Enhancer of split

DEVELOPMENTAL BIOLOGY

Embryonic

E(spl) is transcribed in the late blastoderm [Image]. It is first concentrated in two strips, each 2-3 cells wide on either side of the embryo in the neurogenic ectoderm ventral midline anlage, immediately dorsal of the mesodermal anlage (precursor), extending along the entire anterioposterior axis.

Transcript is present in the entire ectoderm during germ band extention, with the exception of some regions in the procephalic lobe, stomodeum and proctodeum [Images]. Transcripts become restricted to the epidermis during the extended germ band stage. In stage 11, transcripts are detected in the entire mesoderm. Only slight differences in the distribution of transcripts are found when comparing E(spl) with the other members of the E(spl)-C (Knust, 1987). Although little attention is paid to these differences, they likely represent some independent regulation of the different genes of the E(spl)-C, but it has been difficult to pinpoint.

E(spl)-C expression in the head starts during stage 8 in the central protocerebral, followed slightly later by the central deuterocerebral domain. E(spl)-C genes remain expressed at a high level in the both domains for several hours after neuroblasts have delaminated from these regions. During stage 10, the expression progresses and ultimately covers most of the procephalic ectoderm (Younossi-Hartenstein, 1996).

In head midline structures, in particular the optic lobe and stomatogastric nervous system, there may be a late phase of EGFR signaling (as assayed by the expression of aos and activated ERK) whose significance is not yet known. EGFR signaling could be involved in modifying the inhibitory feed-back loop between neurogenic and proneural genes that exists in other neurectoderm cells. In the head midline neurectoderm, regulation of proneural and neurogenic genes has to be different. Thus, instead of a short burst of proneural gene expression in proneural clusters that is resolved into expression in individual neuroblasts, proneural genes are expressed for a long period of time; at the same time, the expression is never restricted to single neuroblasts. Since genes of the E(spl) complex are expressed in the same cells that express l’sc, the inhibitory loop between E(spl)-C and proneural genes must be interrupted at some level. It is possible that Egfr signaling is causing the interruption of this inhibitory loop. Based on genetic studies of Notch and Egfr signaling in the compound eye, it has been speculated that one of the consequences of Egfr activation (which ultimately is required for all ommatidial cell types to differentiate) is to inhibit N signaling, since constitutively active N inhibits ommatidial cell differentiation by preventing response to differentiative signals. However, the same effect could be achieved if Egfr signaling, similar to what is proposed here for the midline neurectoderm, interrupts the inhibition of proneural genes by E(spl). Although this would not prevent N signaling, it would cancel the effect of N signaling on downregulating proneural genes and thereby keep cells in a state of competency to respond to signals (Dumstrei, 1998).

Many cell fate decisions in higher animals are based on intercellular communication governed by the Notch signaling pathway. Developmental signals received by the Notch receptor cause Suppressor of Hairless [Su(H)] mediate transcription of target genes. In Drosophila, the majority of Notch target genes known so far is located in the Enhancer of split complex [E(spl)-C], encoding small basic helix-loop-helix (bHLH) proteins that presumably act as transcriptional repressors. The E(spl)-C contains three additional Notch responsive, non-bHLH genes: m4 and malpha are structurally related, whilst m2 encodes a novel protein. All three genes depend on Su(H) for initiation and/or maintenance of transcription. The two other non-bHLH genes within the locus, m1 and m6, are unrelated to the Notch pathway: m1 might code for a protease inhibitor of the Kazal family, and m6 for a novel peptide. The five genes described in this paper are arrayed between mbeta and m7, both coding for bHLH proteins. Two other bHLH genes, m3 and m5 are intermingled with the five. Bearded and M4 are 16% identical. Furthermore, in transcripts of both Brd and m4 there are three common regulatory sequence motifs within the 3' UTR. These are known as the 'Brd box', the 'GY box' and the 'K box'. As in m4, the sequence motif of the Brd box is found twice in the 3'-UTR of malpha mRNA at similar positions but without a GY box. None of the other four non-bHLH E(spl)-C genes contains either Brd or GY box. The K box appears to be more common. It is found twice in the 3'-UTR of malpha and once each in the 3' UTRs of m2 and m6 (Wurmbach, 1999).

malpha and m4 embyonic expression patterns are nearly indistinguishable, and appear very similar to those of E(spl)-C bHLH genes, particularly m5, m7 and m8. The expression patterns suggest that both genes are under the same regulatory control as are the E(spl) bHLH genes and thus, might serve a role in Notch mediated cell differentiation. Surprisingly, also m2 transcripts accumulate in a pattern reminiscent of the transcript distribution of E(spl) bHLH genes, although there are no structural similarites with either the bHLH or the m4/malpha genes. Therefore m2 might serve as a Notch target gene. Unlike the other E(spl)-C genes, the gene is expressed within neuronal cells in the embryo. m6 mRNA accumulates in the CNS, brain and PNS, and in imaginal tissues. m1 is expressed in the digestive tract. Su(H) is shown to be the transmitter of Notch signaling to malpha, m4 and m2. Thus there are three types of Notch responsive genes. The bHLH genes are represented by m8 and others. m4 and malpha share structural similarity with Bearded. These Bearded family proteins share a presumptive basic amphipatic alpha-helical domain but differ with regard to other conserved sequence elements. m2, coding for a novel protein, represents the third class of Notch responsive genes (Wurmbach, 1999).

Robustness and flexibility and the role of lateral inhibition in the neurogenic network

Many gene networks used by developing organisms have been conserved over long periods of evolutionary time. Why is that? A model is presented of the core neurogenic network in Drosophila. This model exhibits at least three related pattern-resolving behaviors that the real neurogenic network accomplishes during embryogenesis in Drosophila. Furthermore, the model exhibits these behaviors across a wide range of parameter values, with most of its parameters able to vary more than an order of magnitude while it still successfully forms these test patterns. With a single set of parameters, different initial conditions (prepatterns) can select between different behaviors in the network's repertoire. Two new measures are introduced for quantifying network robustness that mimic recombination and allelic divergence and these were used to reveal the shape of the domain in the parameter space in which the model functions. Lateral inhibition yields robustness to changes in prepatterns and a reconciliation of two divergent sets of experimental results is suggested. Finally, it is shown that, for this model, robustness confers functional flexibility. It is concluded that the neurogenic network is robust to changes in parameter values, which gives it the flexibility to make new patterns. The model also offers a possible resolution of a debate on the role of lateral inhibition in cell fate specification (Meir, 2002).

The experimental literature includes both support for, and refutation of, an important role for lateral inhibition in neural determination. The results of this analysis can account for both sets of experiments. If the prepattern that initiates neuroblast selection is well tuned, the prepattern plus a constant level of inhibition could select the winner, absent lateral inhibition. But lateral inhibition buffers the patterning against perturbations in the initial prepatterning (e.g., due to genetic or environmental variation, or 'developmental noise'). Seugnet (1997) reported that, with only constant production of Dl, 80% of proneural clusters developed normally, but 20% produced an extra NB. These experiments are interpreted to say that the prepattern is well tuned in most proneural clusters, but in 20%, either a poorly tuned prepattern or noise causes errors in the absence of lateral inhibition. This is a testable idea. One could remove lateral inhibition as Seugnet did. It would then be predicted that the embryo would be much more sensitive to hyper- and hypo-morphs in prepatterning genes such as extramachrochaete and hairy. It would also be predicted that such embryos would be more sensitive to mutations in genes within the network itself, such as missing or extra copies of Dl or N. The latter prediction is made because those mutations should change the threshold to which the prepattern is tuned. In the absence of lateral inhibition, a prepattern that was well tuned to the former threshold could not also be well tuned to the new threshold (Meir, 2002).

From these results, it is deduced that E(spl) greatly reduces the percentage of random parameter sets that enable lateral inhibition. It is believed this is because E(spl) acts as a homeostat. As the expression levels of the proneural genes (ac and sc) rise, their products activate E(spl). E(spl) then downregulates the proneural genes. As with the thermostat in a house, this negative feedback loop tends to keep the proneural genes at an intermediate level rather than allowing them to switch to either a high or low state. Both E(spl) autoinhibition, and to a lesser extent cis-Dl inhibition of N activation, help overcome this homeostat. On the face of it, this seems a strange design. The ac/sc network itself is a bistable switch that tends to go in the direction it is pushed and remain there. The switch and homeostat mechanisms are exact opposites. Removing the homeostat (the reduced model) makes it easier to find parameter sets that pass various tests (which all involve throwing the switch). Why incorporate counteracting mechanisms in the same circuit (Meir, 2002)?

It is, of course, possible that this is simply a vestige of the network's evolutionary history, with no design rationale. But electrical engineering suggests one possible advantage. An op-amp is a famous circuit that amplifies the difference between two inputs. Good op-amps can amplify a voltage difference more than a million-fold. Usually, though, engineers add a negative feedback circuit (that is, a homeostat). This greatly attenuates the gain but makes the amplifier much more stable; noise generated internally inside the op-amp will not affect the output signal. Reducing function to gain stability is common in other electrical circuits as well. These electrical circuits do not make good direct analogies to genetic networks, but the concept of adding negative feedback to increase stability might still apply. Perhaps the E(spl) homeostat reduces the network's sensitivity to developmental noise such as stochastic changes in transcription or translation rates, in the prepattern, or in the concentrations of modulators such as Da and Emc (Meir, 2002).

A related design benefit might be that the E(spl) homeostat prevents the network from switching individual cells on or off before the prepattern has a chance to decree the winner. A simple bistable switch consisting of ac and sc alone could not help but be thrown in one direction or the other by noise (as apparently takes place in C. elegans anchor cell specification). Adding E(spl) leads to a new, neither-on-nor-off steady state, which could enable the proneural switch to procrastinate until some extrinsic cue forces the system to choose one or the other switched state (Meir, 2002).

Larval

Enhancer of split complex genes manifest distinct patterns of expression in the wing imaginal disc. m8 and m7 mRNAs are detected in clusters of cells that correspond to the locations where sensory organ precursors (SOPs) develop. In addition m8 is also detected in cells at the dorsal/ventral boundary throughout the third instar. The expression of mgamma and mdelta is at times associated with the same SOPs, and at other times, with different SOPs. However mdelta and mgamma mRNAs are only detected in a subset of proneural clusters. Like m8, mgamma is also present at high levels at the dorsal/ventral boundary in early stages. The domain of mß is the most distinctive. It is expressed in the wing blade associated with developing veins, and is also present at the dorsal/ventral boundary and wing margin, and is expressed in a complex pattern elsewhere in the disc, with no simple association with developing sensory organs (de Celis, 1996). In the eye disc, m8 and m7 are expressed spanning the morphogenetic furrow, whereas mgamma and mdelta are expressed just posterior to the furrow. mgamma, mdelta and mß are expressed in the more posterior portions of the disc, where the recruitment of undifferentiated cells into ommatidial units occurs; there is little expression of m8 and m7 in this region (de Celis, 1996).

Effects of Mutation or Deletion

Ectopic expression of m5 or E(spl), both members of the E(Spl)-C, before bristle precursor division results in loss of sensory bristles from all parts of the adult fly. Ectopic expression after bristle precursor division produces bristles with aberrant cuticular structures. Reducing E(spl) gene function using mitotic recombination de-represses the neural fate and produces supernumerary sensory bristle neurons. Thus E(spl) inhibits neural fate during the selection of neural precursors, and also plays a role in restricting the neuronal fate to one of the four progeny cells of the bristle precursor (Tata, 1995).

Comparison between the phenotypes produced by Notch, Suppressor of Hairless and Enhancer of split mutations in the wing and thorax indicate the Su(H) and Notch requirements are not indistinguishable, but that Enhancer of split activity is only essential for a subset of developmental processes involving Notch function. For example Enhancer of split function is required for the segregation of a single sensory organ precursor in in wing morphogenesis but not for the correct differention of the progeny from each sensory organ precursor, requiring Notch and Su(H). Likewise, the ectopic expression of Enhancer of split proteins does not reproduce all the consequences typical of ectopic Notch activation. For example, no ectopic acitvation of wingless occurs when Enhancer of split proteins are ectopically expressed. It is suggested that the Notch pathway bifurcates after the activation of Su(H) and that Enhancer of split activity is not required when the consequence of Notch function is the transcriptional activation of downstream genes. Transcriptional activation mediated by Su(H) and transcriptional repression mediated by Enhancer of split could provide greater diversity in the response of individual genes to Notch activity (de Celis, 1996)

In the mesoderm of Drosophila embryos, a defined number of cells segregate as progenitors of individual body wall muscles. Progenitors and their progeny founder cells display lineage-specific expression of transcription factors but the mechanisms that regulate their unique identities are poorly understood. The homeobox genes ladybird early and ladybird late are shown to be expressed in only one muscle progenitor and its progeny: the segmental border muscle (SBM) founder cell and two precursors of adult muscles. lb activity is associated with all stages of SBM formation, namely the promuscular cluster, progenitor cell, founder cell, fusing myoblasts and syncytial fiber. The segregation of the ladybird-positive progenitor requires coordinate action of neurogenic genes and an interplay of inductive Hedgehog and Wingless signals from the overlying ectoderm. The SBM progenitor corresponds to the most superficial cell from the promuscular cluster, thus suggesting a role for the overlying ectoderm during its segregation. . Since epidermal Wg and Hedgehog (Hh) signaling has been shown to influence muscle formation, the SBM-associated lb expression was examined in embryos carrying hh and wg thermosensitive mutations. Wg and Hh signalings, mutually dependent at this time, are shown to be required for the promuscular lb activity and/or the segregation of SBM progenitors. The initial influence of these signals is no longer observed later in development. In addition to signals from the epidermis, the activity of the mesodermal gene tinman, initially expressed in the whole trunk mesoderm, is involved in the early events of myogenesis. In tin - embryos, the formation of SBM promuscular clusters and segregation of lb-positive progenitor cells are strongly affected, leading to the absence of the majority of SBM fibers. During promuscular cluster formation, since tin expression becomes restricted to the dorsal mesoderm, its influence on ventrolaterally located SBMs is likely to be indirect and mediated via an unknown factor. The lack of neurogenic gene function, known to be involved in cell-cell interactions during lateral inhibition, generates the opposite phenotype. Mastermind - and Enhancer of split - embryos fail to restrict promuscular lb expression to only one cell; in consequence, they display a hyperplastic lb pattern in later stages (Jagla, 1998).

Eye development in Drosophila involves the Notch signaling pathway at several consecutive steps. At first, Notch signaling is required for stable expression of the proneural gene atonal (ato), thereby maintaining the neural potential of the cells. Subsequently, in a process of lateral inhibition, Notch signaling is necessary to confine neural commitment to individual photoreceptor founder cells. Later on, the successive addition of cells to maturing ommatidia is under Notch control. In contrast to previous assumptions, the recessive Notch allele split (N^spl) specifically involves loss of the early proneural Notch activity in the eye, which is in agreement with bristle defects as well. As a result, fewer cells gain neural potential and fewer ommatidia are founded. N^spl alleles are characterized by a smaller number of ommatidia, which usually contain less than the normal set of photoreceptors. Enhancement of this phenotype by the dominant mutation Enhancer of split [E(spl)^D] happens within the remaining proneural cells (in which Ato expression has been abolished). In line with genetic data, this process occurs primarily at the protein level due to altered protein-Protein Interactions and Post-transcriptional Regulation between the aberrant E(spl)^D and proneural proteins. Indeeed, in a yeast two-hybrid assay, the mutant M8*, representing the E(spl)^D alteration, binds significantly more strongly to proneural proteins, especially Achaete and Atonal. The mutant M8* protein does not interfere with the establishment of high Ato levels within intermediate-group cells. In contrast, m8* transcripts accumulate to a much higher level due to increase of mRNA stability caused by the deletion. Heterodimerization of M8* with other E(spl) bHLH proteins is indistinguishable from that of the wild-type M8 protein. Therefore the N^spl mutation reduces the inductive, proneural activity of N, which is normally required to stabilize expression of the proneural gene atonal. As a consequence fewer intermediate groups arise: these groups serve as reservoirs for future R8 cells. E(spl)^D potentiates the deficits of N^spl because in the compromized background the already lowered Ato levels in most presumptive R8 cells now drop below the threshold required to maintain neuronal fate. N^spl is the first Notch mutation known to specifically affect Notch inductive processes during eye development (Nagel, 1999).

The organization and function of the Notch signaling pathway in Drosophila are best understood with respect to the role of this pathway in the process of selection of neural progenitor cells. However, there is evidence that, in addition to neurogenesis, the Notch signaling pathway is involved in several other developmental processes, one of which is the selection of muscle progenitor cells. Thus, the number of these progenitor cells is increased in neurogenic mutants. It has been proposed that muscle progenitor cells are selected from clusters of equivalent cells expressing genes of the achaete-scute gene complex (AS-C). Additional elements of the Notch signaling pathway participate in myogenesis. Gal4 mediated expression of a Notch variant, E(spl) and Hairless shows that the selection of muscle progenitor cells obeys principles apparently identical to those acting at the selection of neural progenitor cells (Giebel, 1999).

To test whether the Notch signaling pathway is involved in myogenesis, the effects of expression of a constitutively active Notch protein (Notch^intra) were examined. A second chromosomal effector line with an UAS-Notch^intra construct was used. This construct led to complete blocking of neural development upon activation with daG32. Embryos carrying that construct driven by daG32 or by 24B-Gal4, respectively, do not express any of the muscle founder cell markers S59 and Kruppel in the mesoderm. Therefore, it is assumed that no muscle progenitor cells are specified in these animals. Confirmation of this assumption is provided by the observation that no muscle fibers differentiate in these embryos, as shown by means of the expression of a myosin heavy chain (MHC) reporter gene. In mutants where fusion of myoblasts is blocked, founder cells express corresponding founder cell markers, while the non-founder myoblasts remain as undifferentiated rounded cells, which express certain muscle specific genes like myosin. Since Notch^intra expressing mesodermal cells are rounded and many of them express the MHC reporter, it is assumed that the MHC expressing cells are non-founder myoblasts that have failed to undergo fusion due to the lack of muscle founder cells (Giebel, 1999).

Further evidence for a Notch pathway role in myogenesis was obtained by overexpressing UAS-E(spl) in the mesoderm. Following Gal4 mediated activation of UAS-E(spl), the number of S59 and Kruppel positive cells is strongly reduced. This correlates with a defect in the number of differentiated muscle cells, as shown by MHC reporter gene expression. Again these data fit well with the results obtained on the development of the neuroectoderm, in which Gal4 driven UAS-E(spl) expression leads to strong reduction of CNS and PNS structures (Giebel, 1999).

echinoid (ed) encodes an immunoglobulin domain-containing cell adhesion molecule that negatively regulates the Egfr signaling pathway during Drosophila photoreceptor development. A novel function of Ed is shown, i.e., the restriction of the number of notum bristles that arise from a proneural cluster. Thus, loss-of-function conditions for ed give rise to the development of extra macrochaetae near the extant ones and increase the density of microchaetae. Analysis of ed mosaics indicates that extra sensory organ precursors (SOPs) arise from proneural clusters of achaete-scute expression in a cell-autonomous way. ed embryos also exhibit a neurogenic phenotype. These phenotypes suggest a functional relation between ed and the Notch (N) pathway. Indeed, loss-of-function of ed reduces the expression of the N pathway effector E(spl)m8 in proneural clusters. Moreover, combinations of moderate loss-of-function conditions for ed and for different components of the N pathway show clear synergistic interactions manifested as strong neurogenic bristle phenotypes. It is concluded that Ed is not essential for, but it facilitates, N signaling. It is known that the N and Egfr pathways act antagonistically in bristle development. Consistently, it is found that Ed also antagonizes the bristle-promoting activity of the Egfr pathway, either by the enhancement of N signalling or, similar to the eye, by a more direct action on the Egfr pathway (Escudero, 2003).

On the mechanism underlying the divergent retinal and bristle defects of M8* (E(spl)D) in Drosophila

Multisite phosphorylation has been implicated in repression by E(spl)M8. It is proposed that these phosphorylations occur in the morphogenetic furrow (MF) to reverse an auto-inhibited state of M8, enabling repression of Atonal during R8 specification. These studies address the paradoxical behavior of M8*, the truncated protein encoded by E(spl)D. It is suggested that differences in N signaling in the bristle versus the eye underlie the antimorphic activity of M8* in N⁺ (ectopic bristles) and hypermorphic activity in N^spl (reduced eye). Ectopic M8* impairs eye development (in N^spl) only during establishment of the atonal feedback loop (anterior to the MF), but is ineffective after this time point. In contrast, a CK2 phosphomimetic M8 lacking Groucho (Gro) binding, M8SDDeltaGro, acts antimorphic in N⁺ and suppresses the eye/R8 and bristle defects of N^spl, as does reduced dosage of E(spl) or CK2. Multisite phosphorylation could serve as a checkpoint to enable a precise onset of repression, and this is bypassed in M8* (Kahali, 2009).

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.