C-terminal Src Kinase Associates with Ligand-stimulated Insulin-like Growth Factor-I Receptor*

Increased expression of the insulin-like growth factor-I receptor (IGF-IR) protein-tyrosine kinase occurs in several kinds of cancer and induces neoplastic transformation in fibroblast cell lines. The transformed phenotype can be reversed by interfering with the function of the IGF-IR. The IGF-IR is required for transformation by a number of viral and cellular oncoproteins, including SV40 large T antigen, Ras, Raf, and Src. The IGF-IR is a substrate for Src in vitro and is phosphorylated in v-Src-transformed cells. We observed that the IGF-IR and IR associated with the C-terminal Src kinase (CSK) following ligand stimulation. We found that the SH2 domain of CSK binds to the tyrosine-phosphorylated form of IGF-IR and IR. We determined the tyrosine residues in the IGF-IR and in the IR responsible for this interaction. We also observed that fibroblasts stimulated with IGF-I or insulin showed a rapid and transient decrease in c-Src tyrosine kinase activity. The results suggest that c-Src and CSK are involved in IGF-IR and IR signaling and that the interaction of CSK with the IGF-IR may play a role in the decrease in c-Src activity following IGF-I stimulation.

The IGF-IR, 1 closely related to the IR, is a receptor proteintyrosine kinase consisting of two heterodimers linked by disulfide bonds (1). Each heterodimer contains an extracellular subunit, the ␣ subunit, responsible for ligand binding and a transmembrane subunit, the ␤ subunit, which contains the protein-tyrosine kinase activity in its intracellular domain. When IGF-I binds to the IGF-IR, the receptor becomes autophosphorylated on several tyrosine residues. Phosphorylation of the receptor allows interactions with phosphotyrosine binding or Src homology 2 (SH2) domain-containing proteins, thereby transducing signals to downstream effectors. Insulin receptor substrate 1 and 2 (IRS-1 and IRS-2) and Shc are major substrates of the IGF-IR that subsequently bind other SH2 domain-containing proteins, including the p85 subunit of phosphatidylinositol 3Ј-kinase, the adaptor proteins, Grb2 and Nck, the protein-tyrosine-phosphatase, Shp 2, and the C-terminal Src kinase, CSK (2)(3)(4)(5)(6)(7)(8)(9)(10). Binding of IGF-I to the IGF-IR activates the Ras-mitogen-activated protein kinase signaling pathway (11,12).
Protein-tyrosine kinases of the Src family have been studied extensively (for reviews see Refs. 13 and 14). Extracellular stimuli activate Src kinases, stimulating cellular DNA synthesis (15), mitosis (16,17), proliferation, adhesion (18), and cytokine production (19). Targeted disruption of the c-src protooncogene (20,21) results in osteopetrosis, suggesting that Src is required for osteoclast differentiation function (22,23). c-Src is involved in transducing signals from several receptor proteintyrosine kinases (13,14), including the platelet-derived growth factor receptor, the epidermal growth factor receptor, the fibroblast growth factor receptor, and the nerve growth factor receptor. Src family kinases have been implicated in insulin and IGF-I signaling pathways. For example, it has been reported that tyrosine-phosphorylated IRS-1 binds to Fyn (24), and the ␤ subunit of the IGF-IR is phosphorylated by Src in v-Srctransformed cells (25).
The c-Src protein has a domain organization consisting of an N-terminal Src homology (SH) 4 domain containing a myristoylation and membrane localization signal, followed by a unique domain, SH3 and SH2 domains, the catalytic SH1 domain, and the C-terminal tail (see Ref. 13 for review). c-Src kinase activity is regulated by phosphorylation of two tyrosine residues, Tyr 416 , the autophosphorylation site, and Tyr 527 in the C-terminal tail, as well as by SH2 and SH3 domain-mediated interactions. The three-dimensional structure of c-Src helps explain the regulatory interactions (26). Phosphorylation of Tyr 527 creates a binding site for the Src SH2 domain, locking the molecule in an inactive state. The SH3 domain contributes to the stability of the inactive state by interacting with a "linker" that joins the SH2 and catalytic domains. The compact organization pushes the two lobes of the catalytic domain together, disabling the active site.
CSK, the C-terminal Src kinase, is a ubiquitously expressed cytoplasmic protein-tyrosine kinase. Its organization resembles that of Src, including an SH3, SH2, and a catalytic domain (27,28). However, CSK lacks the catalytic domain autophosphorylation site, the C-terminal regulatory tyrosine, and the N-terminal myristoylation signal of Src. CSK phosphorylates Src Tyr 527 , down-regulating Src kinase activity (27,29). The regulation of CSK activity is not well understood. Several studies (30 -33) have suggested that relocalization of CSK allows the molecule to phosphorylate Src C-terminal tyrosine. Howell and Cooper (30) suggested a model in which CSK would interact with Src substrates via its SH3 or SH2 domain in order to down-regulate Src activity.
The observations that IGF-IR is a substrate of Src in v-Srctransformed cells and that IGF-IR expression is required for Src-mediated transformation in mouse fibroblasts suggested that Src might participate in IGF-IR signaling. In this study, we observed that CSK bound to the IGF-IR in vitro and in vivo. By using a yeast two -hybrid system to characterize the IGF-IR-CSK interaction, we found that the CSK SH2 domain interacted with Tyr 943 and Tyr 1316 of the IGF-IR. By contrast, the  CSK SH2 domain interacted only with Tyr 1322 of the insulin  receptor (IR). We further observed that Src kinase activity was rapidly decreased following IGF-I and insulin stimulation of human or mouse fibroblasts. Therefore, CSK is a new interacting protein of the IGF-IR and the IR. Our observations suggest a role for CSK and c-Src in IGF-IR and IR signaling.

EXPERIMENTAL PROCEDURES
Cell Lines and Vectors-pCLXSN-hIGF-IR was constructed as described previously. 2 Hemagglutinin (HA)-tagged wild type CSK was generated by polymerase chain reaction (PCR) using convenient cloning sites for insertion into the pEF-HA vector. HA-tagged CSK mutant S109C was made using the QuickChange procedure (Stratagene) from wild type CSK subcloned into pBluescript followed by subcloning into the pEF-HA vector. The pGEX-KG-CSK from which these constructs were derived was a generous gift from K. Neet (see Ref. 35). Both wild type and S109C mutant CSK were verified by DNA sequencing.
Antibodies-Anti-IGF-IR, a rabbit polyclonal antibody, directed against the C terminus of the ␤ subunit, was described previously. 2 Antibody directed against the C terminus of CSK was a gift from K. Neet (see Ref. 35); anti-phosphotyrosine antibody (4G10) was from Upstate Biotechnology Inc. (New York); the hybridoma for the monoclonal antibody to the hemagglutinin epitope tag (anti-HA, monoclonal antibody 12CA5) was kindly provided by Dr. I. Wilson (The Scripps Research Institute); anti-IR ␤ subunit antibody was from Santa Cruz Biotechnology, Inc.; the hybridoma for the monoclonal peptide antibody directed against amino acids 2-17 of v-Src, which recognizes both c-Src and v-Src, was from Microbiological Associates; and anti-Src hybridoma monoclonal antibody 327 was a generous gift from Joan Brugge (see Ref. 36).
Similar amounts of lysate (about 1 ml), adjusted to contain equal amounts of protein (0.5-1.0 mg), were incubated with 1 g of antibody for 2 h at 4°C. For c-Src immunoprecipitation, rabbit anti-mouse secondary antibody was added for an additional 30 min. Protein A-agarose (Repligen) was added for 1 h at 4°C and washed 3 times with lysis buffer at 4°C. Immune complexes were boiled for 5 min in sample buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 10% glycerol, 20% ␤-mercaptoethanol) and separated on a 10% SDS acrylamide gel.
Immunoblotting-After gel electrophoresis, proteins were transferred to Immobilon polyvinylidene fluoride membrane (Millipore, Bedford, MA) using a semi-dry transfer apparatus (Acrylictech) for 90 min at 200 mA. The membranes were blocked overnight at 4°C in TBS-T (50 mM Tris-HCl, pH 8, 150 mM NaCl, 0.05% Tween 20) containing 5% bovine serum albumin. The membranes were incubated with primary antibodies at room temperature for 2 h in TBS-T plus 5% bovine serum albumin, washed in the same buffer without bovine serum albumin, and incubated with horseradish peroxidase-conjugated secondary antibody for 1 h. Antibodies were detected by enhanced chemiluminescence (Amersham Pharmacia Biotech). Conjugated horseradish peroxidaseprotein A antibodies were used for anti-IGF-IR, anti-IR, and anti-Csk blotting and horseradish peroxidase anti-mouse antibodies for 4G10, 2-17, and 12CA5 blotting.
In Vitro Kinase Assay-c-Src and Csk immunoprecipitates were washed three times with lysis buffer and once with cold 20 mM PIPES, pH 7.0, 1 mM dithiothreitol. Twenty l of kinase buffer (20 mM PIPES, pH 7.0, 10 mM MnCl 2 , 1 mM dithiothreitol), containing 20 Ci of [␥-32 P]ATP (specific activity 1 mCi/mM) and 2.5 g of acid-denatured enolase (38) or 50 ng of baculovirus-expressed kinase-inactive Src K297R (39), was added to the beads. After 15 min at 30°C the reaction was stopped by adding 2ϫ sample buffer and analyzed by SDS-polyacrylamide gel electrophoresis. The conditions used in this study are in the linear range of the assay according to Cooper et al. (38). The presence of equivalent amounts of enolase was verified by Coomassie staining of the gel and equivalent amounts of K297R Src by immunoblotting. Finally, the gel was dried and protein phosphorylation revealed by autoradiography. Tryptic phosphopeptide mapping and cyanogen bromide cleavage were done according to Boyle et al. (40).
Yeast Two-hybrid System-The yeast expression plasmids pBTM116 and pLexA-lamin were provided by A. Vojtek (Seattle) (see Ref. 41) and J. Camonis (Paris, France), respectively. The yeast expression plasmid encoding the Gal4 activation domain (AD) pACTII was from CLON-TECH (Palo Alto, CA). The cytoplasmic domains of the IR and IGF-IR, subcloned in pBTM116 in frame with the DNA binding domain of LexA, were described previously (9,42). Plasmid construction, cloning, and DNA sequencing were carried out according to standard protocols (43). When necessary, PCR using thermostable Pfu polymerase (Stratagene) was performed to generate adequate cloning sites to allow the in-frame insertion into the two-hybrid plasmids. The rat IRS-1 cDNA (a gift from J. Pierce, National Cancer Institute, Bethesda) (residues 5-1235) was cloned in frame with the Gal4 AD. The full-length human CSK cDNA was PCR-amplified from pGST-CSK (35) and fused to the Gal4 ADencoding sequence in pACTII. The different subdomains of CSK (SH3/2, SH3, and SH2 constructs) were PCR-amplified from pACTII CSK and fused to pACTII. All point mutations of different proteins were generated by site-directed mutagenesis using the Transformer TM Kit (CLON-TECH). Mutations were verified by DNA sequencing.
The two-hybrid system used in this study was described previously (41,42). The yeast strain L40, which is MATa, trp1, leu2, his3, LYS2::lexA-HIS3, URA3::lexA-lacZ, was provided by A. Vojtek (Seattle) and J. Camonis (Paris, France). Growth conditions and yeast maintenance were as described previously (44). Yeast L40 was transformed simultaneously with the two indicated plasmids by the lithium acetate method of Gietz et al. (45). The transformants were grown on plates lacking tryptophan and leucine to select for the pBTM116 and pACTII derivatives, respectively. After 3 days at 30°C, 3 colonies of each transformation were tested for histidine prototrophy after plating on medium lacking tryptophan, leucine, and histidine and for ␤-galactosidase activity using a liquid assay, performed essentially as described previously (46). In brief, 1 ml of yeast cell extract was incubated with the substrate o-nitrophenyl-␤-D-galactopyranoside, and the increase in A 420 was measured after 10 or 30 min (47). Similar results were obtained by analyzing growth on plates lacking histidine and ␤-galactosidase activity in a filter color assay using X-gal as substrate. Results expressed in Tables I-III represent a summary of the data obtained with these three methods.

CSK Associates with the IGF-IR in Vivo-
To examine the role of c-Src or CSK in IGF-I and insulin signaling, we tested whether c-Src or CSK associated with the IGF-IR in co-immunoprecipitation studies. First, using several different conditions, we tried to co-immunoprecipitate c-Src with the IGF-IR. We were unable to detect c-Src in IGF-IR immunoprecipitates or the IGF-IR in c-Src immunoprecipitates (data not shown).
Then, we tested whether CSK could be co-immunoprecipitated with the IGF-IR. Human 293 cells were starved for 24 h in Dulbecco's modified Eagle's medium supplemented with 0.1% FCS and stimulated with IGF-I. In these experiments we used an analog of IGF-I, LR3-IGF-I, which is not bound and sequestered by IGF-binding proteins (48). We confirmed that LR3-IGF-I activates the IGF-IR by immunoprecipitating the IGF-IR from stimulated cells and immunoblotting the receptor with anti-phosphotyrosine antibody (see Fig. 1B). IGF-IR immunoprecipitates were tested for the presence of CSK by immunoblotting with anti-CSK antibodies. As shown in Fig. 1A (upper panel), CSK was detected in association with the IGF-IR in stimulated cells but not in unstimulated cells. The amount of CSK bound to the IGF-IR compared with the total amount present in 293 fibroblasts, estimated from immunoprecipitation of CSK from the remaining lysates after the IGF-IR immunoprecipitation, represented approximately 5% of total cellular CSK (Fig. 1A, lower panel).
To study further the association between CSK and the IGF-IR, 293 cells were cotransfected with IGF-IR and CSK cDNAs, incubated for 24 h, starved for a further 24 h in Dulbecco's modified Eagle's medium supplemented with 0.1% FCS, and stimulated for various times with IGF-I. As shown in Fig. 1B, CSK was detected in association with the IGF-IR only in stimulated cells. The association was rapid, occurring within 1 min of addition of IGF-I. The IGF-IR also became phosphorylated on tyrosine within 1 min after stimulation with IGF-I. Similar amounts of IGF-IR were present in immunoprecipitates at all time points.
To characterize further the CSK-IGF-IR interaction, we tested whether the IGF-IR could be observed in CSK immunoprecipitates. Since co-immunoprecipitation of CSK with the IGF-IR was observed only after IGF-I stimulation (Fig. 1A), suggesting that tyrosine phosphorylation of the IGF-IR was required for IGF-IR-CSK association, we also tested whether the interaction depended on a functional SH2 domain of CSK.
Human 293 cells were cotransfected with the IGF-IR and a vector encoding either wild type CSK or a CSK mutant with a defect in the FLVRES motif of the SH2 domain, CSK S109C, which fails to bind phosphotyrosine-containing proteins (35). Both CSK wild type and the CSK S109C mutant contained a hemagglutinin epitope tag allowing them to be separated from the endogenous CSK protein during immunoprecipitation. The transfected cells were incubated, starved, and stimulated with IGF-I as described above. HA-CSK immunoprecipitates made with the 12CA5 anti-HA antibody were assayed for the presence of the IGF-IR by immunoblotting. As shown in Fig. 1C, IGF-IR was not observed in unstimulated immunoprecipitates. IGF-IR was present in wild type CSK immunoprecipitates from IGF-I-stimulated cells but not in CSK immunoprecipitates of the CSK S109C mutant, defective in SH2 domain binding to phosphotyrosine. We conclude that CSK associates in vivo with the IGF-IR. Furthermore, the interaction requires a functional SH2 domain of CSK and occurs only after activation of the receptor by IGF-I.
CSK Associates through Its SH2 Domain with the IGF-IR and the IR in Yeast-To characterize further the association between the IGF-IR and CSK and to compare the IGF-IR and the IR with regard to the sites required for CSK binding, we used the yeast two-hybrid system of Fields and Song (49). This system has been used to study the interaction of the IGF-IR and the IR with their two major substrates, IRS1 and Shc, in yeast (42). Tyrosine phosphorylation of these receptors in the yeast two-hybrid system has been demonstrated previously (42). Constructs encoding the intracytoplasmic domain of the IGF-IR or the IR fused to the DNA binding domain of LexA, and full-length CSK fused to the transcriptional activation domain of Gal4 (GAD) were used. Interaction between the LexA-receptor hybrid and the CSK-GAD hybrid in the yeast strain stimulates transcription of two reporter genes, HIS3 and lacZ, which are flanked upstream by LexA-binding sites. The results of cotransfection with various constructs are presented here as a summary of the growth on plates lacking histidine and ␤-galactosidase activity obtained from yeast liquid culture or filter color assay.
We verified that the IGF-IR-LexA or the IR-LexA and the CSK-GAD hybrids were expressed in yeast by immunoblotting of yeast lysates transfected with these constructs (Ref. 42 and data not shown). We confirmed that the constructs did not activate transcription of the reporter genes alone or in combination with an unrelated protein, lamin, or the kinase interaction domain, KID, of cyclic AMP response element-binding protein (Table I). We also confirmed, as an internal positive control, that the IGF-IR and the IR wild type and mutants interacted with IRS-1 (Tables I-III). We used wild type IGF-IR, or a kinase-inactive mutant of the IGF-IR (K1003A), and the wild type IR, or a kinase-inactive mutant of the IR (K1018A), to test the interaction with wild type CSK. Table I shows that

TABLE I
Interaction between CSK and the IGF-IR or the IR Yeast two-hybrid interaction between various domains of CSK and the wild type or kinase-deficient IGF-IR and the IR. The results represent a summary of the data obtained by analyzing growth on plates lacking histidine and ␤-galactosidase activity in a filter color assay using X-gal as substrate or in a yeast liquid culture using o-nitrophenyl-␤-D-galactopyranoside as substrate.
␤-galactosidase activity was stimulated, and growth on plates lacking histidine was observed when wild type IGF-IR or IR was cotransfected with wild type CSK, reflecting the interaction of CSK with the receptors. In contrast, no interaction was observed when the kinase-inactive mutants of the IGF-IR K1003A and the IR K1018A were used. The results demonstrate that CSK interacts with the IGF-IR and the IR in yeast and that the protein-tyrosine kinase activity of the receptors is required for the interaction.
To test the ability of the SH2 and SH3 domains of CSK to bind to the receptors, we fused the SH2 and SH3 domains, alone or in combination, to the GAD, and we tested the ability of the hybrid proteins to interact with wild type or kinaseinactive mutants of the IGF-IR and IR when coexpressed in yeast. Table I shows that ␤-galactosidase activity was stimulated, and growth on plates lacking histidine was observed when the CSK SH2 or CSK combined SH3 and SH2 (CSK SH3/2) domains were cotransfected with the wild type IGF-IR and IR receptors, but not when the kinase-inactive receptors were used. The CSK SH3 domain did not stimulate ␤-galactosidase activity or growth on plates lacking histidine when cotransfected with either wild type or kinase-inactive receptors. These results demonstrate that CSK associates with tyrosinephosphorylated IGF-IR or IR through its SH2 domain in yeast.
The SH2 Domain of CSK Associates with Tyr 943 and Tyr 1316 of the IGF-IR and Tyr 943 of the IR-The recognition specificities of the SH2 domain of CSK (50) suggested that amino acids Thr, Ala, or Ser in position ϩ1 after the phosphorylated tyrosine and Val, Ile, Met, or Arg in position ϩ3 could constitute a high affinity binding motif for the SH2 domain of CSK. Examination of the amino acid sequences of the IGF-IR and the IR revealed two such motifs in the IGF-IR, Tyr 943 -Ala-Ser-Val and Tyr 1316 -Ala-His-Met, and one in the IR, Tyr 1322 -Thr-His-Met. To test the importance of these tyrosine residues for CSK binding to the receptors, we mutated each tyrosine to phenylalanine, incorporated the mutated sequences into vectors containing the LexA DNA binding domain, and tested them for interaction with CSK in the yeast two-hybrid system. The IGF-IR mutants, Y943F, Y1316F, and the double mutant, Y943/1316F, interacted with IRS1 in the two-hybrid system demonstrating that the mutant receptors had normal tyrosine kinase activity and could interact with their substrate in this system (Table II). The mutants were cotransfected with fulllength CSK, the CSK SH2 domain, or the combined CSK SH3 and SH2 domains (Table II). The single mutants of the IGF-IR, Y943F and Y1316F, stimulated ␤-galactosidase activity and growth on plates lacking histidine in the transfected cells. However, when the double mutant, IGF-IR Y943/1316F, was used, ␤-galactosidase activity and growth on plates lacking histidine was not observed. These results confirmed that the SH2 domain of CSK interacts with phosphorylated Tyr 943 and Tyr 1316 of the IGF-IR.
We performed similar experiments using mutants of the IR, Y960F, Y1316F and Y1322F, cotransfected with full-length CSK, the CSK SH2 domain, or the combined CSK SH3 and SH2 domains (Table III). The Y960F (binding site for IRS1) and Y1316F mutants stimulated ␤-galactosidase activity and growth on plates lacking histidine when cotransfected with wild type CSK. The Y1316F mutant also stimulated ␤-galactosidase activity and growth on plates lacking histidine when cotransfected with the SH2 or the SH3/SH2 domains of CSK. However, both a deletion mutant of the IR lacking the 7 amino acids containing the two tyrosines, 1316 and 1322 (42), and the Y1322F mutant failed to stimulate activity or growth with any of the CSK constructs. We conclude that phosphorylated Tyr 1322 of the IR is required for interaction with the CSK SH2 domain in yeast.

c-Src Kinase Activity Decreases after IGF-I and Insulin
Stimulation-To determine whether c-Src kinase activity was modified following IGF-I or insulin stimulation, human 293 or mouse NIH 3T3 fibroblasts (Fig. 2) were treated with both hormones. c-Src was immunoprecipitated and its activity assayed in an in vitro kinase assay using enolase as substrate. Stimulation with either LR3-IGF-I or insulin decreased the kinase activity of Src by about 40 or 30%, respectively, in human 293 and 50 or 20%, respectively, in NIH 3T3 fibroblasts, as measured by the amount of 32 P incorporated into enolase or into c-Src itself (Fig. 2). For each reaction, the amount of immunoprecipitated c-Src was assayed by immunoblotting, and the amount of enolase was determined by Coomassie staining. The amounts of both Src and enolase were similar in all samples.
To characterize further the down-regulation of c-Src by IGF-I, human 293 cells were incubated in medium containing 0.1% FCS for 24 h and then stimulated for various times with LR3-IGF-I at a concentration of 1.3 ϫ 10 Ϫ7 M. c-Src enzymatic activity was tested by in vitro kinase assay as described above. A maximum 35% decrease in c-Src kinase activity, measured  Yeast two-hybrid interaction between CSK and various mutants of the IR. The results represent a summary of the the data obtained by analyzing growth on plates lacking histidine and ␤-galactosidase activity in a filter color assay using X-Gal as substrate or in a yeast liquid culture using o-nitrophenyl-␤-D-galactopyranoside as substrate. the CSK SH2 domain Yeast two-hybrid interaction between CSK and various mutants of the IGF-IR. The results represent a summary of the data obtained by analyzing growth on plates lacking histidine and ␤-galactosidase activity in a filter color assay using X-Gal as substrate or in a yeast liquid culture using o-nitrophenyl-␤-D-galactopyranoside as substrate. NA, not applicable.
Ϫ ϩϩϩ by 32 P incorporation into enolase, was observed after 2 min of IGF-I stimulation and a return to basal activity occurred after about 20 min (Fig. 3A). c-Src autophosphorylation, as measured by 32 P incorporation into c-Src, showed a maximum decrease of 25% and followed a similar time course (Fig. 3B). Human 293 cells were stimulated with increasing concentrations of LR3-IGF-I ranging from 1.3 ϫ 10 Ϫ10 to 6.5 ϫ 10 Ϫ7 M. Fig. 3C shows that the kinase activity of c-Src, measured by 32 P incorporation into enolase, was decreased to a plateau of about 65% of the untreated cell level by concentrations of LR3-IGF-I of 1.3 ϫ 10 Ϫ7 M or higher. c-Src autophosphorylation, measured by 32 P incorporation into c-Src, was decreased to a plateau of about 75% of the untreated cell level (Fig. 3D). Human 293 cells were also stimulated with insulin under the same conditions used in Fig. 3, C and D. The kinase activity of c-Src, measured by 32 P incorporation into enolase, was decreased to about 65% of the untreated cell level after stimulation with concentrations of insulin of 8.6 ϫ 10 Ϫ7 M, without reaching a plateau as for IGF-I stimulation (Fig. 3E). c-Src autophosphorylation, measured by 32 P incorporation into c-Src, was decreased to a plateau of about 80% of the untreated cell level (Fig. 3F).
From these results, we conclude that stimulation of human 293 or mouse 3T3 fibroblasts by IGF-I or insulin causes a rapid and transient decrease in the kinase activity of c-Src which is dependent on the concentration of the hormone.
c-Src Down-regulation after IGF-I Stimulation Requires CSK Binding to the IGF-IR but Increased CSK Activity Is Not Detected after Stimulation-To determine whether c-Src downregulation induced by IGF-I was dependent on the interaction between CSK and the IGF-IR, we tested whether disruption of this interaction would affect c-Src down-regulation. Wild type CSK, SH2 mutant S109C of CSK, and the vector alone as a control were transfected in 293 cells. After 24 h starvation in 0.1% FCS, cells were stimulated with 1.3 ϫ 10 Ϫ7 M IGF-I for 2 min, and c-Src was immunoprecipitated and its activity tested in an in vitro kinase assay using enolase as substrate. As shown in Fig. 4A (upper panel), c-Src activity of cells transfected with the vector alone decreased to 82% after stimulation in comparison to unstimulated cells. As expected, overexpression of wild type CSK further decreased c-Src activity to 64% after IGF-I stimulation in comparison to unstimulated cells. However, in 293 cells transfected with the S109C SH2 mutant of CSK, c-Src activity after stimulation was decreased to 80%, a level similar to that observed in cells transfected with the vector. Protein levels of the wild type and SH2 CSK mutant S109C were similar and did not change after stimulation, as shown by CSK immunoblot (Fig. 4A, lower panel).
Because the kinase activity of c-Src can be decreased by phosphorylation of Tyr 527 by CSK, we also tested whether CSK activity was changed following stimulation of human 293 cells by IGF-I. CSK activity was measured by an in vitro kinase assay of immunoprecipitated CSK using a baculovirus-expressed kinase-inactive mouse c-Src mutant, K297R, as substrate. c-Src activity was measured as described above, using the same 293 cell lysates. The results of three separate experiments, summarized in Fig. 4B, showed significant decreases of the kinase activity of c-Src, measured by ␥-32 P incorporation into enolase and c-Src. In all of the samples showing a decrease in c-Src activity, a small but reproducible increase in CSK activity was observed following stimulation with IGF-I, but this difference was not statistically significant. However, this experiment, testing the activity of total cellular CSK after IGF-I stimulation, cannot rule out that CSK activity changes significantly in the fraction of CSK bound to the IGF-IR, which represents approximately 5% of total cellular CSK (Fig. 1A).
We attempted to examine further the molecular mechanisms involved in the down-regulation of the kinase activity of c-Src following IGF-I treatment. In order to determine whether the decrease in c-Src activity could be attributed to phosphorylation of c-Src Tyr 527 , we tried to detect differences in c-Src phosphorylation in vivo following stimulation of 293 cells with IGF-I. Human 293 cells, radiolabeled overnight with [ 32 P]orthophosphate, were stimulated with IGF-I. Immunoprecipitated c-Src was analyzed by tryptic phosphopeptide mapping or cyanogen bromide cleavage. No significant changes were observed in the peptide maps of Src isolated from these cells after stimulation with IGF-I compared with unstimulated cells (data not shown).
From these data, we conclude that the association between the activated IGF-IR and CSK is likely to be required for c-Src activity down-regulation by IGF-I. DISCUSSION The results presented here show that CSK associates with the IGF-IR and the IR. For both receptors, the association requires a functional protein-tyrosine kinase activity and oc- LR3-IGF-I and Src activity was assayed by ␥-32 P incorporation into enolase. B, in the same experiment as in A, we also determined the ␥-32 P incorporation of Src. C, 293 cells were stimulated with various amounts of LR3-IGF-I for 2 min. Src activity was assayed as above. D, in the same experiment as in C, Src ␥-32 P incorporation was assayed. E, 293 cells were stimulated with various amounts of insulin for 2 min. Src activity was assayed as above. F, in the same experiment as in E, Src ␥-32 P incorporation was assayed.
curs through the SH2 domain of CSK, which binds phosphorylated Tyr 943 and Tyr 1316 of the IGF-IR and Tyr 1322 of the IR in a yeast two-hybrid assay. The ligand-stimulated binding of CSK to the IGF-IR and the IR suggests that CSK participates in IGF-I and insulin signaling. We observed a concomitant decrease in the kinase activity of c-Src which was rapid and transient following stimulation of human 293 cells with IGF-I or insulin. Also, CSK interaction with the activated IGF-IR is likely to be required for c-Src down-regulation, although CSK activity does not change after hormone stimulation.
It is likely that the interaction of CSK with the receptors observed in the yeast two-hybrid system is a direct one, although it is possible that a yeast protein might mediate the interaction. Assuming that the interaction is direct, CSK is a new binding protein of the IGF-IR and the IR.
It may be significant that the SH2 domain of CSK interacts with two tyrosine residues in the IGF-IR, Tyr 943 and Tyr 1316 , but only one in the IR, Tyr 1322 . It is generally accepted that the IR plays a metabolic role in glucose homeostasis and that the IGF-IR is responsible for cell growth and proliferation. Therefore, it is possible that the interaction of CSK with Tyr 943 of the IGF-IR might be responsible for IGF-IR-specific events such as cell proliferation. It is intriguing that CSK also binds to phos-phorylated IRS-1 following insulin stimulation and is involved in insulin-stimulated dephosphorylation of focal adhesion proteins, possibly through inactivation of Src family kinases (8). This observation is in agreement with observations suggesting that tyrosine phosphorylation of the focal adhesion kinase is decreased after insulin treatment (8), although focal adhesion kinase phosphorylation level was increased after IGF-I (34). These differences suggest that focal adhesion kinase tyrosine phosphorylation may not be a direct effect of insulin or IGF-I and may require further investigation.
c-Src catalytic activity is regulated by phosphorylation of two tyrosine residues (13,28). Phosphorylation of Tyr 416 , the autophosphorylation site, is required for full activation of the kinase (13,14). Phosphorylation of Tyr 527 by CSK decreases the kinase activity of Src kinase by allowing intramolecular binding of Tyr 527 to the Src SH2 domain, leading to the inactive "closed" conformation of c-Src. Howell and Cooper (30) examined the features of CSK required for suppression of c-Src activity. They found that mutations in the SH2, SH3, and catalytic domains of CSK rendered CSK incapable of suppressing c-Src activity. They suggested that colocalization of CSK with c-Src, rather than a change in CSK activity, was responsible for regulation of the kinase activity of c-Src by CSK and FIG. 4. CSK interaction with the IGF-IR is required for c-Src downregulation following stimulation with IGF-I but its activity remains unchanged. 293 cells were serumstarved for 24 h and stimulated for 2 min with 1.3 ϫ 10 Ϫ7 M LR3-IGF-I. A, cells were transfected with cDNA from HAtagged either wild type CSK or SH2 domain mutant, S109C, CSK. c-Src activity was tested in in vitro kinase assays using enolase as substrate (upper panel). Expression of the various CSK constructs was detected after anti-HA immunoprecipitates and anti-HA immunoblot (lower panel). B, c-Src and CSK immunoprecipitates from three different experiments were tested for c-Src activity and CSK activity, respectively, in in vitro kinase assays using enolase or baculovirus-expressed kinase-inactive c-Src, K295R, as substrates. *, p Ͻ 0.05; **, p ϭ 0.1 in t test analysis, n ϭ 3. that colocalization might be facilitated by binding of CSK to substrates of c-Src.
In a similar way, binding of CSK to ligand-stimulated IGF-IR and IR could lead to localization of CSK to the membrane and to subsequent effects on the tyrosine-protein kinase activity of c-Src. Indeed, when we used an SH2 domain mutant of CSK, S109C, that does not bind the activated IGF-IR, we observed that c-Src activity was similar to the control (Fig. 4A). Our results suggest that the interaction between CSK and the activated IGF-IR is required for the IGF-I induced c-Src downregulation. We also observed that this interaction does not result in a significant change in the kinase activity of total cellular CSK, although an increase in activity of the fraction of CSK bound to the IGF-IR cannot be excluded. Therefore, we suggest that colocalization of CSK and c-Src following IGF-I treatment is responsible for the transient decrease in the kinase activity of c-Src in fibroblasts. The similar timing of the binding and the decrease in c-Src activity is consistent with this notion. However, since the interaction between c-Src and CSK is not well understood, more complex mechanisms for c-Src down-regulation following IGF-I stimulation are possible.
We were unable to observe a significant change in the phosphorylation pattern of c-Src following stimulation of cells with IGF-I, which might have implicated CSK further in the downregulation of c-Src activity. The lack of an observable change was not unexpected, since c-Src is already phosphorylated on Tyr 527 in unstimulated cells, and observing a small change in phosphorylation of a specific peptide following stimulation is difficult. Furthermore, the results from Fig. 1A, showing that only 5% of CSK binds to the IGF-IR, suggest that it would be very difficult to detect a change in phosphorylation of a particular peptide, using non-quantitative methods such as phosphopeptide mapping or cyanogen bromide cleavage. Also, the absence of an observable increase in kinase activity in CSK may reflect the fact that only a fraction of the cellular CSK is binding to the IGF-IR. Indeed, it is possible, according to the observations of Howell et al. (30), that binding of CSK SH2 domain to the IGF-IR changes its catalytic activity. In this case, the change in activity should involve only a fraction of cellular CSK, thereby resulting in phosphorylation of c-Src Tyr 527 in only a fraction of c-Src molecules. Therefore, observation of a significant increase in CSK activity or observation of a major decrease in the kinase activity of c-Src would be unlikely.
In conclusion, we demonstrated that CSK is a new binding protein of the IGF-IR and the IR, associating with activated IGF-IR and IR through binding of the SH2 domain of CSK to phosphorylated tyrosines in the activated receptors. To explore further the role of CSK and Src in signaling from the IGF-IR and the IR, it will be important to identify downstream targets that are directly affected by IGF-I or insulin-induced CSK binding to the IGF-IR or the IR and down-regulation of c-Src.