all tested for their effects on somite formation, myogenesis and on the pattern of transcription of putative downstream genes. In embryos injected with any of these RNAs, with the exception of groucho2 RNA, the paraxial mesoderm differentiated normally into somitic tissue, but failed to segment correctly. Activation of Notch results in ectopic activation of her1 and her4. This misregulation of the expression of her genes might be causally related to the observed mesodermal defects, since her1 and her4 mRNA injections led to effects similar to those seen with notch1a-intra. deltaC and deltaD seem to function after subdivision of the presomitic mesoderm, since the her gene transcription pattern in the presomitic mesoderm remains essentially normal after misexpression of delta genes. Whereas Notch signaling alone apparently does not affect myogenesis, zebrafish groucho2 is involved in differentiation of mesodermal derivatives (Takke, 1999b).

Results regarding the effect of misexpressing wild-type deltaD have suggested a function for this gene in somite development. Misexpression of a dominant negative variant of deltaD, wild-type deltaC, or an activated form of NOTCH1a, or misexpression of the hairy-E(spl) homolog her1 or her4, leads in all cases to considerable disruption of somitogenesis. However, whereas the mesodermal effects of perturbing deltaC and deltaD activity are similar, those observed following either Notch activation or coinjection of her1 and her4 appear to have a different basis. In the former case, patterning defects are evident, but the presomitic mesoderm seems to be subdivided into somitomeres, as incomplete somite borders are visible; in the latter case, somites apparently do not form, because no somite borders can be seen. Since both her1 and her4 are ectopically activated by notch1a-intra within the presomitic mesoderm and misexpression of both her genes causes defects similar to those seen with notch1a-intra mRNA, it is proposed that both her genes are targets of Notch during somitogenesis. It follows then that, during normal development, Notch-mediated activation of her genes may be causally related to the initial subdivision of the paraxial mesoderm into somitomeres. In contrast, in the case of deltaC or deltaD misexpression the defects seem to be independent of the activity of the two her genes because it fails to perturb the transcription pattern of her1 and her4. The same applies to misexpression of a truncated variant of DELTAD. This result is surprising and leads to two important corollaries: (1) it suggests that neither DELTAC nor DELTAD acts as a ligand to trigger NOTCH-dependent activation of her genes, and (2), it suggests that the DELTAC/D-dependent somitic defects do not depend directly on the activity of her genes. Accordingly, the delta function in somitogenesis that appears to operate downstream of the component of Notch function was assayed; namely, her gene activity. Double in situ hybridizations with her1 or her4 and MyoD probes following misexpression of delta variants suggest that the latter act within the somites once the presomitic mesoderm has been subdivided into somitomeres. By analogy to the situation in Drosophila, where the Notch regulatory network is required for maintenance of the epithelial state in several different instances, as well as for the formation of borders in the wing disc, it is proposed that DELTAC and DELTAD act during the definition and/or maintenance of somitic borders in zebrafish embryos (Takke, 1999b).

At least three important questions remain open in this scenario: (1) whereas the proposed function of Delta in controlling boundary development may rely on a mechanism similar to that operating in the wing margin of Drosophila, the mechanism by which Notch contributes to subdivide the presomitic mesoderm is unclear. (2) The ligand that activates the NOTCH1a receptor (and, consequently, the her genes) during the subdivision of the presomitic mesoderm is unknown. Although there are no less than four delta genes in the zebrafish, only deltaC and deltaD are expressed in the mesoderm and apparently neither one is capable of activating her genes under these experimental conditions. Therefore, there is no obvious candidate for this function. (3) The receptor required for the DELTAC/D-mediated function during later stages of somite development is also unknown (Takke, 1999b).

Previous expression data had suggested a conserved role for the Hes genes in the Notch signaling pathway, but not in segmentation. Here, Hes3 expression during mouse embryogenesis is described. During early development of the central nervous system, Hes3 is expressed specifically in the region of the midbrain/hindbrain boundary, and in rhombomeres 2, 4, 6 and 7. This pattern occurs at approximately the same time that Krox20 expression appears in r3 and r5 and precedes the morphological appearance of rhombomeres. The regulatory interactions between Krox20 and Hes3 are currently unknown. The segmental pattern of Hes3 suggests that it may have a conserved role as a segmentation gene. Later in development, Hes3 is co-expressed with other neurogenic gene homologs in the developing central nervous system and epithelial cells undergoing mesenchyme induction (Lobe, 1997).

Mammalian hairy and Enhancer of split homolog 1 (HES1), a basic helix-loop-helix factor gene, is expressed in retinal progenitor cells, and its expression decreases as differentiation proceeds. Retinal progenitor cells infected with HES1-transducing retrovirus do not differentiate into mature retinal cells, suggesting that persistent expression of HES1 blocks retinal development. In contrast, in the retina of HES1-null mutant mice, differentiation is accelerated, and rod and horizontal cells appear prematurely and form abnormal rosette-like structures. Lens and cornea development is also severely disturbed. There is extensive bipolar cell death in the mutant retina, to the point of complete disappearance of such cells. These studies provide evidence that HES1 regulates differentiation of retinal neurons and is essential for eye morphogenesis (Tomita, 1996).

Hes6 is a basic helix-loop-helix transcription factor homologous to Drosophila Enhancer of Split [E(spl)] proteins. Hes6 is known to promote neural differentiation and to bind to Hes1, a related protein that is part of the Notch signaling pathway, affecting Hes1-regulated transcription. Hes6 is expressed in the murine embryonic myotome and is induced on C2C12 myoblast differentiation in vitro. Hes6 binds DNA containing the Enhancer of Split E box (ESE) motif, the preferred binding site of Drosophila E(spl) proteins, and represses transcription of an ESE box reporter. When overexpressed in C2C12 cells, Hes6 impairs normal differentiation, causing a decrease in the induction of the cyclin-dependent kinase inhibitor, p21^Cip1, and an increase in the number of cells that can be recruited back into the cell cycle after differentiation in culture. In Xenopus embryos, Hes6 is co-expressed with MyoD in early myogenic development. Microinjection of Hes6 RNA in vivo in Xenopus embryos results in an expansion of the myotome, but suppression of terminal muscle differentiation and disruption of somite formation at the tailbud stage. Analysis of Hes6 mutants indicates that the DNA-binding activity of Hes6 is not essential for its myogenic phenotype, but that protein-protein interactions are essential. Thus, a novel role for Hes6 in multiple stages of muscle formation has been demonstrated (Cossins, 2002).

Mash-2 expression begins during preimplantation development, but is restricted to trophoblasts after the blastocyst stage. Within the trophoblast lineage, Mash-2 transcripts are first expressed in the ectoplacental cone and chorion, but not in terminally differentiated trophoblast giant cells. After day 8.5 of gestation, Mash-2 expression becomes further restricted to focal sites within the spongiotrophoblast and labyrinth. Downregulation is probably important for normal development, since overexpression of Mash-2 reduces giant cell formation. The role that the Notch signaling pathway may play in trophoblast development has been investigated. Mash-2 is a homolog of Drosophila achaete/scute complex genes. In the developing mouse placenta, all elements of the Notch pathway are expressed. In particular, the Notch-2, HES-2, and HES-3 genes are coexpressed in trophoblast giant cells and in foci within the spongiotrophoblast at day 10.5 when Mash-2 transcription becomes restricted. Two members of the mammalian Groucho family are expressed in trophoblasts; TLE3 is expressed broadly in the giant cell, spongiotrophoblast, and labyrinthine regions, whereas TLE2 is limited to giant cells and focal regions of the spongiotrophoblast. These data suggest that Notch signaling through activation of HES transcriptional repressors may play a role in murine placental development (Nakayama, 1997).