The EGF receptor and notch signaling pathways control the initiation of the morphogenetic furrow during Drosophila eye development

During Drosophila larval development, the presumptive eye imaginal disc is transformed from an unpatterned and undifferentiated epithelium into a precise, regular array of ommatidia by the passage of the morphogenetic furrow (Heberlein and Treisman, 2000; Ready et al., 1976). The initiation of this wave of morphogenesis along the posterior-lateral margins of the eye disc is under the positive control of Hedgehog (Hh) and Decapentaplegic (Dpp), and the negative control of Wingless (Wg; Chanut and Heberlein, 1997a; Chanut and Heberlein, 1997b; Domínguez and Hafen, 1997; Heberlein and Moses, 1995; Heberlein et al., 1995; Heberlein et al., 1993; Ma and Moses, 1995; Ma et al., 1993; Pignoni and Zipursky, 1997; Treisman and Rubin, 1995). Hh is expressed at the posterior margin just before the initiation of the furrow, and its removal leads to the total and irreversible inhibition of pattern formation (Borod and Heberlein, 1998; Domínguez and Hafen, 1997). In contrast, ectopic expression of Hh ahead of the endogenous furrow results in ectopic furrow initiation and retinal development (Heberlein et al., 1995). Dpp is expressed along the posterior and lateral margins of the eye disc and ectopic expression leads to the initiation of new precocious furrows along the anterior margins (Chanut and Heberlein, 1997b; Heberlein et al., 1993; Pignoni and Zipursky, 1997). The negative regulator Wg is expressed on the lateral margins just anterior to the advancing morpogenetic furrow (Ma and Moses, 1995; Treisman and Rubin, 1995). Ectopic expression of Wg within the morphogenetic furrow halts its progression, while loss of Wg at the margins induces ectopic furrows and precocious retinal development (Ma and Moses, 1995; Treisman and Rubin, 1995).

It has been shown that ectopic Ras pathway signaling can induce ectopic photoreceptor development anterior to the furrow (Dominguez et al., 1998). We now show that Epidermal Growth Factor Receptor (Egfr) and Notch signaling act cooperatively and are necessary and sufficient for the initiation of the endogenous morphogenetic furrow: removal of either signaling pathway results in a block in furrow initiation, while ectopic expression along the margins ahead of the furrow leads to ectopic retinal differentiation. The initiation of the furrow at the posterior margin and its re-initiation along the lateral margins has been thought to be a single process. We show that these two events are temporally and genetically separable, and now refer to the initiation of the furrow at the intersection of the posterior margin and midline as ‘birth’ and the continued re-initiation along the lateral margins as ‘reincarnation’.

The Egfr is a transmembrane receptor tyrosine kinase (RTK) that acts through the Ras cascade (Nilson and Schüpbach, 1999; Schweitzer and Shilo, 1997). In the developing eye, Egfr signaling has been shown to control cell fate specification, inhibit programmed cell death and modulate cell cycle progression (Bergmann et al., 1998; Freeman, 1998; Kumar and Moses, 2000; Kurada and White, 1998). Notch is a transmembrane receptor activated by ‘DSL’ class ligands and transduces signals to the nucleus by means of a pathway that includes the Enhancer of Split Complex genes (E(spl)C; Artavanis-Tsakonas et al., 1995). During eye development, Notch is involved in setting up the dorsal-ventral compartment boundary, establishing planar polarity, spacing ommatidial clusters and cell fate specification (Baker, 2000; Blair, 1999; Cagan and Ready, 1989).

In this report, we now add a novel function to the Egfr and Notch signaling pathways: the initiation of the morphogenetic furrow. Our results suggest that both these pathways act upstream of both Hh and Dpp, but downstream of Wg. We also show that the processes that initiate pattern formation and those that propagate retinal development across the eye imaginal disc are genetically distinct from each other.

The following stocks were used: dpp-GAL4 (Staehling-Hampton et al., 1994), UAS-Egfr^DN, UAS-sspi, UAS-mspi (Freeman, 1996), UAS-vn (Schnepp et al., 1996), UAS-aos (Freeman, 1994), UAS-spry (Kramer et al., 1999), UAS-Egfr^tsla (a temperature-sensitive strain) (Kumar et al., 1998), UAS-Egfr^topCO (a null strain) (Clifford and Schüpbach, 1989), UAS-Egfr type I, UAS-Egfr type II, UAS-Egfr^Elp type I, UAS-Egfr^Elp type II (in which Elp forms are activated) (Lesokhin et al., 1999), UAS-Egfr^{top 4.2} (Queenan et al., 1997), UAS-btl, UAS-btl^act (Lee et al., 1996), UAS-btl^DN (Reichman-Fried et al., 1994), UAS-htl, UAS-htl^act, UAS-htl^DN (Michelson et al., 1998), UAS-Dras1^V12 (activated), UAS-Dras1^N17 (dominant negative) (Scholz et al., 1997), UAS-Draf^gof (Li et al., 1997b), UAS-hep (Boutros et al., 1998), UAS-rl, UAS-rl^sem (in which sem is activated=MAPK[act]) (Martin-Blanco, 1998), UAS-PntP1, UAS-PntP2 (Klaes et al., 1994), UAS-ttk69, UAS-ttk88 (Li et al., 1997a), UAS-aop, UAS-aop^act (Rebay and Rubin, 1995), UAS-BarH1, UAS-BarH2 (Hayashi et al., 1998), UAS-N^DN (Brennan et al., 1997), UAS-Dl^DN (Huppert et al., 1997), UAS-Ser^DN (Fleming et al., 1997), UAS-mam^DN, UAS-Su(H) (Helms et al., 1999), UAS-m8, UAS-m8^DN (Giebel and Campos-Ortega, 1997), UAS-m4, UAS-mα, UAS-mβ, UAS-mδ, UAS-mγ (Alifragis et al., 1997), UAS-m7 (Tata and Hartley, 1995), UAS-pros (Manning and Doe, 1999), UAS-wg (Lawrence et al., 1995), UAS-arm (Boyle et al., 1997), UAS-E cad (a gift from I. Davis), UAS-sgg^act (Hazelett et al., 1998), UAS-slp1, UAS-slp2 (Riechmann et al., 1997), UAS-grk, UAS-sgrk (gifts from T. Schüpbach), UAS-fz^DN, UAS-Dfz2^DN (Zhang and Carthew, 1998), UAS-dsh, UAS-Dsh[PDZ], UAS-Dsh[DIX] (Axelrod et al., 1998), UAS-put^DN (Ruberte et al., 1995) and UAS-sax^DN (Brummel et al., 1994). All UAS-GAL4 crosses were done at 18°C, 25°C and 29°C. All figures shown in the text are from crosses made at 25°C, unless specified otherwise in the text and figure legends. ^DN, ^act and ^gof mean dominant negative, activated and gain of function, respectively.

Egfr^tsla/CyO and Egfr^topCO/CyO adults were mated and allowed to lay eggs for 1 hour at 18°C. Batches of their progeny were raised at 18°C, shifted to 28°C for 12 hours and then returned to 18°C. The 12 hour up-shift periods used were: 0-12, 12-24, 24-36, 36-48, 48-60, 60-72, 72-84, 84-96, 96-108, 108-120, 120-132, 132-144, 144-156, 156-168, 168-180, 180-192, 192-204 and 204-216 hours after egg deposition (AED). Temperature shifts that affected furrow initiation or disrupted compound eye structure are referred to as temperature-sensitive periods (TSP). Eye imaginal discs or adult eyes from at least 20 individuals were examined for each time point described below.

Antibodies used were rat α-Elav (O’Neill et al., 1994), mouse α-Wg (Brook and Cohen, 1996) and rabbit α-Dpp (Hoffmann and Goodman, 1987). Secondary antibodies were conjugated to FITC (Jackson Labs). F-actin was visualized with phalloidin conjugated to TRITC (Molecular Probes). Immunohistochemistry was performed essentially as described previously (Tomlinson and Ready, 1987).

We have previously reported eye imaginal discs in which the initiation of the morphogenetic furrow appeared to be inhibited when Egfr signaling was removed (Kumar et al., 1998). Before we could confirm and study this in more detail, we first determined the time point at which the morphogenetic furrow initiates during normal development. After a short egg collection (1 hour at 18°C), animals were allowed to mature at 18°C and imaginal discs were examined for both the presence of the furrow and for developing clusters (Fig. 1A,E-G). Eye discs show no signs of either ommatidia or the furrow 180 hours AED (Fig. 1E) and the first column of ommatidia that are positive for the neuron specific Elav protein are seen at 192 hours AED (Fig. 1F). As Elav expression lags about three to four columns behind the furrow and each column is laid down every 1.5-2 hours, the initiation of the furrow occurs at about 184-186 hours AED (red line in Fig. 1A). At 204 hours AED discs show a well-advanced furrow and about eight columns of Elav-positive ommatidia (Fig. 1G). Discs from at least 20 individual larvae were examined for each time point.

To determine when and where loss of Egfr signaling affects furrow initiation, we used a conditional temperature-sensitive allele of the receptor (Kumar et al., 1998) and removed Egfr function for consecutive 12 hour intervals beginning immediately at egg deposition and continuing through the larval stages. After 12 hours at the restrictive temperature (29°C), the animals were returned to the permissive temperature (18°C) until either the adult eyes or imaginal discs were examined (see Materials and Methods). Discs from at least 20 individual larvae were examined for each time point. We identified two temperature-sensitive periods (TSPs) in which the loss of Egfr signaling affected the structure of the adult eye through the regulation of furrow initiation (Fig. 1A,C,D).

Removal of Egfr signaling during TSP1 (168 hours – 180 hours AED) resulted in an adult eye that was devoid of ommatidia (compare Fig. 1B with 1C). We examined imaginal discs from TSP1 animals at both 192 hours and 204 hours AED. Instead of seeing a single Elav-positive column at 192 hours AED, and 8 at 204 hours AED, these discs lacked both ommatidia and the morphogenetic furrow, despite being returned to the permissive temperature for 12 and 24 hours, respectively (Fig. 1H,I). Removal of Egfr signaling during TSP2 (192 hours-204 hours AED) resulted in an adult eye with severe structural defects along the posterior-lateral margins (compare Fig. 1B with 1D). The ommatidia at the intersection of the equator and the posterior margin appear not to be affected. Imaginal discs from TSP2 animals were examined immediately at the end of TSP2 (204 hours AED) and while the furrow had clearly initiated at the posterior margin, its continued re-initiation along the lateral margins was inhibited (Fig. 1J). Subsequent removal of Egfr signaling also affects the development of the eye but these defects are not related to the birth or reincarnation of the morphogenetic furrow and have been described in an earlier report (Kumar et al., 1998).

It is important to note that between TSP1 and TSP2 there are 12 hours for which there is no requirement for Egfr function – and that this period contains the time of furrow initiation (red line in Fig. 1A). Thus Egfr signaling is required before initiation to set up some mechanism that will act a few hours later and again subsequently for correct propagation of new columns along the lateral margins. We thus call these two temporally separable aspects of furrow initiation ‘birth’ and ‘reincarnation’.

The furrow initiating protein Dpp is expressed along the posterior-lateral margins (Heberlein and Moses, 1995; Heberlein and Treisman, 2000). We used a dpp^blk-GAL4 driver to express components of the Egfr pathway in this restricted expression domain for all of the experiments described (Fig. 2A; Blackman et al., 1991). By the late third instar, the furrow has progressed more than half way across the eye with about 20 columns of ommatidia in the wild type (Fig. 2B). Expression of wild-type and activated forms of the Egfr using the a dpp^blk-GAL4 driver resulted in the generation of ectopic retinal development along the lateral margins (Fig. 2C). This suggests that Egfr signaling is sufficient to direct the initiation of the morphogenetic furrow at this time and place and may be the critical regulator of a limiting checkpoint in normal development.

Egfr encodes two isoforms (I and II; Livneh et al., 1985) and we expressed both wild type and constitutively active (Egfr^Elp) versions of both isoforms. Only type I Egfr can induce ectopic marginal furrow initiation (Fig. 2D, type II not shown). The dominant negative receptor driven by dpp^blk-GAL4 inhibits furrow reincarnation (along the lateral margins), but not furrow birth (at the posterior, Fig. 2E). Similarly activated Ras and Raf direct ectopic furrow formation, whereas dominant negative versions inhibit reincarnation (but not birth, Fig. 2F-H). The positive pathway elements MEK (not shown), MAPK (Fig. 2I) and the transcription factor isoform PntP2 (Fig. 2J) did not induce ectopic furrows in this assay but the PntP1 isoform did (Fig. 2K). Both Ras pathway target transcription factors Ttk69 and Aop proteins are known to negatively regulate neural development (Dickson, 1998; Lai et al., 1997; Rebay and Rubin, 1995). In our assay both affect furrow reincarnation but not birth (Fig. 2L,M). Other targets of Ras signaling have either no effect on furrow initiation (Ttk88) or block the furrow at various stages (BarH1/2, not shown). These results are consistent with our temperature shift experiments (described above) and support the contention that furrow initiation actually includes two genetically and temporally separable steps: birth and reincarnation.

The Egfr ligand Vein acts in wing vein patterning (Guichard et al., 1999) and another ligand Gurken acts in dorsal patterning of the oocyte (Nilson and Schüpbach, 1999), while the Spitz ligand functions along the embryonic midline and within the developing eye (Freeman, 1998; Perrimon and Perkins, 1997; Schweitzer and Shilo, 1997). We expressed all three ligands with the dpp^blk-GAL4 driver and only Spitz produced ectopic furrows (Fig. 2N,O). Star and Rhomboid (Rho) are thought to be upstream regulators of the positive acting ligand Spitz (Freeman, 1998; Schweitzer et al., 1995; Schweitzer and Shilo, 1997). We expressed Star and Rho individually and neither could induce ectopic furrow initiation (not shown) but together they act synergistically to cause ectopic initiation (Fig. 2P).

The published reports that describe the expression of the dpp^blk-GAL4 that is used in this report suggest that it would be the ideal driver for manipulating signaling at the posterior edge and margins of the eye imaginal disc. We confirmed the expression of the driver by crossing flies carrying the dpp^blk-GAL4 to flies harboring a UAS-lacZ construct and then assaying lacZ expression in both L2 and L3 eye discs (Fig. 3A,B). In both cases, lacZ is restricted to the posterior-lateral margins. This is in contrast to the wild type Dpp expression pattern in which Dpp leaves the margins after initiation and is found in cells within the advancing morphogenetic furrow (Curtiss and Mlodzik, 2000; Heberlein et al., 1993; Ma et al., 1993). We then induced ectopic furrows by expressing contitutively active Egfr along the margins via the dpp^blk-GAL4 driver and again assayed lacZ expression in this background. As we expected, lacZ remained along the posterior lateral margins. The expression of lacZ was highly elevated at the places along the dorsal and ventral margins where ectopic furrows had initiated (Fig. 3C). These results confirm that we are indeed restricting our analysis to the posterior-lateral margins and the elevated levels of lacZ at the sites of ectopic furrow initiation suggest that Egfr signaling is upstream of Dpp.

We used dpp^blk-GAL4 to express wild-type, activated and dominant negative versions of three different RTKs (Egfr, Fig. 2; and Btl and Htl, Fig. 4). Both wild-type and activated Egfr induce ectopic furrow initiation (Fig. 2C), and dominant negative Egfr inhibits furrow initiation (Fig. 2E). In contrast, the expression of wild-type, constitutively active or dominant negative versions of the Drosophila FGF receptor homologs, Btl and Htl, have no effects (Fig. 4). RTKs can be made constitutively active or dominant negative by mutating conserved residues (Hubbard, 1999) thus their mechanism of activation or downregulation is similar. All of these constructs contain the same activating or dominant negative coding changes in their tyrosine kinase domains. This suggests that some other domain and activity of the three receptors confers this functional specificity.

Notch functions during many stages of eye development (Artavanis-Tsakonas et al., 1995; Blair, 1999) and we have recently shown that it also functions in eye disc determination (Kumar and Moses, 2001). We now also find that it cooperates with Egfr signaling to initiate the furrow (Fig. 5). dpp^blk-GAL4 driven dominant negative Notch or its ligands inhibit furrow reincarnation but not birth (Fig. 5B-D). Of nine downstream genes tested, only dpp^blk-GAL4 driven expression of E(spl)C-m8 can generate ectopic furrows (Fig. 5E,F) and dominant negative mutants of E(spl)C-m8 (not shown) and Mam inhibit furrow reincarnation (Fig. 4G; Helms et al., 1999). In this assay, the Notch antagonist Prospero (Pros; Reddy and Rodrigues, 1999) inhibits furrow reincarnation in the same way as loss of Notch signaling (Fig. 5H). The results from these experiments indicate that not only does Notch signaling function in the initiation of the morphogenetic furrow, but it also fails to antagonize EGFR signaling.

How is precocious retinal differentiation via Egfr signaling prevented during normal development? Three direct negative regulators of RTK signaling have been identified in Drosophila: Argos, Sprouty and Kekkon (Casci et al., 1999; Ghiglione et al., 1999; Golembo et al., 1996; Kramer et al., 1999; Reich et al., 1999). Competition between agonist and antagonist ligands may fine-tune the signaling gradient, thereby producing different developmental decisions. We drove the expression of Argos and Sprouty with dpp^blk-GAL4 but saw no effects on furrow initiation (not shown). This suggests that there may be an alternate mechanism that antagonizes Egfr signaling along the eye field margins. This is likely to be via Wg signaling (see below).

The Hh, Dpp and Wg pathways are required for the initiation and progression of the morphogenetic furrow (Heberlein and Moses, 1995; Heberlein and Treisman, 2000). We have shown above that Egfr and Notch signaling also act in this process. dpp^blk-GAL4 driven Wg results in a block in both furrow birth and reincarnation (Fig. 6A). This inhibition is mediated by the receptor Frizzled2 (Fz2, Kennerdell and Carthew, 1998). Expression of a dominant negative Fz2 receptor along the margins blocks Wg signaling and results in ectopic furrow initiation (data not shown). Surprisingly, expression of a dominant negative Frizzled (Fz) receptor has no effect (data not shown), suggesting that the block on furrow initiation by Wg is mediated through Fz2 but not Fz.

Is the inhibition of furrow initiation by Wg mediated by the downstream component Dishevelled (Dsh; Cox and Peifer, 1998)? Overexpression of wild-type Dsh is sufficient to inhibit furrow initiation, but the expression of just the DIX and PDZ domains of Dsh has no effect (data not shown). This suggests that the Dsh plays a role in inhibiting furrow initiation and that its activity in this process maybe mediated by a yet undescribed protein subdomain. We used dpp^blk-GAL4 to drive furrow-inhibiting Wg, together with furrow-inducing Egfr, Ras or Raf. In each case, co-expressed Wg was unable to block the induction of ectopic furrows. This suggests that the Wg signal is upstream of the Egfr pathway in this instance (Fig. 6B). Shaggy (Sgg) normally acts negatively to block the Wg signal downstream of the receptor, while the Drosophila E-cadherin homolog Shotgun (Shg) acts positively (Cox and Peifer, 1998; Dierick and Bejsovec, 1999). dpp^blk-GAL4 driven Sgg (activated) and Shg (dominant-negative) activate reincarnation, in the same way as loss of Wg (Fig. 6C,D). Other Wg pathway elements tested (Armadillo, Slp1 and Slp2) had no effect in this assay (data not shown).

Hh and Dpp signaling act in the birth of the morphogenetic furrow at the posterior margin (Heberlein and Moses, 1995). However, it is not clear which is upstream of the other in this instance. We expressed these proteins with dpp^blk-GAL4 and neither could induce ectopic furrow. dpp^blk-GAL4 driven expression of the repressor isoform of Cubitus interuptus protein (Ci^R, which acts downstream of Hh) was unable to block furrow initiation (not shown) but a dominant negative version of Thick Veins (Tkv), the Dpp type I receptor, was sufficient to block furrow reincarnation but not birth (Fig. 6E). This effect was also achieved when dominant negative versions of Saxaphone (Sax), the other Dpp type I receptor, and Punt (Put) the Dpp type II receptor were expressed either individually or in combination. The effects are more dramatic if either type I receptor is removed simultaneously with the type II receptor, suggesting that there is some redundancy among these receptors during furrow initiation (data not shown). These results together with published reports (Chanut and Heberlein, 1997a; Curtiss and Mlodzik, 2000; Pignoni and Zipursky, 1997) suggest that that Hh role in furrow initiation is limited to birth and that Dpp may function in furrow reincarnation.

We have shown that loss of Notch signals blocks furrow initiation, while activated Egfr signaling promotes furrow initiation (Fig. 1). To establish epistasy we used dpp^blk-GAL4 to co-express dominant negative Notch with activated Egfr, Ras or Raf; in all cases furrow initiation was blocked (Fig. 6F). This suggests that the Notch pathway acts downstream of Raf in the Egfr pathway in this instance.

During the progression of the furrow across the eye disc, Dpp is expressed in cells that are found within the furrow (Fig. 7A,A′, Heberlein and Moses, 1995). In situations in which an ectopic furrow has been generated by overexpression of Egfr, cells within the ectopic furrows also express Dpp (Fig. 7B,B′) indicating that Egfr is acting genetically upstream of Dpp. Using the Egfr conditional allele, we removed Egfr function from the advancing furrow and saw no effect on Dpp expression (Fig. 7C,C′). Normally, Wg is expressed in cells along the lateral margins just anterior to the morphogenetic furrow (Fig. 7D,D′; Ma and Moses, 1995). We have shown that Egfr appears to act genetically downstream of Wg signaling (Fig. 6A,B). In discs that contain ectopic furrows retinal development proceeds from the lateral margins despite the continued expression of Wg, further suggesting that Egfr signaling is downstream of Wg (Fig. 7E,E′). Surprisingly, when Egfr signaling was removed via our conditional allele, Wg expression was seen within the morphogenetic furrow (Fig. 7F,F′).

We have shown that furrow initiation is in fact the sum of at least two genetically separable processes: the genesis of pattern formation at the posterior margin (birth) and the reinitiating of retinal differentiation along the lateral margins (reincarnation). Our studies have identified several functions that act selectively at birth or reincarnation. These results are consistent with reports that in dachshund (dac) mutants, the birth of the morphogenetic furrow is blocked in roughly half of cases, but reincarnation is blocked in the other half (Mardon et al., 1994). This is also consistent with reports of a specific block in furrow reincarnation in dpp mutants (Chanut and Heberlein, 1997a; Curtiss and Mlodzik, 2000). We have identified (1) the point in normal development at which the morphogenetic furrow is born, (2) that Egfr signaling is necessary and sufficient for both birth and reincarnation, (3) that Wg signaling is genetically upstream of the Egfr pathway in this instance, and (4) that Egfr and Notch signaling are upstream of both Hh and Dpp in initiating the morphogenetic furrow.

The Notch signaling pathway is known to antagonize Egfr signaling in many developmental contexts, even within the developing fly eye. However, we find that Notch does not antagonize Egfr signaling in furrow initiation and that Notch signaling is integrated within Egfr signaling downstream of Raf in this instance. We propose that Hh expression is the target of Egfr and Notch regulation at furrow birth, and that Hh but not Dpp constitutes the executive signal at this stage (Fig. 8). This is consistent with published observations of the role of Hh in this first phase of furrow induction (Borod and Heberlein, 1998; Dominguez and Hafen, 1997). We also suggest that Egfr and Notch target Dpp expression for reincarnation, and that Dpp but not Hh is the executive signal for this later phase of furrow induction (Fig. 8). This is consistent with published observations of the role of Dpp in furrow initiation (Burke and Basler, 1996; Chanut and Heberlein, 1997a; Chanut and Heberlein, 1997b; Curtiss and Mlodzik, 2000; Greenwood and Struhl, 1999; Heberlein et al., 1993; Horsfield et al., 1998; Ma et al., 1993; Pignoni and Zipursky, 1997; Wiersdorff et al., 1996).

Our experiments have added the Egfr and Notch pathways to the list (Hh, Dpp and Wg) of signals that regulate furrow initiation (Fig. 8). It is not clear, however, how these pathways interact with each other in this developmental context. However, our data are consistent with a pathway in which Egfr signals to Notch which in turn signals to Hh during the birth of the furrow. During furrow reincarnation, Egfr signals to Notch, which now signals to Dpp. Our data are also consistent with the idea that the Spitz signal that initiates these two events is potentially regulated by the Wg pathway.

There are several examples of these pathways converging in some developmental decisions, while antagonizing each other in other instances. For example, during embryogenesis the Hh and Egfr pathways interact to specify head development (Amin et al., 1999). In the development of the adult abdomen, Wg, Dpp and Egfr pathways are integrated to produce a stereotyped dorsoventral pattern (Kopp et al., 1999), while Egfr and Dpp signals oppose each other to set up the operculum boundary (Dobens et al., 2000) and to distinguish between wing and leg disc fates (Kubota et al., 2000). We have shown that Notch signaling is genetically integrated into the Egfr pathway downstream of Raf. Our results indicate that during morphogenetic furrow initiation they cooperate. This provides one more example of the crucial interaction of the Notch and Egfr pathways in the regulation of development.

We thank S. Artavanis-Tsakonas, J. Axelrod, N. Baker, K. Basler, E. Bier, E. Martin-Blanco, J. Campos-Ortega, S. Campuzano, R. Carthew, S. Cohen, C. Delidakis, C. Doe, J. Fischer, R. Fleming, M. Freeman, U. Gaul, U. Heberlein, Y. Hiromi, M. Hoffman, R. Holgrem, T. Kornberg, M. Krasnow, M. Mlodzik, D. Montell, M. Muskavitch, Z. C. Lai, M. Leptin, I. Livne-Bar, M. O’Connor, N. Perrimon, G. Rubin, K. Saigo, T. Schüpbach, B. Z. Shilo, A. Simcox, J. Treisman and B. Yedvobnick for gifts of fly stocks and antibodies; and B. Yedvobnick, S. L’Hernault and R. Reifegerste for suggestions, advice and support. This work was supported by NIH and NSF grants (RO1 EY-12537 and IBN-9807892) to K. M. and a NIH postdoctoral fellowship (5 F32 EY06763) to J. K.