Constitutive activation of c-Jun N-terminal kinase by a mutant epidermal growth factor receptor.

Epidermal growth factor receptor (EGF) variant type III (EGFRvIII) is a constitutively active, naturally occurring mutation of the EGF receptor that is found in many types of human tumors. When overexpressed in NIH3T3 fibroblasts, EGFRvIII induces transformation by enhancing cell growth and reducing apoptosis. Analysis of downstream signaling pathways has revealed that extracellular signal-regulated kinase activity is down-regulated, raising doubt as to the significance of this pathway in promoting transformation. We investigated whether the c-Jun N-terminal kinase (JNK) pathway was affected by EGFRvIII. NIH3T3 cells expressing EGFRvIII exhibited a high basal level of JNK activity, which was not present in cells overexpressing the normal EGF receptor. Treatment of cells overexpressing EGFRvIII with inhibitors of the EGF receptor or phosphatidylinositol 3-kinase resulted in the down-regulation of JNK activity. Furthermore, the down-regulation of JNK activity was associated with a loss of properties related to transformation, and there was no evidence for JNK activity in the promotion of apoptosis in these cells. These findings implicate constitutive activation of the JNK pathway in transformation by EGFRvIII.

The immediate activators of JNK, generically known as MAP kinase kinases, are MKK4 (4 -6) and the recently identified MKK7 (7). Although both ERKs and JNKs are activated in a similar manner, they define distinct pathways that respond to different extracellular stimuli. ERKs are preferentially activated in response to mitogens, whereas JNKs are potently activated by cytokines and environmental stresses (1). JNK activity in response to stress is associated with the induction of cell cycle arrest and apoptosis (8 -10). Although different stimuli preferentially activate specific MAP kinase pathways, a modest level of cross-activation of different MAP kinase pathways can result from the same stimulus (10 -14).
Aside from the use of a particular pathway, the duration of MAP kinase activity can also determine the physiological response of a cell. For example, transient induction of JNKs provides a growth enhancement signal, whereas persistent activity due to UV light or ␥-irradiation induces apoptosis (10). Consistent with this finding was the observation that ERK and JNK activation had opposing effects on cell survival in PC12 cells. Differentiated PC12 cells deprived of nerve growth factor underwent programmed cell death that was dependent on the down-regulation of ERK activity and was coupled with increased JNK activity (8).
Overexpression of the EGF receptor has been implicated in the progression of numerous types of human cancers, including breast carcinomas and astrocytic neoplasms (15,16). Transfection of the normal EGF receptor into fibroblast cell lines induces ligand-dependent transformation (17), providing a model in which to study the oncogenic effects of the receptor. The addition of EGF to cells overexpressing the normal EGF receptor induces a transient but potent activation of ERK (17,18). This signaling scheme has been established for many receptor tyrosine kinases and is associated with positive growth responses in cells (19).
In addition to overexpression of the normal EGF receptor, many rearrangements of the EGF receptor have been identified in human glial tumors (20,21). The most common of these spontaneously occurring mutant EGF receptors, known as EGFRvIII, has been studied extensively (20 -22). EGFRvIII arises from an in-frame deletion of nucleotides 276 -1075 in the normal EGF receptor cDNA (20). The resulting truncated receptor is similar to the normal EGF receptor in its ability to dimerize and autophosphorylate (18,23). However, EGFRvIII is distinct from the normal receptor in that it does not bind EGF and thus is an unregulated, constitutively active kinase (18,23).
Expression of EGFRvIII in NIH3T3 cells causes a highly transformed phenotype (17). Despite the fact that this receptor activates MEK, the immediate activator of ERK, there is a low basal level of ERK activity in cells expressing EGFRvIII (17,18). This implies that the down-regulation of ERK occurs at the MAP kinase level itself, most likely by increased expression of a phosphatase. Down-regulation of the ERK MAP kinase pathway in other types of transformed cells has also been reported (24,25).
Recently, we found that phosphatidylinositol (PI) 3-kinase is constitutively active in EGFRvIII-transformed cells and that this activity is required for colony formation in soft agar (26). Interestingly, PI 3-kinase has been implicated in the activation of the JNK pathway (27,28). We have investigated JNK activity in NIH3T3 cells expressing EGFRvIII and have found a high basal activity. Inhibition of EGFRvIII or PI 3-kinase activity down-regulated JNK activity, which was correlated with a loss of transformed morphology. Despite this chronic high level of JNK activity and the low level of ERK activity, these cells have a low rate of apoptosis.

EXPERIMENTAL PROCEDURES
Materials-Cell culture media and recombinant human EGF were from Life Technologies, Inc. Tyrphostin AG 1478 (Calbiochem) and LY 294002 (BIOMOL Research Laboratories Inc., Plymouth Meeting, PA) were dissolved in Me 2 SO to a concentration of 20 mM and stored at Ϫ80°C. [␥-32 P]ATP, 125 I-labeled goat anti-mouse IgG, and 125 I-labeled goat anti-rabbit IgG were from NEN Life Science Products. Unlabeled ATP was from Pharmacia Biotech Inc. The anti-phosphotyrosine monoclonal antibody PY20 and the anti-pan-ERK monoclonal antibody were from Transduction Laboratories (Lexington, KY). The anti-JNK1(C-17) polyclonal antibody was from Santa Cruz (Santa Cruz Biotechnology, Santa Cruz, CA), and the anti-phospho-p44/p42 MAP kinase monoclonal antibody was from New England Biolabs Inc. (Beverly, MA). All other materials were from Fisher unless otherwise indicated.
Western Blot Analysis-Total cell lysate (30 g) of each sample was subjected to SDS-polyacrylamide gel electrophoresis on 8.5% acrylamide gels for resolution of MAP kinases and 4 -20% Tris/glycine gels for resolution of the EGF receptor. Protein was transferred to nitrocellulose filters and blocked for 2 h at room temperature in 100 mM Tris, pH 7.5, 0.9% NaCl, and 0.1% Tween 20 with 5% nonfat dry milk. The antiphospho-MAP kinase monoclonal antibody was used at a 1:1000 dilution, whereas the anti-phosphotyrosine monoclonal antibody, the antipan-ERK monoclonal antibody, and the anti-JNK1 polyclonal antibody were used at 1 g/ml in the same blocking solution. 125 I-Labeled goat anti-mouse IgG (5 ϫ 10 5 cpm/ml) was used to detect the monoclonal antibodies, and 125 I-labeled goat anti-rabbit IgG was used to detect the polyclonal antibodies. Quantitation of signals was performed on a Phos-phorImager using ImageQuant software (Molecular Dynamics, Inc., Sunnyvale, CA).
Anchorage-independent Growth Assays-Soft agar assays were performed as described previously (17) with slight modification. Briefly, cells were fed 1 ml of fresh medium once a week for 3 weeks and treated with 0.1 or 1 M tyrphostin AG 1478 every 3 days. After 3 weeks of growth, the colonies were counted.
Solid-phase c-Jun Kinase Assay-GST-c-Jun (amino acids 1-79) was bound to glutathione-Sepharose beads (Pharmacia Biotech Inc.). JNK was precipitated from 300 g of whole cell extracts with GST-c-Jun for 3 h at 4°C. Precipitates were washed three times in phosphate-buffered saline containing 1% Nonidet P-40 and 2 mM NaVO 4 and rinsed once in kinase reaction buffer (25 mM Hepes, 25 mM MgCl 2 , 2 mM dithiothreitol, 0.1 mM NaVO 4 , and 25 mM ␤-glycerophosphoric acid). JNK activity was assayed by resuspending the pellets in 30 l of kinase reaction buffer containing 5 Ci of [␥-32 P]ATP and 20 M unlabeled ATP. Following a 20-min incubation at 30°C, samples were separated by SDS-polyacrylamide gel electrophoresis and transferred to nitrocellulose, and kinase activity was quantitated using the PhosphorImager.
Assay for Apoptosis-Cells were seeded on coverslips in 6-well dishes and grown in complete medium for 3 days. The cells were then incubated in serum-free DMEM containing either 2 M tyrphostin AG 1478 or 5 M LY 294002 for 1 or 3 days and fixed and stained with 4,6diamidino-2-phenylindole (2 g/ml) for viewing by fluorescence microscopy. Apoptotic cells were identified by condensed nuclei.

RESULTS
EGFRvIII Constitutively Activates JNK-Previous work on the signaling effects of EGFRvIII in NIH3T3 fibroblasts showed that ERK activity was down-regulated (17,18), suggesting that signaling though the Ras/ERK pathway was not required for transformation. We wished to investigate whether an additional MAP kinase pathway was affected by the signaling of EGFRvIII. We chose to study the effect of the mutant receptor on the JNK pathway. JNKs are defined in general by their ability to phosphorylate two serine residues in the aminoterminal domain of c-Jun in vitro (29,30). These enzymes have a high affinity for this substrate, so a solid-phase kinase assay utilizing GST-c-Jun (amino acids 1-79) as a substrate for JNK allows measurement of the kinase activity associated with a cell lysate. We employed this technique to compare JNK activity in NIH3T3 cells transfected with an empty expression vector (LTR b2); the normal human EGF receptor (CO12 20c2/b); and the mutant EGF receptor, EGFRvIII (HC2 20d2/c). Unstimulated LTR b2 and CO12 20c2/b cells exhibited very low levels of JNK activity that was modestly enhanced by the addition of EGF (100 ng/ml). In contrast, unstimulated HC2 20d2/c cells had a 13-fold higher JNK activity relative to unstimulated LTR b2 cells (Fig. 1, Kinase Assay), indicating a high constitutive JNK activity in these cells. The addition of EGF to HC2 20d2/c cells did not alter the level of JNK activity, which is consistent with the report that EGFRvIII cannot bind EGF and so is not influenced by this mitogen (18). Western blot analysis showed that JNK expression was nearly equivalent in all three cell lines (Fig. 1, ␣JNK), indicating that the increased JNK activity is not due to the up-regulation of protein expression.
Osmotic stress is known to potently induce JNK activity and can be used to measure the maximum potential in vivo activity of this kinase. Osmotic stress was induced by incubation of cell lines for 15 min in DMEM containing 500 mM sorbitol. This potently increased JNK activity in the transfected cell lines relative to EGF stimulation (Fig. 1). Sorbitol-treated HC2 20d2/c cells achieved the highest level of JNK activation, indicating that the signaling pathway leading to JNK activation was also enhanced in these cells.
ERK activities were determined for LTR b2, CO12 20c2/b, and HC2 20d2/c cell lines under the same conditions used to analyze JNK activity. ERK activity was measured using an antibody that specifically recognizes the phosphorylated forms of ERK1/2. Consistent with earlier findings, EGF stimulation of CO12 20c2/b cells activated ERK, whereas unstimulated or EGF-treated HC2 20d2/c cells had undetectable levels of ERK activity (Fig. 1, ␣Active MAPK) despite equivalent protein expression (Fig. 1, ␣ERK). The anti-ERK antibody also revealed a slower mobility form of ERK in this blot (Fig. 1, ␣ERK) that corresponds to the active form of ERK, but the anti-active MAP kinase antibody appeared to be more sensitive at detecting activated ERK. Incubation of cells with sorbitol had no detectable effect on ERK activation in HC2 20d2/c cells and showed only a slight increase in LTR b2 cells (Fig. 1, ␣Active MAPK). However, sorbitol-treated CO12 20c2/b cells did show substantial ERK activation. The exaggerated JNK and ERK activation observed in sorbitol-treated CO12 20c2/b cells relative to LTR b2 cells suggests that the EGF receptor may enhance the response of NIH3T3 cells to osmotic stress. It has been reported that sorbitol treatment of HeLa cells can result in EGF receptor phosphorylation (13). We tested if the increases in ERK and JNK activities induced by osmotic shock in our cell lines were due to receptor activation. Anti-phosphotyrosine blots revealed that EGF receptor activity in sorbitol-treated CO12 20c2/b and LTR b2 cells was not elevated above basal levels (Fig. 1, ␣pTyr), whereas EGF stimulation of CO12 20c2/b and LTR b2 resulted in EGF receptor phosphorylation. Taken together, these results indicate that overexpression of the EGF receptor indirectly enhances JNK and ERK responses to stress and further demonstrate the differences in JNK versus ERK activation in cells overexpressing the mutant receptor.
Continuous Stimulation of the Normal EGF Receptor Does Not Result in Elevated Levels of JNK Activity-Since the constitutive activity of EGFRvIII resulted in a high basal level of JNK activity, we wanted to determine if the normal EGF receptor could produce the same effect when continuously stimulated with ligand. Quiescent CO12 20c2/b cells were stimulated with EGF (100 ng/ml) and assayed at certain intervals for the activation of the EGF receptor, JNK, and ERK. CO12 20c2/b cells stimulated with EGF showed a maximal activation of the EGF receptor at 5 min, which slowly declined to basal levels by 24 h of stimulation (Fig. 2, ␣pTyr). ERK induction coincided with EGF receptor activation, which was maximal at 5 min of ligand stimulation and declined to undetectable levels by 24 h (Fig. 2, ␣Active MAPK). Maximal JNK activity was detected after 5 min of EGF stimulation, but was greatly reduced by 1 h (Fig. 2, Kinase Assay). JNK activity was above basal levels up to 48 h of EGF stimulation, but was still well below that seen in HC2 20d2/c cells (Figs. 1 and 2).
Inhibition of EGFRvIII Down-regulates JNK, and Receptor Reactivation Leads to Transient ERK but Sustained JNK Activation-Other studies (11,12) and our data have shown that the normal EGF receptor does not strongly activate JNK. We wished to determine if the constitutive JNK activity associated with the HC2 20d2/c cell line was an effect of signaling from the mutant receptor. To examine this, we utilized a specific inhibitor of the EGF receptor, tyrphostin AG 1478. Previous work has shown that HC2 20d2/c cells maintained in 2 M tyrphostin AG 1478 for 3 days and then serum-starved for 10 h showed a nearly complete loss of tyrosine phosphorylation of EGFRvIII (26). Tyrphostin AG 1478 inhibited ERK activation of EGFtreated, but not platelet-derived growth factor-treated, CO12 20c2/b cells, indicating the specificity of the inhibitor for the EGF receptor (26). When HC2 20d2/c cells were treated with tyrphostin AG 1478, there was a reduction in JNK activity in association with inhibition of receptor activity (Fig. 3A, ␣pTyr and Kinase Assay, 0 time point).
To study responses after receptor reactivation, cells were then switched to serum-free DMEM. A transient activation of ERK occurred with the initial reactivation of the mutant receptor. Interestingly, ERK activity was maximal 5 min after release from tyrphostin AG 1478, was dramatically reduced 6 h after release, and declined to undetectable levels within 24 h (Fig. 3A, ␣Active MAPK). In contrast to ERKs, JNK activity remained at levels similar to those in HC2 20d2/c cells that had not been released from tyrphostin AG 1478 until after 6 h (Fig.  3A, Kinase Assay). Increased levels of JNK activity became evident 12 h after release and reached levels that were equivalent to those in untreated HC2 20d2/c cells by 24 h of release. The increased JNK activity following release from tyrphostin AG 1478 correlated with the reestablishment of maximal activity of EGFRvIII (Fig. 3A, ␣pTyr).
Transformed Morphology of HC2 20d2/c Cells Parallels JNK Activity-We examined the morphology of these cells after tyrphostin AG 1478 treatment and then following release. Cells treated as described above showed a reversion to the normal planar appearance of fibroblasts. Following release from tyrphostin AG 1478, HC2 20d2/c cells displayed a similar morphology for the first 12 h of release (Fig. 3B). However, between 12 and 24 h of release, these cells regained their transformed appearance and resembled untreated HC2 20d2/c cells, which coincided with the reactivation of JNK. agar when grown in complete medium (Fig. 3C, CM). Treatment of HC2 20d2/c cells with 0.1 M tyrphostin AG 1478 resulted in a 37% reduction in colony formation, whereas treatment with 1.0 M tyrphostin AG 1478 resulted in a 94% inhibition of colony growth. Tyrphostin AG 1478 treatment of CO12 20c2/b cells yielded the same profile seen in tyrphostin AG 1478-treated HC2 20d2/c cells. Colony formation of CO12 20c2/b cells in the presence of 0.1 M tyrphostin AG 1478 was reduced by 64% and by 92% with 1.0 M tyrphostin AG 1478 (Fig. 3C). These data further suggest that the relationship between the constitutive activities of EGFRvIII and JNK is critical for the transformation state of cells expressing this mutant EGF receptor.
Constitutive JNK Activity Is Dependent on PI 3-Kinase in HC2 20d2/c Cells-Our laboratory has recently found that one of the effects of chronic signaling from EGFRvIII is constitutive PI 3-kinase activity (26). Recently, it was shown that constitutively active forms of PI 3-kinase activate the JNK pathway (27,28). This led us to ask if the high level of PI 3-kinase activity was responsible for the high basal level of JNK activity in HC2 20d2/c cells. HC2 20d2/c and CO12 20c2/b clones were treated with LY 294002, a specific inhibitor of PI 3-kinase activity, in the same manner as described for tyrphostin AG 1478. The inhibition of PI 3-kinase activity in HC2 20d2/c cells resulted in a corresponding reduction in JNK activity without affecting ERK activity (Fig. 4A). The addition of LY 294002 to HC2 20d2/c cells resulted in a partial morphological reversion and growth inhibition (Fig. 4B). In a previous publication (26), we also found that treatment of HC2 20d2/c cells with 2 and 5 M LY 294002 caused 71 and 99% reductions in soft agar colony efficiency. Interestingly, the reduction of JNK activity induced in LY 294002-treated HC2 20d2/c cells was nearly equivalent to that seen in tyrphostin AG 1478-treated HC2 20d2/c cells, indicating that the high basal level of JNK activity was dependent on the up-regulated PI 3-kinase activity. JNK and ERK activation levels were also determined in LY 294002-treated CO12 20c2/b cells. EGF-stimulated CO12 20c2/b cells showed similar levels of ERK activation regardless of treatment with the PI 3-kinase inhibitor (Fig. 4A, ␣Active MAPK). However, EGF-stimulated CO12 20c2/b cells showed a decrease in JNK activation when compared with cells treated with LY 294002 (Fig. 4A, Kinase Assay).
JNK Activity Does Not Correlate with the Induction of Apoptosis in Cells Expressing EGFRvIII-In PC12 cells, prolonged JNK activity and a loss of ERK activity correlate with the induction of apoptosis (8). HC2 20d2/c cells possess the same enzymatic profile, but, in contrast, are actively growing. Since this is in the presence of serum, we examined HC2 20d2/c cells following 1 and 3 days of serum starvation for the induction of apoptosis and compared the results with LTR b2 cells. Cells were assayed for the presence of nuclear condensation and JNK activity. A small percentage of apoptotic cells was noted in the LTR b2 cell line following 1 day of serum starvation, whereas there was negligible apoptosis in the HC2 20d2/c cell line (Fig.  5, bar graphs). The low basal JNK activity in LTR b2 cells was further decreased after 1 day of serum starvation, whereas the robust activity in HC2 20d2/c cells remained unaltered (Fig. 5, Kinase Assay). After 3 days of serum starvation, there was a large increase in the number of apoptotic cells in the LTR b2 cell line, but only a slight increase in the HC2 20d2/c cell line. JNK activity in LTR b2 cells remained extremely low, whereas there was actually a 77% decrease in HC2 20d2/c cells. Since EGFRvIII and PI 3-kinase contribute to the enhanced growth rate in HC2 20d2/c cells, we examined the effect of tyrphostin AG 1478 and LY 294002 on apoptosis. After 1 day of treatment, tyrphostin AG 1478 showed a small effect on LTR b2 cells, but by 3 days, there was a high percentage of apoptotic cells in both the LTR b2 and HC2 20d2/c cell lines (Fig. 5, bar graphs). LY 294002 also increased the incidence of apoptosis in both cell lines, although not to the same extent. There was no JNK activity in LTR b2 cells in the presence of either inhibitor, whereas there was a 60% decrease in HC2 20d2/c cells (Fig. 5, Kinase Assay). Taken together, these data suggest that EGFR-vIII and PI 3-kinase contribute to cell survival, but that JNK activity does not contribute to cell death. DISCUSSION Expression of mutant forms of the EGF receptor has been implicated in the progression of many types of human tumors (20,21). EGFRvIII is the most frequently occurring natural alteration associated with this receptor (20 -22). We studied NIH3T3 fibroblasts transfected with EGFRvIII to elucidate the signaling mechanisms responsible for transformation in these cells. Work in this and other laboratories showed that activation of the normal EGF receptor resulted in a transient but potent ERK induction, whereas EGFRvIII generated minimal ERK induction (17,18). This low level of ERK activity raised questions as to the significance of the ERK pathway in promoting transformation. Analysis of the JNK pathway showed that EGFRvIII-transfected cells displayed a high constitutive level of JNK activity, which was not present in cells overexpressing the normal EGF receptor. Although stimulation of the EGF receptor can transiently activate JNK, the constitutive activation of JNK seen in EGFRvIII-transfected NIH3T3 cells may define a novel mechanism associated with transformation by this receptor.
We have further defined the pathway by identifying PI 3-kinase as the upstream activator of JNK. Interestingly, EGFR-vIII was recently shown to constitutively activate PI 3-kinase, and this activity was essential for the transforming potential of the mutant receptor (26). Inhibition of PI 3-kinase activity associated with HC2 20d2/c cells resulted in a reduction in JNK activity and partial morphological reversion of the transformed appearance of HC2 20d2/c cells. The ability of PI 3-kinase to influence both cell morphology and JNK activity could be explained by its interaction with members of the Rho family of proteins (31,32), which are known to regulate the actin cytoskeleton (33,34) and to activate the JNK pathway (35)(36)(37)(38). Taken together, we speculate that the constitutive PI 3-kinase activity induced by EGFRvIII expression is responsible for the high basal level of JNK activity and that this contributes to the transformed phenotype induced by EGFRvIII. Following the incubation period, cells were assayed for apoptosis as described under "Experimental Procedures." Cells with condensed nuclei were classified as apoptotic, and the percentage of cell death was determined by calculating the ratio of apoptotic to non-apoptotic cells. The apoptotic assay results for LTR b2 and HC2 20d2/c cells are represented in the bar graphs. To analyze the JNK activity associated with the apoptotic response of LTR b2 and HC2 20d2/c cells under the same culture conditions, a duplicate experiment was performed, and cell lysates were collected at the various time points. JNK activity was determined by an in vitro c-Jun kinase assay of the whole cell lysates. An autoradiogram of the kinase assay is shown.
The low level of ERK activation exhibited by cells expressing the mutant receptor has been suggested to be regulated by expression of a phosphatase (17). Interestingly, in NIH3T3 fibroblasts, expression of MAP kinase phosphatase-1, a phosphatase specific for ERKs, is induced by activation of the JNK pathway (39). The constitutive JNK activity associated with EGFRvIII-transfected cells may induce constant expression of MAP kinase phosphatase-1, resulting in the low basal ERK activity exhibited in these cells. We are currently determining if this is the mechanism responsible for down-regulating ERK activation in HC2 20d2/c cells and if such down-regulation is essential for the transformation potential of the mutant receptor.
In support of our data implicating continuous JNK activity with a growth advantage in EGFRvIII transfectants, recent studies found that the JNK pathway was required for transformation by both the Trp-Met (40) and Bcr-Abl (41) oncoproteins. Like EGFRvIII, the Trp-Met oncoprotein is a constitutively active receptor tyrosine kinase that is not influenced by ligand (42). Dominant-negatives forms of Grb2 transformed into cells expressing the Trp-Met protein resulted in a reverted morphology that was associated with both down-regulated PI 3-kinase and JNK activities (40). It is interesting to speculate that preferential activation of the JNK pathway may be a characteristic of unregulated kinases with high transformation potential, such as seen in fibroblasts expressing EGFRvIII, Trp-Met, or Bcr-Abl.
Analysis of EGFRvIII in NIH3T3 cells has provided a model in which the transforming effects of this mutant receptor can be elucidated. We have shown that the transformed state of EGFRvIII-expressing cells is associated with the preferential and constitutive activation of the JNK pathway. These findings also demonstrate that cells are capable of sustained JNK activity and low ERK activity without the induction of apoptosis. In light of our findings and recent reports by others (40,41) implicating the JNK pathway in the promotion of transformation, it is becoming clear that certain oncoproteins and/or cell types actually utilize JNK and not ERK for transformation.