ABSTRACT
We examined the function of secreted forms of the two known Drosophila Notch ligands, DELTA and SERRATE, by expressing them under various promoters in the Drosophila developing eye and wing. The phenotypes associated with the expression of secreted Delta (DlS) or secreted Serrate (SerS) forms mimic loss-of-function mutations in the Notch pathway. Both genetic interactions between DlS or SerS transgenics and duplications or loss-of-function mutations of Delta or Serrate indicate that DlS and SerS behave as dominant negative mutations. These observations were extended to the molecular level by demonstrating that the expression of Enhancer of split mδ, a target of Notch signaling, is down-regulated by SERS. The antagonistic nature of the two mutant secreted ligand forms in the eye is consistent with their behavior in the wing, where they are capable of down-regulating wing margin specific genes opposite to the effects of the endogenous ligands. This analysis uncovers secreted molecular antagonists of Notch signaling and provides evidence of qualitative differences in the actions of the two ligands DLS and SERS.
INTRODUCTION
The Notch signaling pathway controls cell fates throughout metazoan development (reviewed by Artavanis-Tsakonas et al., 1995). Accumulated evidence from vertebrate and invertebrate systems indicates that Notch signaling modulates the ability of non-terminally differentiated cells to respond to differentiation and proliferation signals in order to progress to the next developmental stage. Genetic analyses in Drosophila uncovered a group of genes which define elements of the pathway. The central player of this signaling pathway is the cell surface receptor NOTCH (N). N has been shown to interact via its extracellular domain with two distinct membrane-bound ligands DELTA (DL) and SERRATE (SER) (Fehon et al., 1990; Rebay et al., 1991). N ligands have also been identified in other species, for example LAG-2 and APX-1 in C. elegans, and DL and JAGGED in vertebrates (Henderson et al., 1994; Henrique et al., 1995; Lindsell et al., 1995; Mello et al., 1994; Tax et al., 1994).
All of the identified N ligands encode transmembrane molecules with variable numbers of Epidermal Growth Factor (EGF)-like repeats in their extracellular domains (Chitnis et al., 1995; Fleming et al., 1990; Henderson et al., 1994; Henrique et al., 1995; Kopczynski et al., 1988; Lindsell et al., 1995; Mello et al., 1994; Tax et al., 1994; Thomas et al., 1991; Vässin et al., 1987). Upstream of their EGF repeats, there is a conserved region named the DSL (Delta, Serrate, Lag-2) domain that, along with the EGF repeats, appear to be a structural hallmark of all the N ligands identified thus far. The DSL domain has been implicated in mediating ligand binding to N (Muskavitch, 1994).
The intracellular domains of the ligands are not conserved, at least at the level of primary structures. Experiments exploring their functional role have yet to provide a clear picture. Studies in Xenopus and Drosophila have shown that deletions of these intracellular domains result in dominant negative ligands (Chitnis et al., 1995; Sun and Artavanis-Tsakonas, 1996), while experiments in C. elegans suggest that the specific sequences of the intracellular domain are not important since replacing the intracellular domain of LAG-2 with β-galactosidase sequence does not affect its normal function (Fitzgerald and Greenwald, 1995; Henderson et al., 1994).
Detailed mutational and expression analysis in Drosophila indicate that while DL and SER can, under certain circumstances, be functionally equivalent, they have different expression patterns and are not redundant in development (Gu et al., 1995; Kooh et al., 1993; Thomas et al., 1991). For example, a study using temperature-sensitive mutants demonstrated that DL is involved in the differentiation of every cell type in the eye (Parody and Muskavitch, 1993). In contrast, SER does not seem to be closely involved in photoreceptor cell fate decisions in the eye (Sun and Artavanis-Tsakonas, 1996). Both DL and SER function in wing development and it has been shown that SER and DL signal from the dorsal and ventral wing compartment, respectively, to activate N at the dorsalventral (D/V) boundary (Doherty et al., 1996; Kim et al., 1995).
In our attempts to further characterize the activity of DL and SER and to explore functional differences and define soluble modulators of N signaling, we examined the in vivo activity of secreted forms of DL and SER. A recent study in C. elegans has shown that secreted forms of LAG-2 or APX-1 can substitute for endogenous LAG-2 activity. Furthermore, the secreted APX-1 DSL domain can function as a dominant gain-of-function ligand (Fitzgerald and Greenwald, 1995). Here we explore the in vivo activity of secreted forms of DL and SER in the developing Drosophila eye and wing, where the wild type requirements of these ligands and the cellular architecture of the tissue are well defined. The phenotypic consequences of the expression as well as their effects on downstream gene expression were assayed. In contrast to what was found in C. elegans, we demonstrate that the secreted forms of the Drosophila ligands act as antagonists of N signaling. Further-more, our analyses revealed qualitative differences in the behavior of DLS and SERS, implying that DL and SER may be interacting with distinct effectors.
MATERIALS AND METHODS
Fly strains
All transgenic flies were generated in w1118 background. The fly strains used in the genetic interaction studies are: Dl9p39/TM3; DlBX6/TM6B; Su(H)T4/CyO; Su(H)16/CyO; Dp(3;3)bxd110/TM6C and UAS-Ser (Speicher et al., 1994). The gal4 lines described are: vg-gal4 (gift from Dr K. Vijayraghavan), ptc-gal4; A9-gal4 (gifts from Dr K. Wharton) and T113-gal4 (gift from Dr J. Urban and Dr N. Perrimon).
Constructs for expression studies and germline transformation
To generate the DlS cDNA, three fragments were ligated together. The 5′ EcoRI-NaeI fragment contains the translation start site to nucleotide 725 of Dl (nucleotide numbers as described by Kopcynski et al., 1988). The middle NaeI-BamHI fragment is a PCR product of extracellular Dl from nucleotides 725 to 1860, followed by an in frame BamHI site. The 3′ BamHI-BamHI fragment contains three tandem repeats of the 11 amino acid myc tag and a stop codon at the 3′ end (from psev-DlTMIC; Sun et al., 1996). The three fragments were first ligated together in pBluescript KS vector (Stratagene) and then shuttled into pUC19 vector (New England Biolabs). The resulting DlS cDNA was removed as an EcoRI-EcoRI fragment and inserted into pCaSpeR-hs vector (Thummel and Pirrotta, 1992) for expression in S2 cells and the sevenless expression vector psev (modified as described by Sun et al., 1996), the glass expression vector pGMR (Hay et al., 1994) and the UAS expression vector pUAST (Brand and Perrimon, 1993) for germline transformation.
To generate the SerS cDNA, three fragments were ligated together. The 5′ EcoRI-XbaI fragment contains the translation initiation to nucleotide 2263 of Ser (nucleotide numbers as described by Fleming et al., 1990). The middle XbaI-BamHI fragment is a PCR product of extracellular Ser from nucleotides 2263 to 3358 followed by an inframe BamHI site. The 3′ BamHI-BamHI fragment contains the myc tag and the stop codon identical to that in DlS cDNA. The three fragments were first ligated in pBluescript KS and then shuttled into pUC19. The resulting SerS cDNA was removed as an EcoRI-EcoRI fragment and inserted into a similar set of vectors as described above for cell and germline transformations.
To generate the SerS2 cDNA, three fragments were ligated together in pUC19 vector. The 5′ HindIII-SfiI fragment is removed from psevSerTM (Sun et al., 1996) and contains the translation initiation to nucleotide 4073 of Ser. The middle fragment is a synthesized SfiI-BamHI linker. The 3′ BamHI-BamHI fragment is the same as the 3′ fragment of SerS. The resulting SerS2 cDNA was removed as an EcoRI-EcoRI fragment and inserted into the pCaSpeR-hs vector and the sevenless vector for cell and germline transformation.
Sample preparation for western blot analyses
S2 cells were transfected with 5 μg of the cDNA in pCaSpeR-hs vector as described by Diederich et al. (1994). 12 hours after transfection, the cells were washed three times in M3 medium (JRH Biosciences). After the last wash, the cells were resuspended in 30 μl of M3 medium and induced with 30 minutes incubation at 37°C and 30 minutes recovery at 25°C. This induction is repeated twice. The cells and medium were separated by low speed centrifugation. The cell pellet was washed with PBS. Western blot analysis of both the medium and the pellet was carried out following standard procedures.
Eye and wing discs were dissected from third instar larvae. They were either immediately put into protein loading buffer with protease inhibitors (1 mM Pefabloc SC, 0.5 μg/ml Leupeptin, 0.7 μg/ml Pepstatin) and boiled, or else they were incubated (the peripodium membrane was opened without damaging the disc tissue) in M3 medium for three hours at 25°C and then separated from the medium by low speed centrifugation. The discs were washed in PBS, and the medium was subjected to high speed centrifugation to get rid of cell debris. Western blot analysis was carried out following standard procedures.
Germline transformation
Germline transformation was performed using standard procedures described by Spradling (1986). Each construct (1.2 mg/ml) was injected with helper plasmid Δ2-3 (0.6 mg/ml) into w1118 embryos.
Scanning electron microscopy (SEM) and sectioning of adult eyes
For SEM, adult flies were dehydrated sequentially in 25%, 50%, 75% and 100% ethanol, for at least 12 hours in each step. The 100% dehydration was repeated three times. The preparation was then subjected to critical point drying before being mounted on stubs and viewed on an ISI-SS40 scanning electron microscope.
Plastic sections were prepared and examined as described by Carthew and Rubin (1990).
Antibody staining of third instar larvae eye and wing discs
Discs were dissected and stained as described by Gaul et al. (1992). The antibody for DL is described by Kooh et al. (1993) and Fehon et al. (1991). The mouse anti-MYC antibody (Oncogene Science) was used at 1 to 100 dilution. The mouse anti-E(spl) Mδ antibody was kindly provided by Dr S. Bray (Jennings et al., 1994) and used at 1 to 1 dilution. The rabbit anti-WG antibody was kindly provided by Dr R. Nusse (Cadigan and Nusse, 1996) and used at 1 to 25 dilution.
RESULTS
N is expressed in all cells posterior to the morphogenetic furrow in the developing eye (Fehon et al., 1991). It is also detected in the majority of the cells in the developing wing with the exception of the bristle mother cells and wing vein precursors. The expression pattern of Dl is more restricted than N. Its protein is detected transiently in the cells undergoing differentiation in a third instar eye disc (Parks et al., 1995). The pattern of DL in the wing disc is dynamic: it is first detected in the ventral compartment (Doherty et al., 1996) and later restricted to wing vein precursors and cells flanking the D/V boundary (Kooh et al., 1993). SER is detected in the dorsal wing compartment at the larval stage (Kim et al., 1995), but it is not clear where it is expressed in the eye. In our effort to define functional domains of DL and SER and uncover possible differences between the two ligands, we expressed the extracellular domains of DL and SER in patterns overlapping those of the endogenous ligands and the receptor. We addressed how these mutant proteins can interfere with endogenous N signaling.
Expression of secreted DL and SER in S2 cells and discs
To construct a secreted form of DL (DLS), a stop codon was inserted just C-terminal to the last EGF-like repeat, approximately 20 amino acids upstream of the transmembrane domain (Fig. 1). Secreted SER (SERS) was created by insertion of a stop codon just C-terminal to the end of the EGF-like repeats, approximately 250 amino acids upstream of the transmembrane domain. Each of these proteins carries three tandem repeats of a MYC tag at its carboxyl terminus.
Schematic diagram of the secreted ligands expressed. Full-length (DL, SER) and secreted (DLS, SERS and SERS2) ligands are shown. The various sequence motifs are indicated by shaded boxes and defined at the top of the figure. Numbers refer to the amino acids of the Drosophila DL and SER sequences as described previously (Fleming et al., 1990; Kopczynski et al., 1988). Three tandem repeats of the MYC-tag sequence are added at the carboxyl terminus of each of the secreted ligands.
Schematic diagram of the secreted ligands expressed. Full-length (DL, SER) and secreted (DLS, SERS and SERS2) ligands are shown. The various sequence motifs are indicated by shaded boxes and defined at the top of the figure. Numbers refer to the amino acids of the Drosophila DL and SER sequences as described previously (Fleming et al., 1990; Kopczynski et al., 1988). Three tandem repeats of the MYC-tag sequence are added at the carboxyl terminus of each of the secreted ligands.
To test their expression, both DlS and SerS cDNAs were first expressed in Drosophila S2 cells under a heat shock promoter (phs). Immunostainings show that high amounts of both protein products are detected in cytoplasmic vesicles (Fig. 2A and data not shown). Very little protein is found associated with the cell membrane.
Expression and secretion of DLS and SERS. (A) Expression of phs-DlS in a S2 cell detected by an antibody against the extracellular domain of DL. No staining is detected on the cell surface. (B) Western blot analysis (an 8% SDS-PAGE gel) of the expression and secretion of DLS and SERS in S2 cells. Proteins of expected sizes are detected (69 ×103 for DLS and 107×103Mr for SERS) both in the supernatant and pellet, suggesting that DLS and SERS are expressed in and secreted from S2 cells. For the control, a full-length DL, with the tandem MYC tags at its C terminus (phs-Dlmyc), is expressed and is mostly found in the cell pellet. All proteins are detected using an anti-MYC antibody. (C) Western blot analysis (a 12% SDS-PAGE gel) of the secreted ligands expressed under the sevenless (sev) promoter in third instar eye discs. Eye discs from the corresponding transgenic flies are dissected and incubated in M3 medium for 3 hours. Note that SERS is processed at a specific site to yield a product of approximately 75 kDa. Proteins are detected using an anti-MYC antibody. Molecular mass markers are indicated on the right of panel B and C.
Expression and secretion of DLS and SERS. (A) Expression of phs-DlS in a S2 cell detected by an antibody against the extracellular domain of DL. No staining is detected on the cell surface. (B) Western blot analysis (an 8% SDS-PAGE gel) of the expression and secretion of DLS and SERS in S2 cells. Proteins of expected sizes are detected (69 ×103 for DLS and 107×103Mr for SERS) both in the supernatant and pellet, suggesting that DLS and SERS are expressed in and secreted from S2 cells. For the control, a full-length DL, with the tandem MYC tags at its C terminus (phs-Dlmyc), is expressed and is mostly found in the cell pellet. All proteins are detected using an anti-MYC antibody. (C) Western blot analysis (a 12% SDS-PAGE gel) of the secreted ligands expressed under the sevenless (sev) promoter in third instar eye discs. Eye discs from the corresponding transgenic flies are dissected and incubated in M3 medium for 3 hours. Note that SERS is processed at a specific site to yield a product of approximately 75 kDa. Proteins are detected using an anti-MYC antibody. Molecular mass markers are indicated on the right of panel B and C.
We tested whether DLS or SERS are secreted by transfecting S2 cells with phs-DlS or phs-SerS and inducing ligand production by heat shock, then separately assaying the cell and the medium. Western blot analysis showed that both DLS and SERS, with the predicted molecular mass of 69×103 and 107×103 respectively, are efficiently secreted into the medium (Fig. 2B). This is in contrast to DLmyc, a membrane attached full-length DL which is only associated with the cell pellet.
We proceeded by asking whether DLS and SERS are also secreted in developing flies. Third instar eye discs from transgenic flies expressing sev-DlS or sev-SerS (see below) were isolated and incubated in M3 medium for 3 hours. Western blot analysis showed that both DLS and SERS were detected in the medium as well as in the discs (Fig. 2C).
We consistently observed that SERS appears to be cleaved to produce an approximately 75×103Mr degradation product. This cleavage event parallels secretion, as the relative intensity of the two SERS bands vary, with the medium containing a higher ratio of the cleavage product than the disc. Similar results were obtained if discs were immediately put into protein loading buffer without the three-hour incubation procedure (data not shown). Interestingly, although the 75×103 product is also observed when SerS is expressed in wing discs (data not shown), it is not the major breakdown product in S2 cells, suggesting that this cleavage is mediated by factors which are present in the discs but absent or inactive in S2 cells. Given the molecular mass of the processed product and the fact that our SERS is tagged at the C terminus, the predicted cleavage site is near the first EGF-like repeat of SER. This separates the putative receptor binding domain, the DSL domain, from the rest of the extracellular domain, raising the possibility that SER signaling is regulated by proteolytic events. However, the lack of appropriate antibodies against SER has prevented us so far from addressing whether the endogenous SER is processed in a similar manner.
sev-DlS and sev-SerS do not affect eye development
To determine possible effects of the secreted ligands on cell differentiation, we expressed the cDNAs using the sevenless (sev) promoter cassette. The sevenless cassette is active in a subset of cells posterior to the morphogenetic furrow in the developing Drosophila eye (Bowtell et al., 1991). Other than occasional inter-ommatidial bristle duplications, the eyes of sev-DlS or sev-SerS flies are mostly wild type (Fig. 3A-C). Similar results were observed in sev-SerS2 transgenic eyes. SERS2 is a form of secreted SER that extends to 11 amino acids N-terminal to the transmembrane domain (Fig. 1). These phenotypes differed from the strong rough eye phenotype obtained when membraneattached DL or SER extracellular domain was expressed under the same promoter (Sun and Artavanis-Tsakonas, 1996).
Adult eye phenotypes of the transgenic flies. All transgenic lines shown carry two copies of the corresponding constructs. Scanning electron micrograph (SEM) of adult eyes (posterior is to the right in all panels): (A,B) A sev-SerS eye. Close to wild type other than occasional mispositioning of bristles and the duplication of bristles. (D,E) A GMR-DlS eye. Aberrant ommatidium shape, duplication of bristles and ommatidia fusion was observed. (G,H) A GMR-SerS eye. Melting and smoothing of lens material and loss of bristle shaft cells were observed. (J,K) A sevhsG4-SerS eye. Duplication of bristles and ‘blueberry’ ommatidia phenotype were observed (Sun and Artavanis-Tsakonas, 1996). B,E,H,K are magnified views of A,D,G,J, respectively. Tangential plastic sections (1 μm) through the apical retina of adult eyes: (C) A sev-SerS eye. Wild-type photoreceptor composition is observed. (F) A GMR-DlS eye. Arrows point to regions missing inter-ommatidial pigment cells. (I) A GMR-SerS eye. Arrows point to ommatidia missing photoreceptor R7. (L) A sevhsG4-SerS eye. Arrows point to ommatidia with extra photoreceptor cells.
Adult eye phenotypes of the transgenic flies. All transgenic lines shown carry two copies of the corresponding constructs. Scanning electron micrograph (SEM) of adult eyes (posterior is to the right in all panels): (A,B) A sev-SerS eye. Close to wild type other than occasional mispositioning of bristles and the duplication of bristles. (D,E) A GMR-DlS eye. Aberrant ommatidium shape, duplication of bristles and ommatidia fusion was observed. (G,H) A GMR-SerS eye. Melting and smoothing of lens material and loss of bristle shaft cells were observed. (J,K) A sevhsG4-SerS eye. Duplication of bristles and ‘blueberry’ ommatidia phenotype were observed (Sun and Artavanis-Tsakonas, 1996). B,E,H,K are magnified views of A,D,G,J, respectively. Tangential plastic sections (1 μm) through the apical retina of adult eyes: (C) A sev-SerS eye. Wild-type photoreceptor composition is observed. (F) A GMR-DlS eye. Arrows point to regions missing inter-ommatidial pigment cells. (I) A GMR-SerS eye. Arrows point to ommatidia missing photoreceptor R7. (L) A sevhsG4-SerS eye. Arrows point to ommatidia with extra photoreceptor cells.
Expression of DlS and SerS under the glass promoter interferes with eye development
The expression analysis of the secreted DL and SER was extended by expressing DlS or SerS under the glass promoter (pGMR), which has a different expression profile from the sev promoter. The GMR promoter is active in all cells posterior to the furrow (Hay et al., 1994). Rough eye phenotypes were observed in our GMR transgenics. About 70% of the independent GMR-DlS transgenic lines show mild eye phenotypes, consisting of irregular ommatidial shape, occasional ommatidial fusion and duplication of bristles (Fig. 3D,E). Tangential sections of these rough eyes show that the number of photoreceptor cells in each ommatidium is not affected (total ommatidia scored n=211). However, pigment cell loss was observed reflecting the ommatidial fusion seen externally (Fig. 3F).
About 65% of the GMR-SerS lines show eye phenotypes, mostly mild and very similar to the GMR-DlS phenotypes. The eyes with relatively stronger phenotypes are glossy as a result of ‘melting’ and smoothening of the lens material over ommatidial boundaries. They also contain large patches of bristles with abnormal shafts (Fig. 3G,H). There are some duplicated bristles at the periphery of the eye, more anteriorly than posteriorly. This exterior phenotype of GMR-SerS is very reminiscent of the eye phenotype of alleles of facet, a hypomorphic mutation in N (Markopoulou et al., 1989). Sectioning the eyes that have severe phenotypes reveals that 76% of the total ommatidia (n=490) have the proper number of photoreceptor cells. Among the 24% with abnormal numbers of photoreceptors, 17% (n=83) have fewer photoreceptor cells than normal. The most common phenotype is missing R7, the last photoreceptor cell to be determined (Fig. 3I).
Expression of secreted ligands in third instar eye discs under gal4-UAS control
Both DlS and SerS cDNAs were also expressed under the control of the UAS promoter (Brand and Perrimon, 1993). These transgenic lines were subsequently crossed to the sevhs-gal4 (sevhsG4) line. We refer to the combination of sevhsG4 and UAS-DlS, or UAS-SerS, as sevhsG4-DlS, or sevhsG4-SerS, respectively. In both cases, rough eye phenotypes were observed (Fig. 3J,K, and data not shown).
The sevhsG4 line drives the expression of the ligands under the control of a sevenless/heatshock hybrid promoter. Even though this promoter has a similar expression pattern as the sev cassette used above, we presume that the expression level is substantially higher leading to the mutant phenotypes. The sevhsG4-DlS eyes are similar to GMR-DlS, except for the higher penetrance of the ommatidial fusion phenotype in sevhsG4-DlS (data not shown). However, expression of sevhsG4-SerS led to a stronger rough eye phenotype than GMR-SerS (Fig. 3J,K). Sections of the sevhsG4-SerS eyes revealed extra photoreceptor cell defects (Fig. 3L). Both the external structure and the photoreceptor composition of sevhsG4-SerS eyes are similar to those observed in sev-SerTM eyes, a dominant negative Ser mutant (Sun and Artavanis-Tsakonas, 1996).
Analysis of the genetic interactions and downstream target gene expression
The above observations indicate that the dominant phenotypes associated with the secreted ligands depend on the timing and level of their expression. Our results suggest that these mutant ligands may behave as dominant negative mutations. To investigate this possibility further, we examined the genetic behavior of DlS and SerS transgenic fly lines by crossing them to various mutants of N pathway genes.
Using the GMR driven transgenic lines, we found that the eye phenotypes of both GMR-DlS and GMR-SerS are enhanced by loss-of-function Dl and gain-of-function Su(H), consistent with the notion that they behave as loss-of-function N mutations in the eye (Fig. 4A-D, and Fortini and Artavanis-Tsakonas, 1994). In addition, sevhsG4-DlS and sevhsG4-SerS are suppressed by a duplication of Dl (Fig. 4E,F), suggesting that the secreted ligands perturb the function of full-length DL in the eye.
Genetic interactions between transgenics and N pathway mutants. SEM of: (A) A GMR-DlS/+ eye; (B) A GMR-DlS/+; Dl9p39/+ eye; (C) A GMR-SerS/+ eye; (D) A Su(H)T4/+; GMR-SerS/+ eye. For interaction crosses involving GMR-SerS, transgenic lines with relatively weak phenotypes were used. (E) An UAS-SerS/+; sevhs-gal4/+ eye; (F) An UAS-SerS/+; Dp(3;3)bxd110/sevhs-gal4 eye. Dp(3;3)bxd110 is a duplication of the chromosomal region containing Dl.
Genetic interactions between transgenics and N pathway mutants. SEM of: (A) A GMR-DlS/+ eye; (B) A GMR-DlS/+; Dl9p39/+ eye; (C) A GMR-SerS/+ eye; (D) A Su(H)T4/+; GMR-SerS/+ eye. For interaction crosses involving GMR-SerS, transgenic lines with relatively weak phenotypes were used. (E) An UAS-SerS/+; sevhs-gal4/+ eye; (F) An UAS-SerS/+; Dp(3;3)bxd110/sevhs-gal4 eye. Dp(3;3)bxd110 is a duplication of the chromosomal region containing Dl.
Previous studies have shown that Enhancer of split (E(spl)) bHLH genes are direct targets of N activity and their expression levels are positively regulated by N (Jennings et al., 1994, 1995; Sun and Artavanis-Tsakonas, 1996). We therefore addressed the behavior of the secreted ligands by assaying for their effects on E(spl) bHLH mδ expression in eye discs. We found that the mδ expression is largely unaffected in GMR-DlS or sevhsG4-DlS eye discs (Fig. 5A). In contrast, it is substantially suppressed posterior to the furrow in GMR-SerS discs, and more so in sevhsG4-SerS discs (Fig. 5B,C). The extent of the suppression seems to parallel the severity of the photoreceptor phenotype and corroborates the conclusion that SERS acts as an antagonist of N signaling.
Staining of E(spl)HLH Mδ in third instar eye discs. Posterior is to the right and the morphogenetic furrow is marked by strong staining on the left in all panels. (A) A GMR-DlS eye disc. Staining is similar to wild type. (B) A GMR-SerS eye disc. Expression is slightly down regulated. (C) A sevG4-SerS eye disc. The expression posterior to the furrow is largely suppressed.
Staining of E(spl)HLH Mδ in third instar eye discs. Posterior is to the right and the morphogenetic furrow is marked by strong staining on the left in all panels. (A) A GMR-DlS eye disc. Staining is similar to wild type. (B) A GMR-SerS eye disc. Expression is slightly down regulated. (C) A sevG4-SerS eye disc. The expression posterior to the furrow is largely suppressed.
Expression of secreted ligands leads to small eyes
We observed a small eye phenotype when DlS or SerS is expressed under T113-gal4 control (T113G4-DlS or T113G4-SerS, Fig. 6A,B). As compared to sevhs-gal4, the expression of T113-gal4 is detected earlier, starting from second instar in a very broad domain of the eye disc (data not shown). Examined more closely, the ommatidia in T113G4-DlS or T113G4-SerS eyes are aligned in a regular array, although each ommatidium is square instead of the normal hexagon shape (Fig. 6B). Sections of these adult eyes reveal wild-type photoreceptor cell composition and orientation (data not shown). SERS appears to be more potent in producing the small eye phenotype than DLS. A similar small eye phenotype was described in the pharate adults of a loss-of-function Ser mutant allele (Speicher et al., 1994), supporting the notion that both DLS and SERS function as dominant negative N ligands.
Adult eye phenotype of T113G4-SerS flies. (A) SEM of a T113G4-SerS adult eye. The overall size of the eye is reduced, while the ommatidia array remains regular. (B) A magnified view showing that each ommatidium is square instead of the normal hexagon shape.
Expression of DlS and SerS in the developing wing
The differing abilities of DLS and SERS to down-regulate E(spl)mδ expression in the eye reveals differences between the effects of the two ligands. This could reflect either a simple quantitative difference between the affinities of the two mutant ligands to influence N signaling, or a qualitative difference in their behavior in the developing eye. We attempted to address these questions and further explore the ligand activity by assaying the effects of DLS and SERS on wing development, a process which is also very sensitive to N signaling (de Celis and Garcia-Bellido, 1994). DlS and SerS were expressed in the wing using the gal4-UAS system (Brand and Perrimon, 1993). Here we describe the results from three gal4 lines: a vestigial-gal4 transgenic line (vgG4), which allows expression along the D/V boundary of the developing wing disc; a patched-gal4 line (ptcG4), which drives expression in a stripe of cells along the anterior/posterior border; and an A9-gal4 line (A9G4), which expresses predominantly in the dorsal wing compartment. We refer to the combination of the above gal4 lines and the UAS-DlS line as vgG4-DlS, ptcG4-DlS and A9G4-DlS. Similar nomenclature is used for combinations involving UAS-SerS. The following results are also supported by our data obtained using several other gal4 lines with distinct expression patterns in the developing eye, wing and the notum.
DLS and SERS perturb N function in wing margin formation
Expression of vgG4-DlS resulted in phenotypes ranging from a missing anterior wing margin to narrow wings (Fig. 7A). Expression of vgG4-SerS led to a more severe margin loss (Fig. 7B). These phenotypes are similar to wing margin loss caused by loss-of-function mutations in N or Ser (Lindsley and Zimm, 1992). They also differ from the phenotype produced when full-length Ser is expressed using the same gal4 line (vgG4-Ser). In vgG4-Ser flies, the wing blades are expanded, particularly at the posterior margin (Fig. 7C). Furthermore, the phenotype of vgG4-SerS is substantially suppressed when full-length Ser is expressed simultaneously under the same gal4 control (Fig. 7D). These findings demonstrate that the expression of either DlS or SerS at the D/V boundary interferes with proper N function during wing margin formation.
Adult wing phenotypes resulting from the expression of the secreted ligands. All flies were raised at 25°C. (A) vgG4-DlS. (B) vgG4-SerS. (C) vgG4-Ser. (D) vgG4-SerS, vgG4-Ser. (E) A9G4-DlS. (F) A9G4-DlS; Dl9p39/+. (G) A9G4-DlS, Dp(3;3)bxd110/+. (H) A9G4-SerS.
DLS and SERS down-regulate the expression of margin specific genes
A set of genes involved in the formation of the wing margin have been implicated as downstream targets of N function in wing development. They are down-regulated when N signaling is lost but are ectopically up-regulated in certain compartments when N is activated by the ectopic expression of full-length DL or SER (de Celis et al., 1996; Diaz-Benjumea and Cohen, 1995; Doherty et al., 1996). Thus, the expression levels of these genes in transgenic wing discs are good parameters of how the secreted ligands affect N activity. Using antibodies and lacZ lines, we found that the expression of wg, vg and cut at the wing margin was down-regulated in vgG4-DlS or vgG4-SerS discs (Fig. 8A,B, and data not shown).
Wing margin gene expression in transgenic wing discs. Anterior is to the left and ventral wing compartment is up in all panels. (A) WG expression in a wild type third instar wing disc detected using an anti-WG antibody. Arrow points to the expression stripe across the D/V border. (B) WG expression in a vgG4-DlS wing disc. Note that the expression is largely disrupted. (C) WG (green) and DLS (red, detected using an anti-MYC antibody) expression in an A9G4-DlS disc. A9G4 expresses predominantly in the dorsal wing pouch. The WG stripe across dorsal/ventral boundary is intact. (D) WG (green) and SERS (red, detected using an anti-MYC antibody) expression in an A9G4-SerS wing disc. Note that WG expression at the D/V boundary is patchy (arrow). (E) WG expression in a ptcG4-SerS disc. Arrow points to the residual WG expression in the anterior edge of ptcG4 expression stripe. Arrowhead points to WG at the dorsal hinge region that is not down-regulated by overlapping SERS. The rest of the WG expression at the hinge is out of the focal plane. No apparent morphological defect was observed in this disc. (F) WG expression (green) and SERS expression (red) in the same ptcG4-SerS disc. Note that SERS is at a higher level at the posterior edge than the anterior edge. (G) WG (green) and SERS (red) in a ptcG4-SerS disc more severely affected by the transgene. (H) Nomarski image of the same disc showing the abnormal constriction created at the anterior/posterior boundary (arrow). Due to this morphological defect, we could not determine whether SERS also affects WG expression in a non-autonomous fashion. (I) WG expression in a ptcG4-DlS disc. Arrow points to the region where WG is reduced. (J) WG (green) and DLS (red) in the same ptcG4-DlS disc. Note that WG protein is not affected in the DLS stripe, while being down-regulated anterior to it. The down-regulation of WG did not correlate with any apparent morphological defect. (K) WG (green) and DLS (red) in a ptcG4-DlS disc more severely affected by the transgene. Almost all the WG expression at the D/V boundary is eliminated, while WG at the hinge region remains. (L) Nomarski image of the same disc showing the size reduction of the wing pouch.
Wing margin gene expression in transgenic wing discs. Anterior is to the left and ventral wing compartment is up in all panels. (A) WG expression in a wild type third instar wing disc detected using an anti-WG antibody. Arrow points to the expression stripe across the D/V border. (B) WG expression in a vgG4-DlS wing disc. Note that the expression is largely disrupted. (C) WG (green) and DLS (red, detected using an anti-MYC antibody) expression in an A9G4-DlS disc. A9G4 expresses predominantly in the dorsal wing pouch. The WG stripe across dorsal/ventral boundary is intact. (D) WG (green) and SERS (red, detected using an anti-MYC antibody) expression in an A9G4-SerS wing disc. Note that WG expression at the D/V boundary is patchy (arrow). (E) WG expression in a ptcG4-SerS disc. Arrow points to the residual WG expression in the anterior edge of ptcG4 expression stripe. Arrowhead points to WG at the dorsal hinge region that is not down-regulated by overlapping SERS. The rest of the WG expression at the hinge is out of the focal plane. No apparent morphological defect was observed in this disc. (F) WG expression (green) and SERS expression (red) in the same ptcG4-SerS disc. Note that SERS is at a higher level at the posterior edge than the anterior edge. (G) WG (green) and SERS (red) in a ptcG4-SerS disc more severely affected by the transgene. (H) Nomarski image of the same disc showing the abnormal constriction created at the anterior/posterior boundary (arrow). Due to this morphological defect, we could not determine whether SERS also affects WG expression in a non-autonomous fashion. (I) WG expression in a ptcG4-DlS disc. Arrow points to the region where WG is reduced. (J) WG (green) and DLS (red) in the same ptcG4-DlS disc. Note that WG protein is not affected in the DLS stripe, while being down-regulated anterior to it. The down-regulation of WG did not correlate with any apparent morphological defect. (K) WG (green) and DLS (red) in a ptcG4-DlS disc more severely affected by the transgene. Almost all the WG expression at the D/V boundary is eliminated, while WG at the hinge region remains. (L) Nomarski image of the same disc showing the size reduction of the wing pouch.
Since it has been previously shown that the vg promoter/enhancer is influenced by N signaling (Kim et al., 1996), we needed to ensure that the effects we observed were not simply reflecting the regulation of the vgG4 expression by N. We therefore used additional gal4 lines to examine the effects of the secreted ligands on the wing margin. For example, the ptcG4 line permits expression in a stripe of cells along the anterior/posterior border of a third instar wing disc, with the highest level at the sharp posterior edge and gradually lower levels toward the anterior edge (Fig. 8F). We found that the expression of wing margin specific genes was disrupted in both ptcG4-DlS and ptcG4-SerS discs.
In ptcG4-SerS discs, wg expression is down-regulated where the SerS expressing stripe crosses the D/V border (Fig. 8E,F). In some cases, we also observed residual wg expression inside of the anterior border of the SERS stripe (Fig. 8E, arrow). This suggests that a high concentration of the mutant SERS protein, expressed by the ptcG4 driver in the posterior edge of the expression stripe, is required for the down-regulation of wing margin gene expression.
The expression of ptcG4-DlS can also suppress wg, but the pattern differs qualitatively from that observed with SERS (compare Fig. 8F with 8J). The expression of wg is not suppressed where the DLS stripe intersects with the D/V boundary. In the majority of the ptcG4-DlS discs, wg expression just anterior to the DLS stripe is down-regulated (Fig. 8I,J, arrow). In some cases, the wg margin expression in the anterior compartment is completely eliminated, but in other cases it resumes at a certain distance away from the DLS stripe (Fig. 8J).
We note that wg was not suppressed where either the DlS or SerS expression stripe intersects with the wg expression domain at the hinge region close to the boundary of the wing pouch (Fig. 8E, arrowhead). This suggests that the suppression of wg is mediated by factors present at the D/V boundary but absent from the hinge region. Additionally, in contrast to the induction of wing margin gene expression by full-length SER or DL in either the dorsal or ventral wing compartments (Diaz-Benjumea and Cohen, 1995; Doherty et al., 1996), SERS and DLS failed to induce their ectopic expression.
Finally, in strongly expressing ptcG4-SerS discs, the transgene frequently leads to the formation of an abnormal restriction at the anterior/posterior boundary (Fig. 8G,H). Both the dorsal and the ventral wing compartments are affected. Although no such ectopic restriction is observed in ptcG4-DlS discs, the overall sizes of the wing pouches in strongly expressing discs are much smaller than wild type (Fig. 8K,L).
We conclude that SERS and DLS act as antagonists of N signaling in wing margin formation. In addition, our analysis reveals qualitative differences in the behavior of the two mutant ligands.
Expression of secreted ligands interferes with wing vein specification
A difference in the activity of the two secreted mutant ligands was also observed when DlS and SerS were expressed under the control of A9G4 in the dorsal wing pouch. Raised at 25°C, A9G4-DlS flies show extra vein material (100% penetrance) and occasional nicking at the distal wing tip (20% penetrance, Fig. 7E), similar to the haplo-insufficient phenotype associated with loss-of-function Dl alleles. Genetically, the extra vein phenotype of A9G4-DlS is strongly enhanced by a loss-of-function Dl mutation and suppressed by a duplication of Dl (Fig. 7F,G), indicating that A9G4-DlS interferes with the function of the endogenous Dl. In A9G4-SerS lines, however, no extra vein material was observed along the length of the wing. Instead, we found extensive scalloping of the wing margin (Fig. 7H). Only in the veins that do reach the residual wing margin do we observe occasional small deltas at the tip.
Thus, when expressed in the dorsal compartment, the two secreted ligands reveal different properties, with DLS more potent in affecting vein restriction and SERS more effective in deleting the wing margin. Their different effects on wing margin parallel their abilities to regulate wg expression at the D/V boundary (Fig. 8C,D). The stripe of wg expression along the D/V border remains largely intact in A9G4-DlS discs, while being disrupted in A9G4-SerS discs. In conclusion, the expression of DlS and SerS, using the A9G4 line, resulted in phenotypes mimicking loss-of-function mutations in the N pathway.
DISCUSSION
Previous studies involving the expression of membrane tethered N ligands lacking the intracellular domains demonstrated that they behave as negative regulators of N signaling in Drosophila and vertebrates (Chitnis et al., 1995; Sun and Artavanis-Tsakonas, 1996). In the developing Drosophila eye, the expression of these truncated forms of Dl and Ser under the sev promoter cassette (sev-DlTM and sev-SerTM) led to extra photoreceptor cells as a result of mispecification of non-neuronal precursors into neuronal cell fates. The results we present here are consistent with the notion that the presence of the intracellular domain is functionally important and that the secreted forms of the ligands act as antagonists of N signaling. These findings are in contrast to results in C. elegans where the secreted extracellular domains can function as full-length ligands (Fitzgerald and Greenwald, 1995).
The phenotypes associated with DLS and SERS were only comparable in strength to the membrane bound forms DLTM and SERTM when expressed under sevhsG4, which affords high levels of expression. Both membrane bound and secreted forms behave genetically as loss-of-function mutations and the differences we observed between them appear to be quantitative rather than qualitative in nature, consistent with the notion that membrane attachment in DLTM and SERTM results in higher effective concentrations of the mutant ligands.
Several distinct lines of evidence demonstrate that the secreted ligand forms examined here behave as antagonists of N signaling. The E(spl) mδ expression in the eye is downregulated by SERS, opposite to the effect of the constitutively activated form of N (Jennings et al., 1994; Sun and Artavanis-Tsakonas, 1996). The expression of DLS or SERS across the D/V boundary of the developing wing results in adult phenotypes mimicking loss-of-function N mutations. This is paralleled by their ability to down-regulate the expression of the wing margin specific genes, vg, wg and cut. In A9G4-DlS wings we observed extra vein material in close vicinity of the endogenous veins, a characteristic of loss-of-function alleles of Dl. The opposite phenotype is seen when full-length Dl is expressed under the control of the same gal4 line (data not shown). It is thought that vein specification is similar to neurogenesis in the embryo. A field of equipotent cells are generated along the presumptive vein and their segregation into a vein-specific pathway is controlled by a lateral inhibition mechanism through the N pathway (de Celis and Garcia-Bellido, 1994). Thus, DlS expression in the wing pouch region may affect the endogenous Dl expressed along the presumptive wing veins, interfering with the lateral inhibition process resulting in extra veins. Consistent with this, the phenotype can be suppressed by a duplication of Dl. Our observations are also consistent with the analysis of Beaded of Goldschmidt, an antimorphic allele of Ser that produces a secreted form of the ligand (Hukriede and Fleming, 1997).
Previous studies have shown that SER signals from the dorsal compartment, whereas DL signals from the ventral compartment to affect wing margin formation through N (Diaz-Benjumea and Cohen, 1995; Doherty et al.; 1996, Kim et al., 1995). This would predict that when the dominant negative DlS mutant is expressed in the dorsal wing pouch (A9G4-DlS), it should not affect N at the wing margin. In contrast, SERS expression (A9G4-SerS) would result in N inactivation across the wing margin. Our results are consistent with this prediction. Apart from occasional nicks at the wing tip, the A9G4-DlS wing margin is intact, while the wing margin in A9G4-SerS is profoundly affected.
While our experiments cannot rigorously distinguish whether the difference we detect in the ability of DLS and SERS to regulate E(spl) mδ expression is qualitative or quantitative, other observations imply that they interact with different effectors. The result that ptcG4-SerS down-regulates wg within its expression stripe, while ptcG4-DlS downregulates wg anterior to the stripe, cannot be explained by a simple difference in affinity. As only wg expression in the anterior compartment is down-regulated in ptcG4-DlS, this suggests that a factor (or factors) that functions in the anterior compartment is required for the suppression. The difference in the behavior of DLS and SERS in suppressing wg expression is paralleled by a difference in disc morphology in the strongly expressing lines. ptcG4-SerS leads to the formation of an ectopic constriction along the expression stripe, in contrast to ptcG4-DlS which leads to an apparent size reduction of the wing pouch.
Unlike in A9G4-DlS, A9G4-SerS, wings do not show an extra vein phenotype, as if SerS expression in the dorsal compartment is not capable of interfering with endogenous N function to affect vein formation. A possible explanation for this behavior is based on a recent study of the regulation of SER function by the secreted protein FRINGE (FNG), (Fleming et al., 1997; Irvine and Wieschaus, 1994). It was shown that FNG, which is normally expressed in the dorsal wing pouch, can suppress SER function at the protein level through the DSL domain (Fleming et al., 1997). Interestingly, this suppression is not effective on the DSL domain of DL. Thus FNG may suppress the function of SERS, which still contains a DSL domain, by preventing it from interfering with N signaling in the dorsal wing compartment. We find that when the ectopic constriction forms in ptc-SerS wing discs, it usually affects both the dorsal and ventral compartments, consistent with the idea that SERS under these experimental conditions escapes the suppression of FNG.
Finally, a difference in the behavior of DLS and SERS is revealed by the observation that SERS is processed in discs, while DLS is not. The functional significance of this biochemical difference is unclear, especially since we do not know whether this cleavage is also carried out with the endogenous SER. We observed similar processing in eye discs (sev-SerS) and in wing discs (A9G4-SerS). This is noteworthy given that FNG is not present in the eye, while it is present in the dorsal wing compartment (Irvine and Wieschaus, 1994), arguing that the putative protein interaction between FNG and SERS does not affect this processing.
The mechanism of the antagonistic action of the secreted ligand forms is not clear, but two possibilities are of particular interest. First, it is possible that the secreted ligands function by binding to N and blocking the binding of endogenous ligands to the receptor, a notion favored by the accompanying study (Hukriede et al., 1997). Alternatively, DLS or SERS may bind to endogenous DL or SER, respectively, and prevent them from signaling to the N receptor. The differences mentioned above in the behavior of DLS and SERS seem to favor the notion that they act differentially on the endogenous ligands or other extracellular molecules such as FNG.
Analyses in C. elegans showed that secreted LAG-2 or APX-1 serve as wild-type or gain-of-function ligands, in contrast to the dominant negative behaviour seen with DLS and SERS. A trivial explanation of this apparent discrepancy is that the phenotypes scored in the two systems reflect fundamentally different processes. For example, if the action of the ligand in one system is mediated by unidentified effectors, then it is conceivable that the deletion of the intracellular domain of the ligand does not affect one system but affects the other. The other possibility is that the different behavior reflects the considerable disparity between the molecular composition of the N ligands in Drosophila and C. elegans (Tax et al., 1994). Overall, the two C. elegans ligands contain fewer EGF-like repeats and are much smaller in size. It is therefore conceivable that the structure of the extracellular domain may be less sensitive to structural alterations in the intracellular domain.
Compared to C. elegans, the Drosophila N ligands have more similarities with the vertebrate N ligands, both in terms of structure and of function (Henrique et al., 1995; Lindsell et al., 1995). It has been shown that membrane bound extracellular DL behaves as a dominant-negative ligand in Xenopus (Chitnis et al., 1995), similar to our results in Drosophila (Sun and Artavanis-Tsakonas, 1996). In this respect the results of two recent studies which link the human syndrome Alagille to human jagged 1 are of particular interest (Li et al., 1997; Oda et al., 1997). Point mutations, which would predict mutant jagged proteins very similar to the secreted forms we characterize here, have been associated with the Alagille syndrome. In this case, the fact that heterozygocity of deletions involving jagged 1 suggests that haploinsuficiency of the gene may underly the syndrome. However our data and the accompanying study, clearly show that secreted, truncated forms of both Notch ligands act as antagonists in Drosophila. It is therefore possibile that the Alagille point mutations are dominant negatives and that the phenotypes associated with the haploinsuficiency of the jagged 1 region may be associated with a synthetic phenotype, since the deletions at hand involve hundreds of kilobases presumably affecting many genes in addition to jagged 1. Additional studies will be necessary to determine if the Alagille mutations are acting as dominant negatives or whether they reflect haploinsufficiency in humans. This issue is of particular importance, especially in order to consider therapeutic avenues.
ACKNOWLEDGEMENTS
We thank Dr S. Bray for the anti-E(spl) Mδ antibody, Dr R. Nusse for the anti-WG antibody, and B. Piekos for SEM assistance. We also thank Chris Day, Deborah Eastman, Mark Fortini, Stewart Frankel, Masahiro Go, Caryn Purcell for critical comments on the manuscript, and Laurent Caron for technical assistance. We are very grateful to Dr Robert Fleming and his lab for fruitful discussions and sharing unpublished results throughout the course of this work. X. S. was supported by a Yale University doctoral fellowship. S. A.-T. is supported by the Howard Hughes Medical Institute and by N.I.H. grant NS26084.