The Interaction between the Drosophila Secreted Protein Argos and the Epidermal Growth Factor Receptor Inhibits Dimerization of the Receptor and Binding of Secreted Spitz to the Receptor

Abstract

Drosophila Argos (Aos), a secreted protein with an epidermal growth factor (EGF)-like domain, has been shown to inhibit the activation of the Drosophila EGF receptor (DER). However, it has not been determined whether Aos binds directly to DER or whether regulation of the DER activation occurs through some other mechanism. Using DER-expressing cells (DER/S2) and a recombinant DER extracellular domain-Fc fusion protein (DER-Fc), we have shown that Aos binds directly to the extracellular domain of DER with its carboxyl-terminal region, including the EGF-like domain. Furthermore, Aos can block the binding of secreted Spitz (sSpi), a transforming growth factor α-like ligand of DER, to the extracellular domain of DER. We observed that sSpi stimulates the dimerization of both the soluble DER extracellular domain (sDER) and the intact DER in the DER/S2 cells and that Aos can block the sSpi-induced dimerization of both sDER and intact DER. Moreover, we have shown that, by directly interacting with DER, Aos and SpiAos (a chimeric protein that is composed of the N-terminal region of Spi and the C-terminal region of Aos) inhibit the dimerization and phosphorylation of DER that are induced by DER's overexpression in the absence of sSpi. These results indicate that Aos exerts its inhibitory function through dual molecular mechanisms: by blocking both the receptor dimerization and the binding of activating ligand to the receptor. This is the first description of this novel inhibitory mechanism for receptor tyrosine kinases.

The epidermal growth factor (EGF) receptor (EGFR) is a member of the ErbB family of receptor tyrosine kinases (RTKs), which are composed of an extracellular domain, a transmembrane region, and a cytoplasmic domain, which includes a tyrosine kinase domain (5, 20) (see Fig. 1A). The binding of EGF to its receptor induces conformational changes in the extracellular domain (18), resulting in rapid dimerization of the receptor (3, 8, 25). In its dimerized state, the activated tyrosine kinase phosphorylates tyrosine in the carboxyl-terminal region of the adjacent receptor through an intermolecular mechanism (23, 29, 57).

FIG. 1.

(A) Schematic representation of the domain structures of native and artificially constructed EGFR proteins. The extracellular domain of hEGFR is divided into four subdomains (I, II, III, and IV). The most striking difference between DER and hEGFR is the insertion of a cysteine-rich subdomain (16 Cys) between the second cysteine-rich (20 Cys) subdomain and the TM domain (solid box) of DER (49). The signal peptide is shown by diagonal lines. The His tag (His) and Fc portion of human IgG1 (Fc) are marked. (B) Schematic representation of the domain structure of native and mutant ligands of DER. Aos possesses an EGF-like domain that differs from that of sSpi in that Aos contains an extended B-loop. AosEGF is the C-terminal region, including the EGF-like domain, of Aos. AosEGF-Fc is a fusion protein composed of the C-terminal region of Aos and the Fc region of human IgG1. A chimeric protein, SpiAos was constructed from sSpi and Aos. A Myc tag was added to the C terminus of Aos and SpiAos, and sSpi was tagged with the Flag epitope. (C) Analysis of the monomeric sDER and dimeric DER-Fc proteins by Western blotting. Baculovirus-expressed sDER, DER-Fc, and control medium were separated on an SDS-PAGE gel (8% polyacrylamide) under nonreducing or reducing conditions and probed with mouse anti-sDER antibody. The molecular mass of DER-Fc under the nonreducing condition appeared to be about two times greater than that under the reducing condition. The molecular mass markers (kilodaltons) are shown to the left.

Like its vertebrate homologues, the Drosophila EGFR (DER) mediates various inductive signaling events in several tissues to regulate normal development (1, 42, 50, 55). DER signaling functions principally through the Ras/mitogen-activated protein kinase (MAPK) signal transduction pathway, which is highly conserved between Drosophila and mammals (14, 40). The loss-of-function mutant phenotypes of DER indicate that DER regulates a variety of developmental processes, including the survival of embryonic ectodermal tissues, the proliferation of imaginal discs, the morphogenesis of several adult ectodermal structures, and neural differentiation (7, 55). Since DER signaling is involved in many different aspects of development, like other members of the ErbB family, its activation must be controlled precisely. Evidence from genetic and biochemical analyses indicates that both activating and inhibitory ligands regulate DER signaling (40, 64).

So far, three activating ligands (Vein, Gurken, and Spitz [Spi]) of DER, each of which possesses a predicated EGF-like domain, have been identified in Drosophila. Vein resembles the mammalian neuregulins, which commonly possess an immunoglobulin (Ig)-like domain in addition to the EGF-like domain (51). The vein mutations show strong genetic interactions with mutations of the gene encoding DER (51). Vein is required for cell proliferation during embryogenesis and for cell fate determination in the embryo and wing (51, 56, 67). Gurken, a transforming growth factor α (TGF-α)-like protein, has been implicated as a DER ligand (35). The gurken gene is maternally active and is expressed in the oocyte, where it signals the somatic follicle cells to establish both the anterior-posterior and the dorsal-ventral axes (17, 36). Another activating ligand for DER is Spi, which is also a TGF-α homolog (43). Spi is a well-characterized DER ligand and appears to cause most of the activation of the receptor in situ. It is expressed widely during development and has been shown to be involved in the developmental processes of the embryo, eye, and wing that are similar to those regulated by DER (12, 43). Biochemical analysis in vitro also showed that Spi activates DER signaling. The addition of secreted Spi (sSpi), but not the membrane-associated form, to cultured cells expressing DER gives rise to a rapid autophosphorylation of DER on tyrosine (53).

Recently, it has become clear that a variety of signal transduction pathways, including the bone morphogenetic protein (BMP) (10, 22, 41, 69) and Wnt (32, 63) pathways, are controlled by both positive and negative extracellular regulators. Such sophisticated regulatory mechanisms enable precise spatiotemporal control of receptor activation. In Drosophila, EGFR signaling is also regulated by the inhibitory diffusible protein Argos (Aos), which inhibits the tyrosine autophosphorylation of DER in a tissue culture assay (54). Aos is the first reported extracellular factor shown to inhibit an RTK (54). Genetic interactions between Aos and members of the DER signaling pathway have indicated that Aos functions as an inhibitor of the DER signaling pathway to repress cell fate determination during eye, wing, and chordotonal organ development (37, 38, 45–47, 54). Aos overexpression induces programmed cell death in the developing eye by inhibiting the DER/Ras pathway (48). Aos is a secreted protein (13), and its expression appears to be triggered directly by the DER pathway (16). In a tissue culture assay, Aos can shut off DER signaling that has been activated by sSpi (16). Thus, Aos may form an inhibitory feedback loop (16) to restrict the duration and level of DER signaling. Aos possesses an EGF-like domain (11, 28, 39), which differs from that of Spi and other EGFR agonists in that it contains an extended B-loop consisting of 20 amino acids, instead of the 10 amino acids seen in the agonists. It is important to elucidate the structure-function relationship of Aos, because it provides an opportunity to determine the molecular basis for the distinct properties of DER regulators. Thus, we expect that the Drosophila Aos pathway may provide an excellent model system for developing an inhibitory factor for the human ErbB receptor family.

The precise mechanisms by which Aos inhibits the DER signaling pathway are not yet clear. Two attractive models for Aos action have been proposed (54, 64). One is that Aos directly acts on DER and inhibits the receptor activation; the other is that Aos interacts with another unknown receptor, which in turn inhibits DER activation. However, neither model has been tested by biochemical and functional analyses. To show whether Aos can function according to the former model, it is important to determine whether Aos binds directly to DER and how it blocks DER activation. Here, we show that Aos binds directly to the DER extracellular domain and that it can inhibit the DER signaling through dual inhibitory mechanisms: Aos not only blocks the binding of sSpi to the receptor, but also suppresses the dimerization of the receptor.

MATERIALS AND METHODS

Cell lines.

The Drosophila S2 cell line and the S2-derived transgenic cell line were grown at 27°C in Schneider's medium supplemented with 10% fetal bovine serum, peptone (5 mg/ml) (Difco), and antibiotics. Since activation of the type II DER (Fig. 1A) by sSpi appeared to be much more pronounced than activation of the type I DER (53, 54), we established a type II DER-expressing S2 cell line, termed DER/S2, by cotransfection of pRmHa-DER II and pV8 plasmids into the S2 cells, followed by selection in 1 mg of Geneticin per ml. To induce the expression of DER, the DER/S2 cells were incubated in medium containing CuSO₄ as described previously (4, 53). The S2 and DER/S2 cells did not express DER, Spi, or Vein endogenously (52–54). Reverse transcription-PCR analysis indicated that gurken was not expressed in these cells either (data not shown).

Plasmids.

pBacMel, a baculovirus transfer vector, was generated by insertion of a PCR product containing the baculovirus polyhedrin promoter and the secretion signal of honeybee melittin (60) into the BglII and EcoRI sites of pBAC-1 (Novagen). All transfer vectors for expression of the recombinant proteins were generated by standard methods with pBacMel (44). The restriction enzyme sites necessary for subcloning were created by PCR-based mutagenesis. All of the PCR products were verified by DNA sequence analysis. pBacMel-Aos is a transfer vector for the expression of Aos (corresponding to Leu 28 to Asp 444 of Aos) with a Myc-His tag at the carboxyl terminus of Aos (Fig. 1B). pBacMel-sSpi expresses the secreted form of Spi (Arg 28 to Lys 129) with a Flag-His tag at the carboxyl terminus of sSpi (Fig. 1B). To express the AosEGF-Fc fusion protein, the C-terminal region, including the EGF-like domain, of Aos (Pro 362 to Asp 444) and the Fc portion of human IgG1 from pJFE14/B61-Fc (9) was inserted in frame into pBacMel (Fig. 1B). pBacMel-SpiAos expresses a chimeric protein that is composed of the N-terminal portion of Spi (Arg 28 to Lys 77) and the C-terminal region of Aos (Asn 346 to Asp 444). SpiAos was tagged with a Myc-His epitope at the carboxyl terminus (Fig. 1B). To express sDER (Fig. 1A), the DER extracellular domain (Gly 43 to Ile 812 of DER), pBacMel-sDER was generated. The Fc portion of human IgG1 was inserted into the XhoI site of pBacMel-sDER to generate pBacMel-DER-Fc, a construct for the expression of the DER-Fc fusion protein (Fig. 1A). For expression in Escherichia coli cells, pGEX-5x-3 (Pharmacia Biotech) and pRSET-B (Invitrogen) were used. A PCR product of DER cDNA (corresponding to a fragment extending from the EcoRI site to the stop codon of the DER cDNA) was subcloned into pGEX-5x-3 to construct pGEX-DERc2. A BamHI-XhoI fragment of pBacMel-sDER was inserted into pRSET-B to generate pRSET-sDER1.3, encoding Asp 404 to Ile 812 of the DER extracellular domain. To express AosEGF (Fig. 1B), an EcoRV fragment of Aos cDNA was subcloned into the PvuII site of pRSET-b.

Production and purification of recombinant proteins.

All secreted recombinant proteins (Aos, sSpi, SpiAos, sDER, and DER-Fc) were generated by the baculovirus expression system. The baculovirus transfer vectors were cotransfected into Sf9 cells together with the BacVector-2000 triple-cut virus DNA (Novagen) by using Cellfectin (GIBCO BRL) to generate recombinant baculoviruses. The recombinant baculovirus clones were isolated by plaque purification, and their high-titer stocks (10⁸ to 10⁹ PFU/ml) were obtained with Sf9 cells. High Five cells (a gift from M. Amagai) were infected with the virus stock to produce recombinant proteins, and serum-free medium (JRH Biosciences) was used at 2 days postinfection for several experiments. The recombinant proteins were purified from the conditioned serum-free media by affinity chromatography with ProBond resin (Invitrogen) or protein A-Sepharose 4 Fast Flow (Pharmacia Biotech). Coomassie brilliant blue staining and Western blotting (Fig. 1C) confirmed their purity and specificity.

Antibodies.

The recombinant proteins expressed in E. coli BL21 pLysS cells (Stratagene) transformed with the pGEX-DERc2 or pRSET-sDER1.3 plasmids were purified with glutathione-Sepharose 4B (Pharmacia Biotech) or ProBond resin (Invitrogen) and injected into rats and mice to generate polyclonal antibodies. The resultant antibodies were a rat antibody against the DER cytoplasmic domain (rat anti-DERc) and a mouse antibody against the DER extracellular domain (mouse anti-sDER). These antibodies were preincubated with the S2 cells before use. The mouse monoclonal antibody to Aos was obtained from the Developmental Studies Hybridoma Bank.

DER signaling assay.

The DER activation assay was performed essentially according to the method of Schweitzer et al. (53, 54). The tyrosine phosphorylation of DER was detected by an antiphosphotyrosine antibody (PY20) (Transduction Laboratories) to show the level of DER activation. To show whether SpiAos and Aos could inhibit the ligand-independent activation of DER, the DER/S2 cells were incubated with SpiAos- or Aos-conditioned medium for 15 min in the absence of sSpi. The MAPK (extracellular signal-regulated kinase [ERK]) activation assay was performed according to previously reported procedures (14, 52). The cell lysates were run on duplicate gels and blotted. One blot was probed with anti-dually phosphorylated ERK antibody (Promega) to show the level of ERK activation, and one blot was probed with anti-ERK antibody (2) to show the total amounts of ERK protein loaded. The intensities of the signals of phosphorylated DER and ERK were normalized to those of the DER or ERK protein, respectively, as determined by densitometric analysis.

Detection of monomeric and dimeric sDER.

Cross-linking of the dimerized soluble receptor was performed essentially as reported previously (25). Briefly, the sDER protein (20 nM) and sSpi (100 nM) were incubated with or without Aos (40 nM) or SpiAos (80 nM) for 1 h in a mixture of 20 mM HEPES (pH 7.3), 150 mM NaCl, and 0.02% bovine serum albumin (BSA) and subsequently with 1 mM disuccinimidyl suberate (DSS) (Pierce) for 30 min. The cross-linking reaction was terminated by adding Laemmli's sample buffer (62.5 mM Tris-HCl [pH 6.8], 2% sodium dodecyl sulfate [SDS], 10% glycerol, and 0.002% bromophenol blue). The monomeric and dimeric sDERs were visualized by Western blot analysis with mouse anti-sDER antibody.

Cross-linking of the DER dimer at the cell surface of living cells.

The DER/S2 cells were preincubated with Aos (40 nM) for 2 min. Subsequently the sSpi (25 nM) was added for 10 min. To show whether Aos can inhibit the ligand-independent dimerization of DER, the DER/S2 cells were incubated with Aos (40 nM) for 15 min in the absence of sSpi. After removal of the ligands, the cells were incubated with the membrane nonpermeable cross-linker bis (sulfosuccinimidyl) suberate (BS³) (Pierce), which was dissolved to a final concentration of 1 mM in pH 7.3 buffer (20 mM HEPES, 150 mM NaCl, 50 nM okadaic acid, 1 mM Na₃ VO₄, 5 mM NaF), for 30 min. The cells were subsequently lysed in ice-cold radioimmunoprecipitation assay (RIPA) buffer (53). After immunoprecipitation with a rat anti-DERc antibody, the proteins were separated on an SDS-polyacrylamide gel electrophoresis (PAGE) (4% polyacrylamide) gel. The monomeric and dimeric DERs were detected by Western blot analysis with the mouse anti-sDER antibody. The amount of tyrosine phosphorylation of monomeric and dimeric DER was detected with PY20.

Binding experiments.

For the binding assay with living cells, the DER/S2 cells were treated with 0.7 mM CuSO₄ for 6 h and incubated with Aos (100 nM), SpiAos (200 nM), or sSpi (250 nM) in pH 7.3 binding buffer (20 mM HEPES, 150 mM NaCl, 0.02% BSA) for 15 min. After the buffer was removed, the cells were incubated for 30 min with 1 mM 3,3′-dithiobis (sulfosuccinimidylpropionate) (DTSSP) (Pierce) to stabilize the binding of ligands with the receptor, washed with the binding buffer, and subsequently lysed in 200 μl of RIPA buffer for 20 min on ice. DER was immunoprecipitated with the rat anti-DERc antibody, and the ligands in the immunoprecipitates were detected by Western blot analysis with an antibody to Myc or to the Flag tag. For the blocking assay of sSpi binding, the sSpi (25 nM) and DER-Fc (5 nM) were incubated with various concentrations of Aos (10 to 40 nM) in binding buffer for 15 min and subsequently with 1 mM DTSSP for 30 min. The cross-linking reaction was quenched by the addition of RIPA buffer containing 50 mM glycine. The ligand was immunoprecipitated directly with protein A-Sepharose 4 beads (Pharmacia Biotech) and was detected with an anti-Flag monoclonal antibody.

Western blotting.

The protein samples mixed with Laemmli's sample buffer with or without β-mercaptoethanol were separated by SDS-PAGE and transferred onto an Immobilon-P membrane (Millipore). The proteins on the membrane were detected by incubation with primary antibody overnight at 4°C, followed by incubation with horseradish peroxidase-conjugated secondary antibody for 1 h and then visualized by using the ECL (enhanced chemiluminescence) system (Amersham). A quantitative analysis of the immunoreactive bands was done by computerized densitometry with a Scanning Imager with ImageQuant software (Molecular Dynamics).

BIAcore measurements.

Real-time analysis of the interaction between ligands and receptor was performed on a BIAcore 2000 instrument (BIAcore AB) at 25°C. The purified Aos, sSpi, SpiAos, or DER-Fc was immobilized on the CM5 sensor chip flow cells 2 to 4 by using the amine coupling reagent (BIAcore AB) as previously described (68). The surface plasmon resonance signal from immobilized Aos, sSpi, SpiAos, or DER-Fc generated 5,292, 1,638, 1,679, and 5,319 response units, respectively. All of the kinetic measurements were performed in HBS-EP buffer (10 mM HEPES [pH 7.4], 3 mM EDTA, 150 mM NaCl, 0.005% polysorbate 20) at a flow rate of 20 μl/min. Five serial dilutions of analytes (ranging from 125 to 2,000 nM) were injected for 3 min followed by dissociation in buffer flow for 4 min. After the dissociation phase, the sensor chips were regenerated with a pulse of 10 mM HCl for 1 min at 20 μl/min. Background nonspecific binding that resulted from the binding of proteins to the dextran matrix of the sensor chip flow cell 1 was subtracted from each curve before analysis. Data analysis and calculation of kinetic constants from the sensorgrams were performed with the BIAcore software.

RESULTS

Generation of monomeric and dimeric DER extracellular domains.

To examine the direct interaction between Aos and DER, we generated a DER extracellular domain (sDER) and a fusion protein of the DER extracellular domain and human IgG1 Fc (DER-Fc) by using a baculovirus expression system (Fig. 1A and C). When sDER and DER-Fc were analyzed by Western blotting, bands of approximately 115 and 130 kDa, respectively, were observed under reducing conditions. Under nonreducing conditions, bands of approximately 115 and 260 kDa were apparent. These results indicate that sDER exists as a monomer, but DER-Fc is likely to form dimers (Fig. 1C). We also generated the wild-type and mutant Aos and sSpi proteins (Fig. 1B).

To analyze the inhibitory mechanisms of Aos upon DER function, we established a DER-expressing Drosophila Schneider S2 cell line, DER/S2, by stable transfection of an inducible DER expression vector (pRmHa-DER II) into S2 cells. Immunohistochemistry using the rat anti-DERc antibody demonstrated the presence of the DER protein on the cell surface of DER/S2 cells (data not shown).

Aos binds directly to the extracellular domain of DER.

As a first step toward understanding the functional mechanism by which Aos inhibits DER activation, we examined whether Aos binds DER directly. The DER/S2 cells were incubated with baculovirus-generated Aos, sSpi, or SpiAos, a chimeric protein consisting of the N-terminal portion of Spi and the C-terminal region of Aos, including the EGF-like domain (Fig. 1B). To isolate the ligand-DER complexes formed at the cell surface, the cells were treated with 1 mM DTSSP, a membrane-nonpermeable cross-linker. The ligand-DER complexes were immunoprecipitated from the cell lysate with a rat anti-DER antibody. Aos, SpiAos, and sSpi in the immunoprecipitates were detected by Western blotting with anti-Myc or anti-Flag antibodies. This result indicates that Aos, SpiAos, and sSpi bound to DER at the cell surface (Fig. 2A).

FIG. 2.

Aos, sSpi, and SpiAos bind to the DER extracellular domain. (A) DER/S2 cells incubated with Aos (lane 1), SpiAos (lane 2), or sSpi (lane 3) were treated with the cross-linker DTSSP, followed by immunoprecipitation (IP) by rat anti-DERc antibody. The cross-linker can be cleaved with 5% β-mercaptoethanol in Laemmli's sample buffer at 100°C for 5 min. Aos, SpiAos, and sSpi in the immunoprecipitates were detected by Western blotting analysis with anti-Myc or anti-Flag monoclonal antibodies. (B) Aos (10 nM) incubated with 0.01% BSA (lane 1) or 5 nM DER-Fc (lane 2), and subsequently with DTSSP, was precipitated by protein A and then probed with the anti-Aos monoclonal antibody. (C) sDER (20 nM), which had been incubated with 20 to 30 nM AosEGF-Fc (lane 1) or IgG1 Fc (lane 2) in the absence of the cross-linker, was precipitated with protein A beads and probed with mouse anti-sDER antibody. AosEGF-Fc and IgG1 Fc in the precipitates were detected by anti-Fc antibody. Experiments were performed twice with similar results.

To determine whether Aos binds DER directly, a binding assay was performed with the baculovirus-generated DER-Fc recombinant protein (Fig. 1A). Aos incubated with DER-Fc, BSA, or IgG1 Fc was precipitated with protein A beads and detected with a monoclonal antibody that recognizes Aos. Aos specifically bound to the soluble DER-Fc protein in the absence of the S2 cell membrane (Fig. 2B). This result shows that Aos directly interacts with the extracellular domain of DER.

A previous study suggested that the EGF-like domain of Aos was essential for its function (52). To examine the binding of the Aos C-terminal region, including the EGF-like domain, to DER, we performed a coprecipitation assay using AosEGF-Fc, sDER, and protein A beads. As shown in Fig. 2C, sDER was specifically coprecipitated with AosEGF-Fc, even in the absence of the cross-linker, but not with IgG1 Fc under the same conditions. Thus, the C terminus of Aos is likely to be essential for the interaction between Aos and the extracellular domain of DER.

We also determined the kinetics of the interaction between ligands and DER by using the BIAcore system. Figure 3 shows the real-time measurements of the association and dissociation phases of the interaction of DER-Fc with immobilized Aos (Fig. 3A) or sSpi (Fig. 3B), performed at several different concentrations of the receptor. The sSpi and Aos immobilized on the sensor chip displayed similar rapid association and dissociation rates to DER-Fc. The human IgG1 Fc portion, used by itself as a control, did not interact with the immobilized Aos and sSpi (data not shown). DER-Fc bound to the immobilized sSpi with a K_d of 67 ± 33 nM, which is similar to the affinity of human EGF (hEGF) for hEGFR (38 to 62 nM) (68). The real-time interaction of AosEGF with immobilized DER-Fc is shown in Fig. 3C. The calculated affinities of DER-Fc for Aos and AosEGF were 41 ± 18 and 17 ± 6 nM, respectively.

FIG. 3.

Kinetic analysis of the association and dissociation rates of the interactions of Aos, sSpi, and AosEGF with DER-Fc. (A and B) Sensorgrams showing the binding of DER-Fc in response units to Aos (A) or sSpi (B) immobilized on the sensor chip surface. For each sensorgram, buffer was pumped over the sensor chip at 20 μl/min. At 120 s, the buffer was replaced by solution containing the indicated concentrations of DER-Fc for 180 s. The increase in response reflects the binding of DER-Fc to the immobilized ligands (the association phase). At 300 s, the DER-Fc solution was replaced with buffer. The decay in response represents the dissociation of bound DER-Fc (the dissociation phase). (C) Sensorgram showing the association and dissociation of AosEGF with DER-Fc immobilized on the sensor chip surface. Injected AosEGF bound to DER-Fc in a concentration-dependent manner. Background nonspecific binding derived from the control sensor chip (no ligands) was subtracted from each curve. Experiments were performed three times with similar results.

Aos inhibits sSpi binding to DER.

The finding that both Aos and sSpi bind to the DER extracellular domain prompted us to examine the hypothesis that Aos may compete with sSpi to bind to the receptor, because both Aos and Spi possess an EGF-like motif that is essential for their function (52, 64). Using the DER-Fc fusion protein, we examined the effect of Aos on the binding of Spi to DER. DER-Fc (5 nM) and sSpi (25 nM) were incubated with increasing concentrations of Aos (10 to 40 nM). The immunoprecipitated sSpi was detected by Western blotting. In the absence of Aos, sSpi was shown to efficiently bind to DER-Fc (left lane in Fig. 4A); however, the amount of sSpi bound to DER-Fc markedly decreased as the amount of Aos was increased (Fig. 4A). The inhibitory effect of Aos on the sSpi binding was specific to Aos, because the amount of sSpi bound to DER-Fc was not decreased by BSA (data not shown).

FIG. 4.

Aos blocks sSpi binding to DER. (A) The DER-Fc and sSpi-Flag incubated with various concentrations of Aos followed by DTSSP treatment were precipitated with protein A and then probed with the anti-Flag antibody to detect sSpi bound to DER-Fc. The bottom panel shows the densitometric analysis giving the relative intensity of the sSpi bound to DER-Fc. The intensity of the Flag-positive band in the absence of Aos was defined as 100% of the relative sSpi-binding to DER indicated in the figure. The amounts of sSpi bound to DER-Fc decreased to 37 and 6% in the presence of 20 and 40 nM Aos, respectively. (B) AosEGF-Fc and sSpi-Flag were incubated in the presence (lane 1) or absence (lane 2) of sDER and then immunoprecipitated (IP) by protein A. The precipitates were detected by Western blotting with the anti-Flag (top panel), anti-Fc (middle panel), or anti-sDER (bottom panel) antibodies. The medium of Sf9 cells infected with baculovirus expressing sSpi-Flag was loaded as a positive control for the Western blotting (lane 3, top panel). Experiments were repeated two times with similar results.

We further examined whether Aos and sSpi can bind to DER simultaneously. AosEGF-Fc and sSpi were incubated in the presence or absence of sDER and then precipitated by protein A beads. sSpi and sDER in the precipitates were probed with the anti-Flag and anti-sDER antibodies, respectively, to examine simultaneous binding of AosEGF-Fc and sSpi to sDER (Fig. 4B). sDER was coprecipitated with AosEGF-Fc, but sSpi was not detectable in the precipitates containing complexes of sDER and AosEGF-Fc (Fig. 4B), suggesting that the C-terminal regions of Aos and sSpi cannot bind to DER simultaneously. One possible explanation for these data is that Aos and sSpi bind to the same site on DER, although the possibility of allosteric inhibition cannot be excluded. These results suggest that the interaction between the EGF-like domain of Aos and the extracellular domain of DER inhibited the binding of sSpi to the receptor.

Aos inhibits the dimerization of the DER extracellular domain.

Although the dimerization of the hEGFR extracellular domain has been shown to be induced by EGF using a covalent chemical cross-linking agent (25, 30), the dimerization of DER has not been studied. We first investigated whether sSpi could induce the dimerization of sDER, a monomeric sDER extracellular domain (Fig. 1A and C). The sDER protein was incubated with or without sSpi and subsequently with 1 mM the cross-linker DSS. Monomeric and dimeric sDERs were detected by Western blotting analysis with mouse anti-sDER antibody (Fig. 5). In the absence of sSpi, monomeric sDER (∼115 kDa) was detected, but sDER dimers were not detectable. In the presence of sSpi, both monomers and dimers (approximately 230 kDa) of sDER linked covalently by DSS were detected. This result suggests that sSpi can induce the dimerization of the DER extracellular domain.

FIG. 5.

Aos and SpiAos inhibit the dimerization of sDER induced by sSpi. sDER and sSpi were incubated with or without Aos or SpiAos, and monomeric sDER and covalently cross-linked dimeric sDER were probed with mouse anti-sDER antibody. In the absence of sSpi, essentially only monomeric sDER (M) was detected. In the presence of sSpi, however, both monomers (M) and dimers (D) of sDER were detected. Addition of Aos or SpiAos resulted in inhibition of the receptor dimer formation. Experiments were performed three times with similar results.

We further examined whether Aos could block the dimerization of sDER induced by sSpi. The formation of sSpi-induced sDER dimers was completely blocked upon the addition of either Aos or SpiAos (Fig. 5).

Aos inhibits DER dimerization in living cells.

Since receptor dimerization is the key event for the activation of RTKs, including the ErbB receptor family (8, 20), we examined the effects of Aos on the dimerization and tyrosine phosphorylation of DER in living cells. After incubating intact DER/S2 cells with ligands (sSpi and/or Aos), covalent cross-linking with BS³ (a membrane-impermeable reagent) was performed to stabilize the interaction between the receptor molecules. Subsequently, the amounts of monomeric and dimeric DER and the amount of tyrosine phosphorylation of DER were assayed by Western blotting using rat anti-DERc antibody and an antiphosphotyrosine antibody (anti-pY). The addition of sSpi resulted in an increase in dimers and in the amount of tyrosine phosphorylation of the dimeric DER (Fig. 6A). The percentage of cross-linked dimers in the total amount of receptor was nearly twofold higher when the cross-linking was carried out after stimulating the cells with sSpi than it was in the untreated cells (based on densitometric scanning). When the cells were preincubated with Aos for 2 min before the addition of sSpi, the percentage of DER dimers decreased from 52% to 28%, and the amount of tyrosine phosphorylation of the dimeric DER was also significantly reduced (Fig. 6A). These results demonstrate that the DER dimer is the active state and that Aos can block the dimerization of DER and the tyrosine phosphorylation of the dimeric DER induced by sSpi.

FIG. 6.

Aos inhibits the dimerization and phosphorylation of DER in the DER/S2 cells. (A) Aos inhibits the sSpi-induced dimerization and phosphorylation of DER. The cells were incubated with the indicated ligands, and the BS³-cross-linked DER was detected by Western blotting with the rat anti-DERc antibody (anti-DER). The filter was reprobed with antiphosphotyrosine antibody (anti-pY) to measure the level of activated DER. The positions of dimeric (D) and monomeric (M) DER are shown. The lower panels represent densitometric analysis showing the relative amount of the dimeric DER protein expressed as a percentage of the total amount of the receptor (left panel) and as the relative amount of tyrosine-phosphorylated DER (right panel). The intensity of the phosphorylated DER bands in the absence of both sSpi and Aos was defined as 100%. The intensities of phosphorylated DER were normalized to the amount of total DER protein. The bar graphs give the mean values ± standard deviations of three independent experiments. (B) Aos inhibits the sSpi-independent dimerization of DER. After overexpression of DER, the cells were incubated with or without Aos, and then the BS³-cross-linked DER was immunoprecipitated with rat anti-DERc antibody and probed with mouse anti-DER (left) or anti-pY (right) antibodies. The lower panel represents the densitometric analysis showing the relative intensity of the band corresponding to the dimeric DER before and after the Aos treatment. Experiments were performed twice with similar results.

Previous studies showed that a significant amount of DER autophosphorylation on tyrosine is induced even in the absence of sSpi upon the overexpression of DER in S2 cells (53, 54). Aos was shown to inhibit this sSpi-independent DER activation (54). However, the mechanism underlying this inhibitory action of Aos has remained obscure. We addressed whether DER is prevented from forming dimers when sSpi-independent DER activation is inhibited by Aos. After incubation of the DER/S2 cells with Aos, the amount of DER dimers and the tyrosine phosphorylation of DER were reduced almost equally (56 and 47% compared with those of untreated cells, respectively) (Fig. 6B). Taken together, these results indicate that Aos inhibited the dimerization as well as the autophosphorylation of DER on tyrosine by directly interacting with the receptor.

The SpiAos chimera shows inhibitory activity.

To determine which region of Aos is necessary for its antagonistic function, we constructed a SpiAos chimera, which is composed of the N-terminal portion of Spi and the C-terminal region of Aos, including the EGF-like domain (Fig. 1B), and assayed its functions in the DER/S2 cells. If the C-terminal region of Aos was sufficient to confer its inhibitory function, the SpiAos chimera might be expected to behave like Aos. When the cells were preincubated for 2 min with Aos or SpiAos before sSpi was added, the sSpi-induced tyrosine phosphorylation of DER was reduced to 47 and 55%, respectively (Fig. 7A). Like Aos, SpiAos also inhibited the overexpression-induced tyrosine phosphorylation of DER in the DER/S2 cells (Fig. 7B). After addition of Aos or SpiAos to the cells, the DER tyrosine phosphorylation was reduced to 31 and 48%, respectively, compared with the control cells (Fig. 7B).

FIG. 7.

SpiAos and Aos inhibit the activation of DER and ERK in the DER/S2 cells. (A and C) SpiAos and Aos inhibit the sSpi-induced activation of DER and ERK. The cells were incubated with the indicated ligands. Cells were preincubated with SpiAos or Aos for 2 min before the addition of sSpi. (B and D) SpiAos and Aos inhibit sSpi-independent activation of DER and ERK. After overexpression of DER, the cells were incubated without or with Aos or SpiAos for 15 min. The levels of tyrosine phosphorylation of DER immunoprecipitated (IP) by rat anti-DERc antibody were detected by anti-pY antibody (top panels of A and B). No differences in the total DER protein detected by mouse anti-sDER antibody in each lane were distinguishable (middle panels of A and B). The bottom panels in A and B show the relative amount of tyrosine phosphorylation of DER. The intensities of tyrosine phosphorylation of DER were normalized to the amount of total DER protein, as determined by densitometric analysis. The activated ERK (dp-ERK) was probed by anti-dually phosphorylated ERK antibody (top panels of C and D). No differences in the amounts of ERK protein in each lane were distinguishable (middle panels of C and D). The bottom panels in C and D show the relative intensity of the dp-ERK. The activation of ERK induced by sSpi (C) or by overexpression of DER (D) was defined as 100%. When the cells were preincubated with Aos or SpiAos for 2 min, the phosphorylation of ERK was reduced to 30 or 38% (the intensities of phosphorylation of ERK were normalized to the amount of total ERK protein, as determined by densitometric analysis), respectively (C). The overexpression of DER in the DER/S2 cells exhibited a modest level of ligand-independent ERK activation. After incubation of the cells with SpiAos or Aos, the normalized ERK activation was reduced to 14 or 17%, respectively (D). Experiments were performed twice with similar results.

Since DER functions principally through the Ras/MAPK signal transduction pathway, we tested the inhibitory effect of SpiAos on the activation of ERK, a member of the MAPK family, in the DER/S2 cells. SpiAos was nearly equivalent to Aos in inhibiting the sSpi-induced ERK activation (Fig. 7C) and DER overexpression-induced ERK activation (Fig. 7D). These results suggest that the C-terminal region of Aos that contains the EGF-like domain is sufficient for Aos functioning.

DISCUSSION

In the present study, we have examined the model that Aos directly inhibits the activation of DER. We established in vitro assay systems for DER dimerization and for Aos-DER interaction and demonstrated that Aos could act directly on the DER extracellular domain with its C-terminal region including the EGF-like domain and that Aos uses two molecular mechanisms to inhibit the activation of DER: inhibition of both dimerization of the receptor and binding of sSpi to the receptor.

If Aos functions as a direct inhibitory ligand of DER, direct interaction between Aos and DER should be the initial event leading to inhibition. The overexpression of DER in S2 cells results in the spontaneous autophosphorylation of the DER protein (52–54). Aos inhibits this ligand-independent activation of DER, suggesting that it acts directly on the DER-expressing cell (54). However, it remained to be shown whether Aos binds directly to DER. Clarification of this point is the first step toward understanding the functional mechanism of Aos in the inhibition of DER signaling. The immunoprecipitation analysis of DER demonstrated that Aos binds to DER in living DER/S2 cells (Fig. 2A). To further analyze the direct interaction between Aos and DER, we used recombinant proteins generated by the baculovirus expression system. We observed that Aos or AosEGF-Fc specifically bound the recombinant DER-Fc or sDER (Fig. 2B and C), respectively, indicating that the binding of Aos to the DER extracellular domain was direct. Real-time analysis of the interaction of wild-type or mutant ligands with the soluble receptor showed that Aos and AosEGF associated with DER-Fc with a K_d of approximately 17 to 41 nM, which is similar to or higher than the binding affinity of sSpi (∼67 nM) to DER-Fc (Fig. 3). Taken together, these results indicate that Aos can directly interact through its carboxyl-terminal region, including the EGF-like domain, with the extracellular domain of DER.

Recently, a second extracellular inhibitory ligand of an RTK has been identified. Angiopoietin-2 (Ang2) inhibits the activation of the Tie2/Tek receptor induced by the activating ligand angiopoietin-1 (Ang1) in endothelial cells (33). The binding of Ang2 to Tie2/Tek does not induce the activation of the receptor in endothelial cells, but blocks the binding of Ang1 to the receptor. The primary structure of Ang2 is similar to that of Ang1. Therefore, Ang2 competes with Ang1 for binding to their receptor (33). Aos is structurally related to Spi in that it contains an EGF-like domain, and the EGF-like domain of Aos is essential for its inhibitory function (52, 64). Although both sSpi and the C-terminal region including the EGF-like domain of Aos can bind to DER (Fig. 2), Aos and sSpi cannot bind to it simultaneously, suggesting Aos inhibits DER with a mechanism that is similar to that of Ang2 on Tie2/Tek receptor signaling.

Numerous studies have shown that hEGF activates the hEGFR by inducing receptor dimerization and that the extracellular domain of the receptor alone is able to undergo ligand-dependent dimerization (25, 30). Dimerization of the receptor is the key step for switching signals on or off (65). In spite of its importance for understanding the regulatory mechanisms of DER signal transduction, the dimerization of DER has not yet been studied. On the basis of our knowledge of the mammalian EGFR, we first analyzed whether sSpi induces DER dimerization. We demonstrated that baculovirus-expressed sSpi binds to DER (Fig. 2, 3, and 4) and stimulates sDER dimerization (Fig. 5). The dimeric form of the mammalian EGFR represents the active form (57), and the activated receptor phosphorylates the tyrosine residues of the adjacent receptor in an intermolecular manner (23, 29). The sSpi-induced dimeric DER was strongly phosphorylated on at least one tyrosine residue in DER/S2 cells (Fig. 6A), suggesting that dimerization is also essential for the activation of DER. These results also provided the means for analyzing the mechanism of Aos function in receptor dimerization. We demonstrated that Aos can inhibit the DER dimerization induced by sSpi (Fig. 5 and 6A).

Two molecular mechanisms for the ligand-induced dimerization of mammalian EGFR have been proposed. The first is that ligand binding induces a conformational change in the receptor that exposes a receptor-intrinsic dimerization site (18, 31). The alternative mechanism assumes that EGF is bivalent (19, 31): a ligand binds two receptors and contributes to their dimerization. Recent studies support the ligand bivalence model (58, 61). In the present study, we showed that Aos can inhibit the overexpression-induced dimerization of DER (Fig. 6B). Furthermore, Aos interacts directly with DER. Thus, if Aos is monovalent, having only one binding site for the receptor, the interaction of Aos and DER should result in the inhibition of the DER dimerization. Clearly, a more detailed structural analysis of the interaction of Aos and DER is required before we can fully understand the inhibitory mechanism of DER dimerization by Aos.

Secreted extracellular inhibitors of other growth factor receptors have also been identified. In Xenopus, three secreted polypeptides, Chordin (41), Noggin (22, 69), and Follistatin (10), have been shown to be inhibitors of the BMP receptor, a receptor serine/threonine kinase (66). These secreted proteins bind directly to BMPs, thereby preventing the binding of BMPs to their natural receptor. Frzb, a secreted protein similar to the Wnt receptor Frizzled, binds Wnt and inhibits Wnt signaling (32, 63). These secreted inhibitors antagonize the receptor's signaling by sequestering the ligands from their receptor. There is a significant distinction between these receptor antagonists and Aos, because Aos appears to act on the receptor itself. Although Ang2 binds its receptor, Tie2/Tek, the relationship between the binding and the dimerization of the receptor has not yet been elucidated.

We combined our results and previous findings on the mammalian EGFR to delineate possible mechanisms for the functions of Aos and sSpi in regulating DER activity (Fig. 8). The binding of sSpi to DER results in dimerization and activation of the receptor (Fig. 8A). At present, we have no experimental evidence to determine whether Spitz is a bivalent ligand like mammalian EGF. The EGF-like domain of Aos binds directly to DER at the sSpi-binding site and interferes with the binding of sSpi to DER (Fig. 8B). The Aos-DER interaction inhibits the DER dimerization that is induced by the receptor overexpression in the absence of sSpi (Fig. 6B). Thus, these two mechanisms by which Aos inhibits DER activation are likely to be distinct from each other. However, it is possible that the two mechanisms act cooperatively to inhibit DER activation in vivo.

FIG. 8.

Models of regulation of DER activation by sSpi and Aos. (A) sSpi binding results in DER dimerization. In its dimerized state, activated DER phosphorylates the tyrosine residue(s) located at the carboxyl terminus of the receptor. (B) Aos binds to the DER extracellular domain with its carboxyl-terminal region, which inhibits the dimerization of the receptor as well as the binding of sSpi to the receptor. See the text for details.

The SpiAos chimera appears to be an inhibitor of DER (Fig. 7A and B), like Aos, suggesting that the C-terminal region of Aos containing the EGF-like domain is important for Aos's functions. Many studies have been performed on the structure-function relationship of EGF-like molecules, aimed at the development of EGFR antagonists (24, 62). Vein is a moderate activator of DER and can be converted into an inhibitor by exchanging its EGF domain for that of Aos (52). Since Aos is a unique extracellular antagonist of DER, this system is a useful tool for analyzing the negative regulatory mechanism of the EGFR. Thus, it will be very interesting to identify the vertebrate homologue of Aos. A detailed understanding of how Aos regulates DER may have important therapeutic implications, because the ErbB family of receptors have been shown to be proto-oncogenes (21, 26).

Recently, the intracellular proteins D-cbl (34) and Sprouty (6, 27) and the transmembrane protein Kekkon-1 (15) have also been identified as negative regulators of the DER pathway. Our genetic evidence also suggested that Sprouty is required for the Aos-mediated inhibition of the DER pathway (59). Thus, it is important to elucidate whether and/or how Aos contributes to the function of these inhibitors of DER.