Characterization of Drosophila Insulin Receptor Substrate*

Insulin receptor substrate (IRS) proteins are phosphorylated by multiple tyrosine kinases, including the insulin receptor. Phosphorylated IRS proteins bind to SH2 domain-containing proteins, thereby triggering downstream signaling pathways. TheDrosophila insulin receptor (dIR) C-terminal extension contains potential binding sites for signaling molecules, suggesting that dIR might not require an IRS protein to accomplish its signaling functions. However, we obtained a cDNA encodingDrosophila IRS (dIRS), and we demonstrated expression of dIRS in a Drosophila cell line. Like mammalian IRS proteins, the N-terminal portion of dIRS contains a pleckstrin homology domain and a phosphotyrosine binding domain that binds to phosphotyrosine residues in both human and Drosophilainsulin receptors. When coexpressed with dIRS in COS-7 cells, a chimeric receptor (the extracellular domain of human IR fused to the cytoplasmic domain of dIR) mediated insulin-stimulated tyrosine phosphorylation of dIRS. Mutating the juxtamembrane NPXY motif markedly reduced the ability of the receptor to phosphorylate dIRS. In contrast, the NPXY motifs in the C-terminal extension of dIR were required for stable association with dIRS. Coimmunoprecipitation experiments demonstrated insulin-dependent binding of dIRS to phosphatidylinositol 3-kinase and SHP2. However, we did not detect interactions with Grb2, SHC, or phospholipase C-γ. Taken together with published genetic studies, these biochemical data support the hypothesis that dIRS functions directly downstream from the insulin receptor inDrosophila.

An insulin receptor homolog was identified in Drosophila by Fernandez-Almonacid and Rosen (7). Two isoforms of the insulin receptor have been identified in Drosophila (8,9). One isoform closely resembles mammalian receptors, and the other contains a 368-amino acid C-terminal extension. Because this extension contains many potential sites for tyrosine phosphorylation, including consensus binding sites for the p85 subunit of PI 3-kinase, it was suggested that the Drosophila insulin receptor might mediate signaling activities normally performed by IRS molecules. However, Yenush et al. (10) found that the Drosophila insulin receptor does not mediate insulinstimulated mitogenesis in 32D cells without the addition of mammalian IRS-1. This raised the question as to whether Drosophila might have an endogenous IRS molecule.
In this study, we have obtained a cDNA encoding Drosophila IRS. 2 The dIRS molecule is similar to mammalian IRS molecules, as it contains a pleckstrin homology (PH) domain, a phosphotyrosine binding (PTB) domain, and a phosphorylation domain with multiple potential sites of tyrosine phosphorylation. While this work was in progress, Böhni et al. (11) reported that dIRS is encoded by the chico locus, a mutant phenotype associated with abnormalities in cell size, cell number, and metabolism. To better understand the Drosophila insulin signaling pathway, we have investigated the interactions of dIRS with both human and Drosophila insulin receptors as well as downstream signaling molecules.

Identification and Sequencing of the EST Clone
By using the complete amino acid sequence of mouse (m) IRS-1, we searched the GenBank TM EST data base and identified a fragment of Drosophila IRS (clone LD16868, accession number AA536319), which we obtained from Genome Systems, Inc. (St. Louis, MO). The nucleotide sequences of both strands of the clone were determined using the ABI PRISM Dye Terminator Cycle Sequencing Ready Reaction Kit (Perkin-Elmer) or the ABI PRISM dRhodamine Terminator Cycle Sequencing Ready Reaction Kit (Perkin-Elmer). Reactions were analyzed using the ABI373A DNA sequencer. Sequence data were analyzed and assembled using the computer program GeneWorks (version 2.5.1, Oxford Molecular Group, Inc., Campbell, CA). * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF092046.
Yeast Plasmid Construction dIRS Constructs-We constructed expression vectors for fusion proteins of dIRS with a yeast activation domain by ligating part or all of dIRS cDNA into the plasmid pB42AD (CLONTECH, San Francisco, CA). Fragments of dIRS cDNA were amplified using oligonucleotide primers that introduced in-frame EcoRI and SalI restriction sites into the 5Ј ends of the upstream and downstream primers, respectively. Pfu DNA polymerase (Stratagene, La Jolla, CA) was used to amplify the fragments for the N-terminal construct (amino acids 1-410) as well as the phosphorylation domain construct (amino acids 235-968); however, we used Expand Long Template Polymerase (Roche Molecular Biochemicals) to amplify the cDNA for the dIRS full-length construct (amino acids 1-968). The N-terminal construct contained all residues up to, but not including, the first YXXM motif. Note that the N-terminal construct contained Ϸ175 amino acid residues after the region of homology to PTB domains of mammalian IRS molecules. The phosphorylation domain construct contained all residues following the end of homology to other known PTB domains.
dIR Constructs-We constructed expression vectors for fusion proteins of dIR with the prokaryotic LexA DNA-binding protein by ligating cDNA encoding portions of dIR into the plasmid pLexA (CLONTECH, San Francisco, CA). dIR cDNAs were amplified from EST AA246263 using oligonucleotide primers that introduced in-frame EcoRI and SalI restriction sites into the 5Ј ends of the upstream and downstream primers, respectively. A combination of Pfu DNA polymerase and Taq DNA polymerase (Roche Molecular Biochemicals) was used to amplify the fragments. The dIRc construct contained amino acid residues 1332-2148 (from the transmembrane domain to the C terminus of the molecule). The transmembrane ϩ juxtamembrane domains construct contained residues 1332-1661. The ⌬4 NPXY construct contained residues 1332-1965, beginning in the transmembrane domain and ending before the four C-terminal NPXY motifs.
hIR Constructs-We constructed the yeast expression vector for the fusion protein of the hIRc with the prokaryotic LexA DNA-binding protein by ligating cDNA encoding portions of hIR into the plasmid pLexA (CLONTECH, San Francisco, CA). hIRc cDNA was amplified from a cDNA construct of the insulin receptor (12) using oligonucleotide primers that introduced in-frame EcoRI and BamHI restriction sites into the 5Ј ends of the upstream and downstream primers, respectively. The construct contained residues 941-1343. The IR juxtamembrane NPXY motif mutant (JMm, residues 941-1343; N957A, Y960A) and the IR kinase dead mutant (K1018A, residues 941-1343) were generated by site-directed mutagenesis.

Yeast Two-hybrid System Assays
Assays were performed according to the MATCHMAKER LexA Twohybrid System protocol (CLONTECH, San Francisco, CA). Briefly, EGY48 cells (mat␣ his3, trp1, ura3-52, leu2::3Lexop-LEU2, LYS2), which were pretransformed with p8op-lacZ, were cotransformed with plasmid constructs by the polyethylene glycol/lithium acetate protocol (CLONTECH, San Francisco, CA). Transformants were grown on appropriate SD (CLONTECH, San Francisco, CA) glucose agar plates for 3 days at 30°C. Several independent colonies were transferred to SD galactose/raffinose agar plates and grown overnight at 30°C to induce expression of B42 fusion proteins. The interactions were then assessed using colony lift ␤-galactosidase assays and liquid culture ␤-galactosidase assays with O-nitrophenyl ␤-D-galactopyranoside (Sigma) as the substrate, according to the CLONTECH protocol. For the liquid culture ␤-galactosidase assays, the data were normalized to the interaction of hIRc with mIRS-3 (defined to be 100%). Experiments were repeated twice in triplicate.

Expression in Mammalian Cells
dIRS-Full-length dIRS cDNA was amplified by PCR from the EST clone using Pfu DNA polymerase. The primers included a site for EcoRI restriction endonuclease and a Kozak consensus sequence (13) at the 5Ј end of the coding sequence. The primers also introduced a site for SalI restriction endonuclease at the 3Ј end of the cDNA. The PCR product was first cloned using the Original TA Cloning Kit (Invitrogen, Carlsbad, CA). The insert was excised using appropriate restriction endonucleases and was ligated into pcDNA3.1 Myc-His version A (Invitrogen, Carlsbad, CA).
Chimeric hIR-dIR-EST clone AA246263 served as a PCR template to amplify residues 1344 -2148 of the Drosophila insulin receptor. We used Pfu DNA polymerase and oligonucleotide primers to introduce in-frame ApaI and SpeI restriction sites into the 5Ј ends of the upstream and downstream primers, respectively. The product was cloned into pCR2.1-TOPO using the TOPO TA Cloning Kit (Invitrogen, Carlsbad, CA). We then used a cDNA clone of the human IR (12) in pGEM4z that had a synonymous substitution in codon 948 to create an ApaI restriction site. This hIR cDNA clone was digested with ApaI and SpeI to remove hIR amino acids 949 -1343. The dIR TA-cloned construct was then digested with ApaI and SpeI; the fragment containing dIR residues 1344 -2148 was ligated to the pGEM4z-hIR fragment containing hIR residues 1-948. The entire chimeric hIR-dIR was then excised by restriction digestion with SalI and SpeI and ligated into pcDNA 3 (Invitrogen, Carlsbad, CA) that had been digested with XhoI and XbaI.
Chimeric hIR-dIR NPXY Mutants-We used PCR-based techniques to introduce mutations at the five NPXY motifs in the cytoplasmic domain of the chimeric hIR-dIR molecule. Phenylalanine was substituted for dIR Tyr 1358 in the juxtamembrane of the Y1358F chimeric receptor. In the ⌬CT mutant receptor, we substituted Phe for Tyr 1969 and deleted dIR amino acid residues 1970 -2148 (containing the remaining three NPXY motifs). In the Y1358F/⌬CT mutant, all five NPXY motifs were either mutated or deleted. Constructs were confirmed by sequencing.
dIRS Association with hIR-dIR NPXY Mutants-Transfected cells were serum-starved for 16 h (in DMEM containing 1% insulin-free BSA, 100 units/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine), stimulated with 100 nM insulin for 5 min, and lysed using a buffer containing 0.5% (v/v) Triton X-100, 50 mM Tris-HCl (pH 7.5), 0.3 M sodium chloride, 10 mM sodium pyrophosphate, 100 mM sodium fluoride, 2 mM sodium vanadate, Complete® protease inhibitor tablet (Roche Molecular Biochemicals), and 10 mM N-ethylmaleimide (NEM). Four-fifths of the lysate for each sample was used for immunoprecipitation. The lysate was incubated with an excess of Ultralink-immobilized protein G (Pierce) and autoantibodies to the IR that were obtained from patient B-19 with clinical type B insulin-resistance syndrome (15). In contrast to other experiments described here, proteins were separated under non-reducing conditions by omitting dithiothreitol and adding 10 mM NEM (16) to the lysate in addition to the Laemmli sample buffer. Samples were boiled 3 min before resolving using SDS-PAGE. Proteins were transferred to PVDF membranes. Immunoblotting was performed using monoclonal anti-Myc antibodies (Santa Cruz Biotechnology), polyclonal antibodies directed against the ␣-subunit of the human insulin receptor (Santa Cruz Biotechnology), and monoclonal anti-phosphotyrosine antibodies (Upstate Biotechnology, Inc.). Secondary antibodies were labeled with horseradish peroxidase, and proteins were visualized using ECL (Amersham Pharmacia Biotech).
Genomic DNA Isolation-DNAzol (Life Technologies, Inc.) was used to lyse cells from a confluent 25-ml culture and isolate genomic DNA according to the instructions provided by the product manufacturer.
Endogenous dIRS Expression-S2 cells were lysed in a buffer containing 0.5% (v/v) Triton X-100, 50 mM Tris-HCl (pH 7.5), 0.3 M sodium chloride, 10 mM sodium pyrophosphate, 100 mM sodium fluoride, 2 mM sodium vanadate, and Complete® protease inhibitor tablet (Roche Molecular Biochemicals). Protein from cell lysates was separated by SDS-PAGE and transferred to a PVDF membrane. Immunoblotting was performed using a polyclonal peptide antibody made against residues 919 -935 of dIRS (Zymed Laboratories Inc., South San Francisco, CA). Antibody specificity was confirmed using lysates from COS-7 cells transfected with empty vector or with Myc-tagged dIRS; both preimmune and postimmune serum were tested (data not shown). Secondary antibodies were labeled with horseradish peroxidase. Proteins were visualized using ECL.

RESULTS
Sequence Analysis-We identified a sequence in the EST data base (AA536319) corresponding to a cDNA encoding Drosophila IRS (dIRS). We determined the full-length coding sequence of dIRS cDNA (2907 bp). The clone also contained a 110-bp 5Ј-untranslated sequence, a Kozak consensus sequence (13) at the translation start site, a polyadenylation site, and a poly(A) tail. dIRS is predicted to contain 968 amino acid residues with a calculated molecular mass of 108,000 kDa. The deduced amino acid sequence of our cDNA clone (GenBank TM accession number AF092046) 3 is identical to that published by Böhni et al. (11). Like mammalian IRS molecules, dIRS contains a pleckstrin homology (PH) domain, a phosphotyrosine binding (PTB) domain, and a C-terminal phosphorylation domain. The PH and PTB domains of dIRS are 45 and 41% identical, respectively, to those of rIRS-1; similar results are obtained when dIRS is compared with other mammalian IRS molecules (Fig. 1). The phosphorylation domain does not display significant homology to other molecules. We identified at least six potential sites of tyrosine phosphorylation that are present in the context of known consensus binding sequences for SH2 domain-containing proteins ( Table I). Four of these sites, including two potential binding sites for the p85 subunit of PI 3-kinase (Tyr 411 and Tyr 641 ), are located in the phosphorylation domain.
Presence of Endogenous dIRS-The EST clone described above was derived from Drosophila embryo, although other ESTs of dIRS have since been obtained from the ovary and head. In order to determine whether the predicted dIRS protein was expressed, we obtained an antibody to dIRS by immunizing rabbits with a peptide specific to the C-terminal region of the protein (amino acid residues 919 -935). Through immunoblotting, we confirmed that dIRS is expressed in Drosophila S2 cells (embryonic epithelial cells), with an M r Ϸ144,000. Interestingly, when expressed in COS-7 cells, Myc-tagged dIRS had an M r Ϸ132,000 (Fig. 2).
Chromosomal Localization-When the nucleotide sequence of the entire dIRS EST insert was used to search the Gen-Bank TM nucleotide data base, we noted that the first 110 bases of our EST insert are identical to the last 110 bases of a dJNK genomic sequence (17). Since these bases occur after the dJNK polyadenylation site, we hypothesized that dIRS is located immediately downstream of dJNK, which has been mapped to chromosome 2, region 31 B-C (17) (Fig. 3). To test this hypothesis, genomic DNA was isolated from S2 cells. We used the genomic DNA as a template to perform PCR. In the first PCR, the amplicon extended from the last exon of dJNK through the middle of the 110-bp sequence. In the second PCR, the amplicon extended from the 110-bp sequence to the dIRS sequence. These reactions confirmed that the 110-bp sequence was located in close proximity to the coding sequences of the genes for both dJNK and our dIRS EST clone. Finally, we performed another PCR with one primer derived from a sequence in the dJNK genomic clone (downstream from the polyadenylation site) and a second primer in the 5Ј end of dIRS. After being ligated into the TA-cloning vector, the sequence of the amplified DNA was determined, thereby confirming that it contained both the dJNK genomic clone sequence as well as the dIRS sequence. This suggests that dJNK is immediately upstream of dIRS.
Yeast Two-hybrid-To investigate the functional interactions of dIRS with the human insulin receptor, we used the yeast two-hybrid system (Fig. 4A). We fused dIRS to the prokaryotic LexA DNA-binding protein in the pLexA expression vector; the human insulin receptor cytoplasmic domain (hIRc) was fused to a yeast transcription activation domain in pB42AD. As judged by the yeast two-hybrid assay, the interaction between dIRS and hIRc was nearly three times the strength of that of mIRS-3 and hIRc. The interaction of dIRS with hIRc was specific in that there was a negligible interaction of dIRS with human lamin C. Furthermore, there was no detectable signal when hIRc was expressed in the absence of dIRS (data not shown). The PTB domains of mammalian IRS molecules bind to a phosphotyrosine residue in the juxtamembrane NPXY motif of the insulin receptor. Therefore, we inquired whether the PTB domain in dIRS had similar binding specificity. To address this question, we constructed two mutants. In one mutant, the Asn 957 and Tyr 960 in the NPXY motif were mutated to alanine residues (JMm); in the other mutant (K1018A), Lys 1018 in the ATP-binding site of hIRc was mutated to Ala in order to inactivate the receptor tyrosine kinase. Use of either of the mutant insulin receptors decreased the strength of the interaction by 85%. Interestingly, unlike the situation with mammalian IRS molecules, dIRS retained a significant (Ϸ15%) ability to bind to the hIRc.
To investigate this interaction, we created two overlapping partial constructs of dIRS (Fig. 4A). One construct (amino acid residues 1-410) contained the PH and PTB domains plus 175 amino acids residues downstream from the PTB domain. The other construct (amino acid residues 235-968) contained the C-terminal portion of the molecule distal to the PTB domain. These two constructs will be referred to as the "N-terminal" and "phosphorylation domain" constructs, respectively. The Nterminal construct interacted nearly as strongly as the fulllength dIRS with the WT hIRc. However, like mammalian IRS molecules, the N-terminal construct did not interact with either the JMm or the K1018A hIRc constructs. Interestingly, the phosphorylation domain of dIRS was able to interact with WT hIRc. The strength of this interaction was Ϸ25% of the strength of the interaction between full-length dIRS and WT hIRc, but did not require intact kinase activity or an intact NPXY motif.
To investigate binding between two molecules derived from the same species, we used the yeast two-hybrid system to study the interaction between dIRS and the Drosophila insulin receptor cytoplasmic domain (dIRc) (Fig. 4B). The strength of the interaction between dIRS and dIRc, although significant, was only 20% of the strength of the interaction between dIRS and hIRc. A strong interaction was detected between the N-termi-nal construct and dIRc; however, we did not detect an interaction between the phosphorylation domain of dIRS and dIRc.
The dIR has a 368-amino acid C-terminal extension that contains multiple potential tyrosine phosphorylation sites, including four NPXY motifs. In addition, dIR has an NPXY motif in its juxtamembrane domain (in the homologous position to the NPXY motif present in mammalian insulin receptors). We created two additional constructs in order to map the site in dIRc where dIRS binds. In these two constructs, variable length sequences (183-and 487-amino acid residues) were de-
leted from the C terminus of dIRc; both mutants lack the four NPXY motifs in the C-terminal extension. For both the fulllength and N-terminal dIRS constructs, deletion of the additional four NPXY motifs in the C-terminal extension reduces the interaction by Ϸ70%; deletion of the entire C-terminal extension abolishes the interaction (Fig. 4B).
Expression in Mammalian Cells-We next inquired whether the insulin receptor could phosphorylate dIRS in a more physiological system. To address this question, we transiently cotransfected COS-7 cells with dIRS and a chimeric hIR-dIR. Serum-starved cells were incubated in the absence or presence of insulin (10 or 100 nM). Cell lysates were used for immunoblots that were probed with anti-phosphotyrosine antibodies and anti-Myc antibodies. In cells coexpressing dIRS and chimeric hIR-dIR, incubation in the presence of insulin led to increased tyrosine phosphorylation of both recombinant dIRS and the chimeric insulin receptor (Fig. 5, lanes 10 and 11). When dIRS was expressed in the absence of recombinant insulin receptor, a weak band was seen corresponding to phosphorylated dIRS in extracts of cells exposed to insulin (Fig. 5, lanes  4 and 5).
Coimmunoprecipitation Experiments with Mutant IRs-Studies in the yeast two-hybrid system suggested that the four NPXY motifs in the C terminus of dIR were more important than the juxtamembrane motif in the ability to bind dIRS (Fig. 4B). To further investigate this observation, COS-7 cells were transiently transfected with hIR-dIR and/or Myc-dIRS cDNA and were stimulated with 100 nM insulin for 5 min. Since the M r of dIRS is nearly identical to the M r of the truncated ␤-subunit of the insulin receptor, we performed SDS-gel electrophoresis of the extracts under non-reducing conditions to facilitate separation of the two molecules.
Immunoblotting of cell extracts with an anti-insulin receptor antibody showed that all four forms of the insulin receptor were expressed at comparable levels (Fig. 6B, lanes 1-4). Furthermore, when the phosphotyrosine content of the various receptors was examined using an anti-phosphotyrosine antibody, we found that the full-length hIR-dIR (FL), a receptor with a mutated juxtamembrane NPXY motif (Y1358F), and a receptor lacking NPXY motifs in its C-terminal extension (⌬CT) exhibited similar levels of tyrosine phosphorylation (Fig. 6A, lanes  2-4). In contrast, receptors with all of their NPXY motifs removed or mutated (Y1358F/⌬CT) displayed a reduced phosphotyrosine content (Fig. 6A, lane 1). Immunoprecipitation using anti-insulin receptor antibodies followed by immunoblotting with either anti-phosphotyrosine antibodies (Fig. 6C, lanes   1-4) or anti-receptor antibodies (Fig. 6D, lanes 1-4) showed similar results.
As with the hIR-dIR molecules, expression of Myc-tagged dIRS in the various extracts was similar (Fig. 6B, lanes 6 -9). However, mutation of the hIR-dIR juxtamembrane NPXY motif resulted in a significant reduction in the tyrosine phosphorylation of dIRS (Fig. 6A, lanes 6 and 9), whereas mutation of the C-terminal NPXY motifs of the receptor did not affect its ability to phosphorylate dIRS (Fig. 6A, lane 7). In contrast, immunoprecipitation of extracts with an anti-insulin receptor antibody followed by immunoblotting with anti-Myc antibody showed that Myc-dIRS associated strongly with both the (FL) hIR-dIR (Fig. 6D, lane 8) and the Y1358F mutant IR in which the Tyr in the juxtamembrane NPXY motif was mutated to Phe (Fig. 6D,  lane 9). However, this interaction was significantly reduced when the NPXY motifs in the C-terminal extension were removed (⌬CT) (Fig. 6D, lane 7) or when the NPXY motifs in the juxtamembrane domain as well as the C-terminal extension were removed (Y1358F/⌬CT) (Fig. 6D, lane 6). Furthermore, the dIRS that was associated with the receptors was tyrosinephosphorylated (Fig. 6C, lanes 6 -9). Thus, the juxtamembrane NPXY motif of dIR was required for optimal phosphorylation of dIRS, but the C-terminal NPXY motifs were necessary for formation of a stable complex that allowed for coimmunoprecipitation of dIRS together with dIR.
Coimmunoprecipitation Experiments with SH2 Domain-containing Proteins-Next we investigated the interactions of dIRS with downstream signaling molecules. COS-7 cells were transiently transfected with Myc-tagged dIRS and/or the chimeric hIR-dIR. In extracts of cells expressing both Myc-dIRS and hIR-dIR, anti-Myc antibodies immunoprecipitated two phosphotyrosine-containing proteins corresponding to dIRS and hIR-dIR. Insulin increased the phosphotyrosine content of both molecules (Fig. 7, A and B, lanes 3 and 4). Furthermore, when extracts from cells overexpressing hIR-dIR and Myc-dIRS were immunoprecipitated with anti-Myc antibodies and used for immunoblotting, we observed that insulin increased association of p85 and SHP2 with Myc-dIRS (Fig. 7A, lanes 3  and 4). We did not detect coimmunoprecipitation of Myc-dIRS with phospholipase C-␥, SHC, or Grb2 (Fig. 7A, lanes 3 and 4). As a control, we confirmed that our antibodies detected these SH2 domain-containing proteins in extracts of the transfected COS-7 cells (data not shown). Because there was only a weak signal corresponding to endogenous p85 in the COS-7 cells (data not shown), we confirmed the association of p85 with Myc-dIRS in experiments using COS-7 cells coexpressing p85, dIRS, and the chimeric hIR-dIR (Fig. 7B, lanes 3 and 4).

Parallels between Insulin Signaling Pathways in Mammals
and Drosophila-Many molecules in the mammalian insulin signaling pathway have been identified in Drosophila. The dIR has been cloned and has been demonstrated to be essential for early development in Drosophila. Mutants lacking functional insulin receptors die during the embryonic or early larval stage (18). Although a ligand for dIR has not been directly identified by cloning, several reports suggest the existence of "insulin-like activity" in extracts of Drosophila (19 -21). Furthermore, several molecules that function downstream in the insulin signaling pathway have homologs in Drosophila, including Grb2 (Drk) (22,23), dSHC (24), Sos (25), PI 3-kinase (26 -28), SHP2 (corkscrew) (29), PKB/Akt (30), PTEN (31, 32), and S6 kinase (33). Indeed, some of these molecules (e.g. Sos) were identified in Drosophila before they were known to exist in mammals.
However, structural differences between dIR and mammalian insulin receptors raised the possibility that there might be a major difference in the pathways of insulin signaling. Unlike its mammalian homologs, dIR contains a C-terminal extension with several potential sites of tyrosine phosphorylation. These C-terminal phosphotyrosine residues may serve as docking sites for signaling proteins containing SH2 domains. If dIR could directly form stable complexes with downstream signaling molecules, this might eliminate the need for a Drosophila IRS. Nevertheless, several lines of evidence suggested the presence of a Drosophila IRS (10,34). For example, when the dIR cytoplasmic domain was expressed in 32D cells, coexpression of IRS-1 was required to mediate insulin-stimulated mitogenesis (10). In the present work, we have directly demonstrated the existence of Drosophila IRS by identifying a cDNA encoding the protein. The existence of an IRS molecule in Drosophila emphasizes the many parallels between the insulin signaling pathways in species as diverse as Drosophila and human. dIRS closely resembles mammalian IRS molecules with respect to both structure and function. Like its mammalian homologs, dIRS contains a PH domain, a PTB domain, and a phosphorylation domain. The N-terminal portion of dIRS (i.e. the PTB domain) binds to phosphotyrosine residues located in NPXY motifs of the insulin receptor. In addition, coimmunoprecipitation experiments confirm that the dIRS phosphorylation domain provides binding sites for signaling molecules that interact with mammalian IRS molecules (e.g. PI 3-kinase and SHP2). By using genetic techniques, Böhni et al. (11) also conclude that dIRS interacts with dIR and PI 3-kinase. Böhni et al. (11) also use sequence analysis to suggest that dIRS binds Drk (the Drosophila ortholog of mammalian Grb2) (11). However, the motif cited beginning at Tyr 243 has an Arg at amino acid residue 246 that does not correspond to the consensus binding site for Grb2 (i.e. Tyr(P) followed by two hydrophilic and then one hydrophobic amino acid residue) (35). Additionally, we did not detect immunoprecipitation of Grb2 with dIRS in COS-7 cells, despite abundant expression of Grb2 in the cells.
Differences between dIRS and Mammalian IRS Molecules-There are several structural features that define the family of IRS molecules including: an N-terminal PH domain, a PTB domain that interacts with phosphotyrosine residues in NPXY motifs (e.g. Tyr 960 of the human insulin receptor), and a Cterminal domain containing multiple sites of tyrosine phosphorylation (1). In addition, IRS-2 contains a second domain that binds elsewhere in the insulin receptor; this interaction also requires intact tyrosine kinase activity (36,37). dIRS resembles IRS-2 in that it also contains a second domain that can bind to the hIR. However, unlike IRS-2, this second binding interaction between dIRS and hIR does not require tyrosine phosphorylation. Paradoxically, although we detected this second binding interaction when the two molecules were derived from different species (dIRS and hIR), we did not detect it when both molecules were derived from the same species (dIRS and dIR). The inability to detect this second interaction with dIR raises questions about its physiological significance; however, we cannot rule out the possibility of a low affinity interaction. Even a low affinity interaction at a second binding site might contribute to stabilizing an interaction that was driven primarily by binding of the PTB domain of dIRS to an NPXpY motif in dIR.  4 and 9), ⌬CT-IR (lanes 2 and 7), and Y1358F/⌬CT-IR (lanes 1 and 6). To better equalize the expression levels of the IR constructs, a ratio of 4:1 for full-length versus truncated hIR-dIR constructs was used. Empty expression vectors were included as required to maintain equal DNA concentrations in all transfections. Serum-starved cells were incubated in the presence of insulin (100 nM) for 5 min. Cells were lysed in a buffer containing Triton X-100 (0.5%) and NEM (10 mM). A and B, the samples were applied directly to the gels. C and D, the samples were immunoprecipitated (IP) using anti-IR antibodies prior to analysis by SDS-PAGE (4%) under non-reducing conditions in the absence of dithiothreitol. Proteins were detected by immunoblotting (IB) with anti-insulin receptor ␣ antibody (B and D, lanes 1-5), anti-Myc antibody (B and D, lanes 6 -10), and anti-phosphotyrosine antibody (A and C, lanes 1-10). The upper band present in lanes 6 -10 of C and D represents immunoglobulin G.
IRS molecules bind through their PTB domains to a highly conserved NPXY motif located in the juxtamembrane domain of mammalian insulin receptors (1). The dIR contains an NPXY motif in its juxtamembrane domain; however, the dIR also contains four additional NPXY motifs in its C-terminal extension. This raises the question, to which NPXY motif(s) does dIRS bind? When we deleted the NPXY motifs in the C terminus of dIR in the yeast two-hybrid system as well as in COS-7 cells, this significantly inhibited the association between dIRS and dIR. Thus, dIRS stably interacts with the four NPXY motifs present in the extension. Similarly, Marin-Hincapie and Garofalo (38) concluded that human IRS-1 binds primarily to the C-terminal extension of Drosophila IR. However, in contrast to the findings of Marin-Hincapie and Garofalo (38), we found that the phosphorylation of dIRS by dIR depends on the juxtamembrane NPXY motif, not the C-terminal extension. It is possible that the differences in the results are explained by differences in experimental methods. Whereas we have used full-length insulin receptors and dIRS, Marin-Hincapie and Garofalo (38) used a truncated insulin receptor lacking the ␣-subunit to phosphorylate mammalian IRS-1. By adding insulin to intact cells, we have activated insulin receptors in the plasma membrane. It is not clear whether recombinant ␤-subunits of dIR (38) are transported normally to the plasma membrane or whether they might be retained in intracellular membranes.
In mammals, IRS molecules require the IR juxtamembrane NPXY motif for phosphorylation. Although an association between mammalian IR and mammalian IRS has been shown, the binding appears to be weak or transient as non-stoichiometric amounts of these proteins are brought down in immunoprecipitation experiments (39). Indeed, Auclair et al. (40) presented evidence suggesting that stable (as opposed to transient) binding of IRS-1 and IRS-2 to the IR can lead to insulin resistance and inhibit insulin signaling.
Our findings suggest that the dIR juxtamembrane NPXY motif functions in a manner similar to that of mammalian insulin receptors. However, the stable interaction of the Cterminal extension with dIRS serves a unique function, as it does not allow for efficient phosphorylation of dIRS. There are many potential implications of this finding. As mentioned above, stable binding of IRS to dIR may alter receptor function by preventing binding of other proteins. For example, binding of dIRS might inhibit binding of phosphatases, thereby inhibiting dephosphorylation of the receptor as has been suggested in the case of mammalian insulin receptors (41). Similarly, as discussed below, it is possible that binding of dIRS might inhibit binding of other signaling molecules to the insulin receptor. For example, the C-terminal extension of dIR contains four consensus sites (YXXM) for binding of the SH2 domain of the p85 regulatory subunit of PI 3-kinase (42). It had been suggested that, upon tyrosine phosphorylation, the dIR C-terminal extension might bind p85, thereby activating PI 3-kinase directly without a requirement for an IRS molecule. Indeed, experimental evidence has been obtained in support of this hypothesis. Fernandez et al. (8) demonstrated binding of a GST-p85 SH2 domain fusion protein to the phosphorylated long form of dIR. Furthermore, Yenush et al. (10) demonstrated activation of p85 by insulin-stimulated dIR in 32D cells (which do not contain endogenous IRS molecules). However, it is noteworthy that the YXXM motifs in the C terminus of dIR are contained within NPXYXXM sequences. Thus, the partial overlap of these two motifs (NPXY and YXXM) creates the possibility that stable binding of IRS molecules to the NPXY might inhibit the ability of p85 to bind to the YXXM motif. At least two laboratories have published data consistent with this interpretation. Marin-Hincapie and Garofalo (38) did not detect binding of p85 to dIR ␤-subunit when IRS-1 was bound to the C-terminal extension. Similarly, Yamaguchi et al. (34) did not detect increased binding to p85 or activation of PI 3-kinase by the long form of the receptor in Chinese hamster ovary cells that contain endogenous IRS molecules.
Previous studies suggest that dIR is expressed as two isoforms, a full-length isoform as well as a truncated isoform (likely resulting from proteolytic cleavage) that lacks the Cterminal extension (8). Both isoforms are present in many cells types, although the ratio of long form to short form varies greatly (7,43). In our experiments, the two truncated forms of  1-4) and a chimeric insulin receptor (the hIR extracellular domain and the dIR cytoplasmic domain; lanes 3, 4, 7, and 8) were overexpressed in COS-7 cells. Empty expression vectors were included as required to maintain equal DNA concentrations in all transfections. Serum-starved cells were incubated in the presence or absence of insulin (100 nM) for 5 min. Cell lysates were immunoprecipitated with anti-Myc antibodies followed by immunoblotting with antibodies as indicated. B, COS-7 cells were cotransfected with Myc-tagged dIRS (lanes 1-4), a chimeric insulin receptor (the hIR extracellular domain and the dIR cytoplasmic domain; lanes 3, 4, 7, and 8), and bovine p85 (lanes 1-8). Empty expression vectors were included as required to maintain equal DNA concentrations in all transfections. Serum-starved cells were incubated in the presence or absence of insulin (100 nM) for 5 min. Cell lysates were immunoprecipitated with anti-Myc antibodies followed by immunoblotting with antibodies as indicated.
the receptor (Y1358F/⌬CT and ⌬CT) were present as a single band on the immunoblot (Fig. 6, B and D, lanes 1 and 2). In contrast, the two full-length constructs (FL and Y1358) migrated as a doublet. It is likely that the low M r band corresponds to the proteolytically cleaved isoform lacking the Cterminal extension, whereas the high M r band corresponds to the full-length isoform. However, under our conditions, the high M r band predominates, suggesting that the majority of the receptors are present as the full-length isoform (Fig. 6, B and  D, lanes 3 and 4). It is not known how the relative levels of the two forms are regulated. Nonetheless, it is possible that the ability of dIRS to bind tightly to C-terminal NPXY motifs in the long isoform may modify the signaling specificity of dIR.
Physiological Significance of dIRS-The ability to apply genetics to analyze gene function is the principal advantage of Drosophila as an experimental model. While our studies were in progress, Böhni et al. (11) published work in which they had identified the dIRS gene as the locus of a mutation causing the chico phenotype. In addition, they used genetic methods to map chico to region 31 B-C of chromosome 2, near basket (the gene encoding dJNK), the same region where we mapped the gene using a molecular approach. After the publication of the chromosomal localization and phenotype of chico, Flybase (44) identified chico as flipper, a gene whose phenotype was first reported by Bridges and Mohr in 1919 (45).
The chico flies, lacking functional dIRS, were smaller and grew more slowly than wild type flies. Interestingly, small size and growth retardation were the most obvious abnormalities in "knock-out" mice lacking IRS-1 (46,47). Furthermore, because the chico phenotype in Drosophila is milder than the phenotype observed in flies lacking insulin receptors (18), this suggests that dIRS may not be the only molecule that functions downstream from the insulin receptor. It is not clear whether the dIR has other physiological substrates that are phosphorylated directly by the receptor, or whether the receptor itself may bind downstream signaling molecules (e.g. SH2 domain-containing proteins that might bind to phosphotyrosine residues in the receptor itself). Furthermore, inactivating mutations in dAkt (48), dPTEN (31,49), and dS6 kinase (50) (three other proteins in the insulin signaling pathway) lead to phenotypes that closely resemble chico flies.
In our biochemical studies, we have analyzed the direct binding interaction of dIRS with upstream and downstream molecules. These results elucidate the biochemical mechanisms for the role of dIRS in Drosophila, complementing the genetic studies of Böhni et al. (11).