Transient Activation of Jun N-terminal Kinases and Protection from Apoptosis by the Insulin-like Growth Factor I Receptor Can Be Suppressed by Dicumarol*

The insulin-like growth factor I receptor (IGF-IR) activated by its ligands insulin-like growth factor (IGF)-I or IGF-II mediates suppression of apoptosis and contributes to tumorigenesis and cell growth. Here we investigated the activation of the stress-activated protein kinases including Jun N-terminal Kinases and p38 MAPK by IGF-I in interleukin-3-dependent FL5.12 lymphocytic cells that overexpress the IGF-IR (FL5.12/WT). We have shown previously that IGF-I protects these cells from apoptosis induced by interleukin-3 withdrawal but does not promote proliferation. IGF-I induced a rapid and transient activation of JNK that peaked at 40 min that was paralleled by a transient and robust phosphorylation of c-Jun. p38 was constitutively phosphorylated in FL5.12/WT cells. Activation of the JNK pathway by IGF-I occurred in the presence of phosphatidylinositol 3-ki-nase inhibitors and could be enhanced by anisomycin. Analysis of a series of FL5.12 cells expressing mutated IGF-IRs and analysis of 32D/IGF-IR cells showed that neither the C terminus of the receptor nor IRS-1 and IRS-2 were required for JNK activation, although tyrosine 950 was essential for full activation. The JNK inhibitor dicumarol suppressed IGF-I-mediated activation of JNK

important role in the survival of many cell types and in response to diverse death stimuli. Epithelial cells and fibroblasts are protected from apoptosis induced by oncogene activation or growth factor withdrawal (1), toxic stimuli such as chemotherapeutic drugs, and irradiation (2). Cells of the nervous system can be protected by IGF-I from osmotic stress or growth factor withdrawal (3). Hematopoietic cells are protected by IGF-I from IL-3 withdrawal (4), and in T lymphocytes stimulation through the CD28 receptor leads to increased IGF-IR expression levels and enhanced IGF-I-mediated protection from Fasinduced apoptosis (5).
In addition to anti-apoptotic activity the IGF-IR also has strong mitogenic and transforming activity, and there is considerable evidence to indicate that the presence of the IGF-IR is necessary for tumorigenesis (reviewed in Refs. 6 -9). Fibroblasts derived from IGF-IR null mice cannot be transformed by a series of oncogenes (6). Inhibition of IGF-IR expression or signaling capacity by antibodies (7), triple helix formation (8), or antisense strategies (9) results in induction of apoptosis, failure to grow in anchorage-independent conditions, as well as failure to form tumors in nude mice.
The signaling pathways activated by the IGF-IR have been extensively studied, and it appears that there is considerable overlap in the pathways used for all of the receptor functions. One major signaling pathway activated by the IGF-IR is through its interaction with IRS-1 or IRS-2 leading to activation of PI 3-kinase and AKT (3,10), which promotes suppression of apoptosis via phosphorylation of several proteins. These include the forkhead transcription factor FKHRL1 (11), cAMPresponsive element-binding protein (12), caspase 9 (13), GSK3-␤ (14), and the Bcl-2 regulatory family member Bad (15). Other signaling pathways leading to Bad phosphorylation can also be activated by the IGF-IR including the Ras MAP kinase pathway via Rsk-1 in neuronal cells (16) and via Shc activation or via translocation of RAF to the mitochondria in 32D myeloid cells (17).
Interestingly, despite the prevalence of the AKT pathway in anti-apoptotic signaling from the IGF-IR and the role of the AKT regulatory phosphatase PTEN in tumorigenesis (18,19), there is also compelling evidence for anti-apoptotic signaling from the IGF-IR that is independent of AKT activation or Bad phosphorylation. Rat-1 fibroblasts exhibited IGF-IR-induced PI 3-kinase-dependent anti-apoptotic activity but also exhibited anti-apoptotic activity that could not be blocked by PI 3-kinase inhibitors or by dominant-negative AKT constructs (20). In the IL-3-dependent cell line FL5.12, which we have previously used as a model to study IGF-IR-mediated suppression of apoptosis in response to IL-3 withdrawal, two mutants of the IGF-IR that could not suppress apoptosis could still activate PI 3-kinase and AKT (21). These studies suggest that the AKT pathway is not sufficient for suppression of apoptosis by the IGF-IR in these cells.
The mitogen-activated protein (MAP) kinase family has been investigated for its contribution to anti-apoptotic signaling. However, although IGF-I induces ERK phosphorylation, this is not sufficient to mediate the PI 3-kinase-independent antiapoptotic activity of the IGF-IR (20,22). Another candidate pathway that has not been investigated much in IGF-IR antiapoptotic signaling is that controlled by the MAP kinase-related stress-activated kinases (SAPKs), consisting of c-Jun Nterminal kinases (JNKs) and p38 kinase. Although the role of SAPKs in either promoting or suppressing apoptosis appears to be quite complex (reviewed in Ref. 14), there is considerable evidence to support a role for these proteins in transformation (23,24), a function that overlaps with the anti-apoptotic function of the IGF-IR. JNKs have also recently been strongly associated with the differentiation of T lymphocyte subsets (25).
The role of SAPKs in promoting apoptosis or cell survival appears to depend on the stimulus a cell receives. Neurons activate JNKs in response to growth factor withdrawal (26), and fibroblasts derived from mice with two isoforms of JNK knocked out (JNK1 and JNK2) are more resistant to UV-induced apoptosis than their wild type littermates (27). However, the JNK kinase SEK can protect thymocytes from apoptosis (28), and in a biphasic response to TNF-␣ there is a correlation with prolonged activation of JNK and induction of apoptosis, whereas there is a correlation with transient activation of JNK and suppression of apoptosis (29). Transient activation of JNK by TNF-␣ in HeLa cells could be inhibited by the quinone reductase inhibitor dicumarol, and this lead to potentiation of apoptosis (30), suggesting that transient JNK activation is associated with the survival of tumor cells. Activation of JNK and p38 by cytokines such as erythropoietin has also been associated more tightly with cell survival than with apoptosis (31). IGF-I has been shown to suppress JNK activity and activate p38 in response to high glucose-induced stress in neuroblastoma cells (32) and to suppress prolonged JNK activation by anisomycin in fibroblasts (33). In contrast transient JNK activation has recently been shown to occur in response to IGF-I stimulation of the breast carcinoma cell line MCF-7 (34). However, the consequences of JNK activation by IGF-I for regulation of apoptosis, cell growth, or transformation by the IGF-IR are not known.
In this work we investigated SAPK activation by the IGF-IR and further investigated the effects of dicumarol on the antiapoptotic activity of the IGF-IR using FL5.12 cells that overexpress the wild type (WT) IGF-IR and that are protected by IGF-I from IL-3 withdrawal. We show that JNK is transiently activated by IGF-I stimulation leading to phosphorylation of c-Jun, whereas p38 is constitutively phosphorylated in these cells. JNK activation by IGF-I does not require PI 3-kinase or AKT activation. Dicumarol could block JNK activation and also abrogated suppression of apoptosis by IGF-I but did not affect p38 activity or AKT activation. Similar concentrations of dicumarol induced apoptosis in the breast carcinoma cell line MCF-7. These data suggest that IGF-I-mediated activation of JNK plays a role in suppression of apoptosis by the IGF-IR and that the JNK inhibitor dicumarol can induce apoptosis in tumor cells that are dependent on IGF-IR-mediated survival signals. IGF-I was purchased from PeproTech  (Rocky Hill, NJ). The anti-ERK2, anti-phospho-ERK, anti-phospho-SAPK, anti-SAPK, anti-phospho-p38, anti-p38, anti-AKT, anti-phospho-AKT antibodies, and the non-radioactive SAPK assay kit were  obtained from New England Biolabs (Beverly, MA). Anti-phospho-Jun, anti-Shc polyclonal, anti-Grb-2, anti-Gab-2 and anti-PI 3-kinase p85 subunit antibodies were all obtained from Upstate Biotechnology, Inc. (Lake Placid, NY). An anti-Shc monoclonal antibody was obtained from Transduction Laboratories (Lexington, KY). The p38, ERK, and PI 3-kinase inhibitors PD98049, SB203580, and LY294002, respectively, were purchased from Calbiochem, and dicumarol was from Sigma. The monoclonal anti-actin antibody and the FITC-conjugated anti-mouse IgG, antibody used for immunofluorescence, was purchased from Sigma.

Induction of Apoptosis by IL-3 Withdrawal and Cell Viability Assays in the Presence of PI 3-Kinase and MAP Kinase Inhibitors-FL5.12/WT
cells were cultured at 3 ϫ 10 5 cells/ml in medium containing IL-3 for 24 h, washed three times in serum-free medium, and then cultured at 5 ϫ 10 5 cells/ml in IMDM containing 5% FBS (2 ml/well) in 24-well plates. IGF-I (100 ng/ml) or IL-3 (WEHI CM 10%) was added to triplicate cultures. At the indicated times, 200-l aliquots were removed from each well, and viability was determined by counting live and dead cells after trypan blue staining. The percentage of viable cells was calculated from the total number of cells per well, and all data are presented as the mean of triplicate cultures for each culture condition. Survival assays were performed in medium supplemented with 5% FBS to reduce the amount of IGF-I and IGF-II that is available in FBS that would mask the effects of exogenously added IGF-I. IGF-I-and IL-3mediated suppression of apoptosis was also assayed in the presence of a number of inhibitors of signal transduction pathways. To inhibit PI 3-kinase, ERKs, p38, or JNK, respectively, LY294002 (10 M), PD98049 (50 M), SB203580 (10 M), or dicumarol (50, 100, or 150 M) was added to cultures that were exposed to IGF-I, IL-3, or 5% FBS. Cell viability was monitored by trypan blue exclusion over a 72-h period, and the data are presented as described above.
Western Blotting and Immunoprecipitations-FL5.12 cells overexpressing WT and mutant forms of the IGF-IR were seeded at 5 ϫ 10 5 cells/ml in IMDM containing 10% FBS plus 10% WEHI CM and cultured for 24 h, washed three times in serum-free medium, and then starved for 3.5 h in serum-free medium. Following serum starvation 7 ϫ 10 6 cells were stimulated with IGF-I (100 ng/ml) for up to 60 min, or not, and then washed in cold PBS. Cells were then lysed in 40 l of ice-cold SDS-lysis buffer (1% Nonidet P-40, 0.1% SDS, 20 mM Tris, 50 mM NaCl, 50 mM sodium fluoride, 1 M pepstatin, 1 mM phenylmethylsulfonyl fluoride, 1 M aprotinin, and 1 mM sodium orthovanadate, pH 7.6). Debris was removed by centrifugation at 15,000 ϫ g at 4°C for 15 min, and samples were then denatured by boiling in 2ϫ SDS-PAGE sample buffer for 5 min. Proteins were resolved by 12 or 4 -20% gradient SDS-PAGE and blotted onto nitrocellulose membranes (Schleicher & Schuell). Blots were blocked in TBS-T (20 mM Tris, 150 mM NaCl, 0.05% Tween 20, pH 7.6) with 5% milk for 30 min at room temperature. Primary phospho-specific antibodies were diluted 1/1000 in TBS-T, 5% goat serum, and all other antibodies were diluted 1/1000 in TBS-T, 5% milk and incubated for 1.5 h or overnight. Horseradish peroxidaseconjugated secondary antibodies (Dako Glostrup, Denmark) were used for detection using chemiluminescence with the ECL reagent (Amersham Pharmacia Biotech) or with the Supersignal reagent (Pierce).
For immunoprecipitation cell lysates (10 7 cells in 500 l per sample) were pre-cleared with 20 l of protein G-agarose beads for 1 h, followed by incubation with specific antibodies (5 g) for 18 h. Protein G-agarose beads (20 l) were then added for 3 h, and the beads were washed three times with lysis buffer before boiling in gel sample buffer and separation of the released proteins by SDS-PAGE. Proteins present in the immunoprecipitates were identified by Western blotting as described above.
A non-radioactive kinase assay kit from PerkinElmer Life Sciences was used to assay JNK activity in FL5.12 cells. This kit utilized gluta-thione S-transferase-Jun beads that pull-down JNK protein from cell lysates. Cell lysates from cells that had been stimulated with IGF-I under different conditions, or that were not stimulated, were incubated with glutathione S-transferase-Jun beads for 18 h, washed, and then incubated at 30°C for 30 min in kinase reaction buffer according to the manufacturer's instruction. The beads were then boiled in SDS sample buffer and pelleted at 3,000 ϫ g for 3 min. Proteins in the supernatants were resolved by 12% SDS-PAGE, and phosphorylated Jun was detected by Western blotting with an anti-phospho-Jun antibody, which specifically detects JNK phosphorylation on serine 63 of c-Jun.
MCF-7 Cell Viability Assays-MCF-7 cells were seeded at 1 ϫ 10 4 cells/ml in 24-well plates and cultured for 24 h. Dicumarol was then added to the cells at various concentrations. These and control cultures cells were incubated in Dulbecco's modified Eagle's medium supplemented with 10% FBS for 24, 48, or 72 h. At the indicated times cultures were photographed and then washed. The remaining cells were stained with Giemsa, and the viable cells in each well were counted by microscopic analysis. Cell architecture was also analyzed by immunofluorescence staining of actin in the cells. Cells were fixed in 100% methanol at Ϫ20°C for 5 min and then blocked in blocking solution consisting of 2.5% goat serum in PHEM buffer (60 mM Pipes, 25 mM Hepes, 10 mM EGTA, 2 mM MgCl 2 , pH 6.9) for 1 h. An anti-actin monoclonal antibody was then added to cells for 1.5 h at room temperature. After washing, a FITC-conjugated anti-mouse IgG antibody, diluted 1/100 in blocking solution, and Hoechst dye (0.5 g/ml) were incubated with cells for a further 1.5 h at room temperature. The cells were washed again and photographed using a SPOT-RT digital camera mounted on a Nikon TE300 inverted microscope equipped for epifluorescence.

JNK and the ERKs Become Phosphorylated in FL5.12/ IGF-IR Cells following IGF-I Stimulation, whereas p38 Is Con-
stitutively Phosphorylated-Activation of the MAP kinase family members in response to IGF-I has been demonstrated in different cell types (35). However, the activation of SAPK by the IGF-IR in cells where IGF-I mediates suppression of apoptosis has not been studied. To investigate this, we used FL5.12 cells overexpressing the wild type IGF-IR (FL5.12/WT) that are efficiently protected from IL-3 withdrawal by IGF-I (21). FL5.12/WT cells were removed from IL-3 for 3.5 h and then stimulated with IGF-I or left unstimulated, and Western blotting was performed using antibodies against the active/phosphorylated forms of the MAP kinase and SAPK proteins. Fig.  1A demonstrates that ERKs 1 and 2 are activated upon IGF-I stimulation, and this is in agreement with published studies (17). Levels of p38 phosphorylation are high in unstimulated cells and remain at similar levels in response to IGF-I stimulation. This finding is in contrast to that observed with IGF-Iinduced phosphorylation of p38 in neuroblastoma cells (32) and suggests that p38 is constitutively active in FL5.12 cells or, alternatively, that its phosphorylation is induced very rapidly by brief starvation from IL-3.
JNK phosphorylation is very low or not detectable in FL5.12/WT cells but is greatly increased in response to IGF-I stimulation of cells at 20 and 40 min (Fig. 1A). Interestingly, although the levels of JNK2 (56 kDa) protein seem to be higher than JNK1 (46 kDa) in these cells, the increase in phosphorylation is predominantly in JNK1 rather than in JNK2 (Fig. 1A). JNK activation is not detectable in mock-transfected FL5.12/ Neo cells compared with FL5.12/WT cells (Fig. 2B). An increase in phosphorylation of c-Jun is also detectable between 20 and 60 min in FL5.12/WT cells, and this is also not detectable in FL5.12/Neo cells (Fig. 2B). JNK and c-Jun are also robustly phosphorylated in response to stimulation with IL-3 for 15 min (data not shown).
To rule out the possibility that the increase in JNK activation and Jun phosphorylation in FL5.12/WT cells was due to a stress response caused by withdrawal of IL-3, cells were monitored for JNK activation at different times up to 60 min in the absence of any stimulation. Results shown in Fig. 1C compare levels of c-Jun phosphorylation in starved cells with levels of c-Jun phosphorylation following IGF-I stimulation for 20 min. This demonstrates that although low levels of JNK activity are detected in FL5.12 cells upon IL-3 withdrawal, these levels do not increase over time, whereas stimulation with IGF-I induces a robust activation of JNK.
By having established that IGF-I induces JNK activation in FL5.12 cells, we were next interested in examining the time course of JNK activation and c-Jun phosphorylation. This was investigated using a non-radioactive assay kit that is based on an antibody that specifically detects phosphorylation on serine 63 of Jun, which is phosphorylated by JNK. Fig. 1D shows that JNK phosphorylation activity peaks between 20 and 40 min of IGF-I stimulation and begins to decrease by 60 min. This is indicative of a transient activation of JNK in response to IGF-I and is different from prolonged activation of JNK induced by anisomycin and stress signals (36). Thus, overall these data demonstrate that IGF-I induces transient phosphorylation of both ERKs and JNK in FL5.12/WT cells and that p38 is constitutively active in these cells.

FIG. 1. Activation of MAPK and SAPK pathways in response to IGF-I stimulation of FL5.12/WT cells.
A, cells were washed in serum-free medium, starved for 3.5 h, and then stimulated with IGF-I for 20 or 40 min or left unstimulated (0). Cells were then lysed as described under "Experimental Procedures," and Western blotting was performed using antibodies against the phosphorylated or native forms of ERK, p38, and JNK. B, comparison of FL5.12/Neo with FL5.12/WT cells for JNK activation and phosphorylation of c-Jun in response to IGF-I stimulation for 20 and 60 min. Western blots were probed with antibodies against phospho-JNK or against phospho-Jun (serine 73). Anti-JNK and -actin antibodies were used to re-probe blots to control for protein loading. C, Jun phosphorylation (serine 73) in starved unstimulated FL5.12/WT cells was investigated using lysates from cells treated the same as in A but without IGF-I stimulation. This is compared with stimulation with IGF-I for 20 min. D, a time course of c-Jun phosphorylation was performed using cells that were treated as in A above and a SAPK kit to detect serine phosphorylation of glutathione S-transferase-Jun fusion peptides (serine 63) by Western blotting. Levels of JNK in the cell lysates are shown in the bottom panel. All of the Western blots are representative of several experiments with similar results.
Inhibition of JNK or p38 Abrogates Suppression of Apoptosis by IGF-I in FL5.12/WT Cells-Transient activation of JNKs has been associated with suppression of apoptosis in response to TNF-␣ in T lymphocytes (29), whereas dicumarol blocks JNK and NF-B activation leading to potentiation of apoptosis in HeLa cells (30). In order to determine whether the activation of JNK by IGF-I in FL5.12/WT cells shown above is associated with suppression of apoptosis, we sought to block JNK activation using dicumarol and also to determine the effects of dicumarol on IGF-I-mediated suppression of apoptosis. Previously, we have shown that IGF-I protects FL5.12/WT cells from IL-3 withdrawal, without promoting a proliferative response (21). To determine if dicumarol had an affect on IGF-I-mediated suppression of apoptosis FL5.12/WT cells were cultured in the presence of IL-3, IGF-I, or 5% FBS (control) with or without dicumarol at different concentrations ranging from 25 to 150 M. The results shown in Fig. 2A demonstrate that IGF-Imediated suppression of apoptosis is drastically inhibited by 100 M dicumarol, as is suppression of apoptosis afforded by 5% FBS in the control cells ( Fig. 2A). The viability of cells cultured in the presence of IL-3 is not altered significantly by incubation with 100 M dicumarol ( Fig. 2A), although we did notice that the proliferation of cells cultured in the presence of IL-3 was suppressed and that higher concentrations of dicumarol (150 and 200 M) lead to loss of viability in IL-3 cultures (not shown). A dose-response curve showing the effects of dicumarol on IGF-I-mediated suppression of apoptosis is shown in Fig.  2B. This demonstrates that in the presence of 25 M dicumarol there is no effect on IGF-I-mediated anti-apoptotic activity, but at 50 M there is some abrogation of IGF-I activity, with a progressive abrogation at 75 M to complete inhibition at 100 M dicumarol.
We next tested the effects of dicumarol on JNK activation by IGF-I as well as on the other signaling pathways that are activated by IGF-I or constitutively active in FL5.12/WT cells. The results shown in Fig. 2C demonstrate that IGF-I-mediated phosphorylation of c-Jun is completely inhibited up to 60 min in FL5.12/WT cells by preincubation of the cells with 100 M dicumarol for 5 min. IL-3-mediated phosphorylation of c-Jun is also inhibited under these conditions. In the same cells induction of AKT phosphorylation by IGF-I is not affected in the presence of dicumarol (Fig. 2C). We also investigated whether dicumarol had any influence on p38 or NF-B activity in these cells. p38 is constitutively phosphorylated in FL5.12/WT cells as shown above in Fig. 1A. The status of NF-B activity in FL5.12/WT cells was determined by measuring IB phosphorylation. As shown in Fig. 2D  phosphorylation in unstimulated cells, and this does not significantly increase with either IGF-I or IL-3 stimulation. Pretreatment of FL5.12/WT cells with dicumarol did not alter the levels of p38 and very slightly increased the levels of IB phosphorylation (Fig. 2D). Taken together these data indicate that inhibition of JNK activation with dicumarol abrogates IGF-I-mediated suppression of apoptosis despite the fact that AKT and presumably the PI 3-kinase pathway as well as p38 and NF-B all remain active. This result suggests that activation of the JNK pathway in these cells occurs independently of PI 3-kinase activation.
We next compared the effects of dicumarol with the effects of other MAPK inhibitors on IGF-I-mediated anti-apoptotic activity in FL5.12 cells. IGF-I-mediated protection from IL-3 withdrawal was measured in the absence or presence of the ERK inhibitor, PD98059, or the p38 inhibitor, SB203580. The ERK inhibitor, PD98059, does not have any affect on the viability of cells cultured in the presence of IGF-I or IL-3 (Fig. 3A), although it does inhibit ERK1 and -2 activation in response to IGF-I stimulation (Fig. 3B). We were also interested to determine whether inhibition of p38 activity could affect IGF-Imediated suppression of apoptosis because this protein is apparently constitutively active in FL5.12/WT cells. Results shown in Fig. 3C indicate that the viability of the cells cultured in the presence of IGF-I is diminished in the presence of the p38 inhibitor to below the levels of cells cultured in FBS alone at 48 h. The viability of cells cultured in the presence of 5% FBS only was also significantly decreased, but the viability or proliferation of cells cultured in IL-3 is not affected by SB203580. These data suggest that although IGF-I does not induce phosphorylation of p38 in FL5.12 cells, p38 activity contributes to IGF-I-mediated anti-apoptotic function. To confirm that the p38 inhibitor is acting on p38, we measured p38 phosphorylation in cells stimulated with IGF-I in the absence and presence of SB203580. p38 phosphorylation levels remain constant in the presence of IGF-I up to 60 min and slightly decrease at 90 min of stimulation (Fig. 3D). This phosphorylation is clearly inhibited by 10 M SB203580 in the absence of IGF-I and also up to 60 min of IGF-I stimulation (Fig. 3D). However, SB203580 does not inhibit JNK activation in response to IGF-I (Fig. 3D), suggesting that inhibition of JNK is not responsible for the loss of IGF-I-mediated anti-apoptotic activity.
Taken together, these data demonstrate that dicumarol can block JNK activation by IGF-I and block IGF-I-mediated suppression of apoptosis, without affecting AKT, p38, or NF-B activation. The activity of dicumarol is more potent than that of a p38 inhibitor and an ERK inhibitor that has no effect. This suggests that dicumarol inhibits a pathway involving JNK activation that is required for IGF-I-mediated suppression of apoptosis in these cells.

Transient Activation of the JNK Pathway by IGF-I Does Not
Require PI 3 Kinase Activity-Recruitment and activation of PI 3-kinase leading to activation of AKT is required for suppression of apoptosis by the IGF-IR in many cell types (10, 15) but is not sufficient in fibroblasts and FL5.12 cells (21,24). Since inhibition of JNK activation by dicumarol did not affect AKT activation by IGF-I, but still blocked suppression of apoptosis by IGF-I, we were interested to determine whether activation of JNK by IGF-I requires PI 3-kinase activity. To investigate this we measured JNK activation and suppression of apoptosis in response to IGF-I stimulation of FL5.12/WT cells in the absence or presence of the PI 3-kinase inhibitor LY294002. As shown in Fig. 4A, LY294002 completely suppresses IGF-I-mediated protection from IL-3 withdrawal in FL5.12/WT cells, whereas IL-3-mediated cell viability is not affected. However, in these cells although AKT activation is suppressed by   FIG. 3. Viability of FL5.12/WT cells in the presence and absence of the ERK inhibitor PD98049 or the p38 inhibitor SB203580. A, FL5.12 cells were cultured at 5 ϫ 10 5 /ml in medium containing 5% FBS (control), FBS plus IGF-I (50 ng/ml), or FBS plus IL-3 (WEHI CM, 10%) in the presence or absence of the MAPK inhibitor PD98049 (50 M), and cell viability was monitored at the indicated times using trypan blue exclusion. Data are presented as the mean and standard deviation from triplicate cultures. B, FL5.12/WT cells were seeded at 5 ϫ 10 5 cells/ml and grown overnight. Cells were then starved for 3 h and either left untreated (left 3 lanes) or incubated in the presence of 50 M PD98049 for 30 min (right 3 lanes) before stimulation with IGF-I for the indicated times. Western blotting was performed using anti-phospho-ERK and anti-ERK2 antibodies. C, viability curve for FL5.12/WT cells treated with the p38 inhibitor SB303580. Cells were cultured as described in A but in the presence or absence of 10 M of the p38 inhibitor SB203580. Data are presented as mean and standard deviation from triplicate cultures. D, FL5.12/WT cells were seeded at 5 ϫ 10 5 cells/ml and grown overnight. Cells were then starved for 3 h and treated with 10 M SB203580 or not for 30 min before 100 ng of IGF-I was added for up to 90 min or not (0). Cells were then washed and lysed as described under "Experimental Procedures," and Western blots were probed with antibodies against phospho-p38, p38, and phospho-Jun. LY294002, JNK activation still occurs in response to IGF-I as shown by phosphorylation of c-Jun in the presence of LY294002 (Fig. 4B).
To determine how PI 3-kinase is activated in response to IGF-I in FL5.12 cells, we investigated recruitment of p85 into IRS-1/2 and to Shc complexes. We could not detect p85 IRS-2 complexes or IRS-1 in FL5.12 cells, but as shown in Fig. 4C we found that IGF-I induced phosphorylation of Shc and the formation of Shc-Grb-2 complexes. A Shc-Grb-2 pathway to PI 3-kinase has been shown to occur via Gab-2 recruitment of p85 in BaF3 cells (37). Therefore, we investigated a role for Gab-2 in FL5.12 cells. We found that IGF-I induces the formation of phosphorylated Shc-Gab-2 complexes and p85-Gab-2 complexes. Grb-2 is in all Gab-2 complexes in agreement with Gu et al. (37). By immunoprecipitation of p85 we also saw IGF-Iinduced formation of p85-Grb-2 complexes. Altogether this indicates that IGF-I stimulation can lead to p85 recruitment and PI 3-kinase activation through recruitment of Shc-Grb-2 and Gab-2 in FL5.12 cells.
In summary these results demonstrate that JNK activation by IGF-I is not dependent on PI 3-kinase activity in FL5.12 cells. In addition, the results suggest that PI 3-kinase and JNK control signaling pathways are differentially activated by the IGF-IR. The effects of LY294002 and dicumarol indicate that each of these pathways contributes to, but neither is sufficient for, suppression of apoptosis by the IGF-IR.

IGF-I-activated Increase in JNK Is Not Dependent on Residues in the IGF-IR C Terminus or on IRS-1 and IRS-2, but
Requires Tyrosine 950 for Maximal Activation-By having established that activation of JNK by IGF-I is through a signaling pathway that is distinct from the PI 3-kinase pathway, we were next interested to determine if any of the domains of the IGF-IR previously associated with suppression of apoptosis were necessary for JNK activation. To do this IGF-I-mediated stimulation of JNK activity was measured in FL5.12 cells transfected with a series of IGF-IR mutants (Table I) and also in 32D cells, which lack IRS-1 and IRS-2. We were particularly interested in mutants of the IGF-IR C terminus. These mutants included receptors that have tyrosine 1251 mutated to phenylalanine (Y1251F) or histidine 1293 and lysine 1294 mutated to phenylalanine and arginine (H1293F/K1294R). Cells expressing either of these mutant receptors do not exhibit IGF-I-mediated anti-apoptotic activity (21) but can activate PI 3-kinase and AKT. However, as shown in Fig. 5A FL5.12 cells expressing these receptors could activate Jun phosphorylation in response to IGF-I stimulation. FL5.12 cells expressing a receptor truncated at amino acid 1245 (1245d) could also acti-

FIG. 4. Effects of PI 3-kinase inhibitor LY294002 on FL5.12/WT cell viability and IGF-I-mediated JNK activation.
A, FL5.12/WT cells were cultured at 5 ϫ 10 5 /ml in medium containing 5% FBS (control), FBS plus IGF-I (50 ng/ml), or FBS plus IL-3 (WEHI CM, 10%) in the presence or absence of 10 M LY294002. Cell viability was determined by counting triplicate samples at the indicated times using trypan blue exclusion. B, FL5.12/WT cells were seeded at 5 ϫ 10 5 cells/ml and grown overnight. Cells were then starved for 3 h and treated with 10 M LY294002 or nothing for 30 min either without or with IGF-I stimulation for the indicated times. Cell lysates were prepared for Western blotting with antibodies against phospho-Jun or phospho-AKT. Blots were also probed with anti-AKT antibody to control for AKT levels in the cells and for protein loading. C, FL5.12 cells were washed in serum-free medium, starved for 3.5 h, and then stimulated with IGF-I for 5 or 10 min or left unstimulated. Lysate were prepared and split for immunoprecipitation (I.P.) (10 7 cells/sample) with Shc or Gab-2. In a separate experiment cells were stimulated with IGF-I for 2 or 5 min, and p85 was immunoprecipitated from lysates prepared in the same way as described above. Immunoprecipitated proteins were separated by PAGE, and Western blotting performed with the antibodies indicated below each panel.

TABLE I
JNK activation in FL5.12 cells overexpressing C-terminal mutants of the IGF-IR Cells expressing mutant IGF-IRs were washed in serum-free medium, starved for 3.5 h, and stimulated with IGF-I for 20 or 40 min or left unstimulated. Cells were then lysed as described under "Experimental Procedures," and the lysates were separated on 4 -20% gradient SDS-PAGE. Western blots were performed using anti-phospho-JNK or anti-phospho-Jun antibodies to detect IGF-I-stimulated phosphorylation and activation of JNK. Blots were re-probed with anti-JNK antibody as a loading control.
ϩ Yes S1280-1283/A ϩ Yes H1293F/K1294R Ϫ Yes Y950F/1245d Ϫ Partial vate JNK (Table I), suggesting that the last 92 amino acids of the IGF-IR C terminus are not required for JNK activation. We then investigated whether an IGF-IR mutated at tyrosine 950 in the juxta-membrane region and also deleted at residue 1245 in the C terminus could activate JNK, and we found that FL5.12 cells expressing this mutant had greatly diminished JNK phosphorylation compared with FL5.12/WT cells (Fig.  5B). This result suggests that proteins that interact with tyrosine 950 are required for JNK activation.
To determine whether JNK activation requires signals from IRS-1 or IRS-2 that could bind to Tyr-950, we investigated IGF-I-mediated activation of JNK in the myeloid cell line 32D that stably expresses a WT IGF-IR. Stimulation of these cells with IGF-I in the absence of IL-3 resulted in activation of JNK with similar kinetics to those seen in FL5.12 cells (Fig. 5C). This suggests that IRS-1 and IRS-2 are not required for IGF-IR-mediated activation of JNK. Therefore, taken together with the results from FL5.12 cells expressing the different mutant receptors, we conclude that Tyr-950 is required for maximal JNK activation and that a major part of the IGF-IR C terminus as well as IRS-1 and IRS-2 are dispensable for JNK activation in response to IGF-I.

IGF-I and Anisomycin
Can Induce Additive SAPK Activity in FL5.12/WT Cells-Anisomycin has been used to study SAPK signaling and has been shown to be an enhancer of both persistent SAPK activity and apoptosis (38). Anisomycin desensitizes cells to further JNK activation in fibroblasts (44), and pretreatment of human embryonic kidney cells with IGF-I re-sulted in inhibition of anisomycin-induced SAPK activation (33). To further characterize JNK activation by IGF-I in FL5.12/WT cells and determine its association with suppression of apoptosis, we were therefore interested to determine whether IGF-I pretreatment could affect anisomycin-induced JNK activation. Cells were deprived of IL-3 and stimulated with IGF-I or nothing for 5 min, after which anisomycin was added to the cultures for a further 20 min. As shown in Fig. 6 anisomycin induced JNK phosphorylation, and this was enhanced in cells that were pre-stimulated with IGF-I. Therefore, in FL5.12/WT cells IGF-I does not block anisomycin-induced JNK activation. This result suggests that the IGF-IR activates a signaling pathway upstream of JNK in FL5.12 cells that is distinct from that activated by anisomycin.
Dicumarol Is Cytotoxic to the Breast Carcinoma Cell Line MCF-7-There is considerable evidence to support the idea that suppression of apoptosis by the IGF-IR contributes to tumorigenesis and to the survival and chemo-resistance of several kinds of tumor cells including those derived from breast, lung, and prostate (39). The results shown above suggest that JNK kinase activation by IGF-I is associated with IGF-I-mediated suppression of apoptosis and that dicumarol, a JNK inhibitor, is an effective inhibitor of IGF-I-mediated suppression of apoptosis. Therefore, we were interested to determine if dicumarol could induce apoptosis in tumor cells. To test this we used the breast carcinoma cell line MCF-7, which expresses high levels of IGF-IR, is very IGF-I responsive, and is quite resistant to chemotherapeutic drugs. As shown in Fig. 7A exposure of cultures of MCF-7 cells to dicumarol for 24 or 48 h results in a dose-responsive decrease in cell viability with more than 75% of the cells dead in the presence of 100 M dicumarol by 48 h. The morphology of remaining cells indicates shrinkage and loss of actin organization as shown by immunofluorescence staining with anti-actin antibodies (Fig. 7B). These results demonstrate that dicumarol potently induces apoptosis in MCF-7 cells and further suggest that targeting this pathway could be a specific way in which to inhibit IGF-IR anti-apoptotic signaling in tumor cells. DISCUSSION The role of the JNK signaling pathway in regulating responses to cellular stress and its association with induction of apoptosis in neuronal cells has been well documented. There has also been some evidence to indicate a role for the JNK pathway in promoting cell survival and tumorigenesis (40). Our findings that JNK activation occurs transiently in response to IGF-I stimulation of FL5.12 cells and that inhibition of JNK by dicumarol in these cells where the IGF-IR mediates anti-apoptotic signaling, but not proliferation, support a role for the stress pathways and for JNK activation in suppression

FIG. 5. IGF-I-induced activation of JNK in FL5.12 cells overexpressing mutant IGF-IRs and in 32D/IGF-IR cells.
A, the PerkinElmer Life Sciences SAPK assay kit was used to determine whether IGF-I activates JNK in FL5.12 cells that express mutant IGF-IRs (Y1251F and H1293F/K1294R). Cells were starved and stimulated as for FL5.12/WT cells described above. B, JNK activation in FL5.12 cells expressing the Y950F/1245d mutant compared with FL5.12/WT by Western blotting with an anti-phospho-JNK antibody. Cells were starved and stimulated as in Fig. 1B. 32D cells that overexpress the WT IGF-IR were assayed for their ability to phosphorylate c-Jun in response to IGF-I stimulation for the indicated times. Cells were treated in the same way as FL5.12 cells, and Western blotting was performed with anti-phospho-Jun antibody. Blots were re-probed with an anti-p38 antibody to control for protein levels.
FIG. 6. Anisomycin has an additive effect on IGF-I-induced JNK/Jun phosphorylation. FL5.12/WT cells were cultured in the presence of IL-3 for 24 h, starved for 3.5 h, pretreated with 100 ng of IGF-I for 5 min (ϩ) or not (Ϫ), and then treated with anisomycin (ϩ) or not (Ϫ) at a final concentration of 10 g/ml. Western blots were probed with anti-phospho-JNK and anti-phospho-Jun antibodies, and blots were probed with anti-actin antibody to control for protein levels.
of apoptosis by the IGF-IR. Our findings also suggest that activation of the JNK pathway by IGF-I is distinct from but might operate in addition to the PI 3-kinase/AKT pathway.
Activation of JNK in response to IGF-I stimulation of FL5.12 cells is reminiscent of the activity of p38 and JNK in TNF signaling described by Roulston et al. (29). This study showed that activated JNK and p38 were protective against apoptosis, during the early part of the signaling response to TNF, an effect that could be inhibited by actinomycin D. Similarly transforming growth factor-␤, which has been shown to prevent cell death upon serum withdrawal, induced a sustained phosphorylation of c-Jun with diminution of JNK activation later in the death response (41). Dicumarol could inhibit JNK activation by IGF-I in FL5.12/WT cells at concentrations where it did not inhibit AKT activation or constitutive p38 and NF-B activity. Dicumarol also inhibited IGF-I-mediated suppression of apoptosis at concentrations that did not inhibit the antiapoptotic activity of IL-3. All these data suggest that transient activation of JNK is involved in mediating anti-apoptotic signaling from the IGF-IR.
The potential molecular mechanisms of anti-apoptotic signals from transient activation of JNK in FL5.12 cells remain unclear. Two possible mechanisms are likely to be enhanced transcription of anti-apoptotic genes in the nucleus or signaling events in the cytoplasm that lead to activation of anti-apoptotic proteins. JNK activation induces the phosphorylation of several transcription factors including c-Jun, Elk-1, and ATF-2, which regulate immediate early gene expression (42,43). A role for c-Jun activity in survival signaling has been supported by the activity of non-phosphorylatable dominant-negative forms of c-Jun (44,45). These rendered fibroblasts incapable of survival following UV irradiation (46) and could also attenuate the anti-apoptotic effects of TGF-␤ upon serum withdrawal in A549 cells (41).
The effects of JNK may not be entirely due to transcriptional responses, and JNK has been shown to regulate signaling events near the plasma membrane and in the cytoplasm. For example JNK has recently been shown to phosphorylate IRS-1 on a regulatory serine in its phosphotyrosine binding domain resulting in inhibition of IRS-1 activity in Chinese hamster ovary cells (47). Other apoptosis-regulating proteins have also been shown to interact with JNK such as Bcl-X L and NFATc1 (48,49). Thus, the mechanisms by which JNK might regulate cell survival could include both transcriptional output in the nucleus and anti-apoptotic signaling events in the cytoplasm.
We did not detect increased p38 activity in response to IGF-I stimulation of FL5.12 cells, which is in contrast to observations in neuroblastoma cells where IGF-I suppressed JNK activity but enhanced p38 activity under conditions of stress (32). However, in FL5.12 cells p38 activity also appears to be a mediator of survival signals from the IGF-IR because inhibition of p38 activity resulted in diminution of IGF-I-mediated anti-apoptotic activity. This is in contrast to the observation that inhibition of ERK activity had no effect on cell viability either in the presence of IGF-I, IL-3, or 5% FBS. p38 has also been shown to have a role in survival signaling in neuronal PC12 cells where a requirement for p38 activity was demonstrated in the IGF-Imediated transcriptional activation of the cAMP-responsive element-binding protein, which can lead to increased Bcl-2 expression and increased cell survival (51). We have not been able to detect enhanced expression of Bcl-2 in FL5.12 cells in response to IGF-I stimulation, but we have detected slight increases in Bcl-X L levels (not shown). It also remains possible that other Bcl-2 family members or other survival genes could be targets of p38 or JNK in these cells.
JNK activation following IGF-I stimulation of FL5.12/ IGF-IR cells is PI 3-kinase-independent, and this result defines the JNK signaling pathway as a candidate to be regulated by residues in the IGF-IR C terminus that abrogate the antiapoptotic activity of the IGF-IR (21). However, an IGF-IR deleted at amino acid 1245 could still induce JNK activation in response to IGF-I (Table I and Fig. 5). Activation of JNK following IGF-I stimulation of the IRS-1 and -2 deficient 32D/ IGF-IR cells indicates that IRS-1 and IRS-2 are not required for JNK activation by the activated IGF-IR. However, mutation of tyrosine 950 almost completely inhibited JNK activation, suggesting that residues that interact with Tyr-950 such as Shc or Crk (52) are required for JNK activation. There are many candidate pathways leading to JNK including the Ste20-like kinase (53), Src, Cas, reactive oxygen species (54), and G i (55). We could not inhibit the IGF-I-mediated activation of JNK using G i inhibitors, but we could detect an increase in the autophosphorylation of the hematopoietic progenitor kinase (HPK), 3 min after IGF-I stimulation (data not shown). HPK has been shown to activate JNK via its interaction with Crk in a Rac and Cdc42-independent manner (56). Thus, a possible scenario is that an activated IGF-IR directly recruits Crk, or recruits SHC, which then forms a complex with Grb-2 and Crk that can associate with HPK. Activated HPK has been shown to activate JNK via MEKK/Tak 1 and MKK4 (57). All of this makes SHC, Crk, and HPK likely to be the molecules that mediate the signaling pathway between the IGF-IR and JNK.
The drastic effect that dicumarol had on MCF-7 cell survival suggests that this could be a useful therapeutic approach for killing tumor cells that overexpress the IGF-IR such as pros- tate, lung, and breast carcinomas. The IGF-IR and c-Jun have both been shown to be associated with cellular transformation, and both c-Jun and AP-1 signaling have been postulated to influence breast cancer phenotype (50). The tightening of the actin bundles that was observed following dicumarol treatment of MCF-7 cells (Fig. 7B) is in contrast to the loss of actin architecture seen in prostate carcinoma cells following the overexpression of PSK, a Ste20-like kinase that activates JNK (53).
In summary, we have demonstrated that transient activation of JNK occurs in response to IGF-I stimulation in FL5.12 cells and that dicumarol can block both IGF-I-mediated JNK activation and IGF-I-mediated suppression of apoptosis. The JNK pathway in these cells is distinct from the PI 3-kinase/ AKT activated by the IGF-IR and may be important in maintaining the survival and transformation of tumor cells.