Roles of the Complex Formation of SHPS-1 with SHP-2 in Insulin-stimulated Mitogen-activated Protein Kinase Activation*

SHPS-1 is a receptor-like protein that undergoes tyrosine phosphorylation and binds SHP-2, an SH2 domain-containing protein tyrosine phosphatase, in response to insulin and other mitogens. The overexpression of wild-type SHPS-1, but not of a mutant SHPS-1 in which all four tyrosine residues in its cytoplasmic region were mutated to phenylalanine, markedly enhanced insulin-induced activation of mitogen-activated protein kinase in Chinese hamster ovary cells that overexpress the human insulin receptor. Mutation of each tyrosine residue individually revealed that the major sites of tyrosine phosphorylation of SHPS-1 in response to insulin are Tyr449 and Tyr473. In addition, mutation of either Tyr449 or Tyr473 abolished the insulin-induced tyrosine phosphorylation of SHPS-1 and its association with SHP-2. Surface plasmon resonance analysis showed that glutathioneS-transferase fusion proteins containing the NH2-terminal or COOH-terminal SH2 domains of SHP-2 bound preferentially to phosphotyrosyl peptides corresponding to the sequences surrounding Tyr449 or Tyr473, respectively, of SHPS-1. Furthermore, phosphotyrosyl peptides containing Tyr449 or Tyr473 were effective substrates for the phosphatase activity of recombinant SHP-2 in vitro. Together, these results suggest that insulin may induce phosphorylation of SHPS-1 at Tyr449 and Tyr473, to which SHP-2 then binds through its NH2-terminal and COOH-terminal SH2 domains, respectively. SHPS-1 may play a crucial role both in the recruitment of SHP-2 from the cytosol to a site near the plasma membrane and in increasing its catalytic activity, thereby positively regulating the RAS-mitogen-activated protein kinase signaling cascade in response to insulin.

SHPS-1 is a receptor-like protein that undergoes tyrosine phosphorylation and binds SHP-2, an SH2 domaincontaining protein tyrosine phosphatase, in response to insulin and other mitogens. The overexpression of wildtype SHPS-1, but not of a mutant SHPS-1 in which all four tyrosine residues in its cytoplasmic region were mutated to phenylalanine, markedly enhanced insulininduced activation of mitogen-activated protein kinase in Chinese hamster ovary cells that overexpress the human insulin receptor. Mutation of each tyrosine residue individually revealed that the major sites of tyrosine phosphorylation of SHPS-1 in response to insulin are Tyr 449 and Tyr 473 . In addition, mutation of either Tyr 449 or Tyr 473 abolished the insulin-induced tyrosine phosphorylation of SHPS-1 and its association with SHP-2. Surface plasmon resonance analysis showed that glutathione S-transferase fusion proteins containing the NH 2 -terminal or COOH-terminal SH2 domains of SHP-2 bound preferentially to phosphotyrosyl peptides corresponding to the sequences surrounding Tyr 449 or Tyr 473 , respectively, of SHPS-1. Furthermore, phosphotyrosyl peptides containing Tyr 449 or Tyr 473 were effective substrates for the phosphatase activity of recombinant SHP-2 in vitro. Together, these results suggest that insulin may induce phosphorylation of SHPS-1 at Tyr 449 and Tyr 473 , to which SHP-2 then binds through its NH 2terminal and COOH-terminal SH2 domains, respectively. SHPS-1 may play a crucial role both in the recruitment of SHP-2 from the cytosol to a site near the plasma membrane and in increasing its catalytic activity, thereby positively regulating the RAS-mitogen-activated protein kinase signaling cascade in response to insulin.
SHP-2, a non-transmembrane-type protein tyrosine phosphatase (PTPase) 1 that contains two SH2 domains (1)(2)(3), is thought to participate in the signal transduction pathways of a variety of growth factors and cytokines. SHP-2 binds directly to the PDGF receptor, EGF receptor, and c-KIT in response to stimulation of cells with the corresponding receptor ligand and undergoes tyrosine phosphorylation (4 -7). A PDGF receptor in which Tyr 1009 , the binding site for SHP-2, was changed to Phe was not able to activate RAS in response to PDGF (8), implicating SHP-2 in PDGF-induced RAS activation. Injection of mRNA encoding a catalytically inactive mutant SHP-2 into Xenopus oocytes blocked fibroblast growth factor-and activininduced induction of mesoderm as well as fibroblast growth factor-induced activation of MAP kinase (9). SHP-2 has also been suggested to mediate EGF stimulation of the RAS-MAP kinase cascade that leads to DNA synthesis (10). Corkscrew (the putative Drosophila homolog of SHP-2) is thought to be required for Ras1 activation or to function in conjunction with Ras1 during signaling by the Sevenless receptor tyrosine kinase (11,12). Thus, in general, SHP-2 appears to play a positive role in growth factor-induced cell proliferation, probably through activation of the RAS-MAP kinase cascade.
SHP-2 is also implicated in insulin signal transduction. The binding of insulin to its receptor stimulates receptor autophosphorylation on tyrosine residues as well as the receptor-mediated tyrosine phosphorylation of intracellular substrates such as IRS-1, IRS-2, and GAB-1 (13)(14)(15). These primary substrates of the intrinsic tyrosine kinase activity of the IR serve as docking proteins that recruit SH2 domain-containing proteins such as the 85-kDa subunit of phosphatidylinositol 3-kinase, GRB2, and NCK (16). SHP-2 also binds to these docking proteins via its SH2 domains (17), and its PTPase activity is thereby increased in response to insulin (18 -20). Expression of a catalytically inactive SHP-2 inhibited the insulin-induced activation of RAS (21), MAP kinase (21)(22)(23)(24), or expression of a c-fos reporter gene (24) in a dominant negative manner. Tyrosine phosphorylation of SHC also plays an important role in insulin-induced activation of the RAS-MAP kinase cascade (25,26). Tyrosine-phosphorylated SHC binds to the SH2 domain of the adapter protein GRB2, which is constitutively associated with the guanine nucleotide exchange protein SOS (27). SOS catalyzes the exchange of GTP for GDP on RAS, resulting in activation of the RAF-MEK-MAP kinase cascade (28). In addition, the GRB2⅐SOS complex also binds to tyrosine-phosphorylated IRS-1 (29,30). Thus, both SHC⅐GRB2 and IRS-1⅐GRB2 complexes are implicated in linking IR stimulation to RAS activation. However, the expression of a catalytically inactive SHP-2 does not affect binding of GRB2 to either IRS-1 or SHC in response to insulin, suggesting that SHP-2 may regulate an upstream factor necessary for RAS activation by insulin and that this upstream factor may be required for the GRB2-or SHC-dependent pathway (21). Because the precise mechanism by which SHP-2 mediates activation of the RAS-MAP kinase cascade in response to insulin is unknown, identification of a phosphorylated substrate of SHP-2 is essential.
In an attempt to identify such a physiological substrate of SHP-2, we recently discovered an ϳ120-kDa receptor-like glycoprotein, termed SHPS-1, the tyrosine phosphorylation of which was greatly increased in cells overexpressing a catalytically inactive SHP-2 (31, 32). Ohnishi et al. (33) prepared a monoclonal antibody to the same protein (which they named BIT) and demonstrated the formation of a complex between tyrosine-phosphorylated SHPS-1 (BIT) and SHP-2 in rat brain lysate. Subsequently, we (34) and others (35) cloned a human homolog of SHPS-1 (also named SIRP␣1), and a family of related proteins was also identified (35). The entire putative extracellular region of SHPS-1 consists of three homologous Ig-like domains with multiple N-linked glycosylation sites, indicating that SHPS-1 belongs to the Ig superfamily. The cytoplasmic region of SHPS-1 contains four tyrosine residues each followed by XX(L/V/I) sequences, characteristic of tyrosine phosphorylation sites. These sequences also match well those to which the SH2 domains of SHP-2 prefer to bind as revealed by in vitro binding studies with a phosphotyrosyl peptide library (36). Insulin rapidly stimulates tyrosine phosphorylation of SHPS-1 and its association with SHP-2, and the IR kinase phosphorylates SHPS-1 on tyrosine residues in vitro, suggesting that SHPS-1 is a direct substrate for the IR kinase (32). Thus, the SH2 domains of SHP-2 may bind to one or more phosphorylated tyrosine residues in the cytoplasmic region of SHPS-1 in response to insulin. Because sequences corresponding to known catalytic domains are not present in its cytoplasmic region, SHPS-1 appears to be a docking protein, such as IRS, GAB-1, or Drosophila DOS (37), that recruits SHP-2 from the cytosol to a region near the plasma membrane in response to insulin. However, little is known of the precise roles of complex formation between SHPS-1 and SHP-2 in insulin regulation of biological activities. In addition, it is not clear which tyrosine residues in the cytoplasmic domain of SHPS-1 are essential for its interaction with SHP-2.
We have now generated cells that express wild-type or various mutant forms of SHPS-1. An increase in complex formation between SHPS-1 and SHP-2 induced by overexpression of wild-type SHPS-1 enhanced MAP kinase activation in response to insulin. Furthermore, of the four potential tyrosine phosphorylation sites of SHPS-1, only Tyr 449 and Tyr 473 are required for the optimal tyrosine phosphorylation of SHPS-1 and the binding of SHP-2 to SHPS-1 in response to insulin.

EXPERIMENTAL PROCEDURES
Cells and Antibodies-CHO cell lines expressing human IRs (CHO-IR cells) were maintained in Ham's F-12 medium supplemented with 10% FBS. CHO-IR cells that overexpress a catalytically inactive SHP-2 (SHP-2-C/S cells) were generated previously (21).
To generate CHO-IR cells that overexpress various mutant SHPS-1 proteins, we introduced point mutations that changed each or all of the tyrosine residues in the cytoplasmic region (Tyr 408 , Tyr 432 , Tyr 449 , and Tyr 473 ) to phenylalanine into the rat SHPS-1 cDNA by site-directed mutagenesis. The full-length wild-type and mutant SHPS-1 cDNAs were then cloned into the EcoRI site of the pSR␣ vector. CHO-IR cells (ϳ5 ϫ 10 5 cells per 10-cm dish) were transfected with both 10 g of pSR␣ containing SHPS-1 cDNA and 1 g of pHyg, which contains the hygromycin B phosphotransferase gene, with the use of LipofectAMINE (Life Technologies, Inc.). The cells were cultured in Ham's F-12 medium containing hygromycin B (200 g/ml) (Wako, Osaka, Japan) and 10% FBS, and colonies were isolated 14 -21 days after transfection. Several cell lines expressing wild-type or mutant SHPS-1 proteins were identified by immunoblot analysis of cell lysates with polyclonal antibodies to SHPS-1 as described below.
NIH 3T3 cells or Rat-1 cells overexpressing human IRs (Rat-1-IR cells) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% FBS. NIH 3T3 or Rat-1-IR cells that overexpress wild-type SHPS-1 were also generated as described above.
To generate a mAb specific to rat SHPS-1, we partially purified SHPS-1 from 100 10-cm plates of confluent SR-3Y1 cells as described previously (31). The resulting SHPS-1 preparation was injected into the hind footpads of two BALB/c mice three times at 1-week intervals, after which lymphocytes were isolated from the draining lymph nodes and fused with P3U1 myeloma cells as described previously (32). Antibodies in culture supernatants of the resulting hybridomas were screened on the basis of their ability to immunoprecipitate SHPS-1 from the membrane fraction of SR-3Y1 cells as assessed by immunoblot analysis with antibodies to phosphotyrosine (PY20). Positive hybridomas were rescreened by the same procedure. Among several positive clones, clone 2F34 was chosen, and the corresponding mAb was purified from ascites fluid of mice with a MAPS II kit (Bio-Rad). We found that mAb 2F34 reacts well with rat SHPS-1 but poorly with the corresponding protein of other species such as hamster or mouse. The detailed properties of this mAb will be described elsewhere. In contrast to mAb 2F34, mAb 4C6, which we generated previously (32), reacts well with hamster SHPS-1 but poorly with the corresponding protein of rat or mouse. Rabbit polyclonal antibodies to SHPS-1 (31) or to SHP-2 (21) were generated with GST fusion proteins containing the COOH-terminal regions of either SHPS-1 or SHP-2 as described previously. Rabbit polyclonal antibodies (␣91) to both p44 and p42 MAP kinase were prepared with a synthetic peptide corresponding to residues 307-327 of rat MAP kinase. The rabbit polyclonal antibodies to MAP kinase, ␣92, were prepared against synthetic peptide corresponding to residues 350 -367 of rat MAP kinase. HRP-conjugated mAb PY20 to phosphotyrosine was obtained from Santa Cruz Biotechnology.
Immunoprecipitation and Immunoblot Analysis-Transfected CHO-IR cells, NIH 3T3 cells, or Rat-1-IR cells were deprived of serum for 16 h and then stimulated with insulin. The culture medium was aspirated, and the cells were immediately washed with PBS and frozen in liquid nitrogen. The cells were subsequently lysed on ice in 1 ml of ice-cold lysis buffer (20 mM Tris-HCl (pH 7.6), 140 mM NaCl, 2.6 mM CaCl 2 , 1 mM MgCl 2 , 1% Nonidet P-40, 10% (v/v) glycerol) containing 1 mM phenylmethylsulfonyl fluoride and 1 mM sodium vanadate. The lysates were centrifuged at 10,000 ϫ g for 15 min at 4°C, and the resulting supernatants (1.0 -1.5 mg) were subjected to immunoprecipitation and immunoblot analysis. Supernatants were incubated for 4 h at 4°C with various antibodies bound to protein G-Sepharose beads (Amersham Pharmacia Biotech). The beads were then washed twice with 1 ml of WG buffer (50 mM Hepes-NaOH (pH 7.6), 150 mM NaCl, 0.1% Triton X-100) and resuspended in SDS sample buffer. Gel electrophoresis and immunoblot analysis with various antibodies and an ECL detection kit (Amersham) were performed as described previously (21,38).
Subcellular Fractionation-CHO-IR cells (in four 10-cm plates) treated with or without insulin were frozen in liquid nitrogen, scraped into 2 ml of ice-cold hypotonic lysis solution (20 mM Hepes-NaOH (pH 7.6), 5 mM sodium pyrophosphate, 5 mM EGTA, 1 mM MgCl 2 ) containing aprotinin (10 mg/ml), 1 mM phenylmethylsulfonyl fluoride, and 1 mM sodium vanadate, and homogenized with a Dounce homogenizer. The homogenate was centrifuged at 100,000 ϫ g for 60 min, and the resulting supernatant was referred to the cytosolic fraction. The pellet was suspended in 0.5 ml of membrane solubilizaton solution (20 mM Tris-HCl (pH 7.5), 1% Triton X-100, 150 mM NaCl, 1 mM MgCl 2 ) supplemented with 1 mM phenylmethylsulfonyl fluoride and 1 mM sodium vanadate. The suspension was centrifuged at 100,000 ϫ g for 60 min, and the resulting supernatant was referred to as the solubilized membrane fraction. All procedures were performed at 4°C.
Determination of MAP Kinase Activation-MAP kinase activation was monitored by immunoblot analysis of cell lysates with antibodies to p44 and p42 MAP kinase (New England BioLabs) that recognize the enzymes only when they are activated by phosphorylation of Tyr 204 .
MAP kinase activation was also monitored by using the direct in vitro kinase assay as described previously (21). Briefly, after insulin stimulation, cells were lysed in 500 l of lysis buffer (25 mM Tris-HCl (pH 7.4), 25 mM NaCl, 1 mM sodium orthovanadate, 10 mM NaF, 10 mM sodium pyrophosphate, 0.5 mM EGTA, 10 nM okadaic acid (Wako Chemicals), and 1 mM phenylmethylsulfonyl fluoride). The lysates were incubated with for 3 h at 4°C with ␣92 anti-MAP kinase antibodies that had been bound to protein G-Sepharose beads. The immunoprecipitates were washed twice with lysis buffer and suspended in 35 ml of assay buffer (25 mM Tris-HCl (pH 7.4), 10 mM MgCl 2 , 1 mM dithiothreitol, 40 M ATP, 1 Ci of [␥-32 P]ATP, 2 mM protein kinase inhibitor (Sigma), and 0.5 mM EGTA, and myelin basic protein (1 g/ml) (Sigma) as substrate). After incubation for 10 min at 20°C, the reaction was stopped by adding 10 l of stop solution, containing 0.6% HCl, 1 mM ATP, and 1% bovine serum albumin. Portions (30 l) of reaction mixtures were spotted on P-81 paper (Whatman), which was then washed three times with 0.5% phosphoric acid and once with acetone, and the associated radioactivity was determined by a liquid scintillation counter.
Measurement of Binding Interactions by SPR-A Biacore 2000 instrument (Biacore, Uppsala, Sweden) was used to measure the interaction between phosphotyrosyl peptides and SH2 domains of SHP-2. In this instrument, the four detection sites are placed on a sensor surface. Immobilization of ligand and all analyses were performed with multichannel detection at a flow rate of 10 l/min with PBS as eluent. Immobilization of phosphotyrosyl peptides to the CM-5 surface was performed essentially as described (40,41). Each phosphopeptide (1 mg/ml) dissolved in 1 M NaCl was injected for 7 min onto activated sites 1 and 3, and nonphosphorylated peptides (blank controls) were injected onto sites 2 and 4. All peptide-coupled sites were blocked with 1 M ethanolamine for 7 min to deactivate the remaining active groups. Various concentrations of GST fusions of the NH 2 -terminal or COOHterminal SH2 domains of SHP-2 dissolved in PBS were injected simultaneously over all four sites of the sensor surface. After binding and elution, 0.1 M NaOH (two pulses of 1-min duration) was injected to regenerate the surface for another round of binding.
The sensorgram for each nonphosphorylated peptide was subtracted from that for the corresponding phosphotyrosyl peptide with BIAevaluation 2.1 software to obtain the sensorgram for the specific interaction. The association constant (K a ) was calculated from the equilibrium binding (R eq ) value by Scatchard analysis (42). After subtraction of the nonspecific SPR response due to changes in bulk refractive index (determined by injection over a blank surface), R eq /C (where C is the concentration of injected peptide) was plotted against R eq . The K a and R max (maximal binding) values were calculated from the slope and intercept, respectively, by linear least-squares curve-fitting based on the following equation,

Effects of Overexpression of SHPS-1 on Activation of MAP
Kinase in Response to Insulin-We generated CHO-IR cell lines expressing either wild-type rat SHPS-1 (SHPS-1-WT) or a mutant SHPS-1 in which all four tyrosine residues in the cytoplasmic domain were replaced by phenylalanine (SHPS-1-4F). This mutant SHPS-1 protein should not be able to bind SHP-2 in response to insulin. We also prepared the following three different antibodies to SHPS-1 (see "Experimental Procedures"): (i) the mAb 4C6 (32), which specifically reacts with endogenous SHPS-1 in CHO cells but not with rat SHPS-1; (ii) the mAb 2F34, which reacts with exogenous rat SHPS-1 but not with endogenous SHPS-1 in CHO cells; thus, we were able to distinguish endogenous and exogenous SHPS-1 in CHO-IR cells; and (iii) polyclonal antibodies that recognize both endogenous SHPS-1 and exogenous rat SHPS-1.
Immunoblot analysis of cell lysates with the polyclonal antibodies to SHPS-1 revealed that the amount of endogenous SHPS-1 in the parent CHO-IR cells was virtually undetectable (Fig. 1A, upper panel), only becoming apparent after long exposure times (data not shown). The same antibodies showed that the amount of SHPS-1-4F appeared similar to that of SHPS-1-WT in their respective cells; the amounts of both were Ͼ10 times that of endogenous SHPS-1 in the parental cells. Each cell line was then stimulated with 100 nM insulin for 5 min, lysed, and subjected to immunoblot analysis with or without prior immunoprecipitation. The extents of overall tyrosine phosphorylation of the IR and IRS-1 were similar among the three cell lines (Fig. 1A, upper panel). Immunoprecipitation with antibodies to SHP-2 revealed that the extents of tyrosine phosphorylation of IRS-1 associated with SHP-2 in response to insulin were similar in CHO-IR and SHPS-1-WT cells but slightly increased in SHPS-1-4F cells (Fig. 1A, lower panel). Immunoprecipitation with mAb 4C6 revealed that the extent of tyrosine phosphorylation of endogenous SHPS-1 and the amount of SHP-2 bound to SHPS-1 in response to insulin were similar in CHO-IR and SHPS-1-4F cells but slightly decreased in SHPS-1-WT cells (Fig. 1B). The same filter was reprobed with polyclonal antibodies to SHPS-1 to confirm that similar amounts of endogenous SHPS-1 were present in each lane (data not shown). The mAb 2F34 immunoprecipitated SHPS-1-WT and SHPS-1-4F but not endogenous SHPS-1 (Fig. 1C). In addition, the wild-type SHPS-1, but not the mutant protein, was phosphorylated on tyrosine residues and formed a complex with SHP-2 in response to insulin (Fig. 1C). Approximately 0.2% of total SHPS-1 was estimated to form a complex with SHP-2 in response to insulin in SHPS-1-WT cells (Fig. 1D, lane  2 versus 7).
Given that SHPS-1 is a transmembrane protein, it would be expected that overexpression of wild-type SHPS-1 would increase the amount of SHP-2 that associates with the plasma membrane in response to insulin. We thus prepared solubilized membrane fractions and cytosolic fractions from each of the three CHO-IR cell lines after incubation in the absence or presence of insulin. Immunoblot analysis of the membrane fractions with antibodies to SHP-2 revealed that insulin increased the amount of SHP-2 associated with these fractions in all cell lines (Fig. 2). In addition, the amount of SHP-2 associated with the membrane fraction of stimulated cells was markedly increased in SHPS-1-WT cells but not in SHPS-1-4F cells, compared with that for parental cells. In contrast, the amount of SHP-2 in cytosolic fractions was not significantly altered by insulin stimulation as compared with that for unstimulated cells in three cell lines.
SHP-2 has been suggested to mediate RAS (21) and MAP kinase (21-24) activation in response to insulin, on the basis of the observation that a dominant negative form of the protein blocked these effects of insulin. Therefore, we next determined the effect of overexpression of SHPS-1 on insulin-induced activation of MAP kinase, as determined by immunoblot analysis of cell lysates with antibodies specific for the tyrosine-phosphorylated enzyme. Insulin induced activation of MAP kinase in a concentration-dependent manner. Activation was detectable at 1 nM insulin and was half-maximal at 3 nM in parental CHO-IR cells (Fig. 3, A and B). Expression of a catalytically inactive SHP-2 markedly inhibited insulin activation of MAP kinase, as described previously (21). Expression of wild-type SHPS-1 significantly increased MAP kinase activation in response to insulin at all concentrations tested (Fig. 3, A and B). In contrast, insulin-induced MAP kinase activation in SHPS-1-4F cells was similar to that in parental CHO-IR cells. We also determined MAP kinase activation by using in vitro kinase assay (Fig. 3C). The in vitro kinase assays seems to be more sensitive than the immunoblot analysis in terms of determination of MAP kinase activation induced by lower concentrations of insulin, since we observed the detectable response with 0.1 nM insulin by the in vitro kinase assay. Expression of wild-type SHPS-1 significantly increased MAP kinase activation in response to insulin at lower concentrations of insulin (0.1-1 nM), whereas it did not significantly affect MAP kinase activation in response to 100 nM insulin (Fig. 3C).
During the course of this study, Kharitonenkov et al. (35) showed that expression of SIRP␣1, a human homolog of SHPS-1, inhibits insulin-or EGF-induced activation of MAP kinase in NIH 3T3 cells. We thus tested the effect of expression of wild-type SHPS-1 on insulin activation of MAP kinase in these cells. Expression of SHPS-1-WT markedly enhanced the activation of MAP kinase in both insulin-stimulated and unstimulated NIH 3T3 cells (Fig. 4A). We also generated Rat-1-IR cells overexpressing wild-type SHPS-1 and showed that insulin-induced activation of MAP kinase was again markedly increased in these cells compared with that apparent in control Rat-1-IR cells (Fig. 4B). The overexpression of SHPS-1-WT showed greater effects on MAP kinase activation in response to lower concentrations of insulin as compared with its effects on maximal MAP kinase activation by insulin in either NIH 3T3 cells or Rat-1-IR cells. The extent of enhancement of insulininduced MAP kinase activation by overexpression of SHPS-1

FIG. 1. Characterization of CHO-IR cells that overexpress SHPS-1-WT or SHPS-1-4F. CHO-IR cells, SHPS-1-WT cells, or SHPS-1-4F
cells were incubated for 5 min in the absence or presence of 100 nM insulin as indicated. Cell lysates were then prepared, and 30 g of each sample was subjected to immunoblot analysis with HRP-conjugated PY20 (␣PY) or polyclonal antibodies to SHPS-1 (␣SHPS-1) (A, upper panel). Alternatively, cell lysates (1.5 mg/each sample) were subjected to immunoprecipitation (IP) with polyclonal antibodies to SHP-2 (A, lower panel), with mAb 4C6 specific for CHO cell SHPS-1 (B) or with mAb 2F34 specific for rat SHPS-1 (C). The immunoprecipitates were then subjected to immunoblot analysis with HRP-conjugated PY20 or with polyclonal antibodies to SHP-2 (␣SHP-2). In A, lower panel and C, a duplicate filter was also probed with polyclonal antibodies to SHP-2 or SHPS-1 to reveal the amount of SHP-2 or SHPS-1 immunoprecipitated from each cell line. SHPS-1-WT cells in a 10-cm plate were incubated for 5 min in the absence or presence of 100 nM insulin as indicated (D). Cell lysates prepared were subjected to immunoprecipitation with polyclonal antibodies to SHP-2 (lane 1 and 2). The immunoprecipitates were then subjected to immunoblot analysis with polyclonal antibodies to SHPS-1. The lanes for whole cell lysate (lanes 3-8) contain samples from 5% (lane 3), 2% (lane 4), 1% (lane 5), 0.5% (lane 6), 0.2% (lane 7), or 0.1% (lane 8) of a 10-cm plate and subjected to immunoblot analysis with polyclonal antibodies to SHPS-1. The resulting supernatants from first immunoprecipitation were then subjected to a second round of immunoprecipitation with polyclonal antibodies to SHP-2. When the second immunoprecipitate was then subjected to immunoblot analysis with polyclonal antibodies to SHPS-1, the SHPS-1 complexed with SHP-2 was virtually undetectable (data not shown). The positions of IR, IRS-1, SHPS-1, and SHP-2 are indicated by arrowheads, and those of molecular size standards are indicated in kilodaltons (kDa). appeared to be greater in NIH 3T3 cells or Rat-1-IR cells as compared with that observed in CHO-IR cells. This could be partly due to either the differences of expression levels of endogenous SHPS-1 or just species difference between three cell lines.

Effects of Mutation of Tyrosine Residues of SHPS-1 on Its Tyrosine Phosphorylation and Interaction with SHP-2 in
Response to Insulin-Four tyrosine residues (Tyr 408 , Tyr 432 , Tyr 449 , and Tyr 473 ) in the cytoplasmic region of SHPS-1 represent potential phosphorylation sites and binding sites for the SH2 domains of SHP-2 (31,34). To determine which of these tyrosine residues are phosphorylated in response to insulin and mediate the subsequent binding of SHP-2, we generated four additional CHO-IR cell lines expressing mutant SHPS-1 proteins in which individual tyrosine residues were replaced by phenylalanine (SHPS-1-Y408F, SHPS-1-Y432F, SHPS-1-Y449F, and SHPS-1-Y473F cells). The extent of tyrosine phosphorylation of IR and IRS-1 in response to insulin in all of these cell lines was similar to those in parental CHO-IR cells (data not shown). The various cell lines were then incubated for 5 min in the absence or presence of 100 nM insulin, after which lysates were prepared and subjected to immunoprecipitation with mAb 2F34 (Fig. 5). Immunoblot analysis of the resulting immunoprecipitates with antibodies to phosphotyrosine revealed that the extent of insulin-induced tyrosine phosphorylation of recombinant SHPS-1 was markedly reduced in SHPS-1-Y449F and SHPS-1-Y473F cells but not in SHPS-1-Y408F and SHPS-1-Y432F cells, compared with that in SHPS-1-WT cells. A duplicate filter was probed with polyclonal antibodies to SHPS-1 to confirm that similar amounts of SHPS-1 were present in each lane. Longer exposures revealed a low level (substantially less than 50% of that in SHPS-1-WT cells) of insulin-induced tyrosine phosphorylation of exogenous SHPS-1 in both SHPS-1-Y449F and SHPS-1-Y473F cells (data not shown). Furthermore, the binding of SHP-2 to SHPS-1 was almost totally abolished in both SHPS-1-Y449F and SHPS-1-Y473F cells. These results suggest that both Tyr 449 and Tyr 473 of SHPS-1 are required for the optimal tyrosine phosphorylation of this protein and its interaction with SHP-2 in response to insulin.

FIG. 2. Effects of insulin on the association of SHP-2 with the membrane fraction of CHO-IR cell lines.
CHO-IR, SHPS-1-WT, or SHPS-1-4F cells in a 10-cm plate were incubated for 5 min in the absence or presence of 100 nM insulin as indicated. The solubilized membrane fraction (Plasma membrane) and the cytosolic fraction (Cytosol) were then prepared as described under "Experimental Procedures," and 20 l of each fraction was subjected to immunoblot analysis with polyclonal antibodies to SHP-2.

FIG. 3. Effects of overexpression of SHPS-1 on insulin-induced activation of MAP kinase in CHO-IR cells.
A, CHO-IR, SHP-2-C/S, SHPS-1-WT, or SHPS-1-4F cells were incubated for 5 min with the indicated concentrations of insulin, after which cell lysates were prepared and subjected to immunoblot analysis with antibodies specific for tyrosine-phosphorylated MAP kinase (␣pMAPK). The same blot was also probed with ␣91 polyclonal antibodies to p44 and p42 MAP kinase to ensure that the same amount of MAP kinase was present in each lane. B, the extent of tyrosine phosphorylation of p42 MAP kinase in A was quantified by scanning densitometry with the NIH image program. Data are expressed as a percentage of the value for parental CHO-IR cells exposed to 100 nM insulin and are means Ϯ S.E. of three separate experiments. *, p Ͻ 0.05 versus the corresponding value for parental CHO-IR cells determined by analysis of variance. C, CHO-IR (open column) or SHPS-1-WT (closed column) cells were incubated for 5 min with the indicated concentrations of insulin, after which cell lysates were prepared and subjected to immunoprecipitation with ␣92 antibodies to MAP kinase, and MAP kinase activity in immunoprecipitates was assayed with myelin basic protein as substrate. Data are means of duplicate determinations and are representative of three separate experiments.

SHPS-1 Phosphotyrosyl Peptides and SH2 Domains of SHP-2
by SPR-Given that Tyr 449 and Tyr 473 of SHPS-1 appeared to be responsible for insulin-induced binding of the SH2 domains of SHP-2 to SHPS-1, we next determined the K d values for the binding of GST fusion proteins containing the NH 2 -terminal or COOH-terminal SH2 domains of SHP-2 to immobilized SHPS-1-pY449 or SHPS-1-pY473 peptides by SPR analysis. Representative sensorgrams for the binding of the NH 2 -terminal SH2 domain of SHP-2 to immobilized SHPS-1-pY449 and that of the COOH-terminal SH2 domain of SHP-2 to SHPS-1-pY473 are shown in Fig. 6. SHPS-1-pY449 bound to the NH 2 -terminal SH2 domain of SHP-2 with a K d of 145 Ϯ 42 nM (mean Ϯ S.D. of three independent experiments), whereas the K d for the binding of the same phosphopeptide to the COOH-terminal SH2 domain of SHP-2 was 9747 Ϯ 2171 nM (Table I). In contrast, the K d for the interaction of SHPS-1-pY473 with the NH 2 -terminal SH2 domain of SHP-2 could not be calculated because no specific binding was detected, whereas the K d for the interaction of the same peptide with the COOH-terminal SH2 domain was 1011 Ϯ 353 nM.
PTPase Activity of SHP-2 toward SHPS-1 Phosphotyrosyl Peptides-We have previously shown that the extent of both the tyrosine phosphorylation of SHPS-1 and its association with SHP-2 in CHO-IR cells is maximal 1-5 min after insulin stimulation and decreases thereafter (32). In contrast, the extent of tyrosine phosphorylation of SHPS-1 continued to increase for up to 30 min after stimulation of SHP-2-C/S cells, presumably because of the lack of PTPase activity of the mutant SHP-2. Thus, SHP-2 may dephosphorylate one or more phosphotyrosine residues of SHPS-1 after interaction of the two proteins in response to insulin stimulation. We therefore evaluated the PTPase activity of recombinant SHP-2 toward various SHPS-1 phosphotyrosyl peptides in vitro. As described previously (21), IRS-1-pY1172 or IRS-1-pY1222, a phosphoty-rosyl peptide corresponding to the sequence surrounding Tyr 1172 or Tyr 1222 of IRS-1, was an effective substrate for SHP-2 (Fig. 7). All phosphotyrosyl peptides corresponding to the sequences surrounding the four tyrosine residues (Tyr 408 , Tyr 432 , Tyr 449 , and Tyr 473 ) in the cytoplasmic region of SHPS-1 were also effective substrates for SHP-2. In contrast, IR-pY51, which does not correspond to a binding site for any known SH2 domain-containing protein, was a poor substrate.

FIG. 4. Effects of overexpression of wild-type SHPS-1 on insulin-induced MAP kinase activation in NIH 3T3 cells (A) or Rat-1-IR cells (B)
. NIH 3T3 cells that overexpress wild-type SHPS-1 (NIH 3T3-SHPS-1 cells) (A), Rat-1-IR cells that overexpress the same protein (Rat-1-IR-SHPS-1 cells) (B), and the corresponding parental cells were incubated for 5 min with various concentrations of insulin, after which cell lysates were prepared and subjected to immunoblot analysis with antibodies specific for tyrosine-phosphorylated MAP kinase. The same blots were also probed with ␣91 polyclonal antibodies to p44 and p42 MAP kinase.

FIG. 5. Effects of insulin on the tyrosine phosphorylation of SHPS-1 and its association with SHP-2 in various transfected CHO-IR cell lines.
CHO-IR cells that overexpress wild-type or various mutant SHPS-1 proteins were incubated for 5 min in the absence or presence of 100 nM insulin as indicated. Cell lysates were then prepared and subjected to immunoprecipitation with mAb 2F34, and the resulting immunoprecipitates were subjected to immunoblot analysis with HRP-conjugated PY20 or polyclonal antibodies to SHP-2. A duplicate filter was probed with polyclonal antibodies to SHPS-1.

FIG. 6. Representative sensorgrams of the binding of immobilized SHPS-1 phosphotyrosyl peptides to GST fusion proteins containing the NH 2 -or COOH-terminal SH2 domains of SHP-2.
Fusion proteins containing the NH 2 -terminal (A) or COOH-terminal (B) SH2 domains of SHP-2 were exposed at the indicated concentrations to immobilized SHPS-1-pY449 or SHPS-1-pY473, respectively, for 7 min at a flow rate of 10 l/min. RU, resonance unit.

DISCUSSION
We have shown that the expression of wild-type SHPS-1 enhanced insulin-induced activation of MAP kinase in CHO-IR cells. Overexpression of a wild-type SHPS-1 showed greater effects on MAP kinase activation in response to low concentrations of insulin as compared with its effects on maximal MAP kinase activation induced by insulin. SHPS-1 is tyrosine-phosphorylated and associated with SHP-2 in even unstimulated cells, possibly because of the effect of cell adhesion (31). In addition, expression of the wild-type protein also increased the amount of SHP-2 that bound to SHPS-1 in response to insulin, an effect that also resulted in an increased amount of SHP-2 associated with the particulate fraction. In contrast, the expression of SHPS-1-4F, which neither underwent tyrosine phosphorylation nor bound to SHP-2 in response to insulin, did not affect insulin activation of MAP kinase. Thus, the increase in the amount of SHP-2 associated with SHPS-1 may contribute to the enhancement of insulin-induced MAP kinase activation in SHPS-1-WT cells. These results are consistent with the notion that SHP-2 mediates insulin-induced activation of RAS and MAP kinase (21)(22)(23)(24), although the precise mechanism by which SHP-2 activates RAS in response to insulin remains unclear. The expression of a membrane-targeted form of Corkscrew, the Drosophila homolog of SHP-2, results in the bypassing of Sevenless tyrosine kinase function in development of the R7 photoreceptor, suggesting that recruitment of Corkscrew to the plasma membrane is crucial for its function (12). Similarly, the overexpression of wild-type SHPS-1 increases the amount of SHP-2 recruited to a region near the plasma membrane, the site of RAS localization, in response to insulin. In addition, a phosphotyrosyl peptide corresponding to the sequence sur-rounding both Tyr 449 and Tyr 473 of SHPS-1 (BIT) stimulated the PTPase activity of recombinant SHP-2 in vitro (33), suggesting a possibility that overexpression of SHPS-1 may enhance insulin-stimulated MAP kinase activation by increasing both the recruited amount and PTPase activity of SHP-2.
During the course of the present study, the expression of SIRP␣1, a human homolog of SHPS-1, was shown to inhibit the insulin-or EGF-induced activation of MAP kinase in NIH 3T3 cells (35). However, in our study, expression of SHPS-1-WT also enhanced insulin activation of MAP kinase in both NIH 3T3 and Rat-1-IR cells. There are at least two possible explanations for this discrepancy. First, it might be attributable to the species difference between rat SHPS-1 and human SIRP␣1, although the amino acid sequences of the two proteins are 65% identical (91% similar) (34). Second, the level of expression of SIRP␣1 may be substantially higher than that of SHPS-1; it is possible that SIRP␣1-mediated recruitment of excessive amounts of SHP-2 to a site near the plasma membrane may result in the down-regulation of other signaling molecules required for insulin-induced activation of MAP kinase. In contrast to the positive role of SHP-2 in insulin signaling, SHP-2 bound to CTLA4 appears to inhibit TCR-mediated activation of RAS and MAP kinase (43), suggesting that SHP-2 may exert a negative effect on these events under certain circumstances. Nevertheless, given that SHP-2 positively regulates insulininduced RAS-MAP kinase activation (44), we believe that formation of the SHPS-1⅐SHP-2 complex also regulates insulininduced activation of RAS and MAP kinase in a positive manner.
SHP-2 binds to IRS-1 in response to insulin, and Tyr 1172 and Tyr 1222 of IRS-1 appear to be responsible for the association with SHP-2 (14,18,19,45). The sequences surrounding Tyr 408 and Tyr 449 of SHPS-1 are similar to that surrounding Tyr 1172 of IRS-1, whereas the sequences surrounding Tyr 432 and Tyr 473 of SHPS-1 resemble that surrounding Tyr 1222 of IRS-1 (31). However, mutational analysis of tyrosine residues in the cytoplasmic domain of SHPS-1 indicated that Tyr 449 and Tyr 473 of SHPS-1 are the major sites of tyrosine phosphorylation induced by insulin. Mutation of either Tyr 408 or Tyr 432 did not affect the extent of insulin-induced tyrosine phosphorylation of SHPS-1. The observation that the extent of insulin-induced tyrosine phosphorylation of the Y449F or Y473F mutants of SHPS-1 was substantially less than half that of the wild-type protein suggests that phosphorylation of either Tyr 449 or Tyr 473 may enhance phosphorylation of the other tyrosine residue. We also showed that mutation of either Tyr 449 or Tyr 473 of SHPS-1 completely abolished the insulin-induced binding of SHP-2, suggesting that simultaneous phosphorylation of both Tyr 449 and Tyr 473 is required for the optimal binding of the SH2 domains of SHP-2. Similarly, mutation of either Tyr 1172 or Tyr 1222 of IRS-1 is sufficient to block its insulin-induced association with SHP-2 (45  respectively, in response to insulin (19), consistent with the present results.
Ohnishi et al. (33) have recently shown that a phosphotyrosyl peptide containing the sequence surrounding Tyr 449 , but not one containing the sequence around Tyr 473 , of SHPS-1 (BIT) induced a 4 -5-fold increase in the PTPase activity of recombinant SHP-2 in vitro. Furthermore, a phosphotyrosyl peptide containing the sequence surrounding both Tyr 449 and Tyr 473 of SHPS-1 increased SHP-2 activity 33-fold, indicating that occupancy of both SH2 domains of SHP-2 by Tyr 449 and Tyr 473 of SHPS-1 is required for full activation of SHP-2. This notion is compatible with the observation by Pluskey et al. (18) that the interaction of SHP-2 with a phosphotyrosyl peptide containing the sequence surrounding both Tyr 1172 and Tyr 1222 of IRS-1, but not that with phosphopeptides corresponding to the sequences surrounding only one of these tyrosine residues, induces full activation of the catalytic activity of SHP-2. SHP-2 has been suggested to bind to tyrosine-phosphorylated SHPS-1 in response to insulin and to dephosphorylate SHPS-1 in vivo, given that the extent of tyrosine phosphorylation of SHPS-1 was greatly increased in cells overexpressing a catalytically inactive form of SHP-2 (31,32). We have now shown that phosphopeptides corresponding to the sequence surrounding either Tyr 449 or Tyr 473 of SHPS-1 are effective substrates for SHP-2 in vitro. We therefore propose the following model for CHO-IR cells: insulin rapidly induces the phosphorylation of Tyr 449 and Tyr 473 of SHPS-1, after which SHP-2 binds to these phosphorylated residues of SHPS-1 through its NH 2 -terminal and COOH-terminal SH2 domains, respectively. As a result of its association with SHPS-1, SHP-2 becomes activated and dephosphorylates the phosphotyrosine residues to which it binds. SHP-2 then dissociates from SHPS-1 and activates RAS and MAP kinase by an as yet unidentified mechanism. Thus, SHPS-1 appears to play a crucial role both in the recruitment of SHP-2 from the cytosol to a site near the plasma membrane and in increasing its catalytic activity in response to insulin stimulation.
There are increasing numbers of SHPS-1-like transmembrane proteins that possess multiple Ig-like domains in their extracellular regions but no catalytic domains in their cytoplasmic regions and bind to an SH2 domain-containing PTPase. One such protein is platelet endothelial adhesion molecule-1 (PECAM-1), which is expressed on the surface of endothelial cells, leukocytes, and circulating platelets and plays an important role in the cascade of cell adhesion that results in the extravasation and migration of leukocytes (46). PECAM-1 contains six Ig-like domains in its extracellular region and two potential sites for tyrosine phosphorylation and binding of SHP-2 SH2 domains in its cytoplasmic portion. It undergoes tyrosine phosphorylation during platelet aggregation and subsequently binds to SHP-2 (47). Another example of an SHPS-1-like protein is biliary glycoprotein, or BGP (also known as CD66a, pp120/HA4, and C-CAM1), which is a member of the carcinoembryonic antigen family and possesses Ig-like structures in its extracellular region (48,49). Although its physiological function is not clear, BGP might act as an intercellular adhesion molecule or a bile salt transporter (50). BGP, which contains two tyrosine residues in its cytoplasmic region, is tyrosine-phosphorylated by SRC kinase or the IR kinase (51,52), and it binds SHP-1, another SH2 domain-containing PTPase (1)(2)(3), in response to treatment of cells with pervanadate (50). Mutation of either of the two tyrosine residues in the cytoplasmic portion of BGP results in a marked decrease in the extent of its tyrosine phosphorylation and the complete loss of its binding to SHP-1, consistent with our data on the insulininduced interaction between SHPS-1 and SHP-2.
The sequences surrounding the tyrosine residues in the cytoplasmic regions of these SHPS-1-like proteins, including SHPS-1, are related to those of immune receptor tyrosinebased activation motifs (ITAMs), which were originally described in the TCR and BCR (53,54). The phosphorylation of two tyrosine residues present in this motif by a tyrosine kinase, such as an SRC family kinase, creates a binding site for SH2 domains of signal molecules such as ZAP70 or SYK and thereby initiates TCR-or BCR-mediated signaling. ITAMs contain single or multiple sets of two tyrosine residues, which are usually separated by only 10 or 11 amino acids, in contrast to the spacing between the two tyrosine residues in SHPS-1 (23 amino acids), PECAM-1 (22 amino acids), or BGP (26 amino acids). Other Ig-like receptor proteins, such as Fc␥RIIB, CD22, and CTLA4, bind to SHP-1 or SHP-2 as a result of the interaction between their tyrosine-phosphorylated cytoplasmic tails and the SH2 domains of the SHP proteins (43,55,56), although these complexes inhibit BCR-or TCR-mediated signaling. Thus, SHPS-1-like membrane proteins may represent a subfamily of the Ig superfamily in terms of specific function; they generally function as docking proteins that undergo tyrosine phosphorylation by a receptor tyrosine kinase or an SRC family kinase and recruit SHP proteins in response to a variety of extracellular stimuli, thereby activating SHP PTPase activity.