echinoid mutants exhibit neurogenic phenotypes and show synergistic interactions with the Notch signaling pathway

Initiation of embryonic neurogenesis in Drosophila melanogasterinvolves the selection of specific groups of neuroectodermal cells in the ventral and procephalic neurogenic regions of the embryo. These groups of neuroectodermal cells or proneural clusters have the potential to develop into neural precursors or neuroblasts. Within such individual proneural clusters,normally only one cell will differentiate into a neuroblast, whereas the remaining cells will develop as epidermoblasts(Campos-Ortega, 1993). The proneural basic helix-loop-helix (bHLH) transcription factors, which include the Achaete-scute Complex (ASC) and atonal gene products, endow the cells of the proneural cluster with neurogenic potential(Bertrand et al., 2002; Modolell and Campuzano, 1998). Initially, all cells of the proneural clusters express these proneural transcription factors, but later on their expression becomes restricted to the future neuroblast. A process of lateral inhibition limits proneural gene expression and hence the neurogenic potential to a single cell within an individual proneural cluster. Lateral inhibition is mediated by the genes of the Notch signaling pathway(Artavanis-Tsakonas et al.,1999). Disruption of this lateral inhibition process allows many more or all proneural cluster cells to differentiate into neuroblasts. The increase in the number of neuroblasts, and parallel loss of epidermal precursors, results in neuronal hyperplasia, which is accompanied by a loss of epidermal structures (Hartenstein and Posakony, 1990; Lehmann et al., 1983).

The neurogenic genes Notch (N), Delta(Dl), Presenilin (Psn), Suppressor of Hairless [Su(H)], mastermind (mam) and Enhancer of split complex [E(spl)C] have been shown to interact genetically and to function in the process of lateral inhibition(Campos-Ortega, 1993; Artavanis-Tsakonas et al.,1999). Proneural genes positively regulate the levels of the N ligand, Dl (Haenlin et al.,1994). Upon binding its ligand, N undergoes Psn-dependent processing which results in the release of the N intracellular domain(N^icd) from the membrane (De Strooper et al., 1999; Struhl and Adachi, 1998; Struhl and Greenwald, 1999). N^icd translocates to the nucleus where it forms a complex with the sequence-specific DNA-binding protein, Su(H) (Fortini and Artavanis-Tsakonas,1994; Furukawa et al.,1992; Schweisguth and Posakony, 1992; Tamura et al.,1995) and Mam (Petcherski and Kimble, 2000), a transcriptional co-activator, to activate the expression of multiple genes of the E(spl)C(Bailey and Posakony, 1995; Furukawa et al., 1995; Lecourtois and Schweisguth,1995). This gene complex [E(spl)m8, E(spl)m7, E(spl)m5,E(spl)m3, E(spl)mδ, E(spl)mγ and E(spl)mβ]encodes seven closely related proteins of the bHLH family of transcription factors (Delidakis and Artavanis-Tsakonas,1992; Klämbt et al.,1989; Knust et al.,1992). E(spl) bHLH proteins act as transcriptional repressors in a complex with the co-repressor protein Groucho(Delidakis and Artavanis-Tsakonas,1992; Fisher and Caudy,1998) and downregulate achaete and scuteexpression leading to suppression of neural cell fate in the cell receiving the signal (Heitzler et al.,1996a).

Notch signaling pathway-mediated lateral inhibition is utilized again during the formation of the adult peripheral nervous system (PNS) during larval development. In the larval wing discs, proneural clusters composed of 20-30 cells are established (Campuzano and Modolell, 1992; Modolell,1997). The Notch signaling pathway limits sensory organ precursor(SOP) cell fate to one or two cells within a proneural cluster(Artavanis-Tsakonas et al.,1995). The SOP gives rise to the adult external sensory organ or bristle, which is composed of four cells (neuron, sheath cell, socket cell and hair cell) (Bodmer et al.,1989; Hartenstein and Posakony, 1989). Disruption of the Notch signaling pathway results in specification of additional SOPs, often resulting in the generation of extra sensory bristles (Dietrich and Campos-Ortega, 1984; Hartenstein and Posakony,1990; Schweisguth et al.,1996).

Neurogenic gene function is not limited to neural cell fate suppression. Requirement for neurogenic gene function has been shown for the development of other tissues, including eye, segmented appendages such as leg, mesoderm,muscle, somatogastric nervous system, wing and during oogenesis(Artavanis-Tsakonas et al.,1995).

Echinoid (Ed) is a transmembrane (TM) cell adhesion molecule with seven Immunoglobulin (Ig) C2 domains (Williams and Barclay, 1988), two Fibronectin type III domains(Hynes, 1986) and a TM,followed by a 315 amino acid intracellular domain with no identifiable structural or functional domain (Bai et al., 2001). Ed has been shown to act as a negative regulator of the EGFR signaling pathway during photoreceptor development in the larval eye disc (Bai et al., 2001). Neuroglian has been identified as an activating ligand for the antagonistic effect of Ed on the EGFR signaling pathway(Islam et al., 2003). Here we report on the neurogenic phenotype of echinoid (ed) mutants. Analysis of embryonic lethal alleles of ed reveals hyperplasia of the central nervous system (CNS) at the expense of the epidermal structures. Furthermore, we show that ed function is required for proper morphogenesis of wing and leg, as well as for the specification of the proper number of adult sensory bristles (macrochaetae and microchaetae). edphenotypes are suppressed by increasing N pathway activity [by overexpressing N^act or E(spl)m7] and are enhanced by mutations in Dl or reduction of E(spl)C activity. Thus ed functions synergistically with the Notch signaling pathway during the processes of neural cell fate specification and wing development.

Df(2L)ed1, Df(2L)M24F11, ed^k01102,E(spl)^8D06/TM3,Sb and T80 GAL4 were obtained from the Bloomington Drosophila Stock Center. ed^2B8,ed^3C2 (Bier et al.,1989) and ed^k01102 are P-element insertions in the ed locus. ed^m1 and ed^tsare EMS-induced ed mutant alleles (this study). 69B GAL4 was kindly provided by A. Garcia-Belido (Universidad Autonoma de Madrid, Spain). For genetic interaction studies the following stocks were used: UAS-N^act: ed^ts/UAS-N^act:ed^ts (the UAS-N^act: ed^tschromosome was generated by standard recombination), ed^2B8/P(actin5C-lacZ)CyO; Kr-GAL4/P(actin5C-lacZ) TM3Sb,UAS-N^act/CyO [E. Giniger (FHCRC, Seattle, WA) and B. Yedvobnick (Emory University, Atlanta, GA)], Dl^via1/TM3,Sb(Vaessin and Campos-Ortega,1987), E(spl)^8D06/TM3,Sb(Klämbt et al., 1989), UAS-EdExt (this study), pnr^MD237-GAL4(Heitzler et al., 1996b),Eq-GAL4 [gift from C. T. Chein (Academia Sinica, Taipei, Taiwan)], UAS-E(spl)m7 (Ligoxygakis et al.,1999) and UAS-Dl(Doherty et al., 1996). Flies were raised at 25°C unless otherwise mentioned.

EMS mutagenesis was performed as described in Vaessin and Campos-Ortega(Vaessin and Campos-Ortega,1987). ed alleles were isolated in a cn bwbackground based on lethality in transheterozygosity over ed^2B8 at 29°C.

The ed cDNA clone LD2669 was sequenced to confirm that it contains the entire predicted ed ORF. EdExtra was generated via PCR, using LD2669 as template. EdExtra contains the entire Ed extracellular region, the entire predicted TM region and an additional 11 amino acids after the predicted TM domain. Thus, EdExtra contains amino acids 1-1028. EdExtra coding region was inserted into the pUASTtransformation vector (Brand and Perrimon,1993). pUAST-EdExtra construct was sequenced and transgenic lines were generated by P-element-mediated transformation(Spradling and Rubin, 1982). Thirteen independent UAS-EdExtra fly lines were obtained. Five lines were further characterized. All five lines produce comparable ectopic expression phenotypes.

The ORF of the ed^m1 and ed^tsalleles were analyzed by direct sequencing of PCR products. Genomic DNA was isolated as described in Schlag and Wassarman(Schlag and Wassarman, 1999). Automated sequencing was performed at the Center for Molecular Neurobiology DNA Sequencing Facility (Ohio State University).

In situ hybridization to whole-mount embryos was performed as described previously (Vaessin et al.,1991), using digoxigenin-labeled antisense and sense RNA probes. The following antibodies were used for immunohistochemistry: Rabbit-anti-HRP(1:3000, Jackson Laboratories), Goat-anti-rabbit (1:3000, Jackson Laboratories), Rabbit-anti-deadpan (Bier et al., 1992) and monoclonal Mouse-anti-beta-Galactosidase (Promega). Two different anti-Ed sera were generated in this study. Polyclonal antibodies were raised (AnimalPharm Services, Healdsburg, CA) against a His-tag fusion protein corresponding to the N-terminal 60 amino acids (in Rabbits) and against a His-tag fusion protein corresponding to the C-terminal 200 amino acids (in guinea pigs). Both anti-Ed Extracellular and anti-Ed Intracellular give a similar embryonic expression pattern for Ed. The specificity of both anti-Ed sera was verified in homozygous Df(2L)ed1 and ed^2B8 embryos. Anti-Ed intracellular antibody signal was amplified using biotin-conjugated secondary antibody and Fluorescein Avidin D(Vectashield). Both light microscopy and confocal microscopy (BioRad MRC 1024)was employed for immunohistological analysis. Cuticle preparations were performed as described in Vaessin and Campos-Ortega(Vaessin and Campos-Ortega,1987). Scanning electron micrographs (SEM) as described in Kimmel et al. (Kimmel et al., 1990)as well as standard stereo dissection microscopy were utilized for analysis and documentation of adult phenotypes.

ed has previously been shown to be expressed uniformly in the developing larval wing and eye discs (Bai et al., 2001). In contrast, our analysis of both ed RNA and protein expression during embryogenesis revealed a dynamic spatial and temporal expression pattern of ed gene products. At stage 5, ed is expressed in all cells except the pole cells (data not shown). During stage 9, both ed RNA and protein expression is eliminated from the delaminated neuroblasts, whereas it is still present in the mesodermal and ectodermal cells (Fig. 1A,C). In addition, subcellular localization of the ed transcript becomes evident during this stage. Thus, ed RNA appears to be highly concentrated in the basal region of the ectodermal cells, whereas no or only low levels of ed RNA is detectable in the apical portions of the ectodermal cells (Fig. 1B). A similar localization of ed protein can be observed during this stage(Fig. 1D). During stage 11 ed expression becomes more dynamic and a reduction of edexpression in the ventral neurogenic region becomes evident. In parallel, high levels of Ed expression is visible in the tracheal pits(Fig. 1E). At late stage 11/early stage 12, ed expression in the ventral neurogenic region is terminated. Later in development, ed expression is detected in a variety of developing tissues. These include muscle, epidermis, the hindgut and foregut (Fig. 1H and data not shown). In addition, ed expression is evident in a subset of cells in the CNS (Fig. 1F). At stage 16 ed protein becomes predominantly localized to axons and appears particularly concentrated in the posterior commissures(Fig. 1G).

The observation of embryonic ed expression suggests an embryonic requirement for ed gene function. In agreement with this prediction,three recessive embryonic lethal ed mutations, ed^2B8, ed^3C2 and ed^m1, were identified and analyzed. Two of these alleles, ed^2B8 and ed^3C2, originated from a P-element screen for recessive lethals(Bier et al., 1989). ed^m1 and the temperature-sensitive ed allele ed^ts were isolated as EMS-induced mutations, which are lethal in transheterozygosity with ed^2B8. For ed^2B8, a single P-element insertion was mapped to the 24D3-4 region on the left arm of the second chromosome by in situ hybridization. The embryonic recessive lethality was mapped to this region using the overlapping deficiencies: Df(2L)ed1 (24A3-4; 24D3-4) and Df(2L)M24F11 (24D3-4; 25A2-3). To test whether the phenotype of ed^2B8 homozygous embryos is associated with the P-element insertion in the ed locus, the P-element was remobilized. A complete reversion of the mutant phenotype was observed (data not shown). In situ hybridization and antibody labeling with anti-Ed antisera does not show detectable levels of ed gene products in ed^2B8homozygous mutant post-blastoderm embryos (data not shown), indicating that ed^2B8 represents a strong hypomorphic or amorphic ed allele. Additional lethal P-element insertion lines mapping to the 24D region were obtained from a collection of recessive lethal P-element insertion lines generated by the Berkeley Drosophila Genome Project (BDGP). The pupal lethal P-element insertion line l(2)k01102 carries a P-element in the first intron of the ed transcription unit and was identified as an ed allele based on non-complementation of ed^2B8 (Bai et al.,2001) (this study). The mutation associated with ed^ts was mapped to a single base change, G¹⁸³⁴to A (data not shown). This results in the replacement of the conserved Asp⁶¹⁰ residue in the IgC2 domain V with an Asn residue. ed^ts is viable in homozygosity or in transheterozygosity with other alleles of ed at temperatures up to 25°C, but is embryonic lethal in transheterozygosity over ed^2B8 at 29°C. The molecular basis of ed^m1 has not yet been determined. Sequencing of the ed^m1 allele ORF did not reveal a mutation. Thus, the molecular change associated with the ed^m1 allele may be located in ed introns or ed cis-regulatory regions.

Immunostaining of homozygous (ed^2B8, ed^3C2 and ed^m1) or transheterozygous ed mutant embryos [Df(2L)ed1/ed^2B8,ed^2B8/ed^m1, ed^2B8/ed^ts, and ed^2B8/ed^3C2] with anti-HRP antibody,which recognizes all neuronal cells of the CNS and PNS(Jan and Jan, 1982), reveals hyperplasia of the CNS. The observed hyperplasia ranges from mild bulges in the CNS to an increase in the size of the entire CNS. Fig. 2 shows an intermediate phenotype (Fig. 2A,B). Analysis of ed mutant embryos with antibodies against a neural precursor marker, Deadpan (Bier et al.,1992), shows that the increase in the number of neurons is preceded by an increase in the number of neural precursors(Fig. 2E,F). The enlargement of the CNS is accompanied by a parallel loss of epithelium, manifested as loss of cuticle, from the ventral and procephalic regions of the embryo. The cuticular defects range from a fusion of denticle belts to a complete absence of ventral and ventrolateral cuticle, as well as procephalic cuticle(Fig. 2C,D and data not shown). The ed mutant phenotype is reminiscent of the phenotypes displayed by loss of function mutations of neurogenic genes that are part of the Notch signaling pathway (Lehmann et al.,1983).

ed has previously been shown to be essential for proper eye and wing formation (Bai et al.,2001; Islam et al.,2003). To further examine ed function during postembryonic development, we analyzed the phenotype of several transheterozygous viable combinations of ed alleles. Flies transheterozygous for ed^k01102 and ed^m1(ed^k01102/ed^m1) generally die as pupae, but escaper flies can be isolated from low-density cultures and their adult phenotype examined. Flies transheterozygous for ed^ts/ed^m1 or ed^ts/ed^2B8, shifted to the nonpermissive temperature of 29°C for three days at the end of larval stage 2, and then shifted back to 18°C, can also survive to adulthood. These flies show defects in the wings, legs, sensory bristles and eye. Frequently,transheterozygous ed mutant flies show extra wing vein material,irregular thickening of wing vein II and notching at the distal tip(Fig. 3A,B and data not shown). The leg phenotypes visible in transheterozygous flies were generally mild. Thus, the legs of ed^k01102/ed^m1 males had proper number of segments, but the sex combs were misarranged(Fig. 3C,D). The sensory bristles, macrochaetae and microchaetae, are arranged in a stereotypical pattern on the adult dorsal thorax (Jan and Jan, 1993). ed^ts/ed^m1 or ed^ts/ed^2B8 flies that were shifted to the nonpermissive temperature as third instar larvae (see above) show extra macrochaetae. These extra macrochaetae were generally found next to macrochaeta normally present in the wild type (WT)(Fig. 3E,F). The microchaetae were also increased in density. In addition, ed^ts/ed^m1 or ed^ts/ed^2B8 flies and the ed^k01102/ed^m1 showed a rough eye phenotype (not shown), similar to the eye phenotype described previously(Bai et al., 2001).

The extracellular region of Ed has seven IgC2-type domains and two fibronectin-type III repeats. Transgenic fly lines expressing a truncated Ed protein (EdExt) containing the entire extracellular and TM region, but lacking most of the intracellular region (except for the first 11 amino acids of the intracellular region) were generated. EdExt was expressed in wing discs of otherwise WT flies using various GAL4 drivers. Ectopic expression of EdExt in the wing disc using either the T80 GAL4 or the 69B GAL4 driver lines resulted in extra wing vein material at the distal end of wing vein II and rare (10%, n=30 wings) notching of the distal wing tip (Fig. 4B,C and data not shown). These phenotypes are similar to those seen in the adult hypomorphic ed mutant flies. In addition, and similar to adult flies carrying hypomorphic ed alleles, these animals also show extra sensory bristles (macrochaetae and microchaetae)(Fig. 4E and data not shown). Extra macrochaetae were always found in close proximity to a normally positioned bristle. The extra sensory bristles phenotype was more prominent when a stronger wing disc driver, pnr-GAL4, was used. pnr-GAL4 mediates expression in the mesothorax region(Heitzler et al., 1996b). Expression of EdExt in this region resulted in an increase in the number of macrochaetae. This was especially evident in the scutellar region. Extra bristle specification was accompanied by an apparent loss of epidermal tissue,resulting in a smaller scutellum (Fig. 4F). The loss of epidermal tissue appears to be more severe than might be expected from inappropriate SOP specification alone. This might indicate additional roles of Ed in epidermogenesis. In addition, microchaetae were duplicated or quadrupled and missing their respective sockets, or missing completely (Fig. 4G). The phenotypes observed upon overexpression of the EdExt are qualitatively similar to those observed with various transheterozygous combinations of edmutant alleles. These observations suggest that the EdExt construct behaves as a dominant-negative protein.

The neurogenic phenotype of homozygous ed mutant embryos as well as the extra sensory bristle phenotype of ed transheterozygous adults suggests that ed gene function, similar to the function of Notch signaling pathway genes, may be required to suppress neural fate. We therefore performed genetic interaction tests to investigate the possibility of functional interactions and/or epistatic relationship between ed and genes of the Notch signaling pathway.

Activation of the Notch signaling pathway results in suppression of neuronal cell fate (Bray,1998; Greenwald,1998). Overexpression of the intracellular region of N(N^act) results in a dominant activated phenotype(Fortini et al., 1993; Rebay et al., 1993). Accordingly, when N^act is expressed in embryos using the Kr-GAL4 driver, an almost complete loss of neurons in parasegments 4-6 (Kr expression domain) is detected(Lieber et al., 1993)(Fig. 5C,G). In contrast, ed^ts/ed^2B8 transheterozygous embryos, grown at the restrictive temperature of 29°C, show a phenotype opposite to that caused by ectopic N^act expression. Here, hyperplasia of the CNS with local bulging of the ventral nerve cord is evident(Fig. 5B,F). Reduced ed activity results in a strong suppression of the N^actoverexpression phenotype, as is evident in the CNS of UASN^act:ed^ts/ed^2B8; Kr-GAL4/+ embryos(Fig. 5D,H). Hence, reduction of ed activity can compensate for ectopic Notch signaling pathway activity.

To test for genetic interactions between ed and other genes of the Notch signaling pathway we performed dosage-sensitive interaction assays. For this purpose, a homozygous viable Dl mutation, Dl^via1 (Vaessin and Campos-Ortega, 1987) and E(spl)^8D06, a deletion of the E(spl) complex(Knust et al., 1987), were used. Flies heterozygous for Dl^via1 (+/+; Dl^via1 /+) have wings indistinguishable from the WT. ed^ts/ed^m1; +/+ flies, grown at 25°C, have wings with irregular thickenings in wing vein II(Fig. 5J). In contrast, the wings of ed^ts/ed^m1; Dl^via1 /+ flies,grown at 25°C, display significantly increased irregular thickenings of wing vein II. In addition, extra wing vein material is evident in the posterior cell of the wing (Fig. 5K). Wings of +/+; E(spl)^8D06/+ flies, raised at 29°C, are generally WT, but may display with low penetrance (<20%)extra vein material in the posterior cell(Fig. 5M). Flies heterozygous for ed^ts (ed^ts/+; +/+), raised at 29°C, have morphologically normal wings(Fig. 5L). The wings of ed^ts/+; E(spl)^8D06/+ flies, raised at 29°C, showed an enhanced phenotype compared with +/+; E(spl)^8D06/+ flies: the extra vein material is expanded in ed^ts/+; E(spl)^8D06/+ flies and is visible with full penetrance (Fig. 5N). Thus, synergistic interactions can be observed between ed and Dl, as well as ed and the E(spl)C.

We also analyzed the ability of the dominant-negative EdExt protein to functionally interact with Notch signaling pathway genes. In these experiments EdExt was ectopically expressed using the Eq-GAL4 driver. Eq-GAL4 mediates expression in the anterior region of the presumptive notum in the wing disc, with a stronger expression in the anterior midline region (Pi et al., 2001). UAS-EdExt/+; Eq-GAL4/+ flies exhibit a slight increase in the number of microchaetae (Fig. 6B). Overexpression of full-length Dl has been shown to cause an increase in the number of sensory organs (Doherty et al.,1997) (Fig. 6C). It has been argued that this is because of the ability of Dl to autonomously inhibit N signal reception in Dl-expressing cells. Hence, when all cells of the proneural clusters express high levels of Dl, N signal reception would be inhibited, causing an increase in the number of sensory organs(Doherty et al., 1996; Doherty et al., 1997). Simultaneous overexpression of Dl and EdExt using the Eq-GAL4 driver resulted in a dramatic increase in the number of microchaetae that exceeds the additive combination of the respective phenotypes(Fig. 6D). Hence,dominant-negative EdExt causes increased reduction in Notch signaling pathway-mediated lateral inhibition caused by ectopic expression of Dl. Accordingly, an increase in the activity of the Notch signaling pathway should suppresses the phenotype caused by EdExt overexpression. Flies overexpressing EdExt in the pnr expression domain (UASEdExt/+; pnr-GAL4/+)show extra sensory organs (Fig. 4F, Fig. 6F). Ectopic expression of E(spl)m7, a downstream target of an activated Notch signaling pathway, suppresses sensory organ specification resulting in the absence of bristles (Ligoxygakis et al.,1999). Flies expressing ectopic EdExt and E(spl)m7 simultaneously are indistinguishable from those expressing only E(spl)m7 ectopically(Fig. 6H), implying that increasing the activity of Notch signaling pathway can compensate for reduced Ed activity.

Using the embryonic lethal alleles of ed, we have shown that ed is essential for normal embryonic neurogenesis. The phenotypes associated with loss of function mutations of ed have characteristic hallmarks of mutant phenotypes shown by neurogenic genes such as N,Dl or the E(spl) gene complex, namely hyperplasia of the nervous system and corresponding loss of epidermal structures. Similarly, reduction of ed function during sensory organ formation in the wing disc results in specification of extra sensory organs. This phenotype implies that ed is also required for the process of lateral inhibition in the proneural clusters during adult sensory organ development. A main difference between the mutant phenotypes of neurogenic genes and of ed alleles lies in the severity of the observed hyperplasia of the embryonic CNS. Embryos homozygous for apparently amorphic ed alleles show a less extensive neural hyperplasia than that caused by loss of genes such as N or Dl and resemble embryos homozygous/hemizygous for hypomorphic alleles of other neurogenic genes. Strong maternal effects contribute to weak phenotypes of various amorphic mutant alleles of neurogenic genes such as N, mam or groucho. Indeed, maternal ed transcripts can be readily detected in Northernblot analysis of 0-2 hour-old embryos (data not shown). However, it remains to be determined if a maternal contribution can account for the relatively weak neural hyperplasia exhibited by ed mutant embryos.

ed RNA and protein expression during early neurogenesis indicates that ed gene products become restricted to the neuroectodermal cell layer, whereas no ed products are detectable in the delaminated neuroblasts. The dynamics of ed RNA and protein distribution during neuroblast delamination implies that ed function might be required in cells that remain in the ectodermal cell layer. In such a scenario, similar to N, ed function would be required in the cells receiving the lateral inhibitory signal. However, it should be noted that at the time when the differential distribution of ed RNA and protein becomes detectable,the neuroblast segregation has already been initiated.

Ed expression during embryogenesis is dynamic and seen in many developing organ systems. The widespread expression of ed indicates that ed might be required for the development of multiple organs. Indeed,analysis of the trachea and muscles in ed mutant embryos reveals defects in the proper formation of these organ systems (data not shown). The requirement of ed for normal development of multiple tissues is not limited to embryogenesis. Adult flies with reduced ed activity show defects in leg, wing and eye development(Bai et al., 2001; Islam et al., 2003). A similar widespread expression and requirement in multiple organs has also been observed for the Notch signaling pathway during Drosophiladevelopment (Artavanis-Tsakonas et al.,1999).

Ed protein missing its intracellular region interferes with the process of lateral inhibition, as overexpression of EdExt in the developing wing disc results in an increase in the number of macrochaetae and microchaetae. Additional phenotypes include the irregular thickening of wing vein II and infrequent notching of the wing margin. These phenotypes are similar to those seen upon reduced ed function. Thus, ectopic expression of EdExt interferes with the function of endogenous Ed. A dominant-negative activity of the extracellular portion is not unusual for receptors that bind to ligands and then transduce a signal intracellularly(Rebay et al., 1993). Thus, it is possible that the EdExt competes with the WT Ed for a limited amount of ligand. Because EdExt is missing its intracellular region, its binding to the ligand may have no functional consequence other than limiting the amount of available ligand. The ability of the extracellular domain to act as a dominant-negative molecule and the observation that the temperature-sensitive allele of ed has a mutation associated with Ig C2 domain V implies that the interaction of the extracellular domain with a putative ligand is an essential component of Ed function. Two isoforms of Neuroglian (Nrg) have recently been identified as activating ligands for the antagonistic effect of Ed on the EGFR pathway in the eye disc(Islam et al., 2003). Both isoforms (Nrg¹⁸⁰ and Nrg¹⁶⁷) are expressed in the wing disc (Hortsch et al., 1990)and thus overlap in their expression with Ed. It has yet to be determined whether Nrg also functions as a ligand for Ed during sensory organ development. Ed has been shown to act as a homophilic cell adhesion molecule(Islam et al., 2003) (A.A.,S.C. and H.V., unpublished). In the eye disc, it has been shown that the Nrg-mediated heterophilic activity of Ed in repressing the EGFR signaling pathway is redundant with the homophilic activity of Ed(Islam et al., 2003). Thus, it is possible that the dominant-negative construct interferes with Ed activity by competing for homophilic binding.

Ectopic expression of an activated form of N results in suppression of neuronal specification (Lieber et al.,1993; Rebay et al.,1993; Struhl et al.,1993). In contrast, reduced ed gene activity results in increased specification of neurons. Ectopic expression of N^act in ed^2B8/ed^ts embryos results in a near WT nervous system. The observation of compensating, as opposed to an epistatic phenotype,does not support the formulation of a straightforward epistatic relationship between ed and N gene function. Rather, although both WT N and ed have a similar antineurogenic function during neurogenesis, they might be acting in functionally synergistic, yet possibly parallel regulatory pathways.

The observation of dosage-sensitive interactions between mutations in two genes can also be indicative of closely related roles. We have observed dosage-sensitive interactions between ed and Dl and ed and E(spl). The mild wing phenotype exhibited by ed^m1/ed^ts flies raised at 25°C is enhanced by loss of one copy of Dl. Similarly, the wing phenotype of E(spl)^8D06/+ flies is enhanced by reduction of Ed activity. These observations imply that, in the wing disc also, the Notch signaling pathway and ed are acting synergistically.

Genetic interaction between ed and the Notch signaling pathway is also observed during the development of the adult PNS. Ectopic expression of EdExt results in specification of extra macrochaetae and microchaetae. Overexpression of Dl results in an increase in the number of sensory bristles(Doherty et al., 1997). Simultaneous ectopic expression of EdExt and Dl has a phenotype much stronger than what would be the result of additive combination of the individual phenotypes. The EdExt protein interferes with the activity of endogenous Ed and the decrease in Ed activity increases the neurogenic phenotype caused by Dl overexpression. These observations imply that Ed acts in concert with Dl. An epistatic interaction was also observed with E(spl)m7. Ectopic expression of E(spl)m7 completely suppressed the extra bristles phenotype obtained upon EdExt expression. The complete suppression of the dominant-negative phenotype would imply that E(spl)m7 functions downstream to the ed. However, it is possible that this suppression is a result of the strong antineurogenic activity of E(spl)m7. Although it is presently not clear whether edand the genes of the Notch signaling pathway function in the same or parallel pathway, our observations establish that ed and the Notch signaling pathway genes act synergistically in both embryonic and postembryonic development.

Ed has previously been shown to be a negative regulator of the EGFR pathway. ed mutations enhance the rough eye phenotype of Elp^B1, a gain-of-function EGFR allele, and ed genetically interacts with several components of the EGFR pathway during eye development. As a consequence, ed mutant phenotypes include the generation of extra photoreceptor and cone cells(Bai et al., 2001; Rawlins et al., 2003; Spencer and Cagan, 2003). We have shown that ed mutations result in neural hyperplasia in the embryo and the formation of extra sensory organs on the notum of adult flies,and that ed, in these processes, interacts synergistically with the Notch signaling pathway. We have recently described an Ed paralog, Fred, and have shown that Fred is required to suppress SOP specification and is required for proper eye development (Chandra et al., 2003). Similar to ed, fred interacts synergistically with the Notch signaling pathway during SOP specification and also during eye development. fred also interacts with the EGFR pathway. Furthermore, ed and fred show a dosage-sensitive interaction during eye development, indicating that the two genes function in close concert(Chandra et al., 2003). Hence, ed and fred define a subgroup of two closely related IgC2-type TM proteins that interact synergistically with the Notch signaling pathway and antagonistically with the EGFR pathway.

Our data point to a role of Ed in the cell-cell communication processes that lead to the selection of the future neural precursor from the proneural cluster. In this process, Ed functions synergistically with the Notch signaling pathway. In this role Ed might be part of the cell-cell communication process itself. However, keeping in mind its cell-cell adhesion function, it could be argued that Ed may also be involved in the execution of the developmental decisions that result from cell-cell communication. In this scenario downregulation of ed expression in the future neuroblast may contribute to neuroblast delamination, whereas continued edexpression in the future epidermal precursors maintains cell adhesion and stabilizes their fate. Thus, Ed might be functioning at the level of cell-cell communication and at the level of coordinating cell-cell signaling with morphogenesis.

A paper by Escudero et al. (Escudero et al., 2003) with similar conclusions appears in this issue.

A.A. and S.C. contributed equally to this work and are both considered first authors. We are grateful to Drs Jan, Delidakis, Garcia-Bellido, Chein,Heitzler, Giniger, Yedvobnick, as well as the Bloomington Drosophila Stock Center for providing us with fly stocks. We thank the Berkeley Drosophila Genome Project for excellent genomic sequence information. We would also like to thank an anonymous reviewer for helpful suggestions. DNA sequencing was performed at the Center for Molecular Neurobiology DNA sequencing facility at The Ohio State University. The Campus Microscopy Imaging Facility at The Ohio State University was used for generating SEM. This work was supported by grants from the American Cancer Society and the National Science Foundation to H.V.