Fly stocks

The following Gal4 lines were used: CY2-Gal4 (Queenan et al., 1997), 55B-Gal4 (Brand and Perrimon, 1994), MS1096-Gal4 (Capdevila and Guerrero, 1994), nos-Gal4 and pCOG-Gal4 (Rorth, 1998). UAST-mbSpi, UAST-S and UAST-rho-1 were a gift from A. Michelson. UAST-brho and the S alleles (S²¹⁸ and S^I) were a gift from A. Guichard (Guichard et al., 2000). Deficiencies for brho: Df(3L)5411, Df(3L)5412, Df(3L)4214, Df(3L)2399 and Df(3L)57, were obtained from the Bloomington stock center.

S²¹⁸ mitotic clones in the follicle cells were generated using the FRT40; T155-UASFLP chromosome (Duffy et al., 1998). S²¹⁸ germline clones were generated as previously described (Chou and Perrimon, 1996). Wings and eggs were mounted in Hoyer’s medium for microscopic examination.

Construction of plasmids and generation of transgenic lines

Cloning details for making these constructs are available by request. The coding regions of rho-1 (Bier et al., 1990), brho (Guichard et al., 2000), spi (Rutledge et al., 1992), secspi, S (Kolodkin et al., 1994), grk (Neuman-Silberberg and Schüpbach, 1993), secgrk, mbgrk^myc and mbgrkΔ19AA^myc were subcloned into the P-element vector pUAST (Brand and Perrimon, 1993) and/or pUASP (Rorth, 1998). mbgrk^myc and mbgrkΔ19AA^myc were made by cloning six consecutive Myc tags at the C terminus. The coding regions of S and brho were subcloned in reverse orientation into pUASp to give antisense UASp. The UAS constructs were then introduced into flies by standard methods of P-element-mediated germline transformation of w¹¹¹⁸ embryos.

Insect cell experiments

Immunoblot analysis

For preparation of ovarian extracts, 10 ovaries were dissected in ice-cold PBS. The PBS was removed and replaced with loading buffer. Ovaries were homogenized using a pestle, then centrifuged briefly to pellet debris. An equivalent of two ovaries per lane was loaded on a 10% SDS-PAGE gel under reducing conditions. The mouse monoclonal anti-Myc antibody 9E10 (Ab-1, Calbiochem) was used at 1:1000 and the HRP-conjugated secondary antibody (Vector) at 1:2000. The bands were visualized using ECL (Amersham).

Antibody staining

The Grk monoclonal antibody, 1D12, was raised against amino acids 53-185 from the Grk extracellular domain. Fixation and staining protocols are described elsewhere (Queenan et al., 1999). The monoclonal mouse anti-Myc antibody (Ab-1, Calbiochem) was used at 1:500, followed by FITC anti-mouse (Vector) at 1:500. Then, ovaries were washed, counterstained with phalloidin TRITC (Sigma) and mounted in Citifluor. Confocal images were collected on a Leica TCS confocal microscope.

Overexpression of either mbGrk or secGrk in follicle cells can activate the Egfr

We noticed that Grk, like TGFα and Spi, possess a putative dibasic signal (R₂₄₀ and K₂₄₁) that resembles a potential cleavage site (Brachmann et al., 1989; Wong et al., 1989; Schweitzer et al., 1995a) between the EGF and the transmembrane (TM) domains. To determine whether the extracellular domain of Grk alone is able to activate the Egfr, we generated a secreted form of Grk (secGrk) by inserting a stop codon after the dibasic putative cleavage site (Fig. 1A). This mutation leads to the deletion of both the TM and cytoplasmic domains. After transient expression in S2 cells, significant amount of secGrk was detected in the tissue culture medium (see below). To test the activity of these different forms of Grk, we analyzed the effect of overexpressing them using the UAS/Gal4 system (Brand and Perrimon, 1993). Several transgenic fly lines for UAS-wild-type Grk (mbGrk) and UAS-secGrk were generated and crossed to various Gal4 lines.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

mbGrk and secGrk overexpression in somatic tissues lead to Egfr activation. (A) Structure of mbGrk and secGrk. SP, signal peptide; EGF, EGF motif; TM, transmembrane domain; C, cytoplasmic domain. The arrow represents the site where the stop codon was introduced to generate secGrk. (B-J) Phenotypes observed after UAS-mbGrk and UAS-secGrk ectopic expression in somatic tissues. (B) Wild-type egg with its two dorsal appendages. (C) UAS-mbGrk/CY2: overexpression of mbGrk in all the follicular epithelium leads to a weak eggshell dorsalization (78% of the eggs, n=352). (D) UAS-secGrk/CY2: overexpression of secGrk leads to a very strong dorsalized eggshell phenotype (85% of the eggs, n=244). (E-G) Embryonic cuticles from: wild-type (E), UAS-mbGrk/CY2 (F) and UAS-secGrk/CY2 (G) females. (H) Wild-type wing with five longitudinal veins and two crossveins. (I,J) Weak and strong wing vein phenotypes associated with UAS-mbGrk/MS1096 (I) and UAS-secGrk/MS1096 wings (J).

Overexpression of UAS-mbGrk in the ovarian follicle cells, using the CY2-Gal4 driver, leads to a weak dorsalization of the eggshell, as shown by the thickened dorsal appendages (Fig. 1C), which are spread further apart than in the wild-type controls (Fig. 1B). Examination of the cuticles reveals that the resulting embryos are also dorsalized (Fig. 1F). When overexpressed in the wing disc, using the MS1096-Gal4, UAS-mbGrk induces ectopic wing veins (Fig. 1I) that resemble moderate Egfr gain-of-function phenotypes. Our results are in contrast to previous reports (Queenan et al., 1999; Guichard et al., 2000) and suggest that mbGrk is able to activate weakly the Egfr signaling pathway in somatic tissues.

In contrast to the weak activation observed with UAS-mbGrk, overexpression of UAS-secGrk, using the same Gal4 lines, causes very strong gain-of-function phenotypes similar to those obtained when UAS-λtop, a constitutively activated Egfr, is misexpressed (Queenan et al., 1997). A strong dorsalization of the eggshell is clearly visible by the presence of a mass of dorsal appendage material around the entire anterior circumference of the egg (Queenan et al., 1999) (Fig. 1D). The cuticle of these embryos is also very strongly dorsalized (Fig. 1G). In addition, a very strong wing vein phenotype is observed after overexpression in the wing disc (Queenan et al., 1999) (Fig. 1J).

Altogether, these results indicate that mbGrk and secGrk are able to activate the Egfr when overexpressed in follicle cells. However, the strength of their activation differs: while mbGrk triggers weakly activation of Egfr, secGrk does it strongly. We also noticed a difference in the frequency of transgenic lines giving phenotypes: whereas all 12 UAS-secGrk lines trigger Egfr signaling, only two out of 20 UAS-mbGrk lines were associated with a weak dorsalization phenotype.

Overexpression of either mbGrk or secGrk in the germline can activate the Egfr

Next we tested the activity of the different Grk forms in the oocyte where Grk is normally expressed. However, because the original UAS/Gal4 system does not work in the female germline (Brand and Perrimon, 1993; Manseau et al., 1997), mbGrk and secGrk were subcloned into pUASp, a modified pUAST vector (Rorth, 1998). Subsequently, UASp-mbGrk and UASp-secGrk were overexpressed using the germline-specific Gal4 lines pCOG-Gal4 and/or nanos-Gal4 (nos-Gal4) (Rorth, 1998).

Overexpression of mbGrk and secGrk in the oocyte leads to activation of the Egfr pathway, as demonstrated by the dorsalization of the resulting eggs (Fig. 2A,B). In contrast to what we observed in somatic tissues, mbGrk gives a fully penetrant and stronger phenotype than secGrk. The frequency of transgenic lines giving Egfr gain-of-function phenotypes is also reversed: whereas all the UASp-mbGrk lines strongly activate the Egfr pathway, only two out of 12 UASp-secGrk lines were associated with a weakly dorsalized phenotype. This is in contrast to a report by Queenan et al. (Queenan et al., 1999), who described that after expression in the oocyte using the endogenous grk promoter, secGrk was not able to activate the Egfr. This difference is probably due to a higher expression level of UASp-secGrk.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

UAS-mbGrk and UAS-secGrk overexpression in the oocyte leads to Egfr activation. (A) 100% (n=122) of the eggs derived from UASp-mbGrk/pCOG females exhibit a strong dorsalized eggshell. (B) Only 11% (n=441) of the eggs derived from UASp-secGrk/pCOG females exhibit a weak dorsalized eggshell.

secGrk triggers Egfr autophosphorylation in vitro, but not mbGrk

Several membrane-bound ligands, such as proTGFα, CSF-1, proEGF and Kit ligand, need to be processed from a membrane-bound to a soluble form to activate their respective receptors; however, uncleaved membrane-bound ligands are also known to have signaling activity (Massague and Pandiella, 1993). Because both mbGrk and secGrk activate the Egfr, we performed tissue culture and biochemical experiments to determine the mechanism by which Grk activates the Egfr.

First, using an aggregation assay in Sf9 cells, we examined whether Grk binds to the Egfr. Cells expressing either mbGrk or the Egfr do not aggregate with themselves, or with cells infected with wild-type baculovirus (data not shown). However, when mbGrk-expressing cells are mixed with Egfr-expressing cells, these two kinds of cells strongly aggregate (Fig. 3A). mbGrk appears to bind selectively to the Egfr because there is no aggregation between mbGrk-expressing cells and cells expressing Kek1, a transmembrane protein that is able to bind and inhibit the Egfr (Ghiglione et al., 1999) (Fig. 3B). Furthermore, aggregation between mbGrk and Egfr-expressing cells could be fully and selectively blocked using conditioned media from cells expressing either secGrk (Fig. 3C) or Aos, a secreted Egfr inhibitor (Schweitzer et al., 1995b) (data not shown). These results indicate that mbGrk, expressed in one cell, is capable of interacting with the Egfr on an adjacent cell.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

mbGrk and secGrk bind Egfr, but only secGrk triggers signaling. (A-C) Aggregation of mbGrk-expressing cells with Egfr cells. Sf9 insect cells were independently infected with wild-type baculovirus, or virus encoding Egfr, mbGrk or Kek1. The two populations of cells were mixed and incubated together for 1 hour and then examined. (A) Cells expressing mbGrk strongly aggregate to cells expressing Egfr. (B) Cells expressing mbGrk and cells expressing Kek1 do not aggregate. (C) Cells expressing mbGrk and cells expressing Egfr, in presence of conditioned medium from cells expressing secGrk, no longer aggregate. (D) secGrk but not mbGrk activates Egfr in vitro. S2:Egfr cells (2×10⁶) were incubated for 3 hours with 60 μM CuSO₄ to induce moderate levels of Egfr expression (Schweitzer et al., 1995a). The cells were then stimulated for 30 minutes with serum free media (SFM), mbGrk, secGrk or secSpi. Egfr precipitates were prepared using rabbit anti-Egfr polyclonal sera. Top: the level of tyrosine phosphorylation of the Egfr was assessed by western blotting with mAb RC-20 (PY). Bottom: Egfr expression was then assessed by western blotting with rabbit anti-Egfr polyclonal sera. The lower amounts of Egfr in secGrk and secSpi lanes are most probably due to incomplete stripping of the RC-20 antibody.

The initial event after activation of the EGFR is autophosphorylation of tyrosine residues on the C-terminal tail (Ullrich and Schlessinger, 1990). Biochemical analysis in vitro showed that Spi activates Egfr signaling: the addition of secreted Spi (secSpi), but not the membrane-associated form (mbSpi), to cultured cells expressing the Egfr is associated with a rapid autophosphorylation of Egfr on tyrosine residues (Schweitzer et al., 1995a). To test whether mbGrk and secGrk activate the Egfr, we used S2 cells stably transfected with the Egfr under the control of the metallothionein promoter. After induction with CuSO₄ (see Materials and Methods), a minimal level of spontaneous tyrosine phosphorylation was observed on the Egfr (data not shown), probably owing to overcrowding of Egfr molecules on the cell surface, which leads to spontaneous dimerization of receptor molecules (Schweitzer et al., 1995a). Addition of serum-free medium (SFM), or of baculovirus-membranes expressing mbGrk to S2:Egfr cells, did not affect the level of phosphotyrosine residues on the Egfr. However, addition of baculovirus-generated secGrk or secSpi to the same cells results in a dramatic increase in the phosphotyrosine content of Egfr (Fig. 3D).

The tissue culture experiments described above confirm that Grk is a ligand of Egfr. Moreover, we have shown that mbGrk and secGrk can bind to the receptor, but only secGrk triggers Egfr autophosphorylation.

Grk is cleaved in the germline

As mbGrk does not activate the Egfr in tissue culture, we presumed that Grk must be cleaved to trigger Egfr signaling in vivo. To follow the cleavage of Grk in vivo, we generated a form of Grk that contains six Myc epitope tags at the C terminus (mbGrk^myc) (Fig. 4A). mbGrk^myc is fully active, as demonstrated by its ability to strongly dorsalize the egg after overexpression in the germline (Fig. 4B). The intracellular domain of Grk was then detected with an antibody against Myc, and compared with the overall distribution of Grk, which was detected using an antibody against the extracellular region (Peri et al., 1999; Queenan et al., 1999).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

The extracellular domain of Grk is released from the oocyte and internalized in follicle cells. (A) Structure of mbGrk^myc. (B) All eggs laid by UASp-mbGrk^myc/nos-Gal4 females have a strongly dorsalized eggshell. (C,D) Confocal sections showing UASp-mbGrk^myc/nos-Gal4 egg chambers stained with an anti-Grk antibody that detects the N-terminal region of the protein (C), or with an anti-Myc antibody that recognizes the intracellular region of the protein (D). While the uptake of the extracellular domain of Grk is clearly visible in follicle cells (C), the intracellular domain of Grk remains confined to the oocyte (D). Grk and Myc are in green, and actin is in red. (E) Western blot probed with anti-Myc antibody reveals a cleavage of mbGrk^myc. Lane 2 is a protein lysate from UASp-mbGrk^myc/nos-Gal4 ovaries in which two new bands are detected (arrows). The higher band corresponds to the full-length mbGrk^myc protein whereas the lower one corresponds to a cleaved form. These two bands are absent in extracts from wild-type ovaries (lane 1).

In the wild-type ovary, after expression of mbGrk^myc in the oocyte, the cytoplasmic portion of the mbGrk^myc protein remains confined to the oocyte (Fig. 4D). By contrast, the Grk antibody labels both in the oocyte and large dot-like structures in follicle cells. These structures presumably correspond to vesicular compartments, such as endosomes or lysosomes, that contain the extracellular domain of Grk (Peri et al., 1999; Queenan et al., 1999) (Fig. 4C).

Comparison of protein lysates from ovaries overexpressing mbGrk^myc in the oocyte and from wild-type, using an anti-Myc antibody, reveals the appearance of two bands in the lane corresponding to the protein lysates from ovaries overexpressing mbGrk^myc (Fig. 4E). The band around 70 kDa corresponds to the full-length mbGrk^myc protein, whereas the band around 32 kDa corresponds to a cleaved form. Strikingly, the majority of the mbGrk^myc protein is cleaved, after overexpression in the oocyte, as shown by the relative small amount of the higher band compared with the lower one.

Altogether, these results indicate that Grk is cleaved in the oocyte. The Grk extracellular domain is taken up by follicle cells, while the intracellular domain remains confined to the oocyte.

Deletion of the region between the EGF and the TM domains creates a dominant-negative Grk protein

To determine whether mbGrk processing is required for Egfr activation, we mutated the putative dibasic cleavage signal (R₂₄₀ and K₂₄₁) of Grk. This mutation does not affect the ability of this mutant form of Grk to be cleaved and to activate the Egfr after overexpression in the oocyte (data not shown). Next, we deleted the sequence encoding 19 amino acids (Y₂₂₄ to V₂₄₂) located between the EGF and TM domains that contains the putative dibasic cleavage site (mbGrkΔ19AA^myc, Fig. 5A). When overexpressed in the oocyte, mbGrkΔ19AA^myc does not activate the Egfr. By contrast, the eggs laid by these females have fused dorsal appendages (Fig. 5B). The fusion of the dorsal appendages reflects a weak ventralization of the eggshell, a phenotype associated with Egfr or grk loss-of-function mutations (Schüpbach, 1987; Price et al., 1989). Western blot analysis of the protein lysates from ovaries expressing mbGrkΔ19AA^myc in the oocyte reveals that only the non-processed form of Grk^myc can be detected (Fig. 5C). The lower band described above, which corresponds to a processed form of Grk^myc (Fig. 4E), was never detected. The abolition of this cleavage also results in absence of uptake of the extracellular domain of this truncated protein in follicle cells (Fig. 5D). Finally, to test whether the absence of processing is the reason why mbGrkΔ19AA^myc is not associated with signaling activity, we generated another form of secGrk in which the stop codon was introduced immediately after the last amino acid (A₂₂₃) of the EGF domain. Overexpression of secGrk(A₂₂₃) in somatic tissues does not lead to any phenotype (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Importance of the 19 amino acids located between the EGF and TM domains for Grk activity. (A) Structure of mbGrkΔ19AA^myc. (B) 7% (n=334) of the eggs laid by UASp-mbGrkΔ19AA^myc/nos-Gal4 females present a ventralized eggshell phenotype as shown by the fusion of the dorsal appendages. (C) Western blot probed with anti-Myc antibody reveals that, unlike mbGrk^myc, mbGrkΔ19AA^myc is not cleaved when overexpressed in the oocyte (compare with Fig. 4E). Lane 2 is a protein lysate from UASp-mbGrkΔ19AA^myc/nos-Gal4 ovaries. Only the higher band around 68 kDa that corresponds to the full-length mbGrkΔ19AA^myc, is detected (arrow). This band is absent in wild-type ovaries (lane 1). (D) Confocal section showing a reduction of the uptake of extracellular Grk in follicle cells after overexpression of UASp-mbGrkΔ19AA^myc in the oocyte (compare with Fig. 4C). Staining was carried out using an anti-Grk antibody.

Altogether, our results indicate that the putative dibasic signal is not the cleavage site. However, we show that the 19 amino acid sequence located between the EGF and TM domains is important for Grk activity and cleavage. Overexpression of mbGrkΔ19AA^myc in the oocyte weakly ventralizes the eggshell, presumably by acting as a dominant-negative molecule.

Star and Brho are involved in the activation of Grk during oogenesis

Activation of the Egfr by the transmembrane ligand Spi, requires two transmembrane proteins, Rho-1 and Star (S). These proteins have been proposed to promote presentation and/or processing of mbSpi to generate an active diffusible form of the ligand (secSpi) (Schweitzer et al., 1995a; Pickup and Banerjee, 1999; Guichard et al., 1999; Bang and Kintner, 2000). Although S and Rho-1 act together at many developmental stages (Mayer and Nüsslein-Volhard, 1988; Schweitzer et al., 1995a; Guichard et al., 1999), this is not the case during oogenesis. Indeed S seems to be expressed exclusively in the germline (Pickup and Banerjee, 1999), whereas rho-1 is expressed in the somatic follicle cells (Ruohola-Baker et al., 1993). However, a new rho-related gene called brother of rhomboid (brho) has been identified recently (Guichard et al., 2000). In contrast to rho-1, which is expressed in complex patterns during many stages of development (Bier et al., 1990; Ruohola-Baker et al., 1993), brho is expressed only in the early oocyte between stage 5 and 8 and in cells that abut the posterior follicle cells (Guichard et al., 2000). Thus, we were interested in examining the role of S and Rho family members in Grk activation.

As germline clones of S mutations do not develop beyond stage 1 of oogenesis (Nüsslein-Volhard et al., 1984) (this study, data not shown), and mutations in brho have not yet been identified (Guichard et al., 2000), we could not test directly the function of these two genes in the germline. Thus, we expressed antisense constructs of either UASp-brho or UASp-S in the germline. Interestingly, UASp-S antisense caused a ventralization of the eggshell, as shown by a complete fusion or disappearance of the dorsal appendages (Fig. 6A). In this assay, we did not obtain any significant phenotype after expressing a UASp-brho antisense construct (data not shown). To further examine the function of these genes in the germline, we overexpressed S, rho-1 or brho using UASp, but did not detect any phenotypes. In addition, we were also unable to detect any significant synergy after co-expressing mbGrk with each of these proteins, or after overexpressing mbGrk in flies heterozygous for either S or a deficiency of brho (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Synergy between mbGrk, Star and Rho-related proteins during oogenesis. (A) Reduction of S by expressing an antisense UASp-S in the germline causes the ventralization of the eggshell (88%, n=225). (B) Dorsal appendages are fused in S mutant follicle cell clones. (C) Misexpression of UAS-mbGrk in S/+ follicle cells results in a complete suppression of the mbGrk phenotype (compare with Fig. 1C). (D) Co-expression of UAS-S and UAS-mbGrk with CY2-Gal4 causes a strong eggshell dorsalization phenotype. (E) Ectopic expression of UAS-brho in the follicle cells using CY2-Gal4 causes a weak dorsalized eggshell phenotype. (F) Co-expression of UAS-mbGrk and UAS-brho in follicle cells induces a strong dorsalization of the eggshell. (G) Strong dorsalization of the eggshell after overexpression of mbSpi in the oocyte using pCOG-Gal4. (H) Ectopic expression of secSpi in the oocyte using the same driver causes a weaker dorsalized eggshell phenotype (33% of the eggs, n=167).

To further examine the function of S and Rho-1/Brho in Grk activation, we turned to the follicle cell epithelium, where we have shown that ectopic expression of either mbGrk or secGrk is able to activate the Egfr (Fig. 1). Although Grk is not expressed in follicle cells, the presence of the Egfr and all the downstream components of the pathway allows us to test the synergy between Grk, S and Brho in this tissue. In follicle cells, there is a low level of rho-1 (Ruohola-Baker et al., 1993) and no expression of brho (Guichard et al., 2000). Despite the recent evidence suggesting that S is not expressed in follicle cells (Pickup and Banerjee, 1999), we analyzed whether S is required for Egfr signaling in these cells. Interestingly, we found that S mutant follicle cell clones are associated with a weak ventralization of the eggshell, ranging from partially fused dorsal appendages to complete fusion (Fig. 6B). The phenotypes caused by loss of S in follicle cells resemble those caused by loss of rho-1 (Ruohola-Baker et al., 1993; Wasserman and Freeman, 1998), and reflect the requirement of S and rho-1 in the Spi-dependent activation of the Egfr.

To examine the relationship between S and Grk, we tested whether hyperactivation of the Egfr, after overexpression of UAS-mbGrk in follicle cells, requires S activity. We overexpressed UAS-mbGrk, using CY2-Gal4, in a S/+ heterozygous background. Reducing by half the dose of endogenous S leads to a complete suppression of the eggshell dorsalization phenotype (Fig. 6C; compare with Fig. 1C). A similar suppression was also observed in the wing (data not shown). Furthermore, we co-expressed S and mbGrk in follicle cells using CY2-Gal4. When this driver was used to overexpress S alone, no phenotype was observed (data not shown). However, co-expression of mbGrk and S caused a strong eggshell dorsalization phenotype (Fig. 6D), thus revealing a potent synergism between these two proteins, as has been previously seen in the wing (Guichard et al., 2000). These results indicate that, in the follicular epithelium, S can activate mbGrk.

In addition, we co-expressed mbGrk and Brho with CY2-Gal4 to test for synergy between these two proteins in the follicle cells. Whereas ectopic expression of UAS-brho causes only a weak dorsalized eggshell phenotype (Guichard et al., 2000) (Fig. 6E), co-expression with UAS-mbGrk leads to a strong dorsalization phenotype (Fig. 6F), thus revealing an interaction similar to what has been previously reported in the wing (Guichard et al., 2000). We also observed similar synergy after co-expressing mbGrk and Rho-1 in the follicle cells (data not shown). Finally, co-expression of mbGrk with S and Brho using CY2-Gal4 leads to lethality (data not shown).

In light of our results obtained in the follicle cells and to explain our observations in the germline, we propose that the oocyte contains high levels of both S and Brho. Additional support for this model was obtained from misexpression experiments with Spi. First, unlike secSpi, mbSpi has been shown to be able to trigger Egfr activation following overexpression in somatic tissues only when simultaneously expressed with S and/or Rho-1 (Pickup and Banerjee, 1999; Guichard et al., 1999). Second, we observed a similar synergy between mbSpi, S and Rho-1/Brho in follicle cells (data not shown). Third, overexpression of UASp-mbSpi alone in the germline is associated with a strong dorsalized phenotype (Fig. 6G), and the average UASp-secSpi misexpression phenotype is similar – although weaker – to the one induced by ectopic mbSpi in the oocyte (Fig. 6H).

Altogether, our results are consistent with the model that S and Brho, which are expressed highly in the germline, collaborate to activate mbGrk in the oocyte.

Grk cleavage in S2 cells

As S and Rho-1/Brho activate mbGrk when co-expressed in flies, we tested whether S and/or Rho-1/Brho are involved in Grk cleavage. To analyze this processing in detail, we used S2 cells that do not express either S or Rho-1/Brho (Schweitzer et al., 1995a; Guichard et al., 2000). S2 cells were transiently transfected with constructs encoding either secGrk or mbGrk, in absence or presence of S or Rho-1/Brho, or S and Rho-1/Brho. The mature Grk proteins of about 70 and 45 kDa could be detected in lysates from cells expressing mbGrk and secGrk, respectively (Fig. 7A). In addition, several secGrk intermediate forms ranging from 40 to 45 kDa, which probably represent glycosylation intermediates of the mature secGrk protein, were detected in cell extracts. Such intermediates have also been described for secSpi (Schweitzer et al., 1995a). Analysis of conditioned media from transfected cells reveals that mbGrk cannot be detected in the medium and that mbGrk is neither cleaved nor secreted in S2 cells. However, by contrast, significant amounts of only the mature form of secGrk can be detected in the conditioned medium (Fig. 7B).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

S and Rho-1/Brho in S2 cells. S2 cells were transiently transfected with actin5c (Ac5c) driving expression of the indicated constructs. 72 hours after transfection, cellular lysates and conditioned media were analyzed by western blot with antibody raised against Grk extracellular domain. mbGrk^myc was used to distinguish easily the membrane-bound Grk and cleaved Grk forms. (A) Cell lysates: secGrk proteins exhibited M_r of about 40-45 kDa (lane 2), while mbGrk^myc displayed Mr around 70 kDa (lane 3). Lysates from S2 cells expressing mbGrk^myc with Rho-1 or Brho^GFP (lanes 6 and 7); or mbGrk^myc with Rho-1 or (Brho^GFP) and S^myc (lanes 4 and 5) contained both the mbGrk^myc form and also a cleaved form of Grk that migrated with the secGrk species (lane 2). Lysates from cells expressing mbGrk^myc with S^myc contained only the larger mbGrk^myc species and not the cleaved form of Grk (lane 8). (B) Immunoprecipitates of conditioned media from the transfections described in A: no Grk was detected in media from cells expressing empty vector (lane 1), mbGrk^myc alone (lane 5), mbGrk^myc with Rho-1 or Brho^GFP (lanes 8 and 9, respectively), mbGrk^myc with S^myc (lane 10). The co-expression of mbGrk^myc with both S and Rho-1/Brho^GFP (lanes 6 and 7, respectively) leads to the accumulation of a soluble cleaved form of Grk in the medium that migrates with secGrk (lane 4). As control, secGrk and mbGrk^myc lysates were loaded on the same gel (lanes 2 and 3, respectively). We note that cleaved Grk released in the medium is slightly higher that the engineered secGrk. (C) S and Brho intracellular localization: S2 cells were transfected with Ac5c-Brho^GFP (C1-3) or Ac5c-S^myc (C4-6), both shown in green. The cells were also stained (all shown in red) with phalloidin to mark the plasma membrane (C1 and C4), anti-Drosophila Golgi mAb (C2 and C5) or anti-KDEL to mark the ER (C3 and C6). Brho is localized in discrete vesicles that colocalize with the Golgi apparatus (C2) and not the ER (C3). S is mostly expressed in a peri-plasma membrane pattern (C4). We note that some of S colocalizes with the ER (C6), however, none co-localizes with the Golgi (C5).

Analysis of the conditioned media from different co-transfection experiments support the model that Rho-1/Brho are required for the cleavage of Grk and that the cleaved form of Grk is detected in the conditioned medium only when S is also present in these cells. First, lysates prepared from cells that co-expressed mbGrk with Rho-1/Brho or with Rho-1/Brho and S, contained both mbGrk and a cleaved form that migrates around the mature secGrk species. Second, only the mbGrk form can be detected in lysates from cells expressing mbGrk and S (Fig. 7A). Third, in cells expressing Rho-1/Brho and mbGrk, Grk is cleaved. However, in the absence of S, this cleaved Grk is not secreted in the medium (Fig. 7B).

To determine the subcellular localization of Brho and S, we transiently transfected S2 cells with constructs containing either Brho or S. We also stained these cells with phalloidin, to label filamentous actin and hence the plasma membrane, with an antibody specific for the Drosophila Golgi, and with a pan-ER antibody. Brho is expressed discretely in cytoplasmic vesicles (Fig. 7C1) that co-localize with the Golgi (Fig. 7C2) and not the ER (Fig. 7C3). S is expressed in a diffuse pattern in the cytoplasm, primarily in a peri-plasma membrane pattern (Fig. 7C4). Some of S colocalizes with the ER (Fig. 7C6) but none with the Golgi (Fig. 7C5).

Altogether, our results suggest that Rho-1/Brho are sufficient to catalyze Grk cleavage, and that S is involved in a trafficking/secretion process. The intracellular localization of S is also consistent with a role for S at a step that follows the Brho-dependent cleavage, as S is predominantly very close to the plasma membrane, while Brho localizes to the Golgi.

Grk processing

Our in vivo and in vitro results clearly indicate that Grk is cleaved in the germline (see Fig. 4). An important question is where exactly the cleavage of the Grk precursor occurs? Bang and Kintner (Bang and Kintner, 2000) concluded that the cleavage of Spi occurs in the TM and depends on the 15 amino acid stretch located between the EGF and TM domains. Our results show that the Grk dibasic signal (R₂₄₀ and K₂₄₁) is not the cleavage site because its mutation does not abolish this event. However, mbGrkΔ19AA^myc, in which the 19 amino acid (Y₂₂₄ to V₂₄₂) located between the EGF and TM domains have been deleted, is no longer cleaved, suggesting that this sequence is directly or indirectly involved in the processing.

Our results do not rule out the hypothesis that Grk cleavage occurs in the TM domain as proposed for Spi (Bang and Kintner, 2000). The high conservation between the Spi and Grk TM domains (Neuman-Silberberg and Schüpbach, 1993), in addition to aberrant Grk localization observed with different grk alleles affecting this TM domain (Queenan et al., 1999), reveal its importance. Moreover, the cleaved product of Grk that is released in the medium, after co-expressing mbGrk+S+Rho-1/Brho in S2 cells, has a slightly higher mobility that the engineered secGrk (Fig. 7B). Thus, it is possible that mbGrk is cleaved within the TM domain and that proteolysis depends on the 19 amino acid interval.

Our results reflect the importance of the Grk TM domain for proper processing and routing through the secretory pathway. mbGrk processing is probably tightly regulated and leads to efficient Grk secretion, contrary to engineered secGrk that is poorly secreted from the oocyte and that acts mainly intracellularly (Queenan et al., 1999) (this study).

Brho and S act in the germline to promote secretion of active Grk

The recent findings that S and Brho, a Rho-related protein, are expressed in the oocyte led us to investigate whether they are involved in Grk activation during oogenesis. S and Rho proteins have been proposed previously to be involved in the processing and activation of Spi (Klämbt, 2000); however, because they have no obvious motifs that predict their biochemical functions, their roles in ligand maturation and/or secretion has remained obscure.

The analysis of these proteins in the context of Grk signaling has provided numerous insights into the relationships between these transmembrane proteins. Our in vivo data strongly suggest that the expression level of S and Brho is very high in the oocyte, thus leading to an efficient cleavage and secretion of Grk. However, S and Rho-1 are probably expressed at low level in the follicle cells. Indeed, Pickup and Banerjee (Pickup and Banerjee, 1999) were unable to detect the presence of S in this epithelium using an anti-S antibody, whereas they clearly showed a strong staining in the germline. The presence of both endogenous S and Rho-1 in follicle cells explains why overexpression of mbGrk in this epithelium leads to a weak dorsalization of the eggs. Nevertheless removing one copy of S is sufficient to completely suppress this phenotype. This confirms the observation that overexpression of mbGrk on its own is not able to activate the Egfr in vivo (Fig. 6C), as supported by our in vitro study (Fig. 3D). Overexpression experiments in follicle cells indicate a strong synergy between mbGrk, S, and Brho, as previously observed for Spi (Guichard et al., 1999; Guichard et al., 2000; Bang and Kintner, 2000) (Fig. 6). Further, co-expression of S and Rho-1/Brho is sufficient for Grk cleavage and secretion in S2 cells, strongly suggesting that they are the only proteins required for this process. In addition, these tissue culture experiments reveal that S and Rho-1/Brho are not obligate cofactors for this cleavage, because co-expression of mbGrk with Rho-1/Brho is sufficient to catalyze this proteolytic event (Fig. 7A). S is not required for Rho-1/Brho-mediated proteolytic cleavage in S2 cells, but the soluble Grk extracellular domain is no longer detected in the medium from these cells, indicating that the function of S is necessary for trafficking/secretion of the ligand. However, S is not able to cleave Grk in absence of Rho-1/Brho (Fig. 7A,B). Altogether, our results show that the functions of Rho-1/Brho and S are distinct, which explain their co-dependence and synergism in vivo (Guichard et al., 1999; Guichard et al., 2000).

How do Rho-1/Brho promote cleavage of mbGrk?

Rho-1/Brho may facilitate Grk proteolysis either by activating or recruiting a yet unknown protease. By analogy to the processing of mammalian Egfr ligands, Grk cleavage may be catalyzed by an ADAM-like metalloprotease (Black and White, 1998). Although these molecules are present in Drosophila (Wasserman et al., 2000), nothing is known yet about their functions. An alternative hypothesis, is that despite the absence of known protease domains in their sequences, Rho-1 and Brho themselves may have proteolytic activity. The subcellular localization of Brho, as observed for mature TACE (ADAM17) (Schlöndorff et al., 2000), is predominantly in intracellular compartments (Fig. 7C1-3). In addition, and directly relevant to this hypothesis, Presenilins, which define another subfamily of seven-pass transmembrane proteins, have been proposed to encode proteases (Wolfe et al., 1999). In Drosophila, Presenilin may be directly responsible for the proteolysis of the intra-transmembrane domain of Notch (Struhl and Greenwald, 1999; Ye et al., 1999).

One of the striking feature of Rho-related proteins is that amino acid sequence conservation is most prominent in the predicted TM regions which contain some invariant charged residues (Guichard et al., 2000; Wasserman et al., 2000). This suggests the presence of a hydrophilic pocket that might constitute an enzymatic active site or a channel, as observed in Presenilins (Wolfe et al., 1999). This model is further supported by the recent finding that the TM domain of Spi is important for its functional interaction with Rho-1 (Bang and Kintner, 2000).

rho-related genes have been found in organisms from diverse kingdoms including C. elegans, rat, human, Arabidopsis, sugar cane, yeast and bacteria. Our data suggest that Brho, like Rho-1, promotes Egfr signaling by activating TGFα-like ligands. As RTKs have not been found in plants (Satterlee and Sussman, 1998), yeast or bacteria, the rho-related genes in these organisms presumably serve other functions. It will be interesting to determine whether the activities of these Rho-related proteins are similar to those of Rho-1 and Brho, such as promoting the processing of proteins.

A possible role for S

Mosaic analysis of S, both in the germline (Nüsslein-Volhard et al., 1984) and in follicle cells (Fig. 6B), together with the S antisense experiment (Fig. 6A), demonstrate that S is required in follicle cells for Spi-dependent Egfr activation, and in the germline for Grk-dependent Egfr activation. Our tissue culture experiments suggest that S is not involved in Grk proteolysis, but instead in post-cleavage trafficking or secretion of the ligand (Fig. 7B). The intracellular localization of S is also consistent with a role for S at a step that follows the Brho-dependent cleavage, because we find that S is predominantly very close to, or at the plasma membrane (Fig. 7C4-C6), while Brho localizes to the Golgi (Fig. 7C2). The role of S, however, is not yet resolved because our results contrast with the ER localization of S in the oocyte described by Pickup and Banerjee (Pickup and Banerjee, 1999). Interestingly, unlike Rho-1 and Brho, S is probably involved in other processes as well. For example, S has been identified as a suppressor of Delta (Klein and Campos-Ortega, 1992), one of the Notch ligand. Delta encodes a transmembrane protein that is cleaved by the Kuzbanian metalloprotease, and the extracellular fragment antagonizes the function of the membrane-bound Delta protein as an activating Notch ligand (Klueg et al., 1998; Qi et al., 1999). In the case of Notch signaling, a reduction of S gene activity might lead to a reduced release of the extracellular Delta fragment, and thus enhance Delta signaling (Klämbt, 2000).

Finally, understanding the function of S and Rho-1/Brho in Grk processing is relevant to studies of the mammalian ligands of the EGFR family as well, because TGFα may also be processed in vivo before receptor binding (Peschon et al., 1998). Thus, although further work is needed to fully understand the biochemical function of S, and Rho-1/Brho, our studies have provided a number of insights into the mechanism of action of these molecules.