Potentiation of Insulin Stimulation of Phosphatidylinositol 3-Kinase by Thiazolidinedione-derived Antidiabetic Agents in Chinese Hamster Ovary Cells Expressing Human Insulin Receptors and L6 Myotubes”

Thiazolidinedione derivatives are insulin-sensitizing agents with proven antidiabetic activities in vivo. To ex-plore the mechanism of action of this class of com- pounds, the effects of pioglitazone, CP-86,325, and AD-5075 on elements of the insulin signal transduction pathways were studied in Chinese hamster ovary cells overexpressing human insulin receptor (CH0.T) and L6 myotubes. In CH0.T cells, the binding of insulin to its receptor and the insulin-stimulated tyrosine kinase activity of the receptor were not altered by pioglitazone or CP-86,325. In contrast, treatment of CHO-T cells with the compounds resulted in significant increases in insulin-stimulated phosphatidylinositol (PI) 3-kinase activity. This insulin-enhancing effect was also observed in L6 myotubes treated with CP-86,325. The augmentations in kinase activity observed in CH0.T cells correlated with increases in the amount of PI 3-kinase (p85 subunit) in anti-phosphotyrosine immunoprecipitates of cell ly- sates. No gross changes in the tyrosine phosphorylation state of the insulin receptor substrate-1 were detected in insulin-stimulated CH0.T cells following treatment with the compounds. Furthermore, the compounds did not enhance insulin stimulation of mitogen-activated protein kinase or DNA synthesis in CHO-T (0.75 pCi/well) was added, and the incubation was continued at 37 "C for 1 h. The cells were washed twice with Hepes-buffered saline containing 0.3 m~ CaCI, and lysed with 0.075% SDS. The lysates were precipitated with 10% trichlo- roacetic acid at 4 "C overnight. The precipitates were collected on glass fiber filter circles and washed twice with 5% trichloroacetic acid, and the radioactivity associated with the filters was quantified by scintillation counting. Other Procedures-Protein concentrations were determined using Bradford reagent (Bio-Rad) following manufacturer's instruction. Sta-tistical analysis was conducted using Microsoft Excel 4.0.

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
'The abbreviations used are: IR, insulin receptor; CHO, Chinese hamster ovary; CHO.T, CHO cells expressing human insulin receptor; IRS-1, insulin receptor substrate-1; M A P , mitogen-activated protein; PAGE, polyacrylamide gel electrophoresis; PI, phosphatidylinositol; PIO, pioglitazone and phosphorylate intracellular substrates is essential for its mediation of the complex cellular responses of insulin (2)(3)(4)(5)(6)(7). One of the immediate substrates of the insulin receptor tyrosine kinase is IRS-1, a cytoplasmic protein with an apparent molecular mass of approximately 185 kDa (8). IRS-1 contains numerous Y" tyrosine phosphorylation motifs that interact with downstream effectors of insulin containing src homology (SH) 2 domains (8). One such SH2-containing protein is phosphatidylinositol (PI) 3-kinase, an enzyme composed of a regulatory subunit (p85) and a catalytic subunit (p110) that phosphorylates the D3 position of the inositol ring of PI and its phosphorylated derivatives (9). The binding of p85 to IRS-1 through interactions between the SH2 domain of the former and tyrosine-phosphorylated Y" motifs of the latter activates the enzyme (10)(11)(12)(13). This activation process is implicated in insulin-signaling transduction within cultured cells and in vivo (14)(15)(16)(17)(18).
Insulin is essential for maintaining glucose homeostasis and regulating carbohydrate, lipid, and protein metabolism (1). Decreased cellular response to insulin or perturbation of the insulin-signaling pathways are associated with a number of pathological states. Indeed, marked insulin resistance is found in patients with non-insulin-dependent diabetes mellitus (NIDDM) (19). Thiazolidinedione derivatives are a class of antidiabetic agents with in vivo insulin-sensitizing activities (20). These agents lower plasma glucose and triglyceride levels in animal models of obesity and NIDDM (21)(22)(23). Furthermore, treatment of insulin-resistant animals with the compounds resulted in enhancement of insulin activity in vitro in the three major target tissues of insulin action. This included stimulation of glucose uptake, glucose oxidation, and lipogenesis in adipose and muscle as well as stimulation of lipogenesis and inhibition of glucose output by the liver (24-26). Pilot clinical trials with one thiazolidinedione-derived compound, CS-045, in patients with NIDDM have demonstrated the efficacy of this class of compounds as oral antidiabetic agents (27,281. Despite more than a decade of research, however, little is known about the molecular mechanism of action of these insulin sensitizers.
The present study was initiated to investigate the effects of these antidiabetic agents on various steps in the insulin signal transduction pathways. The results indicate that the thiazolidinediones derivatives may augment cellular insulin sensitivity by potentiating insulin stimulation of PI 3-kinase activity. This potentiation correlates positively with the amount of the PI 3-kinase associated with tyrosine-phosphorylated proteins. Presumably, these compounds augment the association of p85 with tyrosine-phosphorylated IRS-1.
Antibody Production-Polyclonal antibodies against IRS-1 were produced in rabbits using a 20-amino acid peptide as an antigen. The sequence of the peptide corresponds to residues 6 2 5 of the IRS-1 protein (8). The antisera were affinity-purified on an antigen-linked A m -Gel (Bio-Rad) column according to published procedures (29).
Cell Culture and IFeatment-Chinese hamster ovary cells expressing human insulin receptor (CHO.T, a gift from Dr. Richard Roth, Stanford University, CA) were maintained in Ham's F-12 media with 10% fetal bovine serum (Hyclone), 100 unitdm1 penicillin, and 100 pg/ml streptomycin (Life Technologies, Inc.) at 37 "C in 5% CO,. L6 cells (kindly provided by Dr. Amira Nip, The Hospital for Sick Children, Toronto, Canada) were cultured in a-minimal essential medium (Life Technologies, Inc.) containing 2% fetal bovine serum (Life Technologies, Inc.), 100 unitdm1 penicillin G and 100 pg/ml streptomycin sulfate at 37 "C in a humidified atmosphere of 8.5% CO,. Cells were allowed to grow and fuse into myotubes as described previously (30). For experiments, the cells were treated in the appropriate serum-free media containing thiazolidinedione derivatives dissolved in dimethyl sulfoxide. Control cells received equivalent amounts of dimethyl sulfoxide and the final concentration of dimethyl sulfoxide was always kept below 0.1%.
Insulin Binding Assay-CH0.T cells were plated in 24-well dishes and cultured to 80% confluency. The cells were treated with thiazolidinedione derivatives for 24 h and then washed twice with ice-cold buffer A (Hepes-buffered saline, pH 7.4, 0.3 m M CaCl,). Next, the cells were incubated at 4 "C for 16 h with 500 pUwe11 buffer B (100 m M Hepes pH 7.8, 120 m M NaCI, 1.2 m~ MgSO,, 1 m M EDTA, 150 m~ sodium acetate, 10 m M glucose, 1% bovine serum albumin) containing various concentrations of unlabeled insulin and 1251-insulin (40,000 cpm/well). The cells were then washed twice with buffer A and lysed with 0.5 ml of 0.05% SDS, and radioactivity in the lysates was quantitated in a y counter. Nonspecific binding was assessed in the presence of 1 PM unlabeled insulin.
5rosine Kinase Assay-The IR tyrosine kinase activity was determined using a previously described procedure (31). Briefly, confluent CHO or CH0.T cells were treated with the compounds for 5 h followed by stimulation with 100 n M insulin for 10 min at 37 "C. The cells were 1% Triton X-100, 1 mg/ml poly(Glu:'&r) (4:1), 2 pCi of carrier-free [Y-~~PIATP) were added to one set of wells, and the incubation was continued at 25 "C for 30 min. The reaction was terminated by spotting 10 p1 of the reaction mixture on Whatman No. 3MM filter strips. The filter strips were treated with trichloroacetic acid, and the radioactivity was determined by Cerenkov counting. The amount of IR captured in the wells was determined by '251-insulin binding to the second set of wells. Forty microliters of '251-insulin (1000 cpdpl) in WGBT were added to each well, and the incubation was continued at 4 "C for 4 h. The wells were then washed and counted. The tyrosine kinase activities were normalized against the amount of '251-insulin specifically bound to captured IR.
PI 3-Kinase Assay-CHO, CHO.T, and L6 cells were pretreated with the compounds for 5 h and subsequently stimulated with 100 n~ insulin for 5 min at 37 "C. Lysates were prepared, and solubilized proteins were subjected to immunoprecipitation with monoclonal anti-phosphotyrosine antibodies conjugated to agarose beads (Sigma). The beads were washed, and PI 3-kinase activity was determined directly in the immunoprecipitates as described previously ( Tubes were vortexed for 1 min, and 30 pl of the lower layer was spotted on a Silica Gel 60 plate (precoated with 1% potassium oxalate and activated at 100 "C for 1 h). Plates were developed in CHCI3/CH,OW H,O/NH,OH (60:47:11.3:2) (32) and visualized with a PhosphorImager (Molecular Dynamics). Radioactivity associated with the spots corresponding to PIP was quantified by Cerenkov counting. SDS-PAGE and Western Blot-CH0.T cells were lysed in buffer D containing 20 m M Hepes pH 7.4,1% Triton X-100,20 m~ P-glycerophosphate, 150 m~ NaCl, 1 m M sodium orthovanadate, 10 m M sodium fluoride, 10 pg/ml aprotinin. Clarified cell lysates were subjected to immunoprecipitation with monoclonal anti-phosphotyrosine antibodies conjugated to agarose beads. The immunoprecipitates were washed three times with Hepes-buffered saline, and the immunoprecipitated proteins as well as the lysate were separated by 10% SDS-PAGE. The proteins were then transferred to nitrocellulose filters (Schleicher & Schuell), which were subsequently incubated with antibodies directed against phosphotyrosine, the p85 subunit of PI 3-kinase or IRS-1. After washing, the filters were incubated with ['2511protein A, and autoradiograms were analyzed utilizing a PhosphorImager.
MAP Kinase Assay-The assay was performed essentially as described previously (33). CH0.T cells were treated with the compounds Insulin Stimulation of Thymidine Incorporation into DNA-CH0.T cells were plated in 24-well plates and grown to 80% confluency. Prior to experimentation, the cells were incubated in serum-free growth media supplemented with 20 m~ Hepes pH 7.2, for 24 h. This media was then removed and replaced with fresh serum-free media (0.5 ml/well) containing various concentrations of insulin in the absence or presence of test agents. After 16 h, [3H]methylthymidine (0.75 pCi/well) was added, and the incubation was continued at 37 "C for 1 h. The cells were washed twice with Hepes-buffered saline containing 0.3 m~ CaCI, and lysed with 0.075% SDS. The lysates were precipitated with 10% trichloroacetic acid at 4 "C overnight. The precipitates were collected on glass fiber filter circles and washed twice with 5% trichloroacetic acid, and the radioactivity associated with the filters was quantified by scintillation counting.
Other Procedures-Protein concentrations were determined using Bradford reagent (Bio-Rad) following manufacturer's instruction. Statistical analysis was conducted using Microsoft Excel 4.0.

Insulin Binding,
IR Tyrosine Kinase Activity, and IR Autophosphorylatiom" whole cell ligand binding assay using '251-insulin indicated that preincubation of CH0.T cells with P I 0 or CP for 24 h had no effect on the ability of cell surface IR to bind to insulin (Fig. 1). The maximum amounts of 1251-insulin bound were 3.96,4.06, and 4.00 fmol/well for control, PIO, and CP-treated cells, respectively. Competition binding studies utilizing increasing amounts of unlabeled insulin demonstrated that the affinity of the receptor for its ligand remained unchanged after treatment with the antidiabetic agents (Fig. 1).
To determine whether the compounds affected IR tyrosine kinase activity, cells were first incubated with P I 0 or CP for 5 h and then stimulated with various concentrations of insulin for 5 min. The receptors were isolated on microtiter wells previously coated with anti-IR monoclonal antibodies, and the kinase activity was assayed using poly(G1u:Tyr) (4:U as exogenous substrate. Parallel wells were assayed for the amount of receptor present by incubation with '251-insulin. In general, the amounts of the '"I-insulin hound to wrlls containing IR from control or compound-trratcd crlls varirtl hy lrss than 1OfY, As expected, insulin caused a dose-deprndrnt stimulation of IR tyrosine kinase activity in CI-l0.T crlls, and thr maximum stimulation was 26-fold compared with unstimulated cells. Preincuhation of C H 0 . T cells with CP or PI0 did not alter thc srnsitivity or maximal responsivrness of their IR tyrosine kinase activity to insulin ( Fig.  2A ). The inahility of the compounds to affrct IR autophosphorylation was demonstrated hv anti-phosphotyrosine immunohlot analysis of CH0.T cell proteins (Fig. 2 8 ). Prctrcatment of insulin-stimulated cells with thiazolidinrdione derivatives caused no further detectable increase in the tyrosine phosphorylation lrvel of the 97 kDa hand corresponding to t,hr rrceptor /3 suhunit. stimulat.rd with incrrasing concentrations of insulin for 5 min a t 37 "C. T h r cells were soluhilized, and cellular proteins were immunoprrcipitated with anti-phosphotyrosine antibodies. In control CH0.T crlls, stimulation with insulin resulted in a dose-drprndent incrrasr in PI 3-kinasc activitv. Maximal activation was 80-fold comparrd with unstimulated cells. Preincubation of the cells with CP ( 1 ph!~ or P I 0 ( 2 5 pht) potentiated insulin activation of PI 3-kinasr activity 2-3-fold (Fig. 3). This potentiation was found to he dependent upon thr dosr of CP used in the prrincuhation (Fig. 4A ). Furthermore, the effect of C P on PI 3-kinasr activity was apparent aftrr pretreating the cells for 1 h prior to insulin stimulation and the maximal effect maintaincd through a 5-h prrincuhation prriod (Fig. 417 ).

17inding of P I 3 -K i n n w t o Phosphot~rosir~c~ Protcins nnd I R S -1 ~r n s i n c~ Phos~)hory/ntio~~-Upon stimulation of cells
with insulin, PI %kinase hecomrs activatrd hy hinding to tyrosine-phosphorylated IRS-1, a major substrate of the IR tyrosine kinase ( 8 ) . To rxamine the rffects of thiazolidinedionrdrrived compounds on the association of PI 8-kinase with IRS-1 and tyrosine phosphorylation of IRS-1, CH0.T crlls were pre- ively ( n = 3 , p < 0.05) (Fig. 6R ). These increases are similar to the aforementioned augmentations in PI %kinase activity.
Taken together, the results support the conclusion that thiazolidinedione-derivatives increase PI %kinase activity in antiphosphotyrosine immunoprecipitates by increasing the amount of enzyme present. To t e s t whether CP or AD potentiate activation of SIAP kinas(. hy insulin, CH0.T cells were prcincuhated alone or with thc cnmpounds and then acutely stimulated with insulin. Cytosolic MAP kinase activity was then determined usinC myolin hasic protein as substrate. As shown in Fig. 8. the compounds did not alter the sensitivity or maximal responsivrnr.ss of MAP kinasr to insulin stimulation.
Finally, the effect of the thiazolidinetlione derivatives on insulin-stimulated mitogenesis w a s c.xnmincd. Quicswnt C H 0 . T cells were pretrrated for lfi h with insulin alone. or in the presence of AD, CP, or PI0 and then incubated uith I ' , H I methylthymidine for 1 h. Incorporation of rndiol;~hclcd thymidine into DNA was then quantified. None of thv compounds potentiated insulin-stimulated DNA synthclsis (Fig. 9 I.

D1SCI:SSIOS
In recent years. extensive studies on the insulin sipal transduction pathways have led to the identification of an important endogenous IR tyrosine kinase substrate, IRS-I (81. Insulin has also been shown to activate PI 3-kinase in cultured cells 11.2. 151 and in animals (17,18). Further evidence supporting n rolr for PI .?-kinase in the insulin transduction pathwxys comes from experiments demonstrating that insulin stimulation of thc-kinase is attenuated in tissue from animals made insulin resista n t by chemical treatment ( 3 8 , 391 and in a genetic animal model of insulin resistance 1401. The activation of P I 3-kinnsr apparently results from t.he binding of its pH.5 rcyulntory suhunit SH2 domain to phosphotyrosines within sevrral IHS-1 Y" sequence motifs ( 10-131. In the present study, we have investigated th(. c.ffrcts trf n novel class of insulin sensitizing agents, thinzolidinrdionr drrivatives, on multiple steps in the insulin signal transdurtion pathways. We demonstrated that these compounds enhance insulin activation of PI 3-kinase in CHO cells overexpressing the human insulin receptor (Fig. 3 ) and that the increments in the insulin-stimulated PI 3-kinase activity closely correlate with the increased amounts of enzyme associated with tyrosine-phosphorylated proteins (Fig. 6). Presumahlv, the major interaction being augmented hy the thiazolidinedione derivatives is between the kinase and tyrosine-phosphorylated IRS-1.
This potentiation appears to he both dose-and time-dependent ( Fig. 4, n and h ) . The compounds exerted a similar activating effect on PI 3-kinase in L6 myotuhes, a model of insulin-responsive skeletal muscle (Fig. 5). This last result demonstrates that the insulin potentiation effect of thiazolidinedione derivatives is not an artifact arising from the high level of insulin receptors expressed hy CH0.T cells hut may he indicative of their primary in O~U O activity.
The mechanism(s) by which the thiazolidinedione derivatives potentiate insulin activation of PI 3-kinase is not yet entirely clear. They appear to act on postreceptor steps since they did not alter insulin binding to its receptor (Fig. 1) and failed to enhance the tyrosine kinase activity of the receptor activate PI 3-kinase.
IRS-1 has been shown to be involved in insulin-stimulated mitogenic signaling (42). The fact that the antidiabetic agents did not alter insulin-induced DNA synthesis in CH0.T cells (Fig. 9) suggests that their effect on PI 3-kinase is relatively specific. In agreement with this observation, the compounds were unable to augment insulin activation of MAP kinase.
The binding of the p85 subunit of PI 3-kinase to IRS-1 appears to result in conformational changes in the enzyme (43,44). Such alterations may serve to activate the enzyme and may be potentiated by the thiazolidinedione derivatives. It has also been demonstrated that PI 3-kinase is a dual-specificity enzyme containing an intrinsic serine kinase activity that phosphorylates Ser6'' of the p85 subunit. This phosphorylation event inhibits the enzyme's activity (45). Furthermore, it has been shown that tyrosine phosphorylation of the p85 subunit of PI 3-kinase may regulate the activity of the enzyme or alter the affinity of its SH2 domain for phosphotyrosine residues (46, 47). Therefore, it is apparent that the activity of PI 3-kinase is modulated by a number of complex processes including proteinprotein interactions, induced conformational changes, and alterations in its phosphorylation state. These events may not be independent but may modify each other through "cross-talk" during the activation process. The thiazolidinedione derivatives may potentiate any or all of these activation processes, perhaps as an affector of the appropriate kinase or phosphatase. Alternatively, the compounds may promote association of PI 3-kinase with another phosphotyrosine protein and, thus, potentiate insulin activation of the enzyme.
The primary physiological action of PI 3-kinase has not yet been delineated. However, based upon the sequence homology of its pll0 subunit to the yeast VI' S34 gene product, the kinase may be involved in mediating intracellular protein sorting. (48). Studies using growth factor receptors containing PI 3kinase binding site mutations support the conclusion that the enzyme plays a role in mediating receptor endocytosis and intracellular protein trafficking (49). Further work is necessary to define the role of PI 3-kinase in the insulin signal transduction pathways regulating metabolism and mitogenesis.
In summary, insulin resistance of peripheral tissue is a major pathological component of NIDDM. The thiazolidinedione-derived antidiabetic agents lower plasma glucose and triglyceride levels in animal models and humans suffering from obesity and NIDDM. These beneficial effects are apparently the result of the compounds acting as insulin sensitizers in insulin-responsive tissue. Here, we have demonstrated that the thiazolidinedione derivatives potentiate insulin stimulation of PI 3-kinase. Intriguingly, the thiazolidinediones appear to be rather selective in their insulin-dependent activation of PI 3kinase, since other insulin-dependent events such as activation of MAP kinase and mitogenesis were unaffected. This finding unveils a putative molecular mechanism of action for this class of antidiabetic agents and demonstrates a potential use for these compounds as tools to study the insulin signal transduction pathways.