Expression of a Dominant Negative SHP-2 in Transgenic Mice Induces Insulin Resistance*

To elucidate the roles ofSHP-2, we generated transgenic (Tg) mice expressing a dominant negative mutant lacking protein tyrosine phosphatase domain (ΔPTP). On examining two lines of Tg mice identified by Southern blot, the transgene product was expressed in skeletal muscle, liver, and adipose tissues, and insulin-induced association of insulin receptor substrate 1 with endogenous SHP-2 was inhibited, confirming that ΔPTP has a dominant negative property. The intraperitoneal glucose loading test demonstrated an increase in blood glucose levels in Tg mice. Plasma insulin levels in Tg mice after 4 h fasting were 3 times greater with comparable blood glucose levels. To estimate insulin sensitivity by a constant glucose, insulin, and somatostatin infusion, steady state blood glucose levels were higher, suggesting the presence of insulin resistance. Furthermore, we observed the impairment of insulin-stimulated glucose uptake in muscle and adipocytes in the presence of physiological concentrations of insulin. Moreover, tyrosine phosphorylation of insulin receptor substrate-1 and stimulation of phosphatidylinositol 3-kinase and Akt kinase activities by insulin were attenuated in muscle and liver. These results indicate that the inhibition of endogenous SHP-2function by the overexpression of a dominant negative mutant may lead to impaired insulin sensitivity of glucose metabolism, and thusSHP-2 may function to modulate insulin signaling in target tissues.

SHP-2 (also referred to as PTP1D, PTP2C, SHPTP2, or SYP) is a ubiquitously expressed protein-tyrosine phosphatase (PTPase) 1 containing a single PTPase domain and two adjacent Src homology (SH) 2 domains near its N terminus which specifically associate with a variety of tyrosine-phosphorylated proteins upon growth factor stimulation (1)(2)(3)(4)(5). SHP-2 is the mammalian homologue of Drosophila Corkscrew, whose gene product potentiates the Drosophila homologue of mammalian c-raf to positively transmit signals downstream of the Torso receptor tyrosine kinase (6). Furthermore, SHP-2 has been reported to play an important role in mesodermal induction in oocyte by regulation of mitogen-activated protein (MAP) kinase activity (7).
Regarding the roles of SHP-2 in tyrosine kinase signaling, several lines of evidence indicate that SHP-2 acts as a positive mediator in growth factor signaling such as that by plateletderived growth factor and epidermal growth factor (4,5,8,9). After stimulation by these ligands, SHP-2 is tyrosine-phosphorylated and bound to Grb2-SOS complex, resulting in activation of p21 ras and MAP kinase cascade. On the other hand, in the case of insulin signaling, SHP-2 is not tyrosine-phosphorylated in response to insulin stimulation. However, insulin induces the association of IRS-1 with SHP-2 (10,11), and the expression of either a catalytically inactive mutant SHP-2 (Cys/Ser) or a deletion mutant lacking PTPase domain in Chinese hamster ovary cells overexpressing insulin receptors (CHO-IR) results in the attenuation of the insulin-stimulated MAP kinase activity, suggesting that SHP-2 is able to potentiate MAP kinase cascade even in the absence of its tyrosine phosphorylation in those cells (12,13). In contrast to these studies in CHO-IR and NIH3T3 cells overexpressing insulin receptors (12)(13)(14), we found that the introduction of a dominant negative ⌬PTP mutant, which lacks PTPase domain, into Rat-1 fibroblasts overexpressing human insulin receptors (HIRc) attenuated insulin-stimulated phosphatidylinositol (PI) 3-kinase as well as the impaired MAP kinase activity by decreasing the phosphorylation state of IRS-1 (15). Furthermore, Kharitonenkov et al. (16) have reported that overexpression of SHP-2 and IRS-1 in baby hamster kidney cells expressing insulin receptor leads to an increased association of IRS-1 with insulin receptor, resulting in increased insulin-stimulated 2-deoxyglucose uptake. Thus, they speculate that SHP-2 potentiates interaction of IRS-1 with insulin receptors as an adapter molecule. In contrast, a recent study reported that in 32D cells expressing a mutant IRS-1 lacking SHP-2-binding sites, the insulin-stimulated IRS-1 phosphorylation is enhanced and results in the potentiation of insulin-stimulated protein synthesis, suggesting that SHP-2 attenuates IRS-1 phosphorylation and modulates metabolic response of insulin in 32D cells (17). Thus, these different effects of SHP-2 on regulation of IRS-1 phosphorylation are dependent on cell types used for the experiments (CHO, NIH3T3, Rat 1, baby hamster kidney, and 32D cells). Thus, we tried to make transgenic mice expressing our dominant negative mutant SHP-2 (⌬PTP) to clarify the physiological roles of SHP-2 in the in vivo insulin action to regulate glucose utilization, especially in its regulation of insulin signaling in skeletal muscle.
Expression Constructs-The eukaryotic expression vector, pCAGGS was a gift from Dr. J. Miyazaki (Osaka University). This vector consists of a strong promoter based upon that of chicken ␤-actin (18). The transgene was under the control of chicken ␤-actin promoter, CMV-IE enhancer, and rabbit ␤-globulin poly(A) signal. For the ⌬PTP transgene, cDNA encoding a pair of SH2 domains of SHP-2 (amino acid 1-216) was ligated into the EcoRI sites in the expression vector pCAGGS as described (15). Linear DNA of ⌬PTP vector was made by double digestion of ScaI and BamHI and microinjected into the fertilized eggs from BDF1 female mice by standard procedures (19). The construct used for transgenic generation is outlined in Fig. 1A.
Screening of Expression of Transgene Product by Southern, Northern, and Western Blotting-Tail biopsies were taken 4 weeks after birth at weaning for isolation of genomic DNA. To check the integrity of transgene, genomic DNA from tail was digested by 2 sets of restriction enzymes (SalI and ScaI or HincII and ScaI) and hybridized with a vector probe (probe A) or cDNA probe (probe B), respectively, as shown in Fig. 1A. Since probe B interacted with three bands in our Southern blotting, one was transgene (560 bp) and the others were 1200 and 800 bp, there may be endogenous SHP-2 (Syp) gene as shown in Fig. 1B. Thus, we determined the copy number of each line by calculating the ratio of intensity (560:1200 bp). For Northern blotting, 10 g of total RNA was used for assay (20). For Western blotting, various organs were homogenized with a motor-driven homogenizer in ice-cold lysis buffer containing 20 mM Tris-HCl (pH 7.5), 50 mM sodium pyrophosphate, 50 mM sodium fluoride, 1 mM EDTA, 140 mM NaCl, 1% Nonidet P-40, 1 mM sodium orthovanadate, 1 mM PMSF, 50 M aprotinin, 5 g/ml leupeptin, and 2 mM benzamidine. After centrifugation at 15,000 rpm at 4°C for 20 min, the supernatant (75 g of protein) was resolved on 12.5% SDS-polyacrylamide gel, electrotransferred to an Immobilon P membrane (Millipore, Bedford, MA), and blotted with anti-PTP1D antibody. Bound antibodies were detected with horseradish peroxidase-conjugated anti-IgG and visualized with an enhanced chemiluminescence detection system (ECL, Amersham Pharmacia Biotech).
All mice used in this study were F2 and F3 siblings. F2 and F3 mice were kept in sterile microisolators and were observed closely throughout the experiment. To avoid the effect of gender, we used male mice for the following studies. For measuring the growth rate, four mice of either genotype were monitored as to their body weight once every week, starting 4 weeks after birth.
To confirm whether ⌬PTP could reveal a dominant negative property in terms of inhibition of IRS-1 association with endogenous SHP-2 (Syp), we assessed insulin-induced IRS-1 association with SHP-2 as follows. Four units of human insulin was injected by inferior vena cava into anesthetized mice according to the modified method of Araki et al. (21). Two or 5 min after insulin injection, liver and hindlimb muscle were removed and immediately frozen in liquid nitrogen, respectively. The tissue was homogenized in lysis buffer as described. Homogenate was allowed to be solubilized for 1 h at 4°C before centrifugation at 15,000 rpm for 20 min. The supernatant (3-4 mg of protein) was incubated with anti-IRS-1 antibody for 3 h and then with protein G-Sepharose for a further 2 h. The bound proteins were resolved by SDS-polyacrylamide gel electrophoresis, transferred onto polyvinylidene difluoride membrane by electroblotting, and then immunoblotted with anti-PTP1D antibody.
Metabolic Characteristics of Tg Mice-Blood glucose and plasma insulin levels of 12-16-week-old mice were measured between 10:00 and 12:00 a.m. after 4 h fasting under anesthetized conditions with sodium pentobarbital. Blood glucose levels were determined by the glucose oxidation method, and plasma insulin levels were determined using rat insulin enzyme immunoassay kit (Morinaga, Tokyo, Japan). Glucose loading test was performed as follows. After overnight fasting, mice received 3 mg/g glucose solution intraperitoneally, and blood samples were obtained from the tail vein at 0, 30, 60, and 120 min after glucose loading.
Assessment of in Vivo Insulin Sensitivity-To assess in vivo insulin sensitivity, we performed the insulin sensitivity test using constant glucose, insulin, and somatostatin infusion (22). This test proposes that endogenous insulin secretion is suppressed with somatostatin coupled with a fixed constant glucose and insulin infusion. At a high concentration of insulin, glucose production of the liver was reported to be negligible, and the insulin-stimulated glucose utilization appeared to be ascribed to glucose disposal in the peripheral tissue, mainly in the skeletal muscle. Briefly, after overnight fasting, male mice were anesthetized by intraperitoneal injection of sodium pentobarbital, and the right jugular vein was exposed and cannulated with a polyethylene tube for administration of the infusate. After a 30-min infusion with saline, mice were administered the infusate containing glucose (1.125 g/kg/h), insulin (1.0 units/kg/h), and somatostatin (100 g/kg/h) at a constant flow rate of 0.3 ml/h for 120 min. Blood samples were obtained from the tail vein at 0, 30, 60, 90, and 120 min after the start of the infusion of mixed solution. Plasma insulin levels were measured at 120 min. The mean of the last three samples was used for the estimation of steady state blood glucose (SSBG) and steady state plasma insulin levels at 120 min.
Activation of Glycogen Synthase during Insulin Sensitivity Test-At the indicated time points, the hindlimb muscle from the Tg mice was removed, frozen, and kept in liquid nitrogen to be used for assays. The muscle was homogenized in glycylglycine buffer containing 25 mM NaF. The glycogen synthase activity was measured according to the modified Thomas' filter paper method (23), and the data were expressed as percent I Ϫ form (Glc-6-PϪ/Glc-6-Pϩ).
Assessment of Glucose Uptake into Isolated Soleus Muscle-Glucose transport activity was assessed by the measurement of 2-deoxyglucose uptake as described (22). In brief, soleus muscle was separated from hindlimb and incubated with insulin (0 -10 nM) at 25°C in 2 ml of Krebs-Ringer phosphate (KRP) buffer containing 10 mM HEPES and 11 mM glucose for 120 min. The muscle was washed with glucose-free buffer for 10 min and then incubated with a fresh buffer containing D-2-deoxy-[ 3 H]glucose (1 mM, 1 Ci/2 ml) and L-[ 14 C]glucose (0.5 Ci/2 ml), in the presence or absence of 1 or 10 nM insulin for another 30 min at 25°C, respectively. Specific 2-deoxyglucose uptake was determined by subtracting L-glucose uptake from total 2-deoxyglucose uptake.
Insulin Action in Isolated Adipocytes-Adipocytes were isolated from epididymal fat pads by collagenase digestion with some modification as described previously (24). The fat pads were removed and minced into KRP buffer containing 5 mM D-glucose and 1 mg/ml collagenase (type I, Worthington) and digested (1 h, 37°C) with gentle shaking (200 cycle/ min). The adipocytes were washed three times in KRP buffer (each wash in volume 10 times the cell volume). The adipocyte layer was finally diluted with KRP buffer to 20% (v/v) suspension as estimated by packed cells (lipocrit). Insulin binding to 10 6 cells was measured at 8.3 nM 125 I-labeled at Tyr A14 -insulin (NEN Life Science Products). The number of adipocytes was determined by counting in hemocytometer. Twenty % cell suspension corresponded to 10 6 cells per ml. The uptake of amount of trace D-[U-14 C]glucose by isolated adipocytes was measured as described previously with some modifications (24). Adipocytes were incubated in 400 ml of KRP buffer containing 2% albumin in the presence of various concentrations of insulin and trace (300 nM) amounts of 0.1 mCi D-[U-14 C]glucose. The cell suspension was incubated at 37°C for 1 h with continuous shaking at 40 cycle/min. The assay was terminated by centrifugation of a 325-ml aliquot on the top of 100 l of silicon oil in a 500-l microtube in a centrifuge for 1 min. The adipocytes remain on the top of the oil layer, and the buffer was below the oil. The tube was cut just below the adipocyte layer, which was transferred into 5 ml of nonaqueous scintillation fluid, and the radioactivity was measured.
Assessment of Alteration of Insulin Signaling in the Skeletal Muscle and Liver-Two or 5 min after a bolus injection of insulin, liver and hindlimb muscle were removed and immediately frozen in liquid nitrogen, respectively. After the standard sample preparation, the resultant supernatant was incubated with a specific antibody for 3 h and then with protein G-Sepharose for a further 2 h. The bound proteins in the immunoprecipitate or aliquot from the soluble fraction of the lysate were analyzed by Western blotting. Effects of bolus injection of insulin on the profile of phosphotyrosine proteins in muscle were analyzed by Western blotting using either phosphotyrosine antibody or anti-IRS-1 antibody.
Measurement of Akt Kinase Activity-Five or 10 min after a bolus injection of insulin, liver and hindlimb muscle were isolated, and Akt kinase activity was assayed in vitro using histone H2B as a substrate. Briefly, tissue was homogenized and lysed, and a sample for kinase assay was prepared as for the MAP kinase assay. The supernatant (1.5 mg of protein) was incubated with anti-Akt1 antibody for 2 h and then for another 1 h with protein G-Sepharose at 4°C. The assay was conducted in a final volume of 20 l containing 1 M protein kinase inhibitor, 50 M ATP, 10 mM dithiothreitol, 2 Ci of [␥-32 P]ATP, and 10 g of histone H2B at 30°C for 15 min. After stopping the reaction by addition of 10 l of 8 N HCl, a 20-l aliquot was spotted onto P81 phosphocellulose paper. The paper was washed, and phosphorylation was quantified by Cerenkov counting.
Measurement of MAP Kinase Activity-Activated MAP was analyzed by Western blotting using anti-phospho-MAP kinase and Erk2 antibodies during insulin sensitivity test. We also measured the effects on bolus injection of insulin on MAP kinase activity in vitro using myelin basic protein (MBP) as a substrate in muscle and liver as described (15,20). Briefly, the tissue was homogenized and lysed with 25 mM Tris-HCl (pH 7.4) containing 25 mM NaCl, 80 mM ␤-glycerophosphate, 1 mM sodium orthovanadate, 10 mM NaF, 10 mM sodium pyrophosphate, 1 mM EGTA, 1 mM PMSF, and 10 g/ml leupeptin. After brief sonication and centrifugation, 10 l of the obtained supernatant was assayed for kinase activity. The assay was conducted in a final volume of 40 l containing 1 M protein kinase inhibitor, 50 M ATP, 2 Ci of [␥-32 P]ATP, and 20 g of MBP at 30°C for 15 min. A 25-l aliquot was spotted onto P81 phosphocellulose paper, 2.3 cm in diameter (Whatman). The paper was washed with 180 mM phosphoric acid and rinsed with acetone. Phosphorylation was quantified by Cerenkov counting.
Statistics-The data are expressed as the mean Ϯ S.E., unless otherwise stated. Scheffe's multiple comparison test was used to determine the significance of any differences among more than two groups, and the unpaired Student's t test was used to determine the significance of any differences between two groups. p Ͻ 0.05 was considered significant.

RESULTS
Established Transgenic Lines-Sixteen founder transgenic mice were successfully identified by Southern blot analysis of tail DNA. Two lines (S6 and S161), which had high copy number, were selected for further study. The transgene expression was detected by Western blot analysis as shown in Fig. 1C. As expected, transgene was expressed in all tissues examined including skeletal muscle, liver, and adipose tissue, three key tissues involved in glucose homeostasis, matching precisely the pattern of the chicken ␤-actin promoter. The expression levels of ⌬PTP mutant were 10 times greater than those of endogenous mouse SHP-2 (Syp). All studies were performed in heterozygous mice expressing a ⌬PTP mutant. In skeletal muscle, the association of IRS-1 with SHP-2 was impaired after a bolus injection of insulin when compared with that in non-Tg mice as shown in Fig. 2B (27.7 Ϯ 0.4% of non-Tg mice, n ϭ 3), indicating that ⌬PTP could have a dominant negative property. Similar results were observed in liver (Fig. 2B). In contrast, the insulin-induced association of Shc with Grb2 was not affected in Tg mice as shown in Fig. 2C.
Glucose Intolerance and Insulin Insensitivity-Growth rate and body weight of male Tg mice were comparable with those of non-transgenic littermates (non-Tg). When these Tg mice received 3 mg/g glucose load intraperitoneally to evaluate their glucose tolerance, blood glucose levels after glucose loading were significantly higher that those in non-Tg mice at each time point, whereas fasting blood glucose levels were comparable between the two groups as shown in Fig. 3A. Furthermore, as shown in Table I, the plasma insulin levels 4 h after fasting were 3 times greater (p Ͻ 0.01) in Tg mice than non-Tg littermates in both lines (S6 and S161), suggesting that these Tg mice are insulin-resistant. To assess further the in vivo insulin sensitivity, we performed the insulin sensitivity test using constant glucose, insulin, and somatostatin infusion. As shown in Fig. 3B, blood glucose levels plateaued after 60 min infusion and maintained constant values for another 60 min. As summarized in Table II, the steady state of blood glucose (SSBG) levels in Tg mice are 2 times higher than those in non-Tg mice, even though steady state plasma insulin levels were comparable among these groups. Thus, these data indicate that these Tg mice are insulin-resistant mainly in skeletal muscle.
Impaired Activation of Muscle Glycogen Synthase by Insulin-Stimulation of glucose storage rates in skeletal muscle by insulin is thought to activate glycogen synthase, the rate-limiting enzyme in glycogen synthesis (25). We next assessed the effects of insulin on muscle glycogen synthase activity, and we found that insulin infusion increased % I form of glycogen synthase in a time-dependent manner in non-Tg mice at 100 microunits/ml insulin concentration (Fig. 3C). Although % I form of glycogen synthase at the basal condition in Tg mice was not different from that of non-Tg mice, % I form of glycogen synthase in Tg mice at the end of a 60-min insulin infusion was significantly lower than that in non-Tg mice.
Impaired Insulin-stimulated 2-Deoxyglucose Uptake in Soleus Muscle-To evaluate the effect of ⌬PTP expression on glucose transport activity in skeletal muscle, we next measured the insulin-stimulated glucose uptake using 2-deoxyglucose in isolated soleus muscle. Although the basal and maximal insulin-stimulated 2-deoxyglucose uptake was comparable between both groups, the insulin-stimulated 2-deoxyglucose uptake in the presence of 1 nM insulin in Tg mice was significantly lower than that in non-Tg mice as shown in Fig. 4A.
Impaired Insulin-stimulated Glucose Uptake in Isolated Adipocytes-We also measured the insulin-stimulated glucose transport into isolated adipocytes from Tg mice. The cell size and the insulin binding affinity were comparable between non-Tg and Tg mice. Although the basal and maximal insulin effects on glucose transport were comparable in both groups (568 Ϯ 64 and 524 Ϯ 26 cpm at the basal state and 3638 Ϯ 584 and 3836 Ϯ 698 cpm at the maximally insulin-stimulated state in non-Tg and Tg mice, respectively), the dose-response curve for stimulation of glucose transport in Tg mice was shifted to right (ED 50 198 Ϯ 14 to 504 Ϯ 48 pM, p Ͻ 0.01) as shown in Fig. 4B.
Impaired IRS-1 Phosphorylation by Insulin-To assess the molecular mechanism for impaired in vivo insulin actions, we next assessed the alteration of phosphorylation of IRS in muscle and liver after a bolus injection of insulin. First, in the hindlimb muscle and liver, the expression levels of IRS-1 and IRS-2 proteins were comparable between non-Tg and Tg mice. When IRS phosphorylation was assessed by Western blotting using phosphotyrosine antibody, the insulin-stimulated IRS phosphorylation (IRS-1 and IRS-2) was attenuated as shown in Fig. 5A. As shown in Fig. 5, B and C, the insulin-stimulated IRS-1 phosphorylation in Tg mice was 77.4 Ϯ 6.1% that of non-Tg mice. In the isolated adipocytes, insulin-stimulated tyrosine phosphorylation was attenuated in Tg mice as shown in Fig. 6A. Furthermore, insulin-stimulated association of IRS-1 with endogenous Syp was also decreased in isolated adipocytes (Fig. 6B) as found in muscle ( Fig. 2A) and liver (Fig. 2B) of Tg mice.
Impaired Activation of both PI3-Kinase and Akt Kinase by Insulin-We measured PI3-kinase activity in muscle and liver from Tg mice after a bolus injection of insulin. Insulin-stimulated PI3-kinase activity that was immunoprecipitated with anti-IRS-1 antibody was assessed. As illustrated in Fig. 7A, insulin enhanced IRS-1-associated PI3-kinase activity by 5-fold even in Tg mice. However, the magnitude of insulin-stimulated PI3-kinase activities in Tg mice was 50% that in non-Tg mice as shown in Fig. 7B. Similar findings were observed in liver from Tg mice as shown in Fig. 7, C and D. Moreover, in the isolated adipocytes, insulin-stimulated PI3-kinase activity was also impaired in Tg mice as shown in Fig. 7, E and F.
We also measured Akt activity in liver and hindlimb muscle from Tg mice 5 or 10 min after bolus injection of insulin. As illustrated in Fig. 8A, in non-Tg mice, insulin stimulated Akt kinase activity by 2.5-fold. However, insulin stimulated Akt kinase activity in Tg mice by only 1.3-fold and was significantly impaired as compared with that of non-Tg mice. We observed the identical impairment in insulin stimulation in Akt activation in liver from Tg mice in Fig. 8B. Impaired MAP Kinase Activation by Insulin-We next assessed the insulin activation of MAP kinase in skeletal muscle during insulin sensitivity test, because SHP-2 was believed to play a crucial role in the activation of MAP kinase cascade. When we assessed the phosphorylation states of MAP kinase (Erk1 and 2) by Western blotting using anti-phospho-MAP kinase antibody, we found insulin infusion increased the content of phosphorylated MAP kinase after 30 min, and the level then plateaued in non-Tg mice as shown in Fig. 9. On the other hand, MAP kinase in Tg mice was phosphorylated at the basal state, and insulin infusion failed to enhance further the phosphorylation of MAP kinase. Furthermore, we also measured MAP kinase activity toward MBP proteins 10 min after a bolus injection of insulin via vena cava. In non-Tg mice, insulin stimulated muscle MAP kinase activity by 52% as shown in Table III. However, in Tg mice, insulin failed to activate MAP kinase, whereas basal MAP kinase activity was higher than in non-Tg mice. These results were consistent with the results on phosphorylation of MAP kinase in skeletal muscle during insulin infusion detected with phospho-MAP antibody. Furthermore, in liver from Tg mice, insulin-stimulated MAP kinase activation was also impaired with an elevated basal activity (data not shown). DISCUSSION To clarify the physiological roles of SHP-2 in vivo, we thus generated transgenic mice expressing a dominant negative mutant SHP-2 lacking PTPase domain (⌬PTP), and we found that these transgenic mice exhibited in vivo dominant negative effects on the association of IRS-1 with SHP-2 after a bolus injection of insulin. These mice exhibited in vivo insulin resistance against insulin stimulation of glucose utilization and impaired activation of IRS-1 phosphorylation and its downstream insulin signaling such as PI3-kinase and Akt kinase in skeletal muscle and liver of Tg mice as compared with non-transgenic mice.
With regard to the molecular mechanism for the impairment of activation of PI3-kinase and Akt kinase by insulin in Tg mice, we observed a small decrease in tyrosine phosphorylation of IRS-1 in Tg mice expressing a dominant negative mutant SHP-2. Thus, it is possible that SHP-2 modulates IRS-1 phosphorylation states to some extent, and inhibition of endogenous SHP-2 (Syp) function may lead to reduction of IRS-1 phosphorylation. We observed the similar alteration in an IRS-2 molecule. Consistent with this notion, we previously found that overexpression of wild-type SHP-2 potentiated IRS-1 phosphorylation, but overexpression of ⌬PTP mutant attenuated the phosphorylation states of IRS-1 in cultured HIRc cells (11). Alternatively, ⌬PTP mutant may interfere with the PI3-kinase activation of insulin by inhibiting the interaction of endogenous SHP-2 with its target molecules such as Grb2-asscoiated binder-1, -2, mammalian homologues of Daughter of sevenless, or SHPS-1 and SIRP (26 -31) except IRS.
Regarding the regulation of the phosphorylation state of IRS-1 by SHP-2, in vitro experiments indicate that SHP-2 enhances PTPase activity through its binding to IRS-1 peptides and dephosphorylates IRS-1 protein (12). Interestingly, in 32D cells that expresses a mutant IRS-1 lacking an SHP-2 binding motif, the association of IRS-1 with SHP-2 is inhibited and then insulin-induced IRS-1 phosphorylation is enhanced (17). This study is supposed to be inconsistent with several previous studies (12)(13)(14). However, phosphorylation states of IRSs (IRS-1 and IRS-2) may be regulated by more than one mechanism. One is the binding of SHP-2 to IRS through SHP-2 binding motif and dephosphorylates IRS-1 protein. Therefore, the deficiency of those interactions in the 32D cells may potentiate IRS-1 phosphorylation under insulin stimulation (17). In contrast, it has been reported that SHP-2 directly activates the Src kinase by interacting with Src kinase itself (32). Furthermore, overexpression of dominant negative ⌬PTP may result in decreased IRS-1 phosphorylation in Tg mice. Thus, SHP-2 may bind to the other target molecules and dephosphorylate those molecules, and if those molecules are tyrosine kinases like Src kinase, IRS-1 is further phosphorylated. Further investigation is necessary to clarify roles of these SHP-2-binding molecules in modulation of IRS-1 phosphorylation or PI3-kinase activation in insulin signaling.
In the current study, we overexpressed two SH2 domains of SHP-2 as deletion mutant SHP-2 lacking PTPase domain (⌬PTP) but not catalytically inactive Cys/Ser mutant. Strict binding specificity of tandem SH2 domains of SHP-2 has been reported based on several binding studies including crystallographic analysis (33)(34)(35) and a study in which the expression of SH2 domains of SHP-1 did not restore the dysfunction of SHP-2 in oocyte development (7). Furthermore, we did not observe a nonspecific inhibition of binding of p85, a regulatory subunit of PI3-kinase to IRS-1 in Rat 1 HIRc cells expressing ⌬PTP (15). Moreover, in the current study, insulin-induced association of Shc with Grb2 was not impaired in Tg mice, suggesting that the effect of expression ⌬PTP was specific for SHP-2.
Several findings obtained in cells expressing dominant negative mutants demonstrate that SHP-2 plays critical roles in activation of MAP kinase cascade (4,5,8,9,(12)(13)(14)36). Furthermore, SHP-2 is important in oocyte differentiation in terms of regulation of MAP kinase activity (7). Consistent with these reports, either knockout of Syp gene or deletion of N-terminal SH2 domain of Syp gene is lethal with the mice dying in embryonic states (37,38). In these mice, the impairment of MAP kinase activation by fibroblast growth factor was responsible for defects of maturation and differentiation. In contrast to these knockout studies, our Tg mice expressing a dominant negative mutant SHP-2 (⌬PTP) were able to grow normally. Nevertheless, homozygous mice expressing ⌬PTP mutant died in 15-day embryonic stage (data not shown). Furthermore, only

FIG. 3. Glucose intolerance and insulin resistance in Tg mice.
A, time course of blood glucose levels after intraperitoneal glucose load in Tg (q) and non-Tg mice (E). After overnight fasting, mice received 3 mg/g body weight glucose solution intraperitoneally, and blood samples were obtained from the tail vein at 0, 30, 60, and 120 min after glucose loading. B, blood glucose levels during insulin sensitivity test using glucose, insulin, and somatostatin infusion. To assess the in vivo insulin sensitivity, we performed insulin sensitivity test using a modified method described under "Experimental Procedures" using glucose, insulin, and somatostatin infusion. Blood samples were obtained from the tail vein at 0, 30, 60, 90, and 120 min after the start of infusion of mixed solution. C, insulin-stimulated glycogen synthase activity in hindlimb muscle of Tg mice during insulin sensitivity test. At the indicated time points, the hindlimb muscle of the mice was removed, and the glycogen synthase activities were measured according to the modified Thomas' filter paper method. The data are expressed as % I Ϫ form (Glc-6-PϪ/ Glc-6-Pϩ). Data are expressed as the mean Ϯ S.E. (n ϭ 4 -5). *, p Ͻ 0.05; **, p Ͻ 0.01 versus non-Tg mice. one strain expressing ⌬PTP mutant showed abnormal eyelid development among 17 independent Tg strains. 2 Thus, the impairment of Syp function in development may exist but may not be so severe in heterozygous mice. Moreover, we found that MAP kinase activity in these Tg mice was not impaired, and the growth rate was normal, although insulin could not activate further the enzyme activity. Some compensatory mechanisms to regulate MAP kinase cascade in our Tg mice might have occurred (39).
Concerning activation of glycogen synthase, it is generally accepted that PI3-kinase pathway is important for mediating insulin activation of glycogen synthase but not MAP kinase cascade (25, 40 -42). Recently, Akt activity has been reported to be essential for activation of glycogen synthase by insulin (43). In the present study, we found that the activation of Akt by insulin was attenuated in accordance with impaired activation of glycogen synthase despite persistent MAP kinase activation in Tg mice. Thus, our results support a role for Akt kinase in insulin-stimulated glycogen synthase activation.
In Tg mice, fasting plasma glucose levels were not elevated, indicating that the inhibitory effect of insulin on hepatic glu-

TABLE II Steady state blood glucose (SSBG) and plasma insulin levels during insulin sensitivity test
To assess in vivo insulin sensitivity, we performed insulin sensitivity test using a modified method of glucose, insulin, and somatostatin infusion. Blood samples were obtained from the tail vein at 0, 30, 60, 90, and 120 min after the start of infusion of mixed solution. Plasma insulin levels were measured at 120 min. The means of the last three samples were used to estimate SSBG levels, and insulin level at 120 min was used for steady state plasma insulin (SSPI) levels. Each value is presented as the mean Ϯ S.E. The number of mice used in the experiments is indicated in parentheses.  4. Insulin-stimulated glucose uptake into isolated soleus muscle and adipocytes from Tg and non-Tg mice. A, soleus muscle was separated from hindlimb and incubated with insulin (0 -10 nM) at 25°C for 120 min and then 2-deoxyglucose uptake was measured in the presence or absence of 1 or 10 nM insulin for 30 min. Each column shows the mean Ϯ S.E. *, p Ͻ 0.01 versus non-Tg mice (n ϭ 5-6). B, adipocytes were isolated and incubated with insulin (0 -10 nM) at 37°C for 60 min, and then glucose uptake was measured. Although the basal and maximal insulin effects on glucose transport were comparable in both groups, the dose-response curve for stimulation of glucose transport in Tg mice was shifted to right (ED 50 , 198 Ϯ 14 to 504 Ϯ 48 pM, p Ͻ 0.01). cose production was not disrupted, even with the impaired insulin signaling in liver. One possible explanation was that the sensitivity of insulin's inhibition of hepatic glucose production in liver might be better than glucose uptake in skeletal muscle and adipose tissue, and hyperinsulinemia might overcome this defect of insulin signaling in liver (44).
With regard to the role of SHP-2 in glucose transport activity, it has been reported that SHP-2 affects the expression of GLUT1 protein but not GLUT4 (45). However, overexpression of a mutant SHP-2 (Cys/Ser) in isolated adipocytes by electroporation significantly impaired glucose transport activity (46). We also found a rightward shift in insulin-stimulated glucose uptake in adipocytes isolated from Tg mice without any change in the maximum insulin-stimulated glucose transport rate, indicating that the lack of a modulator for insulin signal transduction causes resistance in insulin sensitivity but not insulin responsiveness. Similarly, in the present study, 2-deoxyglucose uptake into isolated soleus muscle was also impaired at a FIG. 7. Insulin-stimulated activation of PI3-kinase in hindlimb muscle, liver, and isolated adipocytes in Tg and non-Tg mice. A and B, 5 min after bolus injection of insulin, hindlimb muscle was isolated, and the PI3-kinase activity that was immunoprecipitated with anti-IRS-1 antibody was measured. Each column presents the mean Ϯ S.E. (n ϭ 5-6). **, p Ͻ 0.01 versus insulin injection in non-Tg mice. C and D, insulin-stimulated activation of PI3-kinase in Tg and non-Tg mice in liver. Two min after a bolus injection of insulin, liver was isolated and the PI3-kinase activity that was immunoprecipitated with anti-IRS-1 antibody was measured. Each column presents the mean Ϯ S.E. (n ϭ 5-6). **, p Ͻ 0.01 versus insulin injection in non-Tg mice. E and F, after isolating adipocytes from Tg mice, adipocytes were incubated with insulin, and then insulin-stimulated activation of PI3-kinase activity that was immunoprecipitated with anti-IRS-1 antibody was measured. PIP, phosphatidylinositol phosphate.
FIG. 8. Insulin-stimulated activation of Akt kinase in muscle and liver in Tg and non-Tg mice. Five or 10 min after a bolus injection of insulin, liver and hindlimb muscle were isolated and then Akt kinase activity that was immunoprecipitated with anti-Akt antibody in muscle (A) and liver (B) was measured in vitro using histone H2B as a substrate. Each column is presented as the mean Ϯ S.E. (n ϭ 4). **, p Ͻ 0.01 versus vehicle injection in non-Tg mice; ##, p Ͻ 0.01 versus insulin injection in non-Tg mice.
FIG. 9. Insulin-stimulated MAP kinase activation during insulin sensitivity test. Activated MAP was analyzed by Western blotting using anti-phospho-MAP kinase (A) and Erk2 antibodies (B) during insulin sensitivity test. physiological concentration of insulin, but the maximally stimulated glucose uptake rate was comparable between Tg and non-Tg mice. Thus, SHP-2 modulates insulin sensitivity of insulin-stimulated glucose uptake in muscle. These findings are different from the previous results on insulin resistance model in lack of key molecules in insulin signal transduction system (21,47,48). Therefore, hyperinsulinemia may overcome these defects and produce normal growth and phenotypes. These mice can be a good model to study the progression of the insulin-resistant state to the diabetic state by modulating environmental factors.
Finally, as recently reported, the disruption of PTP1B, a negative regulator of insulin signaling, leads to an increase in insulin sensitivity and becomes resistant to obesity by high fat feeding (49). In the current study, we demonstrate that the inhibition of functions of SHP-2, a positive regulator of insulin signaling, also induces in vivo insulin resistance. Therefore, these findings indicate that PTPases have critical roles for regulation of insulin signaling, and PTPases are important molecules to modulate insulin sensitivity in vivo.
In conclusion, the inhibition of endogenous SHP-2 (Syp) function by overexpression of mutant SHP-2 may lead to impaired insulin action in in vivo glucose metabolism. Therefore, SHP-2 modulates insulin signaling in skeletal muscle, liver, and adipose tissue and contributes to potentiation of insulin sensitivity in in vivo glucose utilization.