Abstract
During development, the Notch receptor regulates many cell fate decisions by a signaling pathway that has been conserved during evolution. One positive regulator of Notch is Deltex, a cytoplasmic, zinc finger domain protein, which binds to the intracellular domain of Notch. Phenotypes resulting from mutations in deltex resemble loss-of-function Notch phenotypes and are suppressed by the mutation Suppressor of deltex [Su(dx)]. Homozygous Su(dx) mutations result in wing-vein phenotypes and interact genetically with Notch pathway genes. We have previously defined Su(dx) genetically as a negative regulator of Notch signaling. Here we present the molecular identification of the Su(dx) gene product. Su(dx) belongs to a family of E3 ubiquitin ligase proteins containing membrane-targeting C2 domains and WW domains that mediate protein-protein interactions through recognition of proline-rich peptide sequences. We have identified a seven-codon deletion in a Su(dx) mutant allele and we show that expression of Su(dx) cDNA rescues Su(dx) mutant phenotypes. Overexpression of Su(dx) also results in ectopic vein differentiation, wing margin loss, and wing growth phenotypes and enhances the phenotypes of loss-of-function mutations in Notch, evidence that supports the conclusion that Su(dx) has a role in the downregulation of Notch signaling.
DURING development, the regulation and integration of cell-to-cell signaling pathways coordinate the program of events that specifies different cell types. Notch is a transmembrane receptor protein that controls the timing and outcome of cell differentiation decisions (Artavanis-Tsakonaset al. 1995). Inappropriate activation of Notch has been associated with oncogenesis in humans and mice, causing T-cell lymphoma and mammary gland tumors (Ellisenet al. 1991; Jhappanet al. 1992). Normally Notch functions pleiotropically throughout development in many tissues. In the Drosophila wing, Notch is involved in specifying the dorsal-ventral margin and the vein-intervein boundary (Kimet al. 1995; de Celis et al. 1996, 1997; Huppertet al. 1997). Loss of Notch signaling results in thickened veins. In contrast, gain-of-function Notch mutations result in failure of vein cell differentiation, causing gaps in the veins. A number of components and regulators of Notch signaling have been identified. Suppressor of Hairless [Su(H)] is a downstream transcription factor whose function is activated by Notch pathway signaling. The nuclear localization, in response to ligand binding, of both Su(H) and the intracellular domain of Notch has been reported (Fortini and Artavanis-Tsakonas 1994; Schroeteret al. 1998; Struhl and Adachi 1998) and one or both of these mechanisms may be involved in the regulation of Su(H) activity. Hairless is a negative regulator of the Notch signal and represses Su(H) function by direct protein-protein interaction (Brouet al. 1994). A positive regulator of Notch signaling, Deltex, has also been identified (Busseauet al. 1994; Matsunoet al. 1995). Deltex is a large cytoplasmic protein that contains a C-terminal zinc finger domain. Notch and Deltex proteins co-localize within the cell and the N-terminal domain of Deltex has been shown to bind to the ankyrin repeat region of the intracellular domain of Notch. The deltex (dx) mutant phenotype resembles Notch loss-of-function wing phenotypes, as it displays thickened veins and wing margin loss. deltex mutations strongly enhance the phenotypes of mutations in other Notch pathway genes.
The Su(dx) mutation was originally identified in Drosophila melanogaster as a second chromosome, dominant suppressor of deltex (Morganet al. 1931; Lindsley and Zimm 1992), and further alleles have been isolated more recently (Busseauet al. 1994; Fostieret al. 1998). Homozygous and heteroallelic combinations of Su(dx) mutations appear nearly wild type at 25°, with some mutant combinations displaying a forked cross-vein phenotype (Fostieret al. 1998). At 29°, however, homozygous Su(dx)sp and heteroallelic combinations of Su(dx) alleles have a wing-vein gap phenotype similar to those resulting from gain-of-function mutations of Notch (Fostieret al. 1998). While mutations in Su(dx) alone appear only to affect wing vein differentiation, a more widespread role has been detected by combining Su(dx) with mutations in other Notch pathway genes (Fostieret al. 1998). Apart from the genetic interaction with deltex, the Su(dx) mutation also suppresses the dominant wing margin loss phenotype of a Notch deficiency and enhances the Notch gain-of-function AxE2 mutant. Thus the Su(dx) protein can be characterized genetically as a negative regulator of Notch receptor signaling (Fostieret al. 1998). At 25°, homozygous Su(dx)sp and heteroallelic combinations with other Su(dx) alleles can rescue the lethal interaction between deltex and notchoid (nd, a hypomorphic Notch allele). Using this genetic assay, we previously have identified deficiency alleles of Su(dx) that allowed the location of the gene to be resolved to 22B4-C2 (Fostieret al. 1998). In this article we describe the cloning of the Su(dx) gene, its expression pattern in the developing wing, and gain-of-function phenotypes resulting from ectopic expression of the protein.
MATERIALS AND METHODS
Mutant stocks: The Su(dx)32 allele was generated in a screen for rescue of the lethality of nd dx/Y; Su(dx)sp/+ flies [protocol as described in Fostier et al. (1998)]. nd and dxsm stocks were as previously described (Busseauet al. 1994; Hinget al. 1994). UAS-Su(dx)IIa and UAS-Su(dx)IIIh are transgenes expressing Su(dx) cDNA under the control of the Gal4-UAS (Brand and Perrimon 1993) and inserted into the second and third chromosomes, respectively. To obtain transgenic lines, 2b1a cDNA containing the full open reading frame was inserted into the pUAST vector and injected into embryos using standard methodology. The MS1096Gal4 dxsm and Su(dx)sp UAS-Su(dx)IIa lines were constructed by recombination. All stocks were grown on standard Drosophila media and crosses were performed at 25°, apart from the complementation test of Su(dx)sp/Su(dx)32, which was at 29°, and the PtcGal4-driven expression, where flies were transferred to 29° during the third instar larval stage. Wings were dissected from adult flies and mounted in a 1:1 mixture of Canadian balsam and methyl salicylate. Bright field images were acquired using a Zeiss Axioskop microscope and Improvision Openlab image capture software.
Cloning of Suppressor of deltex: Genomic Southern blots were performed on wild-type or mutant DNA, extracted, and blotted using standard methodologies. Clones 2b1a, 31a, and 10a1a were obtained following screening of 500,000 clones from a random primed λ-Gt10 Drosophila imaginal disc library (provided by Tian Xu, Yale University) with a genomic fragment PA1 derived from P1 clone 42-59 [Berkeley Drosophila Genome Project (BDGP resource)]. Clone 15-2a was obtained by screening an oligo(dT)-primed Drosophila larval cDNA library (provided by Nick Brown, University of Cambridge). Twenty-seven pools, each containing ∼40,000 clones, were digested with HindIII, separated by electrophoresis, blotted, and probed with 32P-labeled 2b1a. A pool containing the longest clone was further screened to obtain clone 15-2a. Homologous cDNA sequences were identified by screening the Swissprot and GenBank databases using Blastx. Sequences were aligned using ClustalW (EBI homepage). The Su(dx)sp mutant gene was amplified in overlapping segments by PCR of genomic DNA and the products were directly sequenced.
All sequencing was performed using an Applied Biosciences (Foster City, CA) automated sequencer. Phylogenetic analysis was performed using the PHYLIP program package (Felsenstein 1989). The sequences were first aligned using Clustal. The sequence corresponding to WWP1 contained on the GenBank database was from a clone that was incomplete at the 5′ and 3′ ends. We have extended this protein sequence further 3′ to include the remainder of the C-terminal end, by sequencing an overlapping expressed sequence tag available from the Merck-Washington University EST project. The aligned sequences were input into PHYLIP. To test the reliability of the tree, bootstrap analysis was performed on the input data set to generate 100 replicate sets with replacement, and a consensus tree was calculated.
In situ hybridization: In situ hybridization was performed on pupal wing discs staged at 29°, 24 hr after puparium formation. Pupal cases were opened at the anterior and posterior side and fixed with 4% formaldehyde (FA; Polysciences, Warrington, PA) in phosphate-buffered saline (PBS) overnight at 4°. The cuticle was then dissected from the wings and the pupae were stored in methanol at -20° until use. Before hybridization, pupae were rehydrated in methanol/PBS 70%/30% and 30%/70% and transferred to PBS, fixed in 4% FA/PBS for 10 min at room temperature (RT), washed in PBT (PBS, 0.1% Tween) three times for 10 min, washed in 50% hybridization solution (HS; 50% formamide, 5× SSC, 0.1% Tween, 0.1 mg/ml tRNA, 50 μg/ml heparin adjusted to pH 4.5 with 1 m citric acid) in PBT for 10 min at RT, washed in HS for 10 min at RT, and prehybridized in HS for 1 hr at 70°. Pupae were hybridized overnight at 70° with a digoxigenin-labeled (Boehringer Mannheim, Mannheim, Germany) 2b1a antisense probe and then washed two times (20 min in HS at 70°, 20 min in 50% HS in PBT at 70°) and three times for 20 min in PBT at RT on a rotating wheel. Next the pupae were incubated with an alkaline phosphatase-conjugated anti-digoxigenin antibody (Boehringer Mannheim) at 1:2000 dilution in PBT/1% normal goat serum (Sigma, St. Louis), for 1 hr 30 min at RT and washed three times for 20 min at RT on a rotating wheel. The antibody conjugate was detected using the substrate NBT/BCIP (Boehringer Mannheim).
RESULTS
Cloning of Su(dx): Previously we have described a genetic assay for Su(dx) alleles based on the rescue of the lethality caused by the combination of notchoid and deltex mutations (Fostieret al. 1998). Mutant alleles derived from this screen had visible polytene chromosome rearrangements or deficiencies mapping to the 22B4-C2 region of chromosome 2. We have screened for further Su(dx) alleles using this method and have isolated a mutation with no visible chromosome rearrangement (data not shown). The mutation dominantly suppressed the deltex phenotype and, in addition to the notchoid deltex rescue assay, this mutation failed to complement Su(dx)sp for its temperature-sensitive wing-vein gap phenotype (Figure 1). Homozygous Su(dx) flies display a wing vein gap phenotype at 25° in the background of heterozygous notchoid. The new mutation failed to complement Su(dx)sp for this interaction phenotype (Figure 4d). Thus we conclude that the new mutation is a Su(dx) allele, which we have designated as Su(dx)32. The Su(dx)32 mutant was recessive viable at 25° with a forked cross-vein phenotype. Flies that were homozygous for the Su(dx)32 mutant chromosome were not viable at 29°; however, in a heteroallelic combination with Su(dx)7, a deficiency allele, viable flies eclosed at 29°, with a wing vein gap phenotype. We conclude therefore that the temperature-dependent lethality observed was due to a second mutation on this chromosome, independent of Su(dx). We have previously mapped the Su(dx) gene to the 22B4-22C2 region of chromosome 2, and Southern blotting of the Su(dx)32 mutant DNA, with genomic clones from the 22B4-22C2 region, identified a chromosome rearrangement within an 18-kb subfragment (PA1) of the P1 clone 42-59 that maps to 22C1-2 (from the BDGP resource). Using the PA1 insert as a probe, three overlapping cDNA clones, 2b1a, 31a, and 10a1a, were obtained by screening of a random primed λGt10 imaginal disc library (Figure 3). A further overlapping clone (15-2a) was obtained by screening an oligo(dT) primed Drosophila larval cDNA library, and this clone contained a poly(A) tailed 3′ end. A Drosophila expressed sequence tag, LD32282 (BDGP) that extended the 5′ end of the cDNA by a further 80 bases was also identified. The five overlapping cDNA clones encode a 949-amino-acid open reading frame (Figure 2) within a 4.4-kb transcript, in agreement with a transcript observed on Northern blots (data not shown). The expression pattern of the cDNA was analyzed at different stages of Drosophila development; however, a specific expression pattern was detected only following in situ hybridization on pupal wings. An increased staining along the boundary between the vein and intervein territories was observed (Figure 3). This expression pattern in the pupal wing is similar to that previously observed for Notch (de Celiset al. 1997; Huppertet al. 1997) and consistent with a role in the regulation of Notch receptor signaling. Expression of the gene elsewhere may be low and ubiquitous but we cannot distinguish this from background staining.
—Su(dx)32 fails to complement the Su(dx)sp allele. (a) Wild-type Drosophila wing. Longitudinal veins I-V are indicated. (b) Wing from Su(dx)sp fly showing failure of vein cell differentiation at the nonpermissive temperature. (c) Wing from Su(dx)sp/Su(dx)32 fly showing failure to complement for the temperature sensitive Su(dx)sp wing-vein gap phenotype.
The cDNA of the candidate gene detected in Southern blots the rearrangement previously observed in Su(dx)32 mutant flies (Figure 3). However, we have been unable to detect any rearrangements in genomic DNA from the Su(dx)sp allele. The Su(dx)sp mutant gene was therefore amplified in overlapping segments by PCR of genomic DNA, and the products were directly sequenced. Using this approach, we identified a seven-codon deletion within the protein coding sequence. This small deletion preserved the reading frame and left the remainder of the protein intact (Figure 2).
Rescue of Su(dx) mutant phenotypes: To obtain further confirmation of the identity of the gene, we have constructed a UAS-Sudx transgenic Drosophila line to express the Su(dx) cDNA under the control of the Gal4 responsive promoter and to enable the rescue of Su(dx) mutant phenotypes. Injection of Drosophila embryos was performed with the pUAST plasmid (Brand and Perrimon 1993) carrying the 2b1a cDNA insert, in a P-element vector. The 2b1a clone contains the complete protein coding sequence, including the initiating methionine. An authentic 3′ end was supplied by the pUAST vector. Following P-element-mediated integration into the germ line, several independently transformed fly lines were obtained, which carried the UAS-Su(dx) transgene on the X, second, or third chromosomes. To allow the rescue experiments to be performed, the second chromosome UAS-Su(dx)IIa transgene was recombined with the Su(dx)sp mutation. Two assays for rescue of Su(dx) were performed using the MS1096Gal4 line (Capdevila and Guerrero 1994) to express the Su(dx) protein (Figure 4). The MS1096Gal4 line expresses Gal4 widely during wing development and for these experiments the MS1096Gal4 transgene was recombined on the X chromosome with either notchoid or deltex mutations. Expression of the cDNA in the developing wing rescued the dominant suppression of deltex by the Su(dx)sp allele (Figure 4c). The expressed cDNA also rescued the longitudinal vein gap phenotype displayed in nd/+; Su(dx)sp/Su(dx)32 flies (Figure 4e). Neither the MS1096Gal4 nor UAS-Su(dx)IIa transgenes alone interfered with the Su(dx) mutant phenotypes. Taken together, the molecular analysis of the mutations and the rescue of the Su(dx) phenotypes enabled us to conclude that the cloned cDNA does correspond to the Suppressor of deltex gene.
Sequence analysis of the Su(dx) gene product: Database searching identified three different classes of motifs within the Su(dx) open reading frame (Figure 2). At the N terminus there is a C2 domain (Nalefski and Falke 1996), which has a membrane-targeting function. This is followed by four WW domains that bind to proline-rich sequences (Sudolet al. 1995) and at the C terminus by an E3 ubiquitin ligase or HECT domain (Huibregtseet al. 1995). Ubiquitin ligases mediate ubiquitination of target proteins and have been shown to regulate their degradation by the proteosome organelle (Palombellaet al. 1994; Nefsky and Beach 1996). The small deletion located in the Su(dx)sp allele lies within the ubiquitin ligase domain, while the lipid-binding and protein-protein interaction motifs remain intact.
—Sequence analysis of the Su(dx) gene. (a) Amino acid sequence of Drosophila Su(dx) protein: Su(dx) is a 949-amino-acid protein containing a C2 domain, four WW modules (open boxes) and a C-terminal HECT domain (filled box). The position of the 7-amino-acid deletion in Su(dx)sp mutant is underlined. The sequence is aligned to the two human proteins WWP1 and WWP2 and the mouse protein Itchy. Positions of sequence identity are marked with an asterisk. We have extended the previously published amino acid sequence of WWP1 up to its C terminus to complete the HECT domain sequence by sequencing an overlapping EST. (b) A phylogenetic tree generated by the program PHY-LIP (Felsenstein 1989) to show relatedness of Su(dx) with mammalian Nedd4 family proteins. The numbers at the branch forks represent the percentage of trees conforming to this consensus tree derived from bootstrap sampling of the data. The Drosophila Su(dx) protein clusters with WWP1, WWP2, Itchy, and aip4. Mouse Nedd4 and the human KIAA0093 protein form a separate cluster. UreB1 is a distant HECT domain-related protein outside of the Nedd4 family and is included as an outgroup.
—DNA rearrangement in the Su(dx)32 mutant allele. (a) Southern blot of PstI-digested genomic DNA from Su(dx) mutants. Lanes i and iii contain wild-type DNA. Lanes ii and iv contain genomic DNA from Su(dx)32 flies. Lanes i and ii are probed with the 2b1a cDNA and lanes iii and iv are probed with the PstI subclone P3. The rearrangement observed for Su(dx)32 flies has been confirmed by analysis with several different enzymes (data not shown). (b) Restriction map of the Su(dx) gene. Underlined are the positions of the PstI restriction fragments (P1 to P5) that correspond to the bands indicated on the Southern blot. The locations of the Su(dx)sp and Su(dx)32 mutations are indicated. (c) Exon and domain structure of the Su(dx) gene. The positions of the introns and exons (solid boxes) are indicated with respect to the restriction map above. The domain structure of Su(dx) is shown with a single C2 and HECT domain and four WW domains. The locations of the domains with respect to the exon structure of the gene are as indicated. The overlapping cDNA clones used to construct the full sequence are shown below, approximately to scale. (d) The Su(dx) expression pattern in the pupal wing obtained by in situ hybridization using a digoxigenin-labeled antisense RNA probe generated from the 2b1a cDNA clone.
The combination of C2, WW, and HECT domains places Su(dx) within the Nedd4 family (Kumaret al. 1992). Members of this family have been associated with the regulation of a variety of important biological functions including cell proliferation, inflammation, and protein localization (Nefsky and Beach 1996; Zoladeket al. 1997; Perryet al. 1998). To compare Su(dx) with known mammalian Nedd4 family proteins we constructed a phylogenetic tree of family members using the PHYLIP program package (Felsenstein 1989). The Drosophila Su(dx) protein clusters reliably in this phylogeny with WWP1, WWP2, the mouse protein Itchy, and its closely related human homologue aip4 (Figure 2). WWP1 and WWP2 are human proteins with no known function, which were identified in a screen for proline-rich peptide-binding domains (Pirrozi et al. 1997). Mutations in the itchy locus result in hyperplasia of lymphoid, hemopoetic, and epithelial cells but the mechanism of action of the Itchy protein and the pathways that it regulates are currently unknown (Perryet al. 1998).
Phenotypes resulting from ectopic expression of Su(dx): We have examined the phenotypes resulting from the overexpression of the Su(dx) cDNA in the developing wing. When the expression of Su(dx) was driven along the anterior-posterior compartment boundary of the wing, under the control of the PatchedGal4 (PtcGal4) line, ectopic wing vein formation and wing-margin-loss phenotypes were observed (Figure 5), as would be expected from a downregulation of the Notch pathway. Expression of Su(dx) throughout the developing wing using the MS1096Gal4 line (Capdevila and Guerrero 1994) also resulted in an ectopic vein phenotype but not in wing margin loss. Su(dx) expression also caused a distinct alteration in wing shape. The PtcGal4-driven expression of Su(dx) resulted in a localized broadening of the wing between the third and fourth longitudinal veins, i.e., along the AP boundary (Figure 5b). This was particularly evident in the proximal portion, near the location of the anterior cross vein, which was frequently absent in these flies. The broadening of this region is due to an increased number of cells (determined by counting of wing hairs), rather than abnormal cell growth.
—Rescue of Su(dx) mutant phenotypes. (a) Wing from MS1096Gal4 dxsm/Y fly showing wing-margin-loss and wing-vein-thickening phenotypes. (b) Wing from MS1096Gal4 dxsm/ Y; Su(dx)sp/+ fly showing suppression of deltex phenotypes. (c) Wing from MS1096Gal4 dxsm/Y; Su(dx)sp/ +; UAS-Su(dx)IIIh/+ fly. The vein-thickening and wing-margin-loss phenotypes are restored. (d) Wing from nd/+; Su(dx)sp UAS-Su(dx)IIa/Su(dx)32 fly showing vein gaps in the longitudinal veins L.IV and L.V in the absence of the Gal4-induced Su(dx) expression. (e) Wing from ms1096Gal4 nd/+; Su(dx)sp UAS-Sudx)IIa/Su(dx)32 fly showing that expression of the Su(dx) protein can rescue the longitudinal wing-vein gap-phenotype that results from the interaction between Su(dx) and nd.
If Su(dx) were downregulating the Notch signal, then a further decrease of the signal should enhance the ectopic expression phenotypes. Using the MS1096Gal4 line, we expressed Su(dx) in a Notch mutant background that removes one copy of the Notch gene. This resulted in enhancement of the vein-thickening phenotype of Notch (Figure 6f). Similarly this Su(dx) gain of function resulted in enhancement of the vein-broadening phenotype of Delta, a mutation of the Notch ligand (Figure 6g). We also tested the consequence of expression of Su(dx) on the phenotype of the hypomorphic notchoid allele of Notch. This resulted in a strong genetic interaction with enhanced wing margin loss and considerably broadened vein territories, the latter encompassing approximately half of the wing surface (Figure 6h). The wing size was also much reduced, similar to other mutations in Notch pathway genes that enhance the notchoid phenotype.
DISCUSSION
Su(dx) was originally discovered as a dominant suppressor of the deltex phenotype. We have shown previously that Su(dx) mutants interact with Notch pathway genes and we have genetically characterized Su(dx) as a negative regulator of Notch signaling (Fostieret al. 1998). Here we have identified the molecular product of the Su(dx) locus as an E3 class ubiquitin ligase and have presented further evidence for its role as a Notch pathway regulator.
Su(dx) belongs to a family of E3 class ubiquitin ligase proteins: Two lines of evidence have enabled us to conclude that we have identified the product of the Su(dx) gene. First, we have identified mutations that disrupt the open reading frame of the Su(dx) gene in different Su(dx) alleles, and, second, we have rescued Su(dx) mutant phenotypes by expressing the cDNA in the developing Drosophila wing. Sequence database searching placed Su(dx) within a family of E3 class ubiquitin ligases that are related to the Nedd4 protein. Nedd4 was originally identified as a protein whose expression was downregulated following differentiation of neural precursor cells (Kumaret al. 1992). This family shares three types of conserved motifs. At the N terminus is the C2 domain that has a lipid-binding function. C2 domains are found in a variety of proteins and fall into two classes. One class has a calcium-dependent, lipid-binding activity while the other class lacks the calcium-binding site and has a constitutive lipid-binding function (Nalefski and Falke 1996). The amino acids that constitute the calcium-binding site are not conserved in the Su(dx) C2 domain and Su(dx) would be therefore be expected to be constitutively localized at cell membranes. The C terminus of Su(dx) contains a HECT domain that has been shown previously to act as an E3 class ubiquitin ligase (Huibregtseet al. 1995). The small deletion that was identified within the Su(dx)sp mutation occurs within the HECT domain and leaves the remainder of the reading frame intact. The molecule is thus likely to be able to compete with wild-type Su(dx) for recognition of interacting proteins, while it may be compromised in its ubiquitin ligase function. This suggests therefore a molecular basis for the previously identified antimorphic nature of this allele (Fostieret al. 1998).
—Ectopic expression of Su(dx). (a) Wild-type wing; (b) PtcGal4/+/; UAS-Su(dx)IIIh/+. The PtcGal4-dependent expression is targeted along the anterior-posterior axis between L.III and L.IV and results in a broadened L.III vein (arrow) and wing-margin-loss phenotypes (arrowhead). It is also interesting to note that the ectopic expression of Su(dx) also results in the loss of the cross vein and the lateral expansion of the intervein territory between L.III and L.IV. This was due to an increased cell number rather than cell growth (data not shown). (c) ms1096Gal4/Y; UAS-Su(dx)-IIIh/+ flies displayed an ectopic vein phenotype (arrowheads).
The middle portion of Nedd4 family proteins contains three or four WW modules and there are four of these domains in Su(dx), arranged as two pairs. WW modules have been previously reported to mediate specific protein-protein interactions (Sudolet al. 1995). This part of the protein is thus likely to be involved in selecting the target protein for ubiquitination. It is also possible that the WW domains may bind to a docking protein that directs Su(dx) to the target protein. It is interesting to speculate on the possible nature of the molecules that are recognized by the four WW domains. A number of in vitro and yeast two-hybrid-binding experiments have shown that WW domains selectively bind proline-rich sequences (Staubet al. 1996; Bedfordet al. 1997; Chenet al. 1997; Pirrozi et al. 1997). Furthermore, two groups of WW domain have been proposed. Group I WW domains bind to sequences containing a consensus of PPXY, and group II WW domains recognize proline-rich sequences containing PPLP cores (Sudol 1998). A number of genes that have been previously associated with Notch function contain one of these consensus motifs. The positive and negative regulators of Notch signaling, Deltex and Hairless, both contain PPLP sequences. Dishevelled, a component of the wingless pathway, contains both a PPLP and a PPXY sequence. It has been shown that Dishevelled can bind to the intracellular domain of Notch and act as a negative regulatory protein for the Notch signal (Axelrodet al. 1996). Thus there are several potential candidate proteins that may be targets or docking proteins for Su(dx) function. The structure of a group I WW domain in a complex with a peptide ligand has been determined and the residues that make contact with the ligand tyrosine have been identified (Maciaset al. 1996). It has been suggested that a critical residue defining group I WW domains is a conserved aliphatic residue at position 6 of the second β-strand of the domain (Sudol 1998). In group II WW domains this residue is substituted by an aromatic amino acid. According to this definition, all the WW domains of Su(dx) fall into group I (valine residues at positions 383, 414, 496, and 540 in the protein sequence). Therefore, proteins containing the PPXY consensus sequence might be expected to be the better candidates. The WW domains from two human Su(dx)-related proteins, WWP1 and WWP2, have been shown in vitro to bind to PPXY-containing peptides (Pirozziet al. 1997) and also to the human atrophin protein that contains five PPXY motifs (Woodet al. 1998). However, an attempt to convert a group II WW domain to a domain with group I specificity, by the mutation of key residues, was not successful (Bedfordet al. 1997). Furthermore, in a two-hybrid screen, the human Huntingtin protein, which does not contain either PPLP or PPXY sequences, was found to bind to several WW domains (Faberet al. 1998) and another nonconforming WW domain ligand (WB10) has also been identified (Bedfordet al. 1997). Therefore the nature of WW domain specificity may be more complicated than initial studies indicate. This may be particularly true when binding occurs in vivo with folded, wild-type proteins rather than with peptides.
—(a) Wing from MS1096Gal4/+ fly showing no phenotype in the absence of Su(dx) expression. (b) Wing from N54l9/+ fly. Flies deficient in one copy of Notch display a wing-margin-loss and a mild vein-broadening phenotype. (c) Wing from Dl171/+ fly displaying a vein broadening phenotype, particularly at the distal tips of the longitudinal veins. (d) The wing phenotype of an nd/nd fly resembles the deficiency of one copy of Notch. (e) Wing from Ms1096Gal4/+; UAS-Su(dx)II-J/+ fly. Expression of Su(dx) results in a Delta-like phenotype. (f) Wing from MS1096/N54l9; UAS-Su(dx)-II-J/+ fly displays an enhanced vein-broadening phenotype. (g) Wing from MS1096Gal4/+; UAS-Su(dx)-II-J/+; Dl171/+ fly. Expression of Su(dx) in a heterozygous Delta background results in strong enhancement of the vein-broadening phenotype. The wings of these flies are frequently blistered as can be seen in this example. (h) Wing from MS1096Gal4 nd/Y; UAS-Su(dx)II-J/+ fly showing a strong synergistic enhancement of the vein-broadening and margin loss phenotypes of notchoid. The wing size is significantly reduced compared to wild-type flies.
Extent of Su(dx) function in Notch pathway regulation: Previously we have reported the isolation of deficiency mutations of Su(dx). Heteroallelic combinations of Su(dx)sp/Su(dx)32 and Su(dx)sp/df(2L)Su(dx)7 have very similar phenotypes and show similar strengths of interactions with Notch mutations such as notchoid and Abruptex. It is therefore likely that Su(dx)32 is a null or nearly null mutation. This correlates with the molecular data, which show that there is a DNA rearrangement within the middle of the open reading frame. While this chromosome is recessive lethal at 29°, it is viable at this temperature over the deficiency with a wing-vein gap phenotype. The lethality therefore presumably corresponds to a second mutation on this chromosome not covered by the deficiency. At 25° homozygous Su(dx)32 flies display the forked cross-vein phenotype that we have previously observed in combinations of Su(dx) alleles. It is possible that Su(dx) function is restricted to the wing veins; however, our previous characterization of Su(dx) phenotypes has suggested a wider role (Fostieret al. 1998). Genetic interactions of Su(dx) with Notch pathway genes affect wing margin, macrochaetae, and microchaetae and flight muscle development (Fostieret al. 1998). It is possible therefore that there may be functional overlap of Su(dx) with related genes in the genome. Redundancy of gene function has a precedent in the Notch signaling pathway, within the members of the Enhancer of split gene complex. This complex contains seven DNA-binding, helix-loop-helix proteins that are targets for Notch signaling and that exhibit partial redundancy of function (Delidakis and Artavanis-Tsakonas 1992). Sequence database searching of GenBank reveals several Drosophila sequences with homology to HECT or WW domains, but currently no entries include sequences of sufficient length to determine whether both types of domain are present in the same gene, which would be diagnostic of a Nedd4 family member. However, a clone of a second Drosophila Nedd4-type gene has recently been obtained (Guy Tear, personal communication). Several mammalian Nedd4 family genes are known, and it is an open question as to how many and which of these genes are the true functional homologues of the Su(dx) protein. The phylogenetic analysis shows Su(dx) to be most closely related to a cluster of similar proteins that include WWP1, WWP2, and Itchy. The in vivo functions of the first two are currently not known, but mutations of itchy result in hyperplasia of lymphoid, hemopoetic, and epithelial cells (Perryet al. 1998) by an unknown mechanism. The similarity of Itchy to Su(dx) makes it tempting to speculate that these developmental phenotypes reflect aberrant regulation of Notch pathway function.
Su(dx) downregulates Notch receptor signaling: Using the temperature-sensitive Su(dx)sp allele, we have previously identified a role for Su(dx) in vein cell differentiation during a period of 20-28 hr after puparium formation, at a time when Notch signaling is involved in refining the wing-vein territories (Fostieret al. 1998). In this article we have demonstrated that there is expression of Su(dx) along the borders of the vein precursor territories during this period of wing development, which is consistent with a role in Notch regulation. Ectopic expression of Su(dx) resulted in too many wing epidermal cells becoming committed to the vein cell fate and also, in the case of PtcGal4-dependent expression, a wing margin loss phenotype. Ectopic Su(dx) expression thus simulates Notch loss-of-function phenotypes. We have previously shown that Su(dx) mutations suppress the phenotypes of Notch and Delta loss-of-function alleles (Fostieret al. 1998). Expression of Su(dx) enhanced the Notch and Delta vein broadening phenotypes. In addition there was a very strong genetic interaction with the hypomorphic notchoid mutation of Notch, resulting in extensive broadening of the veins, loss of wing margin, and reduction in wing size. This resembles the genetic interactions of notchoid with mutations in other Notch pathway genes such as deltex. Under the control of the MS1096Gal4 line, Su(dx) cDNA was expressed widely in the developing wing (data not shown) and it is significant that the additional vein tissue was observed adjacent to the existing veins, rather than throughout the domain of Su(dx) expression. Broadening of the existing vein precursor territories, rather than new vein initiation, would be expected from a downregulation of Notch signaling (de Celiset al. 1997). It is interesting to contrast this with the ectopic expression of the vein promoting gene dpp, which is capable of initiating vein differentiation throughout the intervein region (de Celis 1997). Recent work has highlighted the importance of specific proteolytic events in the generation of functional Notch and in the activation of the Notch signal (Schroeteret al. 1998; Struhl and Adachi 1998). The identity of Su(dx) indicates that a further level of control, via ubiquitination, may act directly or indirectly on the Notch pathway. Sel-10, a negative regulator of a C. elegans homologue of Notch (Lin-12), has been shown to be related to the budding yeast CDC4 protein (Hubbardet al. 1997). The latter is a protein known to be involved in targeting yeast cell cycle proteins for proteosome dependent degradation, further supporting a role for ubiquitination in Notch pathway regulation.
Ectopic expression of Su(dx) regulates wing growth: In addition to wing margin loss and ectopic vein differentiation, ectopic expression of Su(dx) in a wild-type background resulted in broadening of the wing through increased cell number. It has previously been shown that accelerating the cell cycle by ectopic expression of the transcriptional regulator dE2F leads to an increase in cell number but not to growth of the wing. This is because there is a compensation of cell growth to maintain the normal wing size (Neufeldet al. 1998). It is therefore unlikely that Su(dx) expression directly regulates the cell cycle and more likely that the increased proliferation is an indirect effect, resulting from misregulation of factors controlling the size of the wing. This is interesting because the factors that control the size of the wing, or other organs, are poorly understood and this phenotype provides a tool to enable the genetic dissection of this control.
Su(dx) contains four WW modules that have previously been shown to mediate protein-protein interactions through binding to specific proline-rich sequences. Because individual WW domains alone are capable of specific binding, it is possible that the wing growth and cell differentiation phenotypes reflect the selection of different protein targets. A growing number of signal transduction pathways have now been shown to be regulated by ubiquitin-dependent proteolysis, including NF-κB (Palombellaet al. 1994), Wingless/β-catenin (Aberleet al. 1997; Jiang and Struhl 1998) and downstream target proteins of the EGF receptor pathway such as Tramtrak (Liet al. 1997). The identification of the target molecule or molecules of Su(dx) and the functions of Su(dx)-related proteins will thus be important avenues for future investigation.
Acknowledgments
We thank Jose de Celis for helpful discussion and for help with in situ hybridization methodology. Nick Brown and Tian Xu provided us with cDNA libraries. Luke Alphey helped with injection of embryos and Jenny Gleason provided valuable advice regarding the phylogenetic analysis. We thank Guy Tear for discussion of unpublished data. The PHYLIP programs are freely available from Dr. J. Felsenstein, University of Washington. We thank Lawrence Hall for assistance with automated DNA sequencing. The MS1096Gal4 line was provided to us by Matthew Freeman. We also acknowledge the Berkeley Drosophila Genome Project for P1 phage genomic clones and the internet resources of FLYBASE. We acknowledge the Biotechnology and Biological Sciences Research Council Commitment and Determination Initiative, the Medical Research Council, Howard Hughes Medical Institute, the Royal Society, and Zeneca Pharmaceuticals for financial support.
Footnotes
-
Communicating editor: K. Anderson
- Received November 30, 1998.
- Accepted March 1, 1999.
- Copyright © 1999 by the Genetics Society of America