Role of c-Abl in Directing Metabolic versus Mitogenic Effects in Insulin Receptor Signaling*

c-Abl is a cytoplasmic tyrosine kinase involved in several signal transduction pathways. Here we report that c-Abl is involved also in insulin receptor signaling. Indeed, c-Abl tyrosine kinase is activated upon insulin stimulation. Inhibition of c-Abl tyrosine kinase by STI571 attenuates the effect of insulin on Akt/GSK-3β phosphorylation and glycogen synthesis, and at the same time, it enhances the effect of insulin on ERK activation, cell proliferation, and migration. This effect of STI571 is specific to c-Abl inhibition, because it does not occur in Abl-null cells and is restored in c-Abl-reconstituted cells. Numerous evidences suggest that focal adhesion kinase (FAK) is involved in mediating this c-Abl effect. First, anti-phosphotyrosine blots indicate that c-Abl tyrosine kinase activation is concomitant with FAK dephosphorylation in response to insulin, whereas c-Abl inhibition is accompanied by FAK phosphorylation in response to insulin, a response similar to that observed with IGF-I. Second, the c-Abl effects on insulin signaling are not observed in cells devoid of FAK (FAK–/– cells). Taken together these results suggest that c-Abl activation by insulin, via a modification of FAK response, may play an important role in directing mitogenic versus metabolic insulin receptor signaling.

Insulin and IGF-I receptors (IR and IGF-IR) 3 are heterotetrameric oligomers sharing a high degree of homology. They consist of two extracellular 135-kDa ␣ subunits that contain the ligand domain and two 95-kDa transmembrane ␤ subunits endowed of tyrosine kinase activity (1,2). Insulin and IGF-I binding to the receptor ␣ subunits stimulates autophosphorylation of the receptor cytoplasmic ␤ tail at multiple tyrosine residues. Activated ␤ subunits interact with and phosphorylate several intracellular substrates at tyrosine residues, including the IRS family proteins (IRS-1/2/3/4) and Gab1 (1,2). These tyrosine-phosphorylated substrates provide docking sites for SH2-contanining proteins, including p85, the regulatory subunits of phosphatidylinositol 3-kinase, and Grb2, a small adapter protein involved in the activation of the Ras pathway (1)(2)(3). Activation of the phosphatidylinositol 3-kinase pathway stimulates Akt and, as a consequence, elicits metabolic effects. On the other hand, the recruitment of Grb2 results in the activation of the Ras/ERK pathway, which in turn regulates cell proliferation and differentiation (1)(2)(3).
Despite the high similarity between IR and IGF-IR, these two receptors display different effects: IR mainly elicits metabolic effects, including glucose uptake and glycogen synthesis, and IGF-IR mainly stimulates cell proliferation, survival, and differentiation. The mechanisms underlying the different actions of IR and IGF-IR have been extensively studied and are believed to depend on the preferential recruitment and activation of different intracellular substrates, including CrkII, Grb10, 14-3-3, and the focal adhesion kinase (FAK) (1)(2)(3).
The FAK is an ubiquitously expressed cytoplasmic tyrosine kinase, involved in integrin and growth factor receptor signaling. It is widely accepted that FAK is a point of convergence in the actions of the extracellular matrix and growth factors (4). FAK tyrosine phosphorylation occurs rapidly in response to growth factor stimulation including PDGF, epidermal growth factor, and IGF-I (4,5). In particular, FAK phosphorylation at tyrosine 925 creates a binding site for the SH2 domain of Grb2, thereby activating the Ras/ERK pathway (6). Moreover, the activated FAK phosphorylates several adapter proteins including the p130 CAS (7) and paxillin (8), which in turn provide docking sites to the small SH2-containing proteins Crk and Nck (7,9,10). Similarly to Grb2, also Crk and Nck recruitment results in the activation of the Ras/ERK pathway (4). Because FAK is able to trigger the Ras pathway in response to several stimuli, it has been proposed that FAK plays a major role in mediating the effect of tyrosine kinase receptors on the ERK activation. Evidence supporting this hypothesis was obtained in FAK Ϫ/Ϫ cells, which exhibit a defect in ERK activation in response to serum and PDGF stimulation (11). Moreover, in NIH3T3 fibroblasts the overexpression of a dominant negative FAK blocks seruminduced activation of ERK (12). At variance with most growth factors, which stimulate FAK phosphorylation, insulin may cause either FAK phosphorylation or dephosphorylation (5,13,14), depending on the cell context and experimental conditions, including ligand concentration, cell adhesion, and IR expression level (13,(15)(16)(17)(18)(19). This suggests that the inhibition of the integrin/FAK/ERK signaling might play an important role in the metabolic, rather than mitogenic, effect of insulin.
c-Abl, like FAK, is a cytoplasmic tyrosine kinase activated upon cell adhesion (20) and exposure to growth factors like PDGF (21) or hepatocyte growth factor (22). Activated c-Abl tyrosine kinase phosphorylates CrkII, a small SH2/SH3-containing protein involved in cell proliferation and migration (23). Tyrosine 221 phosphorylation by c-Abl results in CrkII inactivation because of intramolecular folding caused by the interaction of the Crk SH2 domain with phosphotyrosine 221 (24). Although the c-Abl role in IR signaling is not known, there is evidence indicating that c-Abl and IR may share common substrates, including the Tub and SORBS1 adapter proteins (25,26). In particular, Tub is phosphorylated by both IR and c-Abl, whereas SORBS1 may interact with both IR and c-Abl, and this association is insulin-dependent.
Here we show that c-Abl is activated by insulin and regulates the response of FAK to insulin stimulation. Indeed, c-Abl activation leads to FAK dephosphorylation in response to insulin, whereas c-Abl tyrosine kinase inhibition results in FAK phosphorylation in response to insulin. As a consequence, under normal conditions c-Abl activation by insulin promotes Akt/ GSK-3␤ phosphorylation and glycogen synthesis. When c-Abl is inhibited the FAK response to insulin stimulation changes and phosphorylation, cell proliferation and migration are promoted. These data suggest that c-Abl exerts a pivotal role on insulin actions by driving IR signaling to predominant metabolic effects.
Plasmids and Transfections-AblFKBP-WT and AblFKBP-KD plasmids were kindly provided by Dr. Richard Van Etten (Harvard Medical School, Harvard, MA) (27). These constructs contain the FKBP domain, which allows Abl dimerization and activation in the presence of the dimerizer AP20187. All transfections were performed in 6-well plates with FuGENE 6 (Roche Applied Science) according to the manufacturer's instructions (DNA:FuGENE ratio 1:3). The cells were processed 24 h after transfection.
Viability Assay-Cell viability was measured by the methyl thiazolyl tetrazolium (MTT) test (Amersham Biosciences). 10 3 cells were seeded in 96-well plates. After 24 h, complete medium was replaced with serum-free medium containing 0.1% bovine serum albumin. The cells were then incubated with 10 nM of either insulin or IGF-I, in the presence or the absence of 1.0 M STI571, for additional 48 h. The cells were then incubated with complete medium containing 0.5 mg/ml MTT at 37°C plus 5% CO 2 . 4 h later cells were dissolved in 100 l of a solution containing dimethyl sulfoxide plus 2.5% complete medium, and formazan absorbance was then read at 405 nm.
Silencing of c-Abl Expression by siRNA-The cells were plated onto six-well plates (10 5 /well), maintained in antibioticfree medium for 24 h, and transfected with a mixture containing Opti-MEM, 8 l/well Lipofectamine 2000 (Invitrogen, San Diego, CA), and either 0.5 g/well scramble small interfering RNA (siRNA) or a mixture of four c-Abl siRNA (Dharmacon Research, Inc., Lafayette, CO) for 5 h. The sequences of these siRNAs are available from the manufacturer, and their specificity has been tested by microarray analysis. The cells were then incubated with fresh medium for 48 h, serum-starved for 24 h and incubated with or without 1.0 M STI571 in serum-free medium for 2 h. Then 10 nM of either insulin or IGF-I were added for 5 min, or longer, as indicated in the different experiments.
Migration Assay-Migration assay was performed as previously described (38) with some modifications. Subconfluent cells were preincubated with or without 1.0 M STI571 for 1 h. Following preincubation, the cells were then detached with Trypsin-EDTA, collected, and resuspended at 10 6 /ml. 10 5 cells, in serum-free medium (100 l), were placed on 6.5-mm-diameter polycarbonate filters (8-m pore size; Costar) and coated at the lower side with 250 g/ml collagen IV. Then filters were placed over bottom chambers containing 500 l of serum-free medium with or without 10 nM of ligand (either insulin or IGF-I) and 1.0 M STI571 and incubated for 6 h at 37°C with 5% CO 2 . At the end of 6 h, the cells on the upper surface of the filters were removed with a cotton swab, and the filters were stained with crystal violet solution (0.05% crystal violet in PBS ϩ 20% ethanol). The filters were then placed in 24-multiwell plates containing 250 l/well of 10% acetic acid and incubated for 30 min at room temperature under agitation to elute the dye. The samples were then measured at 590 nm. The row numbers obtained were reported as percentages of basal.
Glycogen Synthesis-The rate of glycogen synthesis was estimated by labeled glucose incorporation into cellular glycogen. The cells were seeded onto 24-multiwell plates. 24 h later complete medium was replaced with serum-free/low glucose medium, and the cells were preincubated with or without 1.0 M STI571 for 2 h and exposed to 10 nM of either insulin or IGF-I for 30 min at 37°C. D-[U-14 C]Glucose (1.25 Ci/well) was then added for 60 min, and the incorporation was stopped by rapidly washing cells three times with ice-cold PBS. The cells were then solubilized in 30% KOH and transferred to separate tubes, and 2 mg of glycogen was added. Glycogen was then precipitated by adding 2.2 volumes of 100% ethanol, and the radioactivity incorporated into glycogen was measured in a Beckman L6000TA ␤counter. The data were normalized for protein content.
Preparation of Cell Lysate and Western Blot-Subconfluent cells were serum-starved for 24 h and incubated with or without 1.0 M STI571 in serum-free medium for 2 h. Then 10 nM of either insulin or IGF-I were added for 5 min, or longer, as indicated in the different experiments. The cells were washed twice with ice-cold PBS, pH 7.4, and lysed in radioimmune precipitation assay buffer (150 mM NaCl, 1% Nonidet P-40, 50 mM Tris, pH 7.4, 10 mM sodium pyrophosphate, 100 mM NaF, 2 mM phenylmethylsulfonyl fluoride, 2 mM sodium vanadate, 10 g/ml pepstatin, 10 g/ml aprotinin, 10 g/ml leupeptin). After being scraped, the samples were rotated for 15 min at 4°C. Insoluble material was separated from soluble extract by microcentrifugation at 10,000 ϫ g for 10 min at 4°C. The protein concentration was determined by the Bradford assay (Bio-Rad). For immunoprecipitation 0.5-1-mg proteins were incubated with 2 g of the indicated antibodies and pro- tein A/G-Sepharose. For Western blot on crude cell lysate, after the addition of 5ϫ sample buffer, samples (50 g of protein) were heated at 95°C for 10 min. Immunoblotting was performed using 1 g/ml of the indicated antibodies. All of the immunoblots were revealed by enhanced chemiluminescence (Amersham Biosciences), autoradiographed, and subjected to densitometric analysis. The densitometric results were corrected for the total protein content.
c-Abl in Vitro Kinase-The cells were incubated as above and lysed with radioimmune precipitation assay buffer. After protein content normalization, immunoprecipitation was performed with anti-c-Abl polyclonal antibody (K12). The beads were then incubated for 40 min with 20 l of a kinase mixture containing 1 g of GST-Crk (kindly provided by Prof. Hidesaburo Hanafusa, Japan), 10 M ATP, in 10 mM Tris, pH 7.4, 10 mM MgCl 2 , 1 mM dithiothreitol. The reaction was then stopped with Laemmli buffer, and the samples were boiled for 10 min. After SDS-PAGE, filters containing c-Abl and phosphorylated Crk were probed with 4G10 monoclonal antibody and reprobed with either monoclonal antic-Abl (8E9) or polyclonal anti-Crk antibody (Santa Cruz). Statistical Analysis-Densitometry results were obtained by using ImageJ 1.37 software (National Institutes of Health). Densitometry results obtained with phosphorylated proteins were normalized for total protein content. The numbers were then expressed as percentages of basal. Cell viability, migration, and glycogen synthesis and densitometry results in response to either insulin or IGF-I, with or without STI571 were compared by two-way analysis of variance. Statistical analysis was carried out with Microsoft Excel software.

Insulin Biological Effects Are Influenced by the Inhibition of c-Abl Tyrosine Kinase
Activity-To explore whether c-Abl tyrosine kinase is involved in insulin biological effects, we incubated HepG2 (Fig. 1, left panel) and MCF-7 cells (Fig. 1, right  panel) in the presence or the absence of the Abl inhibitor STI571 and evaluated both metabolic (glycogen synthesis) and nonmetabolic effects (cell proliferation and migration) in response to insulin. In these experiments insulin effects were compared with the effects induced by IGF-I (Fig. 1). c-Abl inhibition by 1 M STI571 decreased the glycogen synthesis in response to insulin (Fig. 1A compare bar 2 with bar 5). In contrast, evaluation of cell proliferation by MTT assay indicated that exposure to STI571 significantly enhanced the effect of insulin on cell proliferation (Fig. 1B, compare bar 2 with bar 5) and migration (Fig. 1C, compare bar 2 with bar 5). STI571 effect was not observed on IGF-I stimulation (Fig. 1, compare bars 3 with bars 6) and was similar in both cell types.
In these experiments STI571 changed cell responses to insulin and abrogated the differences existing between insulin and IGF-I stimulation (Fig. 1, A and B, compare bars 2 and 3 with 5 and 6). These data suggest that c-Abl activation may contribute to differentiate IR from IGF-IR signaling by selectively increasing the metabolic effects and decreasing the nonmetabolic effects of the IR signaling pathway.
A Functional c-Abl Kinase Influences IR Biological Effects-To assess by a different procedure whether the effect of STI571 on insulin actions is specifically mediated by c-Abl tyrosine kinase inhibition, fibroblasts derived from abl and arg double knock-out mice (abl Ϫ/Ϫ arg Ϫ/Ϫ cells) were also used (28). These cells are devoid of both c-Abl and the related kinase Arg (abl related gene), a cytoplasmic tyrosine kinase with overlapping functions with c-Abl, which is also a target of STI571. As a control, abl Ϫ/Ϫ arg Ϫ/Ϫ cells reconstituted with murine c-Abl by retroviral gene transfer (abl WT arg Ϫ/Ϫ cells) were also used. STI571 did not affect glycogen synthesis in response to insulin in abl Ϫ/Ϫ arg Ϫ/Ϫ cells ( Fig. 2A, compare bar 2 with bar 5), whereas it inhibited glycogen synthesis in response to insulin in abl WT arg Ϫ/Ϫ cells ( Fig. 2A, compare bar 8 with bar 11) in a manner similar to that observed in HepG2 cells (Fig. 1A). In abl Ϫ/Ϫ arg Ϫ/Ϫ cells, STI571 did not enhance the effect of insulin on cell proliferation and migration (Fig. 2, B and C, compare bars 2 with bar 5). In contrast, STI571 enhanced cell proliferation and migration in response to insulin in abl WT arg Ϫ/Ϫ cells (Fig. 2, B and C, compare bars 8 with bars 11) in a manner similar to that observed in HepG2 (Fig. 1B). Similar results were observed with abl ϩ/ϩ arg ϩ/ϩ cells, which are control fibroblasts derived from littermates (Fig. 2, compare bars 14 with bars 17). Moreover, STI571 did not influence insulin effects in abl Ϫ/Ϫ arg Ϫ/Ϫ cells retrovirally infected with a kinase defective c-Abl construct (abl KD arg Ϫ/Ϫ , not shown). These data confirm that the effect of STI571 on insulin biological activities is specifically based upon the presence of a functional c-Abl tyrosine kinase and is not dependent on the STI571 influence or other tyrosine kinases. Cell responses to IGF-I were not affected by STI571 (Fig. 2, compare bars 3-6 and 9 with bar 12).
c-Abl Changes the Effect of Insulin on Akt, GSK-3␤, and ERK Activation-To explain the mechanisms underlying the effect of c-Abl on insulin actions, intracellular signaling molecules downstream IR were studied by Western blot with anti-phosphotyrosine and phosphospecific antibodies (Figs. 3 and 4).  SEPTEMBER 7, 2007 • VOLUME 282 • NUMBER 36

JOURNAL OF BIOLOGICAL CHEMISTRY 26081
To explore whether STI571 may interfere with IR and IGF-IR proximal signaling events, HepG2 cells were incubated or not with the Abl inhibitor STI571 and signaling pathway activation evaluated by Western blot (Fig. 3). STI571 did not affect IR and IGF-IR autophosphorylation in response to their ligands (insulin: Fig. 3A, compare lanes 2 and 3 with lanes 8 and 9; IGF-I: Fig.  3A, compare lanes 5 and 6 with lanes 11 and 12). In addition, STI571 did not influence IRS-1 phosphorylation in response to either insulin (Fig. 3B, compare lanes 2 and 3 with lanes 7 and 8) or IGF-I (Fig. 3B, compare lanes 4 and 5 with lanes 9 and 10), and as a consequence, it did not affect the recruitment of p85 (Fig. 3B, compare lanes 2-5 with lanes 7-10) or Grb2 (Fig. 3C,  compare lanes 2 and 3 with lanes 5 and 6). Moreover, reblotting with anti-c-Abl 8E9 antibody did not detect c-Abl in anti-IRS-1 immunoprecipitates (Fig. 3B, bottom panel).
In contrast, STI571 significantly attenuated Akt and GSK-3␤ phosphorylation in response to insulin, whereas it enhanced ERK response (Fig. 4A, compare lane 2 with lane 5, ϳ2-fold increase by densitometric analysis). The responses to IGF-I remained unchanged (Fig. 4A, compare lane 3 with lane 6). Because STI571 may inhibit not only c-Abl tyrosine kinase but also other tyrosine kinases such as PDGF-R and c-Kit, we used also additional, specific, and independent mechanisms to inhibit c-Abl and evaluate the consequences on the phosphorylation of downstream mediators. HepG2 and MCF-7 cells were subjected to Abl silencing by siRNA (Fig. 4B), and Western blot experiments confirmed that in both HepG2 and MCF-7 cells Abl inhibition reduced Akt and GSK-3␤ phosphorylation in response to insulin, whereas it enhanced ERK activation (Fig. 4B, compare lanes 2 with lane 5). In accordance with the results shown in Fig. 4A, the responses to IGF-I remained unchanged (Fig. 4B, compare lane 3 with lane 6). Experiments performed in abl Ϫ/Ϫ arg Ϫ/Ϫ and abl WT arg Ϫ/Ϫ confirmed these results (Fig. 4C, compare lane 2 with lane 5 and lane 3 with lane 6). Taken together, these data indicate that c-Abl tyrosine kinase inhibition results in reduced Akt/GSK-3␤ phosphorylation by insulin and enhanced ERK activation, whereas no effect is observed on IGF-I activation.
Insulin Stimulates c-Abl Tyrosine Kinase Activity-We then explored whether c-Abl tyrosine kinase activity is affected by insulin stimulation (Fig. 5). HepG2 cells were stimulated with insulin for the indicated times, and the c-Abl tyrosine kinase activity was evaluated in Abl immunoprecipitates by an in vitro kinase assay, using GST-CrkII as a substrate (see "Experimental Procedure"). Anti-Tyr(P) Western blot shown in Fig. 5A indicated that after 2 min of insulin stimulation, an increased Crk phosphorylation by c-Abl is observed and declines within 30 -60 min (Fig. 5A). Exposure of HepG2 cells for 60 min to 1 M of the Abl inhibitor STI571 completely abolished Crk phosphorylation, indicating that the tyrosine kinase activity observed in those immunoprecipitates is Abl-specific. In those experiments K562 cells, expressing the constitutively active BCR-Abl oncoprotein, were used as a positive control (Fig. 5A,  left panel). Interestingly, no increase in c-Abl tyrosine kinase activity was observed in HepG2 cells after stimulation with IGF-I (Fig. 5A, right panel).
We then studied CrkII phosphorylation by c-Abl in living cells (Fig. 5B). Although CrkII is a substrate of IGF-IR but not of IR tyrosine kinase (29,30), insulin is able to stimulate CrkII tyrosine phosphorylation in living cells maybe for the intervention of an unknown kinase (31). We therefore studied CrkII phosphorylation in response to either insulin or IGF-I in HepG2 cells in the presence or the absence of the c-Abl inhibitor STI571 (Fig. 2B). Anti-Tyr(P) Western blot performed on anti-CrkII immunoprecipitates revealed that STI571 abolished insulin-induced CrkII phosphorylation in vivo, whereas it was without effect on IGF-I-dependent CRKII stimulation (Fig. 5B).
To confirm the role of c-Abl in mediating CrkII phosphorylation in response to insulin in vivo, HepG2 cells were subjected to c-Abl silencing by siRNA (Fig. 5C). In accordance with data obtained with STI571, preincubation of HepG2 cells with Abl-siRNA resulted in the abrogation of CrKII phosphorylation in response to insulin (Fig. 5C), whereas the response to IGF-I was not affected (Fig. 5C). These results suggest that insulin, but not IGF-I, activates c-Abl, which in turn activates CrKII.
c-Abl Changes FAK Response to Insulin Stimulation-On the basis of the results obtained in cells subjected to c-Abl inhibition/silencing, we hypothesized that c-Abl could affect IR signaling and biological effects, making them more similar to those of IGF-IR signaling. A divergent point between IR and IGF-IR signaling is the FAK, whose phosphorylation level may be differently regulated by insulin and IGF-I and may be influenced by Crk activation (13,14). We therefore evaluated FAK activation in response to either insulin or IGF-I stimulation in HepG2 cells incubated in the presence or the absence of STI571 (Fig. 6A). Anti-phosphotyrosine Western blot performed in HepG2 cells showed that insulin induced FAK dephosphorylation (Fig. 6A, compare lane 1 with lane 2), whereas IGF-I caused FAK phosphorylation as previously reported (13). However, in the presence of the c-Abl inhibitor STI571, insulin stimulated a relevant FAK phosphorylation (Fig. 6A, compare lane 4 with lane 5). To confirm the role of c-Abl tyrosine kinase in mediat-  SEPTEMBER 7, 2007 • VOLUME 282 • NUMBER 36

c-Abl Role in IR Signaling
ing the effect of insulin on FAK phosphorylation, HepG2 and MCF-7 cells were subjected to c-Abl silencing by siRNA (Fig.  6B). In accordance with data obtained with STI571, in cells treated with Abl-siRNA we observed FAK phosphorylation (instead of a dephosphorylation) in response to insulin (Fig. 6B,  compare lane 1 with 2 and lane 4 with lane 5), whereas FAK response to IGF-I remained unchanged (Fig. 6B, compare lane  1 with lane 3 and lane 4 with lane 6). Experiments performed in abl Ϫ/Ϫ arg Ϫ/Ϫ and abl WT arg Ϫ/Ϫ also confirmed these results (Fig. 6C, compare lane 1 with lane 2 and lane 4 with lane 5). To confirm the effect of c-Abl activation on FAK phosphorylation, abl Ϫ/Ϫ arg Ϫ/Ϫ cells were transfected with AblFKBP-WT constructs that encode for an Abl protein dimerized by the presence of the AP20187 compound (27). Abl Ϫ/Ϫ arg Ϫ/Ϫ cells were also transfected with AblFKBP-KD (kinase defective) as a control. Anti-Tyr(P) Western blots revealed that AP20187 incubation resulted in AblFKBP-WT activation in transfected abl Ϫ/Ϫ arg Ϫ/Ϫ cells (Fig. 6D, left panel) and caused a dramatic FAK dephosphorylation in response to insulin stimulation (Fig. 6D,  right panel). In contrast, AP20187 incubation did not change FAK phosphorylation in response to insulin in empty and in AblFKBP-KD-transfected cells.
Taken together our data suggest that insulin may induce FAK dephosphorylation via the activation of c-Abl tyrosine kinase. As a consequence, when c-Abl is inhibited, FAK phosphorylation (rather than a dephosphorylation) occurs in response to insulin, an effect similar to that observed with IGF-I stimulation.
FAK Mediates the Effect of c-Abl on Downstream IR Signaling Effectors-To confirm the role of FAK in the c-Abl tyrosine kinase signaling after insulin stimulation, we performed experiments in FAK Ϫ/Ϫ cells, fibroblasts derived from FAK knockout mice. As a control R6 cells, fibroblasts derived from littermates (11), were used.  Blots with a phosphospecific anti-ERK antibody revealed that in FAK Ϫ/Ϫ cells c-Abl inhibition by STI571 did not enhance ERK activation in response to increasing doses of insulin, whereas it did in R6 cells (Fig. 7A, compare lanes 2 and 3  with lanes 8 and 9). Again, an increased ERK phosphorylation was observed in R6 cells exposed to STI571 only in response to insulin, but not to IGF-I (Fig. 7A, compare lanes 5 and 6 with  lanes 11 and 12).
We then tested the effect of STI571 c-Abl inhibition on insulin biological effects (Fig. 7B). Insulin effects were not affected in FAK Ϫ/Ϫ cells (Fig. 7B, compare bar 2 with bar 5), whereas glycogen synthesis was inhibited, and cell proliferation and migration were enhanced in response to insulin in R6 cells (Fig.  7B, compare bar 8 with bar 11). Also in this cell model, therefore, c-Abl inhibition provided results similar to those observed in HepG2, MCF-7, and abl WT arg Ϫ/Ϫ cells. Finally, as in all other cell types, STI571 did not affect cell responses to IGF-I (Fig. 7B, compare bar 3 with bar 6 and bar 9 with bar 12). Taken together, these results confirm in several cell models and by different experimental procedures that c-Abl is involved in IR signaling by affecting FAK phosphorylation in response to insulin stimulation.

DISCUSSION
Our results, obtained by independent experimental approaches including the evaluation of the insulin stimulation effect on c-Abl kinase activity, inhibition of c-Abl tyrosine kinase by STI571, and the use of c-Abl-null cells, indicate the involvement of c-Abl tyrosine kinase in IR signaling. At the same time, these data exclude the c-Abl involvement in IGF-IR signaling. The latter observation suggests that c-Abl may contribute to the difference in signaling and the biological effects of the two cognate receptors.
As far as c-Abl activation by insulin is concerned, c-Abl has already been reported to be involved in other tyrosine kinase receptor signaling, including PDGF-R, Met (hepatocyte growth factor receptor), and Trk (NGF receptor) (21,22,32,33). Both activated PDGF-R and Met stimulate c-Abl tyrosine kinase activity (21,22); whereas c-Abl activation by PDGF results in increased cell proliferation (21), c-Abl activation by Met exerts a negative regulatory effect and is concomitant with reduced cell migration in response to hepatocyte growth factor in thyroid cancer cells (22). c-Abl activation by Trk results in CrkII phosphorylation, axonal outgrowth, and guidance (34). These observations highlight the role of c-Abl tyrosine kinase in growth factor receptor signaling, although the biological role and consequences of c-Abl activation may be different (opposite) depending on the receptor involved, the signaling pathway, and the cell context. Here we describe the role of c-Abl in insulin receptor signaling and metabolic effects; by inhibiting insulin-induced ERK activation and enhancing, at the same time, Akt activation, c-Abl confers specificity to the IR signaling, leading to predominant metabolic effects. It is noteworthy that c-Abl does not influence the predominant mitogenic effects of the cognate IGF-IR. The exact pathway leading to the activation of c-Abl tyrosine kinase after IR stimulation by insulin, however, remains to be elucidated.
Here we report that cell exposure to insulin stimulates c-Abl tyrosine kinase activity and that, in turn, this causes FAK dephosphorylation, and as a consequence, it leads to a predominant phosphorylation of the Akt/GSK-3␤ and to glycogen synthesis. In contrast, when c-Abl is inhibited or absent, FAK is phosphorylated after insulin stimulation, and this results in an increased effect of insulin on the ERK pathway activation, cell proliferation, and migration. We also provide evidence that the effect of c-Abl is not dependent on proximal events of IR signaling, including receptor ␤-subunit and/IRS-1 phosphorylation and the recruitment of the small adapter protein p85 and Grb2 but rather on the distal effect on FAK phosphorylation. When c-Abl is inhibited or absent, a dramatic change in the response of FAK to insulin stimulation occurs, and insulin stimulates FAK phosphorylation in a manner similar to that of IGF-I. c-Abl inhibition, therefore, selectively modifies IR signaling and effects in such a way that IR signaling will mimic IGF-IR signaling in respect to FAK phosphorylation. c-Abl-dependent changes of FAK phosphorylation are a necessary step for the effect of c-Abl on insulin signaling; in cells devoid of FAK (FAK Ϫ/Ϫ cells) c-Abl inhibition is without effect on insulin signaling and actions. These results are in accordance also with previous observations that a functional FAK is required for the metabolic effects of insulin in HepG2 cells (35) and L6 myotubes (19).
The specific role of FAK phosphorylation or dephosphorylation, however, requires further investigations. It has been shown that insulin is not able to stimulate FAK dephosphorylation in cells transfected with an IR mutant, where Tyr 1210 is replaced by Phe (36). Interestingly, in these cells, insulin is more effective in stimulating ERK rather than phosphatidylinositol 3-kinase when compared with cells transfected with the wild type IR (36). These results further remark the importance of FAK dephosphorylation in restraining the mitogenic effect and enhancing the metabolic effects of insulin.
Because c-Abl is activated by cell adhesion (20), it is reasonable to suppose that it may play a relevant role in the complex interplay between integrin and insulin signaling, where inte- grins determine the cell relationship with the extracellular matrix and insulin regulates the cell energy metabolism and storage. Indeed, it has been reported that cell detachment from the extracellular matrix results in a dramatic switch of the FAK phosphorylation in response to insulin, similar to that observed when c-Abl is inhibited (13). When cells attach, c-Abl tyrosine kinase is activated, and this may cause FAK dephosphorylation in response to insulin, and consequently, inhibition of the cell ability to respond to insulin stimulation in terms of further migration and proliferation, whereas the cell metabolic activity is potentiated. In contrast, when c-Abl activity is down-regulated, as it occurs when cells detach, metabolic effects of insulin (like glycogen synthesis) are blunted, whereas effects on cell proliferation and migration are potentiated to an extent similar to that elicited by IGF-I. By this mechanism detached cells may obtain the biochemical instruments favoring their relocalization and proliferation. These data may also explain why insulin can cause FAK phosphorylation or dephosphorylation depending on different cell context and experimental conditions (13,(15)(16)(17)(18)(19).
We have not investigated the mechanism underlying c-Abl tyrosine kinase effect on FAK response to insulin stimulation. Some indirect evidence supports a role for CrkII as a possible link between c-Abl and FAK phosphorylation; carboxyl deletion mutants of CrKII enhance FAK phosphorylation (37), reproducing the condition when c-Abl activity is suppressed.
In conclusion the present results suggest that c-Abl is an important component of the intracellular signaling pathway activated by insulin, able to confer specificity to IR signaling by determining FAK phosphorylation or dephosphorylation. When c-Abl is inhibited or absent, insulin stimulation causes FAK phosphorylation, thereby mimicking IGF-I stimulation. Under this condition the metabolic effects of insulin are attenuated, while cell proliferation and migration are enhanced. c-Abl deregulation, therefore, could be involved in a variety of pathological conditions including insulin resistance, abnormal proliferative effects, and cancer, a condition in which this mechanism may provide a selective advantage to malignant cells in terms of both proliferation and invasiveness. c-Abl may be regarded, therefore, as a molecular switch or rheostat that determines the quality of cell response to insulin by restraining insulin signaling output to metabolic effects. Further studies of this mechanism may result from interest in different pathological conditions, and c-Abl targeting by specific molecules may prove useful for designing new therapeutic strategies aimed at controlling cell responsiveness to insulin.