In human glioblastomas, the most common mutation of epidermal growth factor receptor (EGFR) is an in-frame deletion of an 801-bp sequence in the extracellular domain of EGFR termed EGFRvIII. The EGFRvIII does not bind ligand EGF but has constitutive tyrosine phosphorylation (pTyr) content and kinase activity that result in enhanced transformation, reduced apoptosis, and resistance to therapy. Here we report that the protein tyrosine phosphatase SHP-2 modulates a mitogen-activated protein kinase (MAPK) kinase (MEK)-mediated signaling pathway that regulates EGFRvIII pTyr and cell survival in U87MG.EGFRvIII cells. Overexpression of the phosphatase-inactive form of SHP-2 inhibited EGFRvIII pTyr by decreasing MAPK phosphorylation. Consistent with this, we observed that the MEK inhibitor PD98059, but not the phosphatidylinositol 3′-kinase inhibitor LY294002, inhibited EGFRvIII pTyr. Furthermore, constitutive EGFRvIII pTyr content observed in U87MG, LN229, and U373MG glioblastoma cells, but not in NR6.EGFRvIII fibroblasts, correlated with elevated MAPK levels in these cells. Interestingly, LY294002, but not PD98059, inhibited wild-type EGFR pTyr in response to EGF treatment in U87MG parental cells and in wild-type EGFR-overexpressing U87MG cells. Inhibition of EGFRvIII pTyr by PD98059 was not observed to be phosphorylation site specific. However, LY294002 more specifically inhibited wild-type EGFR pTyr at residues Tyr992 and Tyr1068 in the COOH terminus. Treatment of U87MG.EGFRvIII cells with PD98059, but not LY294002, also resulted in increased cell death in response to cisplatin. Collectively, a distinct MEK-mediated pathway in human glioblastoma cells appears to differentially modulate EGFRvIII and wild-type EGFR pTyr, and inhibition of the MAPK pathway sensitizes EGFRvIII-containing human glioblastoma cells to cisplatin-induced cell death.
The epidermal growth factor receptor (EGFR) plays an important role in transducing signals in normal cells to regulate cell proliferation and differentiation (1 , 2) . Deregulation of EGFR by elevated expression, mutation, and/or gene rearrangement has been observed in human cancers, including human glioblastomas (3, 4, 5) . The most common mutation of the EGFR is an in-frame deletion of an 801-bp sequence in the extracellular domain of EGFR that results in a truncated receptor termed EGFRvIII (ΔEGFR or del2–7EGFR) (6 , 7) . The EGFRvIII oncoprotein does not bind ligand EGF but has constitutive tyrosine phosphorylation (pTyr) content and kinase activity that results in enhanced transformation, reduced apoptosis, and resistance to therapy (8 , 9) . These properties distinguish the naturally occurring EGFRvIII oncoprotein from the wild-type EGFR.
The EGFRvIII oncoprotein is expressed only in human cancers and has not been observed in normal tissues (10 , 11) . This differential expression suggests its importance in tumorigenicity. EGFRvIII transforms NIH-mouse fibroblast cells and NR6 cells (12 , 13) and enhances human breast cancer and human glioblastoma malignancy (14 , 15) . EGFR mutations, including EGFRvIII, are found in glioblastomas but not typically in low-grade astrocytomas, suggesting a role for this mutation in the progression of astrocytoma tumorigenicity (16) . It has been observed that EGFRvIII is constitutively phosphorylated in U87MG cells transfected with this oncoprotein (17 , 18) , and the elevated phosphorylation and associated kinase activity enhance U87MG transformation (17) . However, it also has been observed that EGFRvIII transforms NR6 fibroblasts without an apparent increase in its phosphorylation under in vitro growth conditions (13) . Moreover, constitutive phosphorylation of the v-erbB oncogene product is not required for transformation of fibroblasts (19) . Therefore, the pTyr content of the EGFRvIII oncoprotein may be cell-type specific and may be required for particular transforming functions in human glioblastoma cells. It is of interest to consider whether EGFRvIII pTyr is regulated by specific intrinsic factors in human glioblastoma cells that may lead to the identification of new and specific targets for the treatment of human glioblastoma multiforme.
SHP-2 is a protein tyrosine phosphatase (PTP) and contains two Src homology-2 (SH2) domains located in its NH2 terminus and a COOH-terminal PTP domain (20) . This specific molecular structure suggests that SHP-2 is involved in regulating signals initiated by receptor tyrosine kinases (RTKs; ref. 21 ). In response to growth factor stimulation, growth factor receptors become transphosphorylated on COOH-terminal tyrosine residues. These phosphotyrosines act as docking sites for recruitment of SH2 molecules, including SHP-2, to activate downstream signaling cascades, such as mitogen-activated protein kinase (MAPK) and/or protein kinase B/Akt pathways. SHP-2 promotes extracellular signal-regulated kinase (ERK) activation stimulated by EGF and insulin, and the phosphatase activity of SHP-2 is required (22, 23, 24) . In response to cellular stresses (e.g., heat shock and UV irradiation), SHP-2 can differentially regulate ERK or c-Jun NH2-terminal kinase (JNK) activities and influence cell survival (25) . In glioblastoma cells, SHP-2 phosphatase activity is important for phosphatidylinositol 3′-kinase (PI3k) activation in response to growth factor treatment (26) . As a protein phosphatase, SHP-2 can dephosphorylate several important molecules, including EGFR and Gab1 (27) . It has been found that the Gab1-SHP-2 interaction is required for the regulation of ERK (28) and PI3k activities (29) , indicating that SHP-2 couples RTKs to downstream signaling modules.
Little is known regarding how cell signaling pathways modulate the activation of upstream oncogenic growth factor receptors. To address whether SHP-2 can modulate EGFRvIII pTyr and function in U87MG cells, we used genetic approaches in our experiments and showed that a phosphatase-inactive form of the SHP-2 PTP (SHP-2 C459S) down-regulated EGFRvIII pTyr by down-modulating MAPK activation in U87MG.EGFRvIII cells. We also showed that MAPK activation was required for EGFRvIII pTyr and cell survival. Conversely, wild-type EGFR pTyr depended on PI3k activity. These data suggest that the basal level of EGFRvIII pTyr, kinase activity, and survival signaling in U87MG glioblastoma cells are modulated by the activation state of the MAPK signaling pathway. These data also suggest a feedback mechanism by which downstream signaling modules affect upstream receptor activation and may have important therapeutic significance for the treatment of human cancers, including glioblastomas that express the EGFRvIII oncoprotein.
MATERIALS AND METHODS
Fetal bovine serum (FBS) was purchased from HyClone (Logan, UT). Dulbecco’s modified Eagle’s medium (DMEM) was obtained from Mediatech Cellgro (Kansas City, MO). Penicillin, streptomycin, and Lipofectamine were purchased from Invitrogen Life Technologies, Inc. (Carlsbad, CA). The inhibitors PD98059 and LY294002 were obtained from Biomol Research Laboratories, Inc. (Plymouth Meeting, PA). Acetyl-Asp-Glu-Val-Asp-p-nitroanilide was purchased from Enzyme System Products (Livermore, CA). Antibodies anti-p-Tyr (PY20), anti-EGFR, and anti-SHP-2 were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-phospho-Akt, anti-Akt, and anti-phospho-EGFR site-specific antibodies (Tyr845, Tyr992, and Tyr1068) were purchased from Cell Signaling Technology (Beverly, MA). Other anti-phospho-EGFR site-specific antibodies were Tyr1173 (Upstate Biotechnology, Inc., Lake Placid, NY) or Tyr1086 and Tyr1148 (Biosource International, Camarillo, CA). Anti-phospho MAPK and anti-ERK1/2 antibodies were purchased from Promega Corporation (Madison, WI). All of the other chemicals were obtained from commercial sources.
Cell Culture and Transfection Procedure.
Parental U87MG and the clonal derivatives U87MG.EGFRvIII (original clone termed U87MG.ΔEGFR) and U87MG.wtEGFR cells, which were stably transfected with either EGFRvIII or wild-type EGFR, were obtained from Dr. Webster Cavenee (Ludwig Cancer Institute, San Diego, CA; ref. 9 ). NR6.EGFRvIII (also termed NR6.ΔEGFR) cells are a subclone of mouse NR6 fibroblast cells, which stably express EGFRvIII and also were obtained from Dr. Webster Cavenee. All of the cell lines were grown in DMEM supplemented with 10% FBS, 100 units/mL penicillin, and 100 μg/mL streptomycin and were cultured at 37°C in a humidified atmosphere containing 5% CO2. The medium for stable transfectants contained 400 μg/mL G418.
Wild-type and various Shp-2 mutant cDNAs (the phosphatase-inactive form of Shp-2, Shp-2 C459S, and two independent SH2 domain point mutants of Shp-2, Shp-2 R32E, and Shp-2 R138E, which impair SHP-2 binding to phosphorylated tyrosine residues) were obtained from Dr. Gibbes Johnson (Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, MD; ref. 26 ) and were subcloned into the pcDNA3.1 myc/his vector (Invitrogen). U87MG.EGFRvIII cells were transfected with either pcDNA3.1 myc/his empty vector or a vector containing an Shp2 mutant cDNA using Lipofectamine (Life Technologies, Rockville, MD). Eighty micrograms per milliliter of hygromycin were used to select for stable clones. After 2- to 4-week selection, clones were examined by Western blot analysis to confirm expression of protein encoded by each cDNA. For transient transfection, U87MG, LN229, and U373MG cells were transfected with EGFRvIII cDNA using FUGENE 6 (Roche Molecular Biochemicals, Basel, Switzerland). Forty-eight hours later, cells were harvested for analysis.
Western Blot and Immunoprecipitation Analysis.
Cells were harvested and lysed in RIPA buffer on ice for 10 minutes [20 mmol/L Tris-HCl (pH 7.5), 0.15 mol/L NaCl, 1% Triton X-100, 1 mmol/L phenylmethylsulfonyl fluoride, 1 mmol/L Na3VO4, 1 mmol/L EGTA, 1 μg/mL leupeptin, 2 μg/mL aprotinin, and 10 μg/mL pepstatin]. Supernatant protein concentration was measured using the Bio-Rad Dc Protein Assay kit (Hercules, CA). For Western blot analysis, an equal amount of protein (25 μg/lane) was loaded. For immunoprecipitation experiments, lysates containing 500 μg protein were incubated with anti-EGFR antibody overnight. Protein A-Sepharose (Sigma, St. Louis, MO) then was added for 1 hour, and the beads were washed three times with lysis buffer and boiled for 5 minutes in sample preparation buffer [250 mmol/L Tris (pH 6.8), 10% SDS, 10% β-mercaptoethanol, and 40% glycerol]. The proteins were electrophoresed in 8% SDS polyacrylamide gels and transferred to nitrocellulose membranes. Membranes were probed with specific antibodies and detected by an enhanced chemiluminescence detection kit (Amersham Bioscience, Piscataway, NJ). The bands in Fig. 5C ⇓ (Lanes 1 and 2) were analyzed using the NIH image program.
Measurement of Caspase-3 Protease Activity.
Caspase-3 activity was assayed by the cleavage of Acetyl-Asp-Glu-Val-Asp-p-nitroanilide. Briefly, cells (1 to 2 × 106/dish) were cultured in 10-cm dishes. After attachment to dishes, cells were treated with either 50 μmol/L PD98059 or 20 μmol/L LY294002 for 48 hours. Inhibitors then were removed, and cells were treated with 50 μmol/L cisplatin for 72 hours in medium as described previously (9) and modified. Cells were harvested, and cell pellets were incubated in hypotonic buffer for 30 minutes on ice. After homogenization and centrifugation, the supernatant was transferred to a new tube, and the protein concentration was measured. The same amount of protein (50 μg) was used to test for protease activity. After a 1-hour incubation at 30°C, p-nitroanaline release was monitored at 405 nm by an ELISA reader and calculated from a standard curve and expressed as picomole per milligram protein per minute (30) .
Lactate Dehydrogenase Assay.
In each experiment, cells were incubated with inhibitors for 48 hours. The inhibitors then were washed away, and the cells were treated with 50 μmol/L of cisplatin for 72 hours. The medium from one of the duplicate wells was collected, and 0.1% Triton X-100 was added to the other well for 20 minutes. The medium and the Triton X-100 lysate were cleared by centrifugation. The supernatant was used for the lactate dehydrogenase (LDH) assay using the Cytotox96 kit (Promega). The percentage of cell death was calculated as the percentage of LDH activity in the medium compared with the Triton X-100-lysed cells.
SHP-2 Protein Tyrosine Phosphatase Enhances Tyrosine Phosphorylation and Activation of the EGFRvIII Oncoprotein.
Three studies have shown that the protein tyrosine phosphatases TCPTP, Cdc25A, and SHP-2 can inhibit EGFRvIII and/or EGFR pTyr in U87MG human glioblastoma cells, Hep3B cells, and Cos-1 cells (27 , 31 , 32) . As part of a broader effort to understand how SHP-2 PTP regulates RTK signals, we examined how SHP-2 modulates activated ErbB signals in human cancer cells, specifically glioblastoma cells, which require EGFR signaling for transformation and survival (26 , 33) . To investigate how SHP-2 PTP modulates EGFRvIII pTyr in glioblastoma cells, we stably transfected U87MG.EGFRvIII cells with different Shp-2 mutant constructs as described in Materials and Methods (26) . Fig. 1A ⇓ shows the expression levels of epitope-tagged Shp-2 constructs stably expressed in these clones. The same blots were probed with an antiphosphotyrosine antibody, and surprisingly we found that basal EGFRvIII pTyr in full serum conditions was dramatically decreased in SHP-2 C459S clones but increased in SHP-2 wild-type clones (Fig. 1B ⇓ , top). The level of EGFRvIII pTyr in the two glioblastoma clones expressing either SHP-2 R32E or SHP-2 R138E was similar to the empty vector control (Fig. 1B ⇓ , top). All of the cell lines were found to have comparable levels of EGFRvIII oncoprotein expression (Fig. 1B ⇓ , bottom). EGFRvIII pTyr and EGFRvIII proteins showed two bands that resulted from different glycosylation of these proteins. These data indicate that SHP-2 increases EGFRvIII pTyr content and that the phosphatase activity of SHP-2 is required for this event.
Phosphatase Activity of SHP-2 Is Required for EGFRvIII-Mediated MAPK Activation.
Because SHP-2 phosphatase activity is required for EGFRvIII pTyr, it is possible that SHP-2-regulated events downstream of RTK activation are involved in regulating EGFRvIII pTyr and kinase activation. SHP-2 can positively regulate MAPK activity (28 , 34) and modulate PI3k activity either positively or negatively in different cell systems (26 , 35) . To further evaluate whether distinct domains in SHP-2 affect MAPK and PI3k activities in EGFRvIII-transformed cells, we used glioblastoma subclones expressing SHP-2 derivatives and containing variable levels of basal EGFRvIII pTyr. Because SHP-2 C459S clone 13 expressed a low level of ectopic SHP-2 C459S, SHP-2 C459S clone 8 was used for these studies (Fig. 2) ⇓ . Cells were cultured in the presence or absence of serum for 24 hours, and ERK, MAPK, and Akt phosphorylation levels subsequently were determined by Western blot analysis. As shown in Fig. 2 ⇓ , ERK phosphorylation, but not Akt phosphorylation (data not shown), was decreased in SHP-2 C459S-containing clones under serum-containing and serum-starved conditions. These data suggest that SHP-2 does not directly dephosphorylate EGFRvIII pTyr but that SHP-2-modulated MAPK regulates EGFRvIII pTyr.
MAPK Activity Is Required for Constitutive EGFRvIII pTyr in Glioblastoma Cells.
To test whether the MAPK pathway is involved in regulating EGFRvIII pTyr, we used the MAPK kinase (MEK) inhibitor PD98059 and the PI3k inhibitors wortmannin and LY294002, in the presence and absence of serum, to examine EGFRvIII pTyr by Western blot analysis. After 48 hours of treatment, we found that PD98059, but not wortmannin or LY294002, inhibited EGFRvIII pTyr in the presence of serum (Fig. 3A ⇓ , Lane 4). Serum withdrawal also caused a decrease in constitutive EGFRvIII pTyr and a modest decrease in total EGFRvIII protein (Fig. 3A ⇓ , Lane 5). Moreover, serum withdrawal plus PD98059 resulted in an additive decrease in basal EGFRvIII pTyr (Fig. 3A ⇓ , Lane 8), whereas serum withdrawal plus either wortmannin or LY294002 resulted in a decrease of basal EGFRvIII pTyr (Fig. 3A ⇓ , Lanes 6 and 7). To further prove that EGFRvIII pTyr is inhibited by PD98059, immunoprecipitation experiments were performed, and the results showed that PD98059, but not LY294002, inhibits EGFRvIII pTyr at 48 hours (Fig. 3B) ⇓ , which was consistent with our Western blot analysis results (Fig. 3A) ⇓ . Previous studies (36 , 37) have shown that short-time incubation with LY294002 can inhibit MAPK phosphorylation. Therefore, a time course experiment following the addition of these inhibitors was performed (Fig. 3C) ⇓ . Inhibition of MAPK phosphorylation by LY294002 was shown only at early time points, which is consistent with the published studies (36 , 37) . Therefore, these data indicate that EGFRvIII-constitutive pTyr depends on MAPK activation in U87MG human glioblastoma cells.
Tyrosine Phosphorylation of EGFRvIII Correlates with Elevated MAPK Levels in Human Glioblastoma Cells.
We next evaluated EGFRvIII pTyr and MAPK levels in different human glioblastoma cells and mouse fibroblasts. We used two additional glioblastoma cell lines, LN229 and U373MG, and NR6 mouse fibroblasts expressing comparable levels of the EGFRvIII oncoprotein. LN229, U373MG, and U87MG cells were transiently transfected with EGFRvIII cDNA and cultured for 48 hours. U87MG.EGFRvIII and NR6.EGFRvIII stable clones were cultured in the presence of 10% FBS for 24 hours, followed by Western blot analysis. An anti-p-Tyr antibody was used to show that EGFRvIII was constitutively phosphorylated in glioblastoma cells but not in NR6 cells (Fig. 4A) ⇓ . Concurrently, an anti-EGFR antibody was used to confirm comparable levels of EGFRvIII in these cell lines (Fig. 4A) ⇓ . Interestingly, MAPK phosphorylation was much higher in glioblastoma cells than in fibroblast cells, whereas all of the cells had comparative levels of total MAPK in full serum growth medium (Fig. 4B) ⇓ .
Regulation and Site-Specific Inhibition of Wild-Type EGFR pTyr by PI3k Pathway Activation in Human Glioblastoma Cells.
Because EGFR and EGFRvIII contain the same COOH-terminal phosphorylation sites, it was of interest to determine whether wild-type EGFR pTyr was regulated by MAPK and/or PI3k pathway activation. We used parental U87MG cells and a wild-type EGFR-overexpressing subclone, U87MG.wtEGFR (17) , for these experiments. U87MG.wtEGFR cells contain a comparable amount of wild-type EGFR to levels of EGFRvIII in U87MG.EGFRvIII cells (17) . Both cell lines were pretreated with either PD98059 or LY294002 for 48 hours. Parental U87MG cells containing lower levels of endogenous EGFR that exhibit ligand-dependent EGFR pTyr (33) then were stimulated with EGF (50 ng/mL) for 10 minutes, followed by Western blot analysis. Interestingly, we found that pTyr of ligand-stimulated endogenous EGFR in parental U87MG cells was inhibited by LY294002 but not PD98059 (Fig. 5A) ⇓ . Consistent with this, the constitutive pTyr of overexpressed wild-type EGFR in U87MG.wtEGFR cells also was inhibited by LY294002 but not PD98059 (Fig. 5B) ⇓ . These results indicate that wild-type EGFR pTyr is regulated by the level of PI3k pathway activation, but not MAPK pathway events, and that this regulation is independent of the expression level of wild-type EGFR. Moreover, both constitutive pTyrs in the setting of overexpressed wild-type EGFR and ligand-activated wild-type EGFR were similarly regulated.
Differential patterns of pTyr of COOH-terminal sites in EGFR may lead to diverse phenotypic effects in cells. Because wild-type EGFR and EGFRvIII have distinctly different patterns of ligand responsiveness, kinetics of internalization, and transforming efficiency (17) , we wished to further examine whether the MAPK and PI3k pathways could differentially modulate pTyr levels in the EGFR/EGFRvIII COOH terminus. We used anti-phospho-EGFR site-specific antibodies for these experiments. U87MG.EGFRvIII and U87MG.wtEGFR clones were cultured overnight, and inhibitors were added for 48 hours. Western blot analysis showed that inhibition of MEK by PD98059 and inhibition of ERK phosphorylation by U0126 (Fig. 5C ⇓ , Lanes 4 through 6), but not LY294002 treatment (Fig. 5C ⇓ , Lane 7), suppressed constitutive EGFRvIII pTyr. However, the reduction of constitutive EGFRvIII pTyr by inhibition of the MAPK pathway was not phosphorylation site specific. All of the tyrosine phosphorylation sites examined showed a reduced level of basal pTyr in the setting of MAPK pathway inhibition. Conversely, treatment with the PI3k pathway inhibitor LY294002 caused a different pattern of inhibition of pTyr in wild-type EGFR (Fig. 5C ⇓ , Lane 2). These phosphorylated bands, except that corresponding to pTyr1086, which consistently exhibited a low basal phosphorylation level (Fig. 5C ⇓ , Lane 1), were quantitatively analyzed by densitometry, and the data showed that inhibition of wild-type EGFR phosphorylation by LY294002 resulted in more selective reduction of phosphorylation at residues Tyr992 and Tyr1068 (Fig. 5D) ⇓ . These data suggest that pTyr may be differentially modulated in wild-type EGFR by the level of PI3k activation and suggest that enzymatic activity of wild-type EGFR may be regulated by the level of PI3k pathway activation.
PD98059 but not LY294002 Sensitizes U87MG.EGFRvIII Cells to Cisplatin-Induced Cell Death.
To further address the biological significance of MAPK/PI3k-mediated pTyr of EGFRs and to examine the role of MAPK activation in cell survival in EGFRvIII-expressing human cancer cells in particular, we used PD98059 and LY294002 in cell survival assays. U87MG.EGFRvIII cells were preincubated with the inhibitors for 48 hours, and cells then were subsequently treated with 50 μmol/L of cisplatin for 72 hours. We then measured caspase-3 protease activity (9) by using Acetyl-Asp-Glu-Val-Asp-p-nitroanilide as a substrate (Fig. 6A) ⇓ . Concurrently, we monitored cisplatin-induced cell death by an LDH assay (Fig. 6B) ⇓ . We found that PD98059 dramatically potentiated cisplatin-induced caspase-3 activity and apoptosis in U87MG.EGFRvIII cells at 72 hours in full growth medium (Fig. 6A and B) ⇓ . However, treatment with LY294002 did not enhance cell death of U87MG.EGFRvIII cells following cisplatin treatment and did not alter caspase-3 activity (Fig. 6A and B) ⇓ . Fig. 6C ⇓ shows that cisplatin treatment induced a morphologic change in U87MG.EGFRvIII cells. A greater number of cells that showed a rounded morphology characteristic of apoptosis were observed under these conditions (Fig. 6C ⇓ , bottom left, with treatment by cisplatin only). PD98059 increased the amount of floating cells exhibiting hallmark blebbing, indicating an increase in the apoptotic fraction under these conditions (Fig. 6C ⇓ , bottom middle, indicated by bold head arrow). However, LY294002 did not enhance cell death and prevented the observed morphology change seen following PD98059 treatment (Fig. 6C ⇓ , bottom right).
In this study, we have shown that SHP-2 phosphatase activity is required for EGFRvIII pTyr and MAPK, but not Akt, activation. Interestingly, we previously have shown that SHP-2 phosphatase activity is required for PI3k activation downstream of wild-type EGFR in parental U87MG glioblastoma cells (26) . Therefore, wild-type EGFR and EGFRvIII may use different mechanisms to activate protein kinase B/Akt signaling in human glioblastoma cells. EGFRvIII pTyr correlated with elevated MAPK levels. Our data indicate that the pTyr content of wild-type EGFR and EGFRvIII is differentially regulated by downstream signaling pathways. Specifically, constitutive EGFRvIII pTyr is modulated by PTP SHP-2-dependent MAPK pathway activation, whereas wild-type EGFR pTyr is modulated by PI3k signaling events. Consistent with this, EGFRvIII pTyr is inhibited by PD98059 but not by LY294002. LY294002, but not PD98059, inhibits wild-type EGFR pTyr following ligand treatment in parental U87MG cells and constitutive pTyr in U87MG.wtEGFR cells. Inhibition of EGFRvIII pTyr by PD98059 is not found to be pTyr site specific, whereas inhibition of wild-type EGFR pTyr is observed to occur more selectively at residues Tyr992 and Tyr1068 in the COOH terminus. These data indicate that downstream MAPK and PI3k/Akt signaling events differentially modulate kinase activation of wild-type EGFR and EGFRvIII and suggest that activated autocrine signaling loops, mediated in part by SHP-2-regulated MAPK signaling events, are involved in modulating EGFRvIII pTyr and kinase activation in human glioblastoma cells. Finally, PD98059, but not LY294002, sensitizes U87MG.EGFRvIII cells to cisplatin-induced apoptosis.
MAPK has been shown to directly phosphorylate wild-type EGFR at Thr669 to regulate its internalization and substrate phosphorylation on tyrosine residues (38) . Although there are no data indicating that MAPK also can phosphorylate EGFRvIII at Thr669, this may be unlikely in our system because the phosphorylation sites in EGFRvIII are COOH-terminal tyrosine residues that are shared by wild-type EGFR. Our results show that PD98059, but not LY294002, can inhibit constitutive EGFRvIII pTyr in the presence of serum. We also noticed that LY294002 and wortmannin decrease EGFRvIII pTyr in the absence of serum, and serum withdrawal alone attenuates EGFRvIII pTyr (Fig. 4A ⇓ , Lane 5). Therefore, it is likely that the inhibition of EGFRvIII pTyr by LY294002 and wortmannin in the absence of serum may be because of a more general effect of serum withdrawal. Notably, the total EGFRvIII protein level is decreased in the presence of PD98059 with and without serum compared with control cells (Fig. 3A) ⇓ . However, the decrease of EGFRvIII pTyr is more obvious than the protein level itself. EGFRvIII pTyr is not even detectable in the absence of serum (Fig. 3A ⇓ , Lane 8). Moreover, the immunoprecipitation experiments also show that PD98059 inhibits EGFRvIII pTyr (Fig. 3B) ⇓ .
Because PD98059 and U0126 do not inhibit specific COOH-terminal sites of EGFRvIII pTyr (Fig. 5C) ⇓ , we speculate that unknown phosphatases and/or kinases that modulate EGFRvIII pTyr and kinase activity are regulated by MAPK signaling events. It also has been reported that chemical reagents, including sphingosine and ammonium sulfate, can increase v-erbB-like receptor kinase activity (39) . Thus, it is possible that similar factors may contribute to mediating EGFRvIII kinase activities. Collectively, these observations indicate that the constitutive kinase activity of ectodomain-deleted erbB family receptors can be modulated, and this may have therapeutic implications.
Tyrosine-phosphorylated growth factor receptors can be dephosphorylated by PTPs. Recent work has shown that three PTPases, including SHP-2, Cdc25A, and TCPTP, can directly dephosphorylate EGFR and/or EGFRvIII (27 , 31 , 32) . In U87MG.EGFRvIII human glioblastoma cells, TCPTP inhibited EGFRvIII pTyr and suppressed cell growth in vivo (31) . It also was reported that PD98059 inhibits U87MG.EGFRvIII cell proliferation (31) . Our results indicate that PD98059 inhibits EGFRvIII pTyr and kinase activation. Therefore, it is possible that the inhibition of U87MG.EGFRvIII cell proliferation by PD98059 (31) may have resulted from the inhibition of EGFRvIII pTyr. It would be interesting to know whether the expression levels or activities of TCPTP are up-regulated following MAPK inhibition in U87MG.EGFRvIII transfected SHP-2C459S cells. Cdc25A has been shown to dephosphorylate EGFR (32) . Because EGFR and EGFRvIII have the same COOH-terminal pTyr sites, it also would be interesting to examine whether Cdc25A can dephosphorylate EGFRvIII. Independently, it also has been shown that SHP-2 can dephosphorylate EGFR, depending on the model system used (27 , 32) . Our results indicate that SHP-2 does not directly dephosphorylate EGFRvIII pTyr in glioblastoma cells, but SHP-2-modulated downstream events are important to regulate EGFRvIII pTyr. Therefore, it would be interesting to identify the substrates of SHP-2 in human glioblastoma cells containing wild-type EGFR and EGFRvIII. Experiments trapping SHP-2 substrates in human glioblastoma cells are presently ongoing in our laboratory.
Overexpression of EGFRvIII in NR6 fibroblasts has not been observed to result in an increase in its intrinsic tyrosine kinase activity (21) . EGFRvIII basal kinase activity was equal to that of wild-type EGFR in EGFR-overexpressing NR6 cells in a protein kinase assay (13) . In studies with human glioblastoma cells, it has been shown that EGFRvIII is tyrosine phosphorylated independent of EGF ligand treatment (15 , 18) . Our results show that constitutive EGFRvIII pTyr was not detectable in NR6 fibroblasts but was observed in glioblastoma cells, consistent with previous observations (17) . There are two possibilities for this result. First, we observed that the basal ERK phosphorylation was higher in parental glioblastoma cells than in NR6 cells in the presence of 10% serum (Fig. 4B) ⇓ . We speculate that this elevated MAPK activity may contribute to EGFRvIII pTyr. Therefore, when EGFRvIII is expressed in glioblastoma cells, EGFRvIII can be more efficiently phosphorylated by the enhanced MAPK-mediated signaling events. Tyrosine-phosphorylated EGFRvIII then may further enhance downstream events, including MAPK activities, to feedback and modulate its own phosphorylation status. Second, glioblastoma cells contain endogenous wild-type EGFR, whereas NR6 cells lack endogenous EGFR and do not bind EGF (Fig. 4A ⇓ ; ref. 40 ). Although endogenous EGFR requires ligand treatment to become tyrosine phosphorylated (18) , it is possible that wild-type EGFR may dimerize with EGFRvIII to cause a conformational change that increases EGFRvIII kinase activity. A physical association between EGFR and EGFRvIII may not be surprising because we and others previously have observed coassociation between erbB-2 receptors and EGFRvIII (18 , 41) .
It has been shown that inhibition of EGFRvIII pTyr by AG1478 in U87MG.EGFRvIII cells sensitized cells to cisplatin-induced cell death (9) . Our results show that PD98059 inhibited EGFRvIII pTyr and also sensitized U87MG.EGFRvIII cells to cisplatin treatment. It suggests that the MAPK pathway is involved in regulating cell survival, and this may occur through inhibition of EGFRvIII pTyr in EGFRvIII-transformed human glioblastoma cells. Interestingly, we noticed that PD98059 treatment resulted in two distinct morphologic cell types. In addition to an unchanged cell morphology (indicated by plain head arrow, Fig. 6C ⇓ , bottom middle), characteristically apoptotic cells were observed (indicated by bold head arrow, Fig. 6C ⇓ , bottom middle). How PD98059 treatment induces these different effects is unknown. Present work is under way to more precisely determine the extent of coupling between SHP-2, MAPK, and EGFR signaling and cell survival phenotypes. Collectively, one potential therapeutic implication of the present report is that MAPK pathway constituents and the SHP-2 PTP may represent rational therapeutic targets in human cancer cells containing oncogenic EGFRvIII receptors.
We thank Drs. Gurpreet S. Kapoor and Dmitri Kapitonov for helpful discussion.
Grant support: Grants from NIH (RO1 CA-90586) and the Department of Veterans Affairs (Merit Review Program) to D. M. O’Rourke.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Requests for reprints: Donald M. O’Rourke, Department of Neurosurgery, University of Pennsylvania School of Medicine, 502 Stemmler Hall, 36th and Hamilton Walk, Philadelphia, PA 19104. Phone: 215-898-2871; Fax: 215-898-9217; E-mail:
- Received October 6, 2003.
- Revision received August 19, 2004.
- Accepted September 8, 2004.
- ©2004 American Association for Cancer Research.