We present a molecular and genetic analysis which elucidates the role of deltex in the Notch signaling pathway. Using the yeast ‘interaction trap’ assay, we define the protein regions responsible for heterotypic interactions between Deltex and the intracellular domain of Notch as well as uncover homotypic interaction among Deltex molecules. The function of the Deltex-Notch interaction domains is examined by in vivo expression studies. Taken together, data from overexpression of Deltex fragments and from studies of physical interactions between Deltex and Notch, suggest that Deltex positively regulates the Notch pathway through interactions with the Notch ankyrin repeats. Experiments involving cell cultures indicate that the Deltex-Notch interaction prevents the cytoplasmic retention of the Suppressor of Hairless protein, which otherwise is sequestered in the cytoplasm via association with the Notch ankyrin repeats and translocates to the nucleus when Notch binds to its ligand Delta. On the basis of these findings, we propose a model wherein Deltex regulates Notch activity by antagonizing the interaction between Notch and Suppressor of Hairless.

The Notch (N) gene encodes a transmembrane protein with an extracellular domain containing 36 epidermal growth factor (EGF)-like repeats and an intracellular domain bearing 6 tandem cdc10/SWI6/ankyrin repeats (ankyrin repeats) (Wharton et al., 1985; Kidd et al., 1986; Breeden and Nasmyth, 1987). Extensive genetic and molecular analyses in Drosophila and other organisms have led to the hypothesis that N functions as a receptor in an evolutionarily conserved cell interaction mechanism (reviewed by Greenwald and Rubin, 1992; Fortini and Artavanis-Tsakonas, 1993; Artavanis-Tsakonas, et al., 1995). It is thought that the N pathway regulates the competence of a broad spectrum of uncommitted cells to respond to specific developmental signals, thus controlling a fundamental step in the progression of undifferentiated cells to a more committed state (Coffman et al., 1993; Fortini et al., 1993; Fortini and Artavanis-Tsakonas, 1993).

Genetic studies in Drosophila have identified several genes that appear to be part of this signaling pathway (Vässin et al., 1985, 1987; de la Concha et al., 1988; Fleming et al., 1990; Xu et al., 1990; Xu and Artavanis-Tsakonas, 1990; Gorman and Girton, 1992; Fortini and Artavanis-Tsakonas, 1994). These genes encode the two membrane-bound N ligands, Delta (Dl) and Serrate (Ser) (Vässin et al., 1987; Kopczynski et al., 1988; Fleming et al., 1990; Thomas et al., 1991), the cytoplasmic protein Deltex (Dx) (Busseau et al., 1994), and the nuclear proteins encoded by mastermind (mam), Hairless (H), the Enhancer of split complex (E(spl)), and Suppressor of Hairless (Su(H)) (Smoller et al., 1990; Bang and Posakony, 1992; Maier et al., 1992; Delidakis et al., 1991; Schrons et al., 1992; Furukawa et al., 1991, 1992; Schweisguth and Posakony, 1992; Fortini and Artavanis-Tsakonas, 1994).

Although the biochemical nature of N signaling is not understood, physical interactions between some of the proteins implicated in this pathway have been documented. In vitro experiments have shown that N interacts extracellularly with Dl and Ser (Fehon et al., 1990; Rebay et al., 1991). The interaction domains of the N receptor and these two ligands have been defined by deletion studies in S2 cells (Rebay et al., 1991; M.A.T. Muskavitch, personal communication) and ectopic expression studies of truncated forms in transgenic flies have shown that deletion of the ligand-binding domain of N results in constitutive activation of the receptor (Rebay et al., 1993; Struhl et al., 1993; Lieber et al., 1993; Fortini et al., 1993). These studies have also established that the N ankyrin repeats are essential for this activation (Rebay et al., 1993; Lieber et al., 1993).

The relationship between ligand binding and N activation is still unclear. However, we have begun to study interactions between the N cytoplasmic domain and putative intracellular components of the N pathway. dx is believed to function in the N signaling pathway because it displays genetic interactions with N, Dl, mam, H, and Su(H) (Xu and Artavanis-Tsakonas, 1990; Xu et al., 1990; Gorman and Girton, 1992; Busseau et al., 1994; Fortini and Artavanis-Tsakonas, 1994). We have previously shown that dx encodes a novel cytoplasmic protein with widespread tissue distribution (Busseau et al., 1994). The involvement of Dx in the N pathway is further supported by the finding that Dx interacts physically with the ankyrin repeats of the N protein (Diederich et al., 1994). These repeats also mediate molecular interactions between N and the Su(H) protein (Fortini and Artavanis-Tsakonas, 1994).

In the present study, we use molecular genetic approaches to address the functional role of Dx in the N signaling pathway. We also identify the existence of two sequence motifs which Dx shares with other proteins: a ring-H2-zinc finger and a putative SH3-binding site. Using the yeast ‘interaction trap’ system (Zervos et al., 1993), we define the interaction domains between N and Dx and provide evidence for a novel homotypic interaction for Dx proteins. Overexpression studies of Dx and an ‘activated’ N receptor suggest that Dx acts as a positive regulator in the N pathway. Functional analyses of the Dx domains demonstrate that overexpression of a fragment including the N-Dx interaction region can rescue dx loss-of-function mutations. Moreover, overexpression of this fragment results in mutant phenotypes similar to those caused by activation of the N receptor. We find that in cultured Drosophila cells, Dx prevents the cytoplasmic retention of Su(H) which otherwise is sequestered in the cytoplasm via association with N and translocates to the nucleus in a ligand-dependent fashion. In light of these findings, we discuss the role of Dx in N signaling and present a model for its action.

Interaction trap assay

The interaction trap assay was performed according to Zervos et al. (1993). The yeast strain, EGY40, and vectors, pLEX202+PL, pEG202, pJG4-5, pSH18-34 and pRFHM1 were generous gifts from Roger Brent. pLEX202+PL and pEG202 were designed to produce fusion proteins with residues 1-202 of the LexA DNA-binding domain. pJG4-5 was designed to produce fusion proteins with a transcriptional activation domain. Basic yeast techniques were carried out as described by Sherman (1991) and yeast transformations were performed according to the method of Gietz et al. (1992). Transformants were obtained initially on standard UraHisTrp-glucose plates and then transferred to UraHisTrpX-gal-galactose or Ura HisTrpX-gal-glucose indicator plates.

β-galactosidase assays were done according to Guarente (1983) in UraHisTrp-galactose and UraHisTrp-glucose media and cultured at 30°C overnight. They were performed using 0.2 ml of overnight culture media (OD600 = 0.2-1.3) and a reaction time of 5 minutes and normalized to OD600 = 1, and activity values were assigned as 1 unit = 0.001 OD420/1.0 OD600.

Constructs for the interaction trap system

The cDNA fragments encoding the products of each gene were obtained by PCR. For all constructs, ligation junctions were checked by restriction analysis and/or by DNA sequencing. The plasmids referred to as pLEX and pJG are the constructs created in pLEX202+PL (or pEG202) and pJG4-5 (or pJGSTOP), respectively. pJGSTOP was modified from pJG4-5, and stop codons were introduced for all three reading frames beyond the XhoI site.

N constructs

N amino acids are numbered according to Wharton et al. (1985). pLEXICN1 and pJGICN1 contain N cDNA encoding amino acids 1827-2259. pLEXICN-2 and pJGICN-2 contain cDNA encoding amino acids 2109-2704. The N deletion constructs, pJGICN1827-2109 (Fig. 2 construct 2), pJGICN1827-2076 (Fig. 2 construct 3), pJGICN1827-1996 (Fig. 2 construct 4), pJGICN1827-1963 (Fig. 2 construct 5), pJGICN1827-1921 (Fig. 2 construct 6), pJGICN1846-2076 (Fig. 2 construct 7), pJGANK1-5(1889-2076) (Fig. 2A construct 8) and pLEXANK1-5(1889-2076), pJGANK3-5(1969-2076) (Fig. 2 construct 9), pJGANK5(2036-2076) (Fig. 2A construct 10) and pJGANK1846-1963 (Fig. 2 construct 11) contain N cDNAs encoding amino acids indicated in the names. pJGICNΔANK (Fig. 2 construct 12) contains an N cDNA encoding amino acids 1827-1884 and 2111-2259. This construct has an internal deletion of the ankyrin repeats.

Fig. 1.

Schematic diagram of N and Dx proteins. (A) Diagram of the N protein. Protein motifs, including a signal peptide (SP), 36 epidermal growth factor-like (EGF) repeats, 3 Notch/lin-12 repeats (N), a transmembrane domain (TM), 2 putative nuclear localization signals (NLS), 6 cdc10/ankyrin repeats (ANK), a polyglutamine stretch (OPA) and PEST-like sequence (PEST) are shown. ICN1 (aa1827-2259) and ICN2 (aa 2109-2704) cover most of the N cytoplasmic domain with an overlap from amino acid 2109 to 2259. ANK1-5 comprises the first five ankyrin repeats (beginning at the N terminus) and is the minimal element necessary to mediate N-Dx interactions. Amino acid numbers are according to Wharton et al. (1985). (B) Diagram of Dx protein. The Dx protein is arbitrarily divided into 3 regions (domains I, II and III), based on the positions of two OPA repeats. The region binding to the N ankyrin repeats, a putative SH3 domain binding site and a ring-H2-zinc finger are shown. Dx1-303 (aa 1-303) encompasses the domains capable of mediating N-Dx and Dx-Dx interactions. DxII-III (aa 306-737) contains domains mediating Dx-Dx interactions. Amino acid numbers are according to Busseau et al. (1994). (C) Putative SH3 domain binding consensus in Dx protein. A putative SH3 domain binding consensus located in Dx domain II (aa 476-484) is shown. Alignments are made against a region of a mouse rho-GAP protein (3BP-1) that has been shown to bind to SH3 domain of the Abelson tyrosine kinase (Ren et al., 1993). The region shown for the Son of sevenless protein is presumed to be the sequence binding to the SH3 domain of Drk (Oliver et al., 1993; Simon et al., 1993). Hairless is a novel basic protein of unkown function (Bang and Posakony, 1992; Maier et al., 1992). Disabled is a substrate for the Drosophila Abelson tyrosine kinase (Gertler et al., 1993). Bold letters show the amino acids matching the consensus (Ren et al., 1993). (D) Ring-H2-zinc finger motif in Dx protein. Ring-H2-zinc finger motif located in Dx domain III is shown (Freemont, 1993). The first and last cysteines in the motif correspond to aa 547 and 599, respectively.

Fig. 1.

Schematic diagram of N and Dx proteins. (A) Diagram of the N protein. Protein motifs, including a signal peptide (SP), 36 epidermal growth factor-like (EGF) repeats, 3 Notch/lin-12 repeats (N), a transmembrane domain (TM), 2 putative nuclear localization signals (NLS), 6 cdc10/ankyrin repeats (ANK), a polyglutamine stretch (OPA) and PEST-like sequence (PEST) are shown. ICN1 (aa1827-2259) and ICN2 (aa 2109-2704) cover most of the N cytoplasmic domain with an overlap from amino acid 2109 to 2259. ANK1-5 comprises the first five ankyrin repeats (beginning at the N terminus) and is the minimal element necessary to mediate N-Dx interactions. Amino acid numbers are according to Wharton et al. (1985). (B) Diagram of Dx protein. The Dx protein is arbitrarily divided into 3 regions (domains I, II and III), based on the positions of two OPA repeats. The region binding to the N ankyrin repeats, a putative SH3 domain binding site and a ring-H2-zinc finger are shown. Dx1-303 (aa 1-303) encompasses the domains capable of mediating N-Dx and Dx-Dx interactions. DxII-III (aa 306-737) contains domains mediating Dx-Dx interactions. Amino acid numbers are according to Busseau et al. (1994). (C) Putative SH3 domain binding consensus in Dx protein. A putative SH3 domain binding consensus located in Dx domain II (aa 476-484) is shown. Alignments are made against a region of a mouse rho-GAP protein (3BP-1) that has been shown to bind to SH3 domain of the Abelson tyrosine kinase (Ren et al., 1993). The region shown for the Son of sevenless protein is presumed to be the sequence binding to the SH3 domain of Drk (Oliver et al., 1993; Simon et al., 1993). Hairless is a novel basic protein of unkown function (Bang and Posakony, 1992; Maier et al., 1992). Disabled is a substrate for the Drosophila Abelson tyrosine kinase (Gertler et al., 1993). Bold letters show the amino acids matching the consensus (Ren et al., 1993). (D) Ring-H2-zinc finger motif in Dx protein. Ring-H2-zinc finger motif located in Dx domain III is shown (Freemont, 1993). The first and last cysteines in the motif correspond to aa 547 and 599, respectively.

Fig. 2.

Functional dissection of the N cytoplasmic domain. Twelve deletion constructs (ACT constructs 1-12) of ICN1 are shown schematically as protein fusions with a transcriptional activation domain (see Materials and methods). These were introduced into the yeast strain EGY40 harboring the lacZ reporter construct (pSH18-34) and the Dx protein fusion with a LexA DNA-binding domain (aa 1-202). β-galactosidase levels were measured for three independent transformants and the average value (with standard deviation) is shown in arbitrary units (see Materials and methods). The activities recorded for induced (galactose) and non-induced (glucose) cultures are represented by hatched and solid bars, respectively. ACT constructs: 1; pJGICN1, 2; pJGICN1827-2109, 3; pJGICN1827-2076, 4; pJGICN1827-1996, 5; pJGICN1827-1963, 6; pJGICN1827-1921, 7; pJGICN1846-2076, 8; pJGANK1-5(1889-2076), 9; pJGANK3-5(1969-2076), 10; pJG ANK5(2036-2076), 11; pJGANK1(1846-1963), 12; pJGICN1ΔANK (aa 1827-1884 and 2111-2259), 13; pJG STOP (no insert).

Fig. 2.

Functional dissection of the N cytoplasmic domain. Twelve deletion constructs (ACT constructs 1-12) of ICN1 are shown schematically as protein fusions with a transcriptional activation domain (see Materials and methods). These were introduced into the yeast strain EGY40 harboring the lacZ reporter construct (pSH18-34) and the Dx protein fusion with a LexA DNA-binding domain (aa 1-202). β-galactosidase levels were measured for three independent transformants and the average value (with standard deviation) is shown in arbitrary units (see Materials and methods). The activities recorded for induced (galactose) and non-induced (glucose) cultures are represented by hatched and solid bars, respectively. ACT constructs: 1; pJGICN1, 2; pJGICN1827-2109, 3; pJGICN1827-2076, 4; pJGICN1827-1996, 5; pJGICN1827-1963, 6; pJGICN1827-1921, 7; pJGICN1846-2076, 8; pJGANK1-5(1889-2076), 9; pJGANK3-5(1969-2076), 10; pJG ANK5(2036-2076), 11; pJGANK1(1846-1963), 12; pJGICN1ΔANK (aa 1827-1884 and 2111-2259), 13; pJG STOP (no insert).

Dx construct

Dx amino acids are numbered according to Busseau et al. (1994). pLEXdx and pJGdx contain the entire Dx coding region, amino acids 1-737. pJGdx1-528 (Fig. 3A construct 2), pJGdx1-362 (Fig. 3A construct 3), pJGdx1-303 (Fig. 3A construct 4) and pLEXdx 1-303, pJGdx1-204 (Fig. 3A construct 5), pJGdx1-28 (Fig. 3A construct 6), pJGdx24-204 (Fig. 3A construct 7), pJGdx83-204 (Fig. 3A construct 8), pJGdx167-204 (Fig. 3A construct 9), pJGdx46-303 (Fig. 3A construct 10), pJGdx83-303 (Fig. 3A construct 11), pJGdx167-303 (Fig. 3A construct 12), pJGdx212-303 (Fig. 3A construct 13), pJGdxII-III(306-737) (Fig. 3A construct 14), pJGdxIII(514-737) (Fig. 3A construct 15) and pJGdxII(306-487) (Fig. 3A construct 16) contain Dx cDNAs encoding amino acids indicated in the names.

Fig. 3.

Functional dissection of Dx domains. (A) Deletion analysis of Dx protein domains. Sixteen Dx deletion constructs (ACT constructs 1-16) are shown schematically as protein fusions with a transcriptional activation domain (see Materials and methods). These were introduced into the yeast strain EGY40 harboring the lacZ reporter plasmid (pSH18-34) and either of two LexA DNA-binding domain fusion constructs, pLEXICN1 (a) or pLEXdx (b). ACT constructs: 1; pJGdx (full-length), 2; pJGdx1-528, 3; pJGdx1-362, 4; pJGdx1-303, 5; pJGdx1-204, 6; pJGdx1-28, 7; pJGdx46-204, 8; pJGdx83-204, 9; pJGdx167-204, 10; pJGdx46-303, 11; pJGdx83-303, 12; pJGdx167-303, 13; pJGdx212-303, 14; pJGdxII-III(306-737), 15; pJGdxIII(514-737), 16; pJGdxII(306-487), 17; pJGSTOP (no insert). β-galactosidase activity levels were measured for 3 independent transformants and the average value (with standard deviation) is shown in arbitrary units. The activities recorded for the induced cultures (galactose) and non-induced cultures (glucose) are represented by stippled and solid bars, respectively. (B) Dx domain I interacts with N in Drosophila S2 cell. Confocal microscope images of Drosophila S2 cells are presented as split images with N protein distribution shown in green and Dx in red. Each panel represents a cotransfection experiment involving N and Dx expression plasmids: a; pMTNcDNA (full-length N) and pCaSpeR-hs dx, b; pMTECN (deletion of cytoplasmic domain of N) and pCaSpeR-hs dx, c; pMtNcDNA and pCaSpeR-hs dx1-303. Cells expressing these constructs were mixed with cells expressing Dl to induce cell surface ‘capping’. Immunofluorescent labeling of cells was performed using mouse anti-N monoclonal antibody and rat anti-Dx polyclonal antibody. Cocapped regions of N and Dl are indicated by arrows.

Fig. 3.

Functional dissection of Dx domains. (A) Deletion analysis of Dx protein domains. Sixteen Dx deletion constructs (ACT constructs 1-16) are shown schematically as protein fusions with a transcriptional activation domain (see Materials and methods). These were introduced into the yeast strain EGY40 harboring the lacZ reporter plasmid (pSH18-34) and either of two LexA DNA-binding domain fusion constructs, pLEXICN1 (a) or pLEXdx (b). ACT constructs: 1; pJGdx (full-length), 2; pJGdx1-528, 3; pJGdx1-362, 4; pJGdx1-303, 5; pJGdx1-204, 6; pJGdx1-28, 7; pJGdx46-204, 8; pJGdx83-204, 9; pJGdx167-204, 10; pJGdx46-303, 11; pJGdx83-303, 12; pJGdx167-303, 13; pJGdx212-303, 14; pJGdxII-III(306-737), 15; pJGdxIII(514-737), 16; pJGdxII(306-487), 17; pJGSTOP (no insert). β-galactosidase activity levels were measured for 3 independent transformants and the average value (with standard deviation) is shown in arbitrary units. The activities recorded for the induced cultures (galactose) and non-induced cultures (glucose) are represented by stippled and solid bars, respectively. (B) Dx domain I interacts with N in Drosophila S2 cell. Confocal microscope images of Drosophila S2 cells are presented as split images with N protein distribution shown in green and Dx in red. Each panel represents a cotransfection experiment involving N and Dx expression plasmids: a; pMTNcDNA (full-length N) and pCaSpeR-hs dx, b; pMTECN (deletion of cytoplasmic domain of N) and pCaSpeR-hs dx, c; pMtNcDNA and pCaSpeR-hs dx1-303. Cells expressing these constructs were mixed with cells expressing Dl to induce cell surface ‘capping’. Immunofluorescent labeling of cells was performed using mouse anti-N monoclonal antibody and rat anti-Dx polyclonal antibody. Cocapped regions of N and Dl are indicated by arrows.

mam, E(spl) m8, and cactus constructs

pLEXmam and pJGmam contain the entire coding region of mam and were generated from pMTmam (D. Bettler and B. Yedvobnick, unpublished data). pLEXm8 and pJGm8 contain the entire coding region of E(spl) m8. pLEXcactus and pJGcactus contain a cactus cDNA fragment encoding amino acids 173-500, including the entire ankyrin repeat coding region. This fragment was obtained from pcact5B (Geisler et al., 1992).

Sequencing

Sequences were determined by the dideoxy method using Sequenase version 2.0 (USB) according to instructions provided by the manufacturer.

S2 cell expression experiments

A colocalization assay of N and Dx was performed using Drosophila Schneider 2 cultured cells according to the method of Diederich et al. (1994). The following expression constructs have been described previously: full-length N, pMTNcDNA; an intracellular deletion of N, pMTECN; and full-length Dx, pCaSpeR-hs dx (Fehon et al., 1990; Rebay et al., 1991; Busseau et al., 1994; Diederich et al., 1994). Deletion construct of Dx, pCaSpeR-hs dx 1-303 contains cDNA encoding amino acids 1-303.

Experiments on the subcellular localization of Su(H) were performed as described previously with slight modifications (Diederich et al., 1994; Fortini and Artavanis-Tsakonas, 1994). For expression of fulllength N and Dx, metallothionein promoter constructs in pRmHa-3a, pMTNMg and pMTdx were used, respectively (Fehon et al., 1990; R. J. D. and S. A.-T., unpublished data). The Su(H) construct, pSu(H)4 was used to express cMyc-epitope tagged full-length Su(H) under the hsp70 promoter (Fortini and Artavanis-Tsakonas, 1994). Drosophila S2 cells or a stable transformant line carrying pMTNMg were transfected with either pSu(H)4 alone or pSu(H)4 and pMTdx. The metallothionein promoter constructs, pMTNMg and pMTdx were induced for 12 hours followed by induction of the hsp70 promoter construct, pSu(H)4 (Diederich et al., 1994; Fortini and Artavanis-Tsakonas, 1994). Detection of Dx and Su(H) proteins has been described previously (Diederich et al., 1994; Fortini and Artavanis-Tsakonas, 1994). The cMyc-epitope tagged Su(H) was detected using mouse anti-Myc monoclonal antibody (Fortini and Artavanis-Tsakonas, 1994). Dx was detected with rat anti-Dx monoclonal antibodies, C579-3L and C579-8D (R. J. D. and S. A.-T., unpublished data).

Germline transformation

Germline transformation was performed using standard procedures described by Spradling (1986). pCaSpeR-hs dx1-204 and pCaSpeR-hs dx1-303 (1 mg/ml) were injected with helper plasmid Δ2-3 (0.2mg/ml) (Robertson et al., 1988) into w1118 (Lindsley and Zimm, 1992) embryos.

Heat-shock conditions

Heat-shocks were performed using a programmable circulating water bath (Fisher Science). One cycle of heat-shock involved 1 hour at 37°C followed by a 2 hour recovery at 25°C. Late third instar larvae were collected and left at 20°C for approx. 6 hours until most had reached the prepupal stage (0-24 hours), or were left for another day (24-48 hours) at 25°C; prepupae and pupae were heat-shocked for 1 to 7 cycles and incubated at 25°C until eclosion.

Wing preparations and scanning electron microscopy

Wings were mounted in Aquamount (BDH Limited) and viewed on a Leitz Orthoplan 2 microscope. Video images were collected as described by Fehon et al. (1990).

Flies were processed for SEM by dehydration in an ethanol series, followed by critical point drying and mounting on stubs, and viewed on an ISI-SS40. SEM photographs were scanned into computer files using a Umax UC630 scanner, and images were assembled in Canvas (Deneba Systems, Inc.).

Molecular interactions involving N and Dx

We have used the ‘yeast interaction trap’ system (Zervos et al., 1993) to test for physical binding between the intracellular domain of N and the products of several genes that have been implicated genetically in the N pathway. In this assay, one protein segment is fused to the DNA-binding domain of the LexA protein, which binds to the promoter of a LexAop-lacZ reporter construct but does not activate transcription. A second foreign protein segment is fused to an acidic transcriptional activation domain that does not bind DNA on its own. Coex-pression of these two fusion proteins in yeast cells results in the functional reconstitution of an active LexA ‘hybrid’ transcription factor only if the foreign proteins physically interact with one another. Activity of the hybrid transcription factor is monitored by transcription of the β-galactosidase reporter gene (Materials and methods).

Constructs expressing full-length Dx as both LexA (LEX) and acidic activation domain (ACT) fusions were tested for interactions with segments of the N protein. For these experiments, the intracellular domain of N was divided into two parts, termed ICN1 and ICN2, each of which was expressed as either a LEX or an ACT fusion (Fig. 1A; Materials and methods). Heterotypic interactions were observed between ICN1 and Dx, irrespective of which of the two proteins was expressed as the LEX or ACT fusion. Moreover, control experiments involving the ankyrin repeats of the Cactus protein (Geisler et al., 1992; Kidd, 1992) indicate that the N-Dx inter-action is specific for the N ankyrin repeats (data not shown), which is consistent with previous observations in Drosophila cultured cells (Diederich et al., 1994). In addition, a Dx-Dx homotypic interaction was detected. This assay did not detect any interactions between N and two additional N pathway elements, Mam or E(spl) m8 (data not shown).

Defining the interaction domain of the N protein

To dissect further the interactions defined above, a deletion study was performed on the N ICN1 and Dx polypeptides. The results are summarized in Fig. 2. The N region responsible for the ICN1-Dx interaction is restricted to the first five of the six ankyrin repeats (Fig. 2, construct 8). Conversely, deletion of all six ankyrin repeats from ICN1 completely abolishes ICN1-Dx interaction (Fig. 2, construct 12). Attempts to further delimit the interaction domain were not informative (Fig. 2, constructs 4 and 9). In this assay, the minimal N protein region that retains the capacity to interact with Dx is composed of the first five ankyrin repeats.

Defining the interaction domains of the Dx protein

Dx encodes a 737-amino acid protein (Busseau et al., 1994). Two opa repeats subdivide the primary structure into three domains, which we arbitrarily designated as domains I, II and III (Fig. 1B). Although a novel protein, Dx contains a putative SH3-binding domain in domain II and a ring-H2-zinc finger in domain III (Ren et al., 1993; Freemont, 1993). This type of zinc finger has been proposed to mediate protein interactions (Freemont, 1993). In light of these observations, we performed a deletion analyses to define further the domains of Dx that mediate ICN1-Dx and Dx-Dx interactions (summarized in Fig. 3A).

A construct containing domain I and the first opa repeat (amino acids 1-303, Fig. 3A, construct 4) is capable of inter-acting with both ICN1 and full-length Dx fusion proteins. Smaller Dx segments that include amino acids 1-204 or 46-303 still bind to ICN1 but fail to interact with full-length Dx (Fig. 3A, constructs 5 and 10, respectively). We also note that Dx fragment 1-303, which corresponds to domain I, displays homotypic interactions with itself but does not bind to Dx domains II or III (data not shown). Moreover, segments of Dx consisting of either domain II alone or domain III alone interact homotypically (data not shown) or with full-length Dx (Fig. 3A, constructs 15 and 16). These results indicate that, while each of the three domains of Dx is capable of homotypic inter-actions in this yeast assay, only domain I mediates associations with the ankyrin repeats of N.

N-Dx interactions in Drosophila cells

Because the above analysis of N and Dx was performed in a heterologous yeast assay, we examined whether the Dx-N interaction domains defined in yeast are also capable of inter-acting in Drosophila cells. S2 cells were cotransfected with plasmid expression constructs that placed N and Dx under the inducible control of the Drosophila metallothionein and hsp70 promoters, respectively (Diederich et al., 1994). N expression was induced first to ensure proper cell surface localization, followed by a brief heat-shock to induce Dx expression. These cells were then aggregated with cells expressing Dl, a pre-sumptive membrane-bound ligand of N, to produce a ‘mutual capping’ of N and Dl at the point of cellular contact (Fehon et al., 1990). Corroborating the results of Diederich et al. (1994), colocalization of full-length Dx with the ‘capped’ N indicates a molecular interaction between the two proteins (Fig. 3B, a). In contrast, deletion of the intracellular domain of N abolishes the colocalization of Dx with N although capping of N still occurs (Fig. 3B, b). Consistent with our studies in the yeast assay, we find that a Dx fragment including domain I (amino acids 1-303) is capable of associating with N (Fig. 3B, c), while a Dx fragment containing domains II and III fails to exhibit any detectable association with N (data not shown). Thus, interactions between N and Dx are qualitatively similar in the heterologous yeast system and in Drosophila cells.

The biological activity of the Dx-N interaction domains

To determine the biological significance of the N-Dx interactions defined by the above biochemical approach, we examined the activity of the Dx protein by in vivo overexpression experiments. These experiments allowed us to define thermocritical periods during which overexpression of full-length Dx or Dx domain I (aa 1-303) resulted in mutant phenotypes in the adult flies. In summary, we found that full-length Dx under heat-shock control is capable of inducing a complex array of mutant phenotypes, affecting eye, wing and bristle morphology (summarized in Figs 4 and 5). Under similar conditions, some of the same phenotypes are produced by expression of only Dx domain I.

Fig. 4.

In vivo effects of ectopically expressed full-length Dx and Dx domain I on thoracic bristle development. (A-I) Scanning electron micrographs of the thoraces of heat-shocked transgenic flies. (A) w1118 (0-24 hours after pupariation). Wild-type thorax. In fewer than 20% of the flies one or two extra macrochaetae are seen (see C and I). (B) AxE2 without heat-shock. A N gain-of-function allele, Ax shows missing microchaetae. (C) Full-length Dx (0-24 hours after pupariation). The ‘bald’ phenotype was observed at 10-80% penetrance. (D,E) Full-length Dx (24-48 hours after pupariation). The ‘double shaft/single socket’ phenotype was observed at approx. 30% penetrance. (F,G) Full-length Dx (24-48 hours after pupariation). The ‘double socket’ phenotype was observed at approx. 80% penetrance. (H) Full-length Dx (prepupal stage). Extra macrochaetae and microchaetae were observed with 10-50% penetrance. (I) pCaSpeR-hs dx1-303 (Dx domain I) (0-24 hours after pupariation). The bald phenotype was observed at ∼50% penetrance. Note the similarity of this phenotype to that shown in C under same experimental conditions. Approximate developmental stages at which heat-shocks were administered are indicated. All four pCaSpeR-hs dx (full-length Dx) and five pCaSpeR-hs dx1-303 transgenic lines examined show the phenotypes described above with different penetrance. Phenotypes of full-length Dx transgenic lines were observed with 1-7 heat-shocks. The phenotype of pCaSpeR-hs dx1-303 transgenic flies (I) was observed with 4-7 heat-shocks. None of the five independent transformant lines of pCaSpeR-hs dxII-III tested showed any detectable mutant phenotype under the conditions described above (data not shown).

Fig. 4.

In vivo effects of ectopically expressed full-length Dx and Dx domain I on thoracic bristle development. (A-I) Scanning electron micrographs of the thoraces of heat-shocked transgenic flies. (A) w1118 (0-24 hours after pupariation). Wild-type thorax. In fewer than 20% of the flies one or two extra macrochaetae are seen (see C and I). (B) AxE2 without heat-shock. A N gain-of-function allele, Ax shows missing microchaetae. (C) Full-length Dx (0-24 hours after pupariation). The ‘bald’ phenotype was observed at 10-80% penetrance. (D,E) Full-length Dx (24-48 hours after pupariation). The ‘double shaft/single socket’ phenotype was observed at approx. 30% penetrance. (F,G) Full-length Dx (24-48 hours after pupariation). The ‘double socket’ phenotype was observed at approx. 80% penetrance. (H) Full-length Dx (prepupal stage). Extra macrochaetae and microchaetae were observed with 10-50% penetrance. (I) pCaSpeR-hs dx1-303 (Dx domain I) (0-24 hours after pupariation). The bald phenotype was observed at ∼50% penetrance. Note the similarity of this phenotype to that shown in C under same experimental conditions. Approximate developmental stages at which heat-shocks were administered are indicated. All four pCaSpeR-hs dx (full-length Dx) and five pCaSpeR-hs dx1-303 transgenic lines examined show the phenotypes described above with different penetrance. Phenotypes of full-length Dx transgenic lines were observed with 1-7 heat-shocks. The phenotype of pCaSpeR-hs dx1-303 transgenic flies (I) was observed with 4-7 heat-shocks. None of the five independent transformant lines of pCaSpeR-hs dxII-III tested showed any detectable mutant phenotype under the conditions described above (data not shown).

Fig. 5.

In vivo effects of ectopically expressed full-length Dx and Dx domain I on wing development. Adult wings from flies of the following genotypes and heat-shock treatments: (A) w1118 (0-24 hours after pupariation). Wild-type wing. Only approx. 1% of flies show very small wing vein gaps after our heat-shock treatment (data not shown). (B) AxE2 without heat-shock. A N gain-of-function allele, Ax shows wing vein gaps. (C)w1118; Dp(1;2)w+51b7/+ without heat-shock. Note the thickened wing veins (Confluens phenotype) caused by increasing the gene dose of N (D)Full-length Dx (0-24 hours after pupariation). Note the wing vein gap phenotype (approx. 10% penetrant) and its similarity to that produced by AxE2 (see B). (E) Full-length Dx (24-48 hours after pupariation). Note the thickened wing veins (70-100% penetrant) and their similarity to that produced by N gene duplications (see C). (F) pCaSpeR-hs dx1-303 (Dx domain I) (24-48 hours after pupariation). Note the wing vein gap phenotype (10-90% penetrant) and its similarity to the N gain-of-function allele AxE2 (see B) and full-length Dx (see D). Approximate developmental stages at which heat-shocks were administered are indicated. All four pCaSpeR-hs dx (full-length Dx) and five pCaSpeR-hs dx1-303 transgenic lines examined show the phenotypes described above with different penetrance. All phenotypes of full-length Dx transgenic lines were observed with 1-7 heat-shocks. The phenotype shown in F was produced by 4-7 heat-shocks. None of the five independent transformant lines of pCaSpeR-hs dxII-III tested showed any detectable mutant phenotype under the condition described above (data not shown).

Fig. 5.

In vivo effects of ectopically expressed full-length Dx and Dx domain I on wing development. Adult wings from flies of the following genotypes and heat-shock treatments: (A) w1118 (0-24 hours after pupariation). Wild-type wing. Only approx. 1% of flies show very small wing vein gaps after our heat-shock treatment (data not shown). (B) AxE2 without heat-shock. A N gain-of-function allele, Ax shows wing vein gaps. (C)w1118; Dp(1;2)w+51b7/+ without heat-shock. Note the thickened wing veins (Confluens phenotype) caused by increasing the gene dose of N (D)Full-length Dx (0-24 hours after pupariation). Note the wing vein gap phenotype (approx. 10% penetrant) and its similarity to that produced by AxE2 (see B). (E) Full-length Dx (24-48 hours after pupariation). Note the thickened wing veins (70-100% penetrant) and their similarity to that produced by N gene duplications (see C). (F) pCaSpeR-hs dx1-303 (Dx domain I) (24-48 hours after pupariation). Note the wing vein gap phenotype (10-90% penetrant) and its similarity to the N gain-of-function allele AxE2 (see B) and full-length Dx (see D). Approximate developmental stages at which heat-shocks were administered are indicated. All four pCaSpeR-hs dx (full-length Dx) and five pCaSpeR-hs dx1-303 transgenic lines examined show the phenotypes described above with different penetrance. All phenotypes of full-length Dx transgenic lines were observed with 1-7 heat-shocks. The phenotype shown in F was produced by 4-7 heat-shocks. None of the five independent transformant lines of pCaSpeR-hs dxII-III tested showed any detectable mutant phenotype under the condition described above (data not shown).

Full-length or domain I Dx overexpression between 0-24 hours after puparium formation (AP) caused a rough eye phenotype characterized by irregular ommatidia and lens disruptions indicative of cell death (data not shown). Dx overexpression is also associated with mutant bristle phenotypes on the adult thorax. Examples are shown in Fig. 4. Both fulllength and domain I Dx overexpressed during the first six hours of pupariation resulted in the loss of external sensory organ structures (Fig. 4C,I), i.e. a ‘bald’ phenotype. The bald phenotype is similar to the gain-of-function Abruptex phenotype of N (Fig. 4B). Overexpression of full-length Dx at later developmental stages (approx. 24 hours after pupariation) results in ‘double socket’ and ‘double shaft/single socket’ (Fig. 4F,G and D,E) phenotypes. Overexpression of full-length Dx at late third instar larval or prepupal stage causes an extra bristle phenotype (Fig. 4H). All of these phenotypes can also be induced by the expression of activated forms of N at the same phenocritical periods (Rebay et al., 1993; I. Rebay and S. A.-T., unpublished data).

Consideration of the phenocritical periods associated with the Dx phenotypes provides further insight into how overexpression of Dx affects N signaling. N has previously been shown to be essential at two distinct stages of bristle development (Hartenstein and Posakony, 1990). Both loss-of-function and gain-of-function N mutations can cause bald phenotypes, albeit at different phenocritical periods (Hartenstein and Posakony, 1990; Rebay et al., 1993). Gain-of-function mutations cause a bald phenotype by interfering with early stages of bristle formation, whereas loss of N function induces a bald phenotype at later stages. Since the bald phenotype, caused by overexpression of Dx, occurs at early stages (0-6 hours AP) we conclude that the mutant effect is associated with activation of the N pathway.

Abruptex-like wing vein phenotypes (Fig. 5B) resulted from heat-shock pulses of full-length Dx (Fig. 5D) or domain I Dx (Fig. 5F) constructs during the early pupal period. Overexpression of full-length Dx also produced a thickening of wing veins (Fig. 5E) similar to that caused by the N Confluens allele, a phenotype that is also observed when an extra dose of the N gene is added to the wild-type chromosomal complement (Fig. 5C). Both phenotypes are also produced by the expression of an activated N protein under heat-shock control (Rebay et al., 1993). We also examined the biological activity of Dx domain II-III which does not interact with N but contains the putative SH3-binding domain and the ring-H2-zinc finger. Overexpression of Dx domain II-III (Fig. 1B) failed to produce any mutant phenotypes (data not shown).

Dx is a positive regulator of the N pathway

The fact that overexpression of Dx and activated N give similar phenotypes is consistent with the idea that Dx acts as a positive regulator of the pathway. We sought to further corroborate this conclusion by comparing the dominant phenotypes produced by overexpression of either Dx or an activated form of the N receptor, termed ΔECN (Rebay et al., 1993), in three different N mutant backgrounds. Synergistic effects between two mutations suggest, albeit not conclusively, that the corresponding genes belong to the same pathway. Therefore, if the observed Dx overexpression phenotypes are due to disruption of the N pathway, changing the N genetic background should affect the overexpression phenotype. Moreover, if Dx acts as a positive regulator of N signaling, we would expect that the phenotypes resulting from the overexpression of Dx and ΔECN would show similar variations in different N backgrounds.

Under modest heat-shock conditions in a wild-type N genetic background, flies transformed with the Dx or ΔECN constructs have wild-type wings (Fig. 6A-C). However, transformant flies of Dx and ΔECN that are also heterozygous for a N deletion (N/+) exhibit thickened wing veins, while nontransformed N/+ flies only show distal wing blade notching (Fig. 6D-F). Flies carrying a duplication of N have essentially normal wings with occasional extra wing vein material (Fig. 6G; Lindsley and Zimm, 1992). In contrast, overexpression of Dx or ΔECN in flies also bearing the N duplication results in missing wing veins (Fig. 6H,I). A similar wing vein phenotype is also observed when either Dx or ΔECN are overexpressed in flies heterozygous for an N mutation notchoid1 (nd1/+), which itself has no dominant wing phenotype (Fig. 6J-L; Xu et al., 1990). Therefore, in all three N mutant backgrounds tested, the phenotypic consequences of overexpression of Dx and activated N are remarkably similar, reinforcing the idea that Dx functions as a positive regulator in the N pathway.

Fig. 6.

Wing phenotypes induced by overexpression of Dx or activated N in various N mutant backgrounds. Adult wings from flies of the following genotypes: (A) Wild-type N genetic background. (B) [pCaSpeR-hs dx]/+. (C) [ΔECN]/+ (Rebay et al., 1993). (D) N64i1b/+. Flies heterozygous for a N deletion. (E) N64i1b/+; [pCaSpeR-hs dx]/+. (F) N64i1b/+; [ΔECN]/+. (G) SM1,Dp(1;2)51b/+. Flies carrying a N duplication. (H) SM1,Dp(1;2)51b/[pCaSpeR-hs dx]. (I) SM1,Dp(1;2)51b/[ΔECN]. (J) nd1/+. Flies heterozygous for the N mutation notchoid1. (K) nd1/+; [pCaSpeR-hs dx]/+. (L) nd1/+; [ΔECN]/+. Heat-shocks were administered twice 0-24 hr AP. All four lines of pCaSpeR-hs dx show the phenotypes with different penetrance. No mutant wing phenotypes were observed in the absence of heat-shock for any of the transformant genotypes used in this study.

Fig. 6.

Wing phenotypes induced by overexpression of Dx or activated N in various N mutant backgrounds. Adult wings from flies of the following genotypes: (A) Wild-type N genetic background. (B) [pCaSpeR-hs dx]/+. (C) [ΔECN]/+ (Rebay et al., 1993). (D) N64i1b/+. Flies heterozygous for a N deletion. (E) N64i1b/+; [pCaSpeR-hs dx]/+. (F) N64i1b/+; [ΔECN]/+. (G) SM1,Dp(1;2)51b/+. Flies carrying a N duplication. (H) SM1,Dp(1;2)51b/[pCaSpeR-hs dx]. (I) SM1,Dp(1;2)51b/[ΔECN]. (J) nd1/+. Flies heterozygous for the N mutation notchoid1. (K) nd1/+; [pCaSpeR-hs dx]/+. (L) nd1/+; [ΔECN]/+. Heat-shocks were administered twice 0-24 hr AP. All four lines of pCaSpeR-hs dx show the phenotypes with different penetrance. No mutant wing phenotypes were observed in the absence of heat-shock for any of the transformant genotypes used in this study.

The involvement of the Dx protein in N signaling appears to be primarily mediated by the ankyrin binding Dx domain I. As shown in Fig. 7, the dx phenotype is rescued by overexpression of either full-length Dx or a fragment of Dx encompassing domain I (Fig 7B,C). In contrast, overexpression of domain II-III does not rescue the dx phenotype (data not shown). These results are consistent with the fact that overexpression of domain I can produce dominant phenotypes whereas domain II-III cannot.

Fig. 7.

The loss-of-function dx phenotype can be rescued by overexpression of ‘activated’ N. Adult wings from flies of the following genotypes: (A) dx24. Small deltas of extra vein material are visible where the veins reach the wing margin (arrows). (B) dx24/Y; [pCaSpeR-hs dx]/+. dxwing phenotype was rescued (30% showed complete rescue). (C) dx24; [pCaSpeR-hs dx 1-303]. dx wing phenotype was rescued (50-100% showed complete rescue). (D) dx24/Y; [ΔECN]/+. dx wing phenotype was rescued (70% shows complete rescue). Heat-shocks were administered 2-4 times 0-24 hr AP. All three independent transformant lines of pCaSpeR-hs dx1-303 showed rescue. Four independent lines of pCaSpeR-hd dxII-III tested did not show any detectable rescue (data not shown).

Fig. 7.

The loss-of-function dx phenotype can be rescued by overexpression of ‘activated’ N. Adult wings from flies of the following genotypes: (A) dx24. Small deltas of extra vein material are visible where the veins reach the wing margin (arrows). (B) dx24/Y; [pCaSpeR-hs dx]/+. dxwing phenotype was rescued (30% showed complete rescue). (C) dx24; [pCaSpeR-hs dx 1-303]. dx wing phenotype was rescued (50-100% showed complete rescue). (D) dx24/Y; [ΔECN]/+. dx wing phenotype was rescued (70% shows complete rescue). Heat-shocks were administered 2-4 times 0-24 hr AP. All three independent transformant lines of pCaSpeR-hs dx1-303 showed rescue. Four independent lines of pCaSpeR-hd dxII-III tested did not show any detectable rescue (data not shown).

Finally, the availability of activated N allowed us to examine the epistatic relationship between N and dx. We asked whether a loss-of-function dx phenotype can be influenced by activation of N. Fig. 7D shows that the dx phenotypes involving wing venation abnormalities and outstretched wings (data not shown) were rescued by overexpression of activated N. This observation is consistent with the role of Dx as a positive regulator of N pathway. Furthermore, this result suggests that activation of N bypasses the requirement of Dx, and that dx acts upstream of N.

Interactions between Dx, N and Su(H)

We have recently reported that, in cultured Drosophila cells, the nuclear protein Su(H) is sequestered in the cytoplasm when coexpressed with N and is translocated to the nucleus when N binds to its ligand Dl (Fortini and Artavanis-Tsakonas, 1994). We asked whether these events might be influenced by Dx; we were especially interested because cytoplasmic retention of Su(H) requires the intracellular ankyrin repeats of N (Fortini and Artavanis-Tsakonas, 1994) and Dx function is mediated by interactions with the same repeats.

To test the effects of Dx on Su(H) subcellular localization, we coexpressed Dx and Su(H) in a cell line that stably expresses Notch. Our results are shown in Fig. 8. Su(H) is localized to the nucleus when expressed in S2 cells that do not express N (Fortini and Artavanis-Tsakonas, 1994 and Fig. 8A). When Su(H) is expressed alone in the N-expressing S2 cells, it is retained in the cytoplasm (Fortini and Artavanis-Tsakonas, 1994 and Fig. 8B). However, when Dx is coexpressed with Su(H) in the N-expressing S2 cells, Su(H) is seen in the nucleus (Fig. 8C). Assuming that the translocation of Su(H) to the nucleus reflects N signaling, then these results are consistent with the idea that Dx acts upstream of N and is involved in regulating the activity of the N receptor.

Fig. 8.

Dx interferes with cytoplasmic retention of Su(H) by N in Drosophila cultured cells. Drosophila S2 cells or an S2 cell line stably transformed with a full-length N construct were transfected with a construct expressing Su(H) or with constructs expressing Su(H) and Dx. Su(H) tagged with a myc epitope (Fortini and Artavanis-Tsakonas, 1994) and Dx were visualized by immunofluorescent confocal microscopy. Staining for Su(H) is shown in green (A-C), and that for Dx is shown in red (C). (A) A S2 cell expressing Su(H) only, shows nuclear localization (nuclear localization: 82.4±2.4 %). (B) A stable transformant cell induced to express full-length N and then induced to coexpress Su(H), shows cytoplasmic localization (nuclear localization: 20.1±1.1%). (C) A stable transformant cell induced to coexpress both full-length N and Dx (red) and then induced to coexpress Su(H) (green) as well, shows nuclear localization (nuclear localization: 61.9±3.4%). For each experiment, 100 cells were scored for predominantly nuclear or cytoplasmic localization of Su(H). The mean and range obtained for duplicated experiments are indicated in parentheses for each case.

Fig. 8.

Dx interferes with cytoplasmic retention of Su(H) by N in Drosophila cultured cells. Drosophila S2 cells or an S2 cell line stably transformed with a full-length N construct were transfected with a construct expressing Su(H) or with constructs expressing Su(H) and Dx. Su(H) tagged with a myc epitope (Fortini and Artavanis-Tsakonas, 1994) and Dx were visualized by immunofluorescent confocal microscopy. Staining for Su(H) is shown in green (A-C), and that for Dx is shown in red (C). (A) A S2 cell expressing Su(H) only, shows nuclear localization (nuclear localization: 82.4±2.4 %). (B) A stable transformant cell induced to express full-length N and then induced to coexpress Su(H), shows cytoplasmic localization (nuclear localization: 20.1±1.1%). (C) A stable transformant cell induced to coexpress both full-length N and Dx (red) and then induced to coexpress Su(H) (green) as well, shows nuclear localization (nuclear localization: 61.9±3.4%). For each experiment, 100 cells were scored for predominantly nuclear or cytoplasmic localization of Su(H). The mean and range obtained for duplicated experiments are indicated in parentheses for each case.

The interaction between dx and N was first noted when it was discovered that loss-of-function dx mutations are dominant suppressors of the lethality associated with certain ‘negatively complementing’ heteroallelic combinations of the Abruptex class of N mutant alleles (Foster, 1975; Portin, 1975; Xu and Artavanis-Tsakonas, 1990; Xu et al., 1990). The idea that dx is a part of the N signaling pathway is also supported by the gene dosage-dependent interactions dx exhibits with N, Dl, mam, H and Su(H) alleles and the finding that the Dx protein interacts physically with the intracellular part of N (Vässin et al., 1985, 1987; de la Concha et al., 1988; Fleming et al., 1990; Xu et al., 1990; Xu and Artavanis-Tsakonas, 1990; Gorman and Girton, 1992; Fortini and Artavanis-Tsakonas, 1994; Diederich et al., 1994). With some exceptions, loss-of-function dx phenotypes are similar to those seen in a variety of N mutants; these phenotypes include thickened wing veins, extra bristles and rough eyes (Gorman and Girton, 1992). However, the molecular and genetic studies carried out to date have not revealed the biochemical properties of Dx or its biological role. The present study addresses both of these questions.

Homology searches have revealed the existence of two distinct sequence motifs in Dx: a region showing homology to SH3-binding domains, and a ring-H2-zinc finger residing in domains II and III, respectively (Fig. 1C and D; Ren et al., 1993; Freemont, 1993). The function of ring-H2-zinc fingers is unknown. However, it has been speculated that they may be involved in protein-protein interactions (Freemont, 1993). Biochemically, we have shown that Dx is capable of binding zinc (R. J. D., unpublished results), but the evidence gathered so far does not allow us to clearly assign a function to the putative Dx SH3-binding domain or the ring-H2-zinc finger. While our transformation experiments demonstrate that only domain I is sufficient to rescue the dx mutation, they also suggest that domains II and III are not redundant for Dx function. For instance, overexpression of full-length Dx induces the dominant phenotypes much more efficiently than overexpression of Dx domain I.

Protein-protein interactions involving the N ankyrin repeats

Previous functional analyses involving truncated forms of N have demonstrated that the intracellular ankyrin repeats are essential for the activation of N (Rebay et al., 1993; Lieber et al., 1993). This result is consistent with expression studies in C. elegans which show that overexpression of the ankyrin repeat domain results in gain-of-function phenotypes of the N-like gene glp-1 (Roehl and Kimble, 1993). In addition to mediating binding to Dx, the ankyrin repeat region also mediates interactions with the Su(H) protein (Fortini and Artavanis-Tsakonas, 1994). Moreover, the domain of Dx that we have found to bind N does show some amino acid sequence similarity to Su(H) (Fortini and Artavanis-Tsakonas, 1994). These observations highlight the essential role the ankyrin repeats play in N signaling. Since it is possible that several ankyrin repeat interacting components simultaneously exist in the same cell, we presume that any two-way interaction will depend on the concentration of the individual components. It is unlikely that a more precise definition of the ankyrin repeat regions involved in interactions with Dx and Su(H) can rely on further deletion analysis in view of our finding that only the most C-terminal (sixth) ankyrin repeat may be deleted without affecting the ability of N to bind Dx. The fragment containing the ankyrin repeats 1-5 can interact with Dx, while fragments containing repeats 1-2 or 3-5 are incapable of interacting (Fig. 2). This observation may reflect either a cooperative effect between these two sets of repeats, or a requirement for all five repeats to form the structural domain which mediates the inter-actions with Dx.

The role of Dx in N signaling

The functional studies we present here show that the dominant phenotypes induced by Dx overexpression are similar to those induced by the overexpression of an activated N receptor. Assuming that activated N phenotypes are the consequence of constitutive N signaling, Dx appears to be a positive regulator of the pathway, an idea reinforced by the finding that the over-expression phenotypes of Dx and activated N are modulated in identical fashion by changing the N activity in the genetic background. Moreover, the fact that activated N can rescue the dx phenotype, suggests that dx acts genetically upstream of N, and that dx is a positive regulator of the N pathway. Dx domain I is capable of binding to the N ankyrin repeats, is sufficient to rescue the loss-of-function phenotypes of dx, and is able to induce the overexpression phenotypes of Dx. These observations suggest that Dx is involved in the N signaling pathway through interactions with N ankyrin repeats.

In the wing, loss-of-function alleles of dx or Dl exhibit similar phenotypes (Xu and Artavanis-Tsakonas, 1990; Gorman and Girton, 1992; Parody and Muskavitch, 1993). Moreover, the interaction of either Dx or Dl with N interferes with the cytoplasmic retention of Su(H) by N. This result is consistent with the idea that both Dx and Dl alter the overall activity of the pathway by affecting the subcellular localization of Su(H). Even though we cannot evaluate the relative contributions of Dx and Dl in the transduction of the N signal in vivo, it is reasonable to speculate that the role of Dx is to maintain the ‘activated’ state of N. Dx may do this by interfering with the retention of Su(H) in the cytoplasm via its inter-action with the N ankyrin repeats.

The phenotypes associated with the overexpression of Su(H) or Dx are similar to those generated by activated forms of the N receptor (Rebay et al., 1993; Struhl et al., 1993; Schweisguth and Posakony, 1994). These observations can be explained if excess Su(H) in the cell cannot be fully retained in the cytoplasm by the available N, resulting in the translocation of Su(H) to the nucleus and consequent activation of the N pathway. Overexpression of Dx could also have the same effect since it may interfere with the cytoplasmic retention of Su(H), as our cell culture data suggest. How the interplay between Dx and Su(H) regulates the N signaling in vivo, and whether it is involved in regulating N activity at all developmental times, remains to be determined. The expression studies performed to date suggest that both Su(H) and dx have broad and overlapping expression patterns (Diederich et al., 1994; Busseau et al., 1994; Schweisguth and Posakony 1992; Furukawa et al., 1992).

A model for Dx action

The results we present here allow us to formulate a model for Dx function. Fig. 9 summarizes the proposed physical interactions and biochemical functions of proteins in the Notch pathway. Dx interacts with the ankyrin repeats of N as an oligomer (Fig. 9A). The interaction of the N ankyrin repeats and Dx domain I is essential for Dx function. The activation of the N receptor via an interaction with its ligand Dl results in the translocation of Su(H) to the nucleus, which is otherwise retained in the cytoplasm by its association with the ankyrin repeats. Dx interferes with the cytoplasmic retention of Su(H) by virtue of its binding to the N ankyrin repeats. This model is undoubtedly an oversimplification, as it fails, for instance, to explain why complete elimination of Notch in vivo does not result in the production of activated phenotypes. In addition it should be kept in mind that aspects of this model are based on overexpression data which must be corroborated by in vivo studies since they can be misleading under certain circum-stances. Nevertheless, in spite of these uncertainties this model provides us with a useful working hypothesis. Additional studies are underway to test and refine this speculative model and to elucidate the complex interactions between Notch, its ligands and its intracellular partners.

Fig. 9.

Model of physical interactions of the cytoplasmic N signaling proteins. (A) Summary of protein-protein interactions involving N and Dx proteins. Protein-protein interactions documented in this study are shown schematically. N is shown as an open rectangle anchored in the cell membrane (double solid line), with six ankyrin repeats (horizontal ovals). The Dx protein is depicted as composed of three separate domains, labeled I, II, and III, as described in the text. Association of the N ankyrin repeats and Dx domain I is depicted. Homotypic interactions of Dx domains are indicated by solid arrowheads. (B) Model of antagonistic interactions of Dx and Su(H). A nuclear protein, Su(H) has been shown to interact with the N ankyrin repeats (Fortini and Artavanis-Tsakonas, 1994). In Drosophila cultured cells, the Su(H) protein is sequestered in the cytoplasm when coexpressed with N via its interaction with the N ankyrin repeats (left; Fortini and Artavanis-Tsakonas, 1994). The Su(H) protein translocates into the nucleus when N binds to its ligand, Dl (middle; Fortini and Artavanis-Tsakonas, 1994). We suggest that the N-Dx interaction is involved in promoting the nuclear translocation of the Su(H) protein (right).

Fig. 9.

Model of physical interactions of the cytoplasmic N signaling proteins. (A) Summary of protein-protein interactions involving N and Dx proteins. Protein-protein interactions documented in this study are shown schematically. N is shown as an open rectangle anchored in the cell membrane (double solid line), with six ankyrin repeats (horizontal ovals). The Dx protein is depicted as composed of three separate domains, labeled I, II, and III, as described in the text. Association of the N ankyrin repeats and Dx domain I is depicted. Homotypic interactions of Dx domains are indicated by solid arrowheads. (B) Model of antagonistic interactions of Dx and Su(H). A nuclear protein, Su(H) has been shown to interact with the N ankyrin repeats (Fortini and Artavanis-Tsakonas, 1994). In Drosophila cultured cells, the Su(H) protein is sequestered in the cytoplasm when coexpressed with N via its interaction with the N ankyrin repeats (left; Fortini and Artavanis-Tsakonas, 1994). The Su(H) protein translocates into the nucleus when N binds to its ligand, Dl (middle; Fortini and Artavanis-Tsakonas, 1994). We suggest that the N-Dx interaction is involved in promoting the nuclear translocation of the Su(H) protein (right).

We thank Robert Geisler and Christiane Nüsslein-Volhard for providing the cactus cDNA. We are especially grateful to Roger Brent, Russ Finley, Steve Hanes, Erica Golemis and Jenö Gyuris for their generosity and advice regarding the yeast interaction trap assay. We thank Laurent Caron, Audrey Hing and Robert Mann for technical assistance, Martin Baron for his insights on protein structure, and Mark Fortini, Iain Dawson and Grace Gray for comments on the manuscript. K. M. and M. J. G. were supported by long-term fellowships from the International Human Frontier Science Program Organization (HFSPO) and postdoctoral fellowships from the Howard Hughes Medical Institute. R. J. D. was supported by a postdoctoral fellowship from the Anna Fuller Fund and the Howard Hughes Medical Institute. S. A. T. is supported by the Howard Hughes Medical Institute and by N.I.H. grant NS26084.

Artavanis-Tsakonas
,
S.
,
Matsuno
,
K.
and
Fortini
,
M. E.
(
1995
).
Notch signaling
.
Science
268
,
225
232
.
Bang
,
A.G.
, and
Posakony
,
J.W.
(
1992
).
The Drosophila gene Hairless encodes a novel basic protein that controls alternative cell fates in adult sensory organ development
.
Genes Dev
.
6
,
1752
1769
.
Breeden
,
L.
, and
Nasmyth
,
K.
(
1987
).
Similarity between cell-cycle genes of budding yeast and fussion yeast and the N gene of Drosophila
.
Nature
329
,
651
654
.
Busseau
,
I.
,
Diederich
,
R.J.
,
Xu
,
T.
, and
Artavanis-Tsakonas
,
S.
(
1994
).
A member of the Notch group of interacting loci, Dx encodes a cytoplasmic basic protein
.
Genetics
136
,
585
596
.
Coffman
,
C.R.
,
Skoglund
,
P.
,
Harris
,
W.A.
, and
Kintner
C.R.
(
1993
).
Expression of an extracellular deletion of Xotch diverts cell fate in Xenopus embryos
.
Cell
73
,
659
671
.
de la Concha
,
A.
,
Dietrich
,
U.
,
Weigel
,
D.
, and
Campos-Ortega
,
J.A.
(
1988
).
Functional interactions of neurogenic genes of Drosophila melanogaster
.
Genetics
118
,
499
508
.
Delidakis
,
C.
,
Preiss
,
A.
,
Hartley
,
D.A.
, and
Artavanis-Tsakonas
,
S.
(
1991
).
Two genetically and molecularly distinct functions involved in early neurogenesis reside within the Enhancer of split locus of Drosophila melanogaster
.
Genetics
129
,
803
823
.
Diederich
,
R.J.
,
Matsuno
,
K.
,
Hing
,
H.
, and
Artavanis-Tsakonas
,
S.
(
1994
).
Cytosolic interaction between Dx and Notch ankyrin repeats implicates Dx in the Notch signaling pathway
.
Development
120
,
473
481
.
Fehon
,
R. G.
,
Kooh
,
P. J.
,
Rebay
,
I.
,
Regan
,
C. L.
,
Xu
,
T.
,
Muskavitch
,
M. A. T.
and
Artavanis-Tsakonas
,
S.
(
1990
).
Molecular interactions between the protein products of the neurogenic loci Notch and Delta, two EGF- homologous genes in Drosophila
.
Cell
61
,
523
534
.
Fleming
,
R.J.
,
Scottgale
,
T.N.
,
Diederich
,
R.J.
, and
Artavanis-Tsakonas
,
S.
(
1990
).
The gene Serrate encodes a putative EGF-like transmembrane protein essential for proper ectodermal development in Drosophila melanogaster
.
Genes Dev
.
4
,
2188
2201
.
Fortini
,
M.E.
, and
Artavanis-Tsakonas
,
S.
(
1993
).
Notch: neurogenesis is only part of the picture
.
Cell
75
,
1245
1247
.
Fortini
,
M.E.
, and
Artavanis-Tsakonas
,
S.
(
1994
).
The Suppressor of Hairless protein participates in Notch receptor signaling
.
Cell
79
,
273
282
.
Fortini
,
M.E.
,
Rebay
,
I.
,
Caron
,
L.A.
, and
Artavanis-Tsakonas
,
S.
(
1993
).
An activated Notch receptor blocks cell-fate commitment in the developing Drosophila eye
.
Nature
365
,
555
557
.
Foster
,
G. G.
(
1975
).
Negative complementation at the Notch locus of Drosophila melanogaster
.
Genetics
81
,
99
120
.
Freemont
,
P.S.
(
1993
).
The RING finger
.
Ann. New York Acad. Sci
.
703
,
174
192
.
Furukawa
,
T.
,
Kawaichi
,
M.
,
Matsunami
,
N.
,
Ryo
,
H.
,
Nishida
,
Y.
,
Honjo
,
T.
(
1991
).
The Drosophila RBP-Jκ gene encodes the binding protein for the immunoglobulin Jκ recombination signal sequence
.
J. Biol. Chem
.
266
,
23334
23340
.
Furukawa
,
T.
,
Maruyama
,
S.
,
Kawaichi
,
M.
, and
Honjo
,
T.
(
1992
).
The Drosophila homolog of the immunoglobulin recombination signal-binding protein regulates peripheral nervous system development
.
Cell
69
,
1191
1197
.
Geisler
,
R.
,
Bergmann
,
A.
,
Hiromi
,
Y.
and
Nüsslein-Volhard
,
C.
(
1992
).
cactus, a gene involved in dorsoventral pattern formation of Drosophila, is related to the IκB gene family of vertebrates
.
Cell
71
,
613
621
.
Gertler
,
F.B.
,
Hill
,
K.K.
,
Clark
,
M.J.
, and
Hoffmann
,
F.M.
(
1993
).
Dosage- sensitive modifiers of Drosophila abl tyrosine kinase function: prospero, a regulator of axonal outgrowth, and disabled, a novel tyrosine kinase substrate
.
Genes Dev
.
7
,
441
453
.
Gietz
,
D.
,
St. Jean
,
A.
,
Wood
,
R.A.
, and
Schiestl
,
R.
(
1992
).
Improved method for high efficiency transformation of intact yeast cells
.
Nucl. Acids Res
.
20
,
1425
.
Gorman
,
M.J
, and
Girton
,
J.R.
(
1992
).
A genetic analysis of Dx and its interaction with the Notch locus in Drosophila melanogaster
.
Genetics
131
,
99
112
.
Greenwald
,
I.
, and
Rubin
,
G. M.
(
1992
).
Making a difference: the role of cell- cell interactions in establishing separate identities for equivalent cells
.
Cell
68
,
271
281
.
Guarente
,
L.
(
1983
).
Yeast promoters and lacZ fusions designed to study expression of cloned genes in yeast
.
Meth. Enzymol
.
101
,
181
191
.
Hartenstein
,
V.
, and
Posakony
,
J. W.
(
1990
).
A dual function of the Notch gene in Drosophila sensillum development
.
Dev. Biol
.
142
,
13
30
.
Kidd
,
S.
(
1992
).
Characterization of the Drosophila cactus locus and analysis of interactions between cactus and dorsal proteins
.
Cell
71
,
623
635
.
Kidd
,
S.
,
Kelley
,
M.R.
, and
Young
,
M.W.
(
1986
).
Sequence of the Notch locus of Drosophila melanogaster: relationship of the encoded protein to mammalian clotting and growth factors
.
Mol. Cell. Biol
.
6
,
3096
3108
.
Kopczynski
,
C.C.
,
Alton
,
A.K.
,
Fechtel
,
K.
,
Kooh
,
P.J.
, and
Muskavitch
,
M.A.T.
(
1988
).
Delta, a Drosophila neurogenic gene, is transcriptionally complex and encodes a protein related to blood coagulation factors and epidermal growth factor of vertebrates
.
Genes Dev
.
2
,
1723
1735
.
Lieber
,
T.
,
Kidd
,
S.
,
Alcamo
,
E.
,
Corbin
,
V.
, and
Young
,
M.W.
(
1993
).
Antineurogenic phenotypes induced by truncated Notch proteins indicate a role in signal transduction and may point to a novel function for Notch in nuclei
.
Genes Dev
.
7
,
1949
1965
.
Lindsley
,
D.
, and
Zimm
,
G.G.
(
1992
).
The Genome of Drosophila melanogaster
San Diego, California
:
Academic Press
.
Maier
,
D.
,
Stumm
,
G.
,
Kuhn
,
K.
, and
Preiss
,
A.
(
1992
).
Hairless, a Drosophila gene involved in neural development, encodes a novel, serine rich protein
.
Mech. Dev
.
38
,
143
156
.
Parody
,
T.R.
and
Muskavitch
,
M.A.T.
(
1993
).
The pleiotropic function of Delta during postembryonic development of Drosophila melanogaster
.
Genetics
135
,
527
539
.
Portin
,
P.
(
1975
).
Allelic negative complementation at the Abruptex locus of Drosophila melanogaster
.
Genetics
81
,
121
133
.
Rebay
,
I.
,
Fleming
,
R. J.
,
Fehon
,
R. G.
,
Cherbas
,
L.
,
Cherbas
,
P.
and
Artavanis-Tsakonas
,
S.
(
1991
).
Specific EGF repeats of Notch mediate interactions with Delta and Serrate: implications for Notch as a multifunctional receptor
.
Cell
67
,
687
699
.
Rebay
,
I.
,
Fehon
,
R.
and
Artavanis-Tsakonas
,
S.
(
1993
).
Specific truncations of Drosophila Notch define dominant activated and dominant negative forms of the receptor
.
Cell
74
,
319
329
.
Ren
,
R.
,
Mayer
,
B.J.
,
Cicchetti
,
P.
, and
Baltimore
,
D.
(
1993
).
Identification of a ten-amino acid proline-rich SH3 binding site
.
Science
259
,
1157
1161
.
Robertson
,
H.M.
,
Preston
,
C.R.
,
Phillis
,
R.W.
,
Johnson-Schlitz
,
D.M.
,
Benz
,
W.K.
(
1988
).
A stable genomic source of P element transposase in Drosophila melanogaster
.
Genetics
118
,
461
470
.
Schrons
,
H.
,
Knust
,
E.
, and
Campos-Ortega
,
J.A.
(
1992
).
The Enhancer of split complex and adjacent genes in the 96F region of Drosophila melanogaster are required for segregation of neural and epidermal progenitor cells
.
Genetics
132
,
481
503
.
Schweisguth
,
F.
, and
Posakony
,
J.W.
(
1992
).
Suppressor of Hairless, the Drosophila homolog of the mouse recombination signal-binding protein gene, controls sensory organ cell fates
.
Cell
69
,
1199
1212
.
Schweisguth
,
F.
, and
Posakony
,
J.W.
(
1994
).
Antagonistic activities of Suppressor of Hairless and Hairless control alternative cell fates in the Drosophila adult epidermis
.
Development
120
,
1433
1441
.
Sherman
,
F.
(
1991
).
Getting started with yeast. Guide to Yeast Genetics and Molecular Biology Methods in Enzymology 194
(ed.
C.
Guthrie
and
G. R.
Fink
), pp.
3
21
.
San Diego, California
:
Academic Press
.
Simon
,
M.A.
,
Dodson
,
G.S.
, and
Rubin
,
G.M.
(
1993
).
An SH3-SH2-SH3 protein is required for p21ras1 activation and binds to sevenless and Sos proteins in vitro
.
Cell
73
,
169
177
.
Smoller
,
D.
,
Friedel
,
C.
,
Schmid
,
A.
,
Bettler
,
D.
,
Lam
,
L.
, and
Yedbovnick
,
B.
(
1990
).
The Drosophila neurogenic locus mastermind encodes a nuclear protein unusually rich in amino acid homopolymers
.
Genes Dev
.
4
,
1688
1700
.
Spradling
,
A.C.
(
1986
).
P element mediated transformation
. In
Drosophila: A Practical Approach
(ed.
D. M.
Glover
), pp.
175
197
.
Washington DC
:
IRL Press
.
Struhl
,
G.
,
Fitzgerald
,
K.
and
Greenwald
,
I.
(
1993
).
Intrinsic activity of the lin-12 and Notch intracellular domains in vivo
.
Cell
74
,
331
345
.
Thomas
,
U.
,
Speicher
,
S.A.
, and
Knust
,
E.
(
1991
).
The Drosophila gene Serrate encodes an EGF-like transmembrane protein with a complex expression pattern in embryos and wing discs
.
Development
111
,
749
761
.
Vässin
,
H.
,
Vielmetter
,
J.
, and
Campos-Ortega
,
J.A.
(
1985
).
Genetic interactions in early neurogenesis of Drosophila melanogaster
.
J. Neurogenet
.
2
,
291
308
.
Vässin
,
H.
,
Bremer
,
K.A.
,
Knust
,
E.
, and
Campos-Ortega
,
J.A.
(
1987
).
The neurogenic gene Delta of D. melanogaster is expressed in neurogenic territories and encodes a putative transmembrane protein with EGF-like repeats
.
EMBO J
.
6
,
3431
3440
.
Wharton
,
K.A.
,
Johansen
,
K.M.
,
Xu
,
T.
, and
Artavanis-Tsakonas
,
S.
(
1985
).
Nucleotide sequence from the neurogenic locus Notch implies a gene product that shares homology with proteins containing EGF-like repeats
.
Cell
43
,
567
581
.
Woodgett
,
J.R.
(
1991
).
A common denominator linking glycogen metabolism, nuclear oncogenes and development
.
Trends Biochem. Sci
.
16
,
177
181
.
Xu
,
T.
and
Artavanis-Tsakonas
,
S.
(
1990
).
Dx, a locus interacting with the neurogenic genes, Notch, Delta and mastermind in Drosophila melanogaster
.
Genetics
126
,
665
677
.
Xu
,
T.
,
Rebay
,
I.
,
Fleming
,
R. J.
,
Scottgale
,
T. N.
and
Artavanis-Tsakonas
,
S.
(
1990
).
The Notch locus and the genetic circuitry involved in early Drosophila neurogenesis
.
Genes Dev
.
4
,
464
475
.
Zervos
,
A. S.
,
Gyuris
,
J.
and
Brent
,
R.
(
1993
).
Mxi1, a protein that specifically interacts with Max to bind Myc-Max recognition sites
.
Cell
72
,
223
232
.