The Protein Tyrosine Phosphatase TCPTP Suppresses the Tumorigenicity of Glioblastoma Cells Expressing a Mutant Epidermal Growth Factor Receptor*

, Glioblastoma multiforme (GBM) is the most aggres-sive type of glioma and GBMs frequently contain ampli-fications or mutations of the EGFR gene. The most common mutation results in a truncated receptor tyrosine kinase known as (cid:1) EGFR that signals constitutively and promotes GBM growth. Here, we report that the 45-kDa variant of the protein tyrosine phosphatase TCPTP (TC45) can recognize (cid:1) EGFR as a cellular substrate. TC45 dephosphorylated (cid:1) EGFR in U87MG glioblastoma cells and inhibited mitogen-activated protein kinase ERK2 and phosphatidylinositol 3-kinase signaling. In contrast, the substrate-trapping TC45-D182A mutant, which is capable of forming stable complexes with TC45 substrates, suppressed the activation of ERK2 but not phosphatidylinositol 3-kinase. TC45 inhibited the proliferation and anchorage-independent growth of (cid:1) EGFR cells but TC45-D182A only inhibited cellular proliferation. Notably, neither TC45 nor TC45-D182A were generated by PCR using cloned Pfu or PfuTurbo (cid:1) DNA polymerases, respectively (Stratagene, La Jolla, CA), and the TC45 and TC45-D182A pBluescript constructs as templates. For gen-erating the TC45 and TC45–182A pCG constructs, the oligonucleotides incorporated a Spe I site immediately 5 (cid:2) to the initiating codons and a Bam HI site immediately 3 (cid:2) to the termination codons. The 5 (cid:2) oligonu- cleotide used was 5 (cid:2) -GGCTCCCACTAGTATGCCCACCACCAT CGAG-CGGGAG-3 (cid:2) , and the 3 (cid:2) oligonucleotide was 5 (cid:2) -CCCAGTCATGGATC- CTTA GGTGTCTGTCAATCTTGGCCT-3 (cid:2) . Spe I/ Bam HI-digested PCR products were cloned into the Xba I/ Bam HI site of the mammalian expression vector pCG. To generate the TC45 and TC45-D182A pWZL-(Hygro) constructs, the TC45 and TC45-D182A cDNAs were excised with Sal I and Eco RI from the respective pBluescript constructs and cloned into the same sites of the retroviral expression vector pWZL(Hy- gro). To generate the (cid:1) EGFR-pCDNA3.1 construct, the (cid:1) EGFR pBluescript construct was digested with Sac II, blunt-ended, digested with Xho I, and cloned into the Eco RV and Xho I sites of pCDNA3.1 (Invitro-gen, San Diego, CA). The structures of the recombinant plasmids gen- erated were confirmed by restriction endonuclease analysis and the fidelity of the cloned cDNAs confirmed by sequencing.

Glioblastoma multiforme (GBM) is the most aggressive type of glioma and GBMs frequently contain amplifications or mutations of the EGFR gene. The most common mutation results in a truncated receptor tyrosine kinase known as ⌬EGFR that signals constitutively and promotes GBM growth. Here, we report that the 45-kDa variant of the protein tyrosine phosphatase TCPTP (TC45) can recognize ⌬EGFR as a cellular substrate. TC45 dephosphorylated ⌬EGFR in U87MG glioblastoma cells and inhibited mitogen-activated protein kinase ERK2 and phosphatidylinositol 3-kinase signaling. In contrast, the substrate-trapping TC45-D182A mutant, which is capable of forming stable complexes with TC45 substrates, suppressed the activation of ERK2 but not phosphatidylinositol 3-kinase. TC45 inhibited the proliferation and anchorage-independent growth of ⌬EGFR cells but TC45-D182A only inhibited cellular proliferation. Notably, neither TC45 nor TC45-D182A inhibited the proliferation of U87MG cells that did not express ⌬EGFR. ⌬EGFR activity was necessary for the activation of ERK2, and pharmacological inhibition of ERK2 inhibited the proliferation of ⌬EGFR-expressing U87MG cells. Expression of either TC45 or TC45-D182A also suppressed the growth of ⌬EGFR-expressing U87MG cells in vivo and prolonged the survival of mice implanted intracerebrally with these tumor cells. These results indicate that TC45 can inhibit the ⌬EGFR-mediated activation of ERK2 and suppress the tumorigenicity of ⌬EGFR-expressing glioblastoma cells in vivo.
Glioblastoma multiforme (GBM) 1 is the most malignant form of brain cancer with 50% of patients dying within the first year of diagnosis. Common in many de novo GBMs is the overexpression of wild type or mutant epidermal growth factor (EGF) receptor. The EGF receptor is a protein tyrosine kinase (PTK) that regulates fundamental cellular processes such as proliferation, migration, differentiation, and survival. A common rearrangement of the EGF receptor gene results in the expression of a truncated protein known as ⌬EGFR (also known as EGFR-vIII and de2-7EGFR) that has an in-frame deletion of 267 amino acids from the extracellular domain (reviewed in Refs. [1][2][3]. ⌬EGFR has been detected in several human tumors, especially in GBMs, breast carcinomas, lung cancers (4,5), and also prostate carcinomas (6). ⌬EGFR does not bind EGF but dimerizes in the absence of ligand and activates the PTK in the intracellular portion of the receptor (1)(2)(3).
Despite the prevalence of ⌬EGFR in many human cancers, little is known about the signaling cascades that ⌬EGFR utilizes to promote tumor growth (1). Previous studies have shown that the expression of ⌬EGFR can transform NIH3T3 fibroblasts and that phosphatidylinositol 3-kinase (PI3K) but not the mitogen-activated protein kinase (MAPK) ERK2 signaling is necessary for this effect (7). In addition to ⌬EGFR, de novo GBMs frequently contain numerous other genetic alterations including the mutation or deletion of the tumor suppressor genes PTEN and INK4A (reviewed in Refs. 1 and 2). The U87MG cell line is derived from a patient diagnosed with GBM and lacks PTEN, p16 INK4A , and p14 ARF (8). U87MG cells express very low levels of wild type EGF receptor, and the introduction of ⌬EGFR significantly enhances the growth of tumor xenografts in nude mice (9 -11). This enhanced tumorigenicity correlates with enhanced cellular proliferation and reduced apoptosis in vivo (12) and is dependent on the PTK activity and the autophosphorylation of tyrosines at the C terminus of ⌬EGFR (10).
T-cell protein tyrosine phosphatase (TCPTP) is an intracellular tyrosine-specific protein phosphatase that is expressed ubiquitously (reviewed in Ref. 13). The human TCPTP transcript is alternatively spliced to generate two variants: a 48-kDa form, which localizes to the endoplasmic reticulum by virtue of a hydrophobic C terminus, and a 45-kDa form (TC45), which lacks the hydrophobic C terminus and under basal conditions localizes to the nucleus. Despite its nuclear localization, TC45 can exit the nucleus in response to certain stimuli to recognize specific substrates in the cytoplasm (14 -16). When TC45 is transiently overexpressed in COS1 cells it dephosphorylates the wild type EGF receptor and inhibits the growth factor-or integrin-induced and EGF receptor-mediated activation of PI3K/Akt (14,15). In this study we have investigated the potential of TC45 to modulate the biological outcomes of EGF receptor activation and to suppress EGF receptor function in cancer. We have generated stable TC45-producing cell lines in ⌬EGFR-expressing U87MG glioblastoma cells to examine whether TC45 can suppress ⌬EGFR-mediated growth and tumorigenicity. TC45 dephosphorylated ⌬EGFR to inhibit the ⌬EGFR-mediated proliferation of glioblastoma cells and the growth of tumor xenografts in nude mice. Although TC45 inhibited the activation of both PI3K/Akt and ERK2, our studies indicate that the suppression of ERK2 may be sufficient to inhibit the proliferation of ⌬EGFR-expressing U87MG cells in vitro and in vivo. Our studies highlight the importance of the MAPK ERK2 in ⌬EGFR-mediated proliferation and shed light on the ⌬EGFR-mediated signaling cascades that promote GBM growth.

EXPERIMENTAL PROCEDURES
Materials-PD98059 and monoclonal anti-EGFR Ab-1 were purchased from Calbiochem Oncogene Research Products (Cambridge, MA), AG1478 from Sigma, polyclonal anti-EGFR used for immunoblotting from Santa Cruz Biotechnology (Santa Cruz, CA), monoclonal PI3K p85 antibody from BD Transduction Laboratories (San Diego, CA), and monoclonal phospho-ERK2, polyclonal phospho-Akt (Ser-473), and Akt antibodies from New England BioLabs (Beverly, MA). ERK2 1B3B9 antibody was provided by M. Weber (University of Virginia) and the monoclonal phosphotyrosine G104 and TCPTP CF4 antibodies by N. K. Tonks (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY).
Plasmid Constructs-The TC45 and TC45-D182A cDNA-pBluescript constructs (17) and the ⌬EGFR cDNA-pBluescript construct (10) have been described previously. pCG (18) constructs encoding TC45 and TC45-D182A were generated by PCR using cloned Pfu or PfuTurbo DNA polymerases, respectively (Stratagene, La Jolla, CA), and the TC45 and TC45-D182A pBluescript constructs as templates. For generating the TC45 and TC45-182A pCG constructs, the oligonucleotides incorporated a SpeI site immediately 5Ј to the initiating codons and a BamHI site immediately 3Ј to the termination codons. The 5Ј oligonucleotide used was 5Ј-GGCTCCCACTAGTATGCCCACCACCAT CGAG-CGGGAG-3Ј, and the 3Ј oligonucleotide was 5Ј-CCCAGTCATGGATC-CTTA GGTGTCTGTCAATCTTGGCCT-3Ј. SpeI/BamHI-digested PCR products were cloned into the XbaI/BamHI site of the mammalian expression vector pCG. To generate the TC45 and TC45-D182A pWZL-(Hygro) constructs, the TC45 and TC45-D182A cDNAs were excised with SalI and EcoRI from the respective pBluescript constructs and cloned into the same sites of the retroviral expression vector pWZL(Hygro). To generate the ⌬EGFR-pCDNA3.1 construct, the ⌬EGFR pBluescript construct was digested with SacII, blunt-ended, digested with XhoI, and cloned into the EcoRV and XhoI sites of pCDNA3.1 (Invitrogen, San Diego, CA). The structures of the recombinant plasmids generated were confirmed by restriction endonuclease analysis and the fidelity of the cloned cDNAs confirmed by sequencing.
Cell Proliferation, Soft Agar, Tumorigenicity, and Survival Studies-Cellular proliferation was measured using the MTT assay as described by the manufacturer (Roche Diagnostics, Mannheim, Germany), and soft agar assays were performed as described previously (20). After 3-4 weeks, colonies in soft agar were stained with crystal violet and photographed. For tumorigenicity and survival studies, TC45 (clones 8 and 16 pooled) or TC45-D182A (clones 11 and 14 pooled) ⌬EGFR-U87MG cells (5 ϫ 10 5 ) were inoculated into the brains of 4 -5-week-old female BALB/c nude mice as described previously (10). For tumorigenicity studies the brains were resected at 14 days post-implantation, and frozen sections were stained with hematoxylin.

RESULTS
⌬EGFR Is a Cellular Substrate of TC45-All protein tyrosine phosphatases contain an Asp residue that is essential for catalysis and mutation of this residue (Asp-182 in TC45) to alanine generates mutants that can form stable complexes with tyrosine phosphorylated substrates (15,21). Using the TC45-D182A substrate-trapping mutant, we have shown previously that the tyrosine-phosphorylated wild type EGF receptor is a substrate for TC45 (15). To determine whether TC45 can also recognize ⌬EGFR as a cellular substrate, we transiently coexpressed ⌬EGFR with either wild type TC45 or the D182A mutant in 293 cells. We assessed the ability of TC45 to dephosphorylate ⌬EGFR and whether TC45-D182A was capable of forming a stable complex with the tyrosine-phosphorylated ⌬EGFR. As seen in lysates of these co-transfected cells, TC45 dephosphorylated ⌬EGFR, whereas TC45-D182A protected ⌬EGFR from dephosphorylation by endogenous phosphatases (Fig. 1). Moreover, the tyrosine phosphorylated ⌬EGFR could be co-immunoprecipitated with the TC45-D182A mutant but not with wild type TC45 (Fig. 1). These results indicate that the TC45-D182A-trapping mutant and the tyrosine-phosphorylated ⌬EGFR interact through the active site of the phosphatase. Similar results were also obtained when ⌬EGFR and TC45 were co-expressed in U251MG glioblastoma cells (data not shown). These results are consistent with ⌬EGFR being a direct substrate of TC45, as we have demonstrated previously for the wild type EGF receptor (14,15).
Effect of TC45 on ⌬EGFR-mediated Proliferation-To assess the ability of TC45 to regulate ⌬EGFR signaling and function, we stably expressed TC45 or the TC45-D182A mutant in either U87MG human glioblastoma cells or in U87MG cells that had been generated previously to stably express ⌬EGFR (⌬EGFR-U87MG) to levels similar to those of GBMs (10,11). To increase the gene delivery efficiency and to attain stable gene integration, vector control amphotropic retroviruses or those expressing TC45 or TC45-D182A were used to infect U87MG or FIG. 1. ⌬EGFR is a cellular substrate of TC45. 293 cells were co-transfected with ⌬EGFR and either vector control, TC45-, or TC45-D182A-expressing constructs. TCPTP immunoprecipitates and lysates were resolved by SDS-PAGE and immunoblotted with antibodies specific for the EGF receptor (to detect ⌬EGFR) or TCPTP. The ⌬EGFR immunoblot was stripped as described previously (15) and reprobed with anti-phosphotyrosine (pTyr) antibodies.
⌬EGFR-U87MG cells. After approximately 2 weeks of selection in hygromycin B, very few colonies were observed in the TC45 or TC45-D182A retrovirus-infected ⌬EGFR-U87MG cells relative to vector control ( Fig. 2A). In contrast, no significant difference was observed for the vector control, TC45, or TC45-D182A infected U87MG cells ( Fig. 2A). The infection of the U87MG and ⌬EGFR-U87MG cells was undertaken at the same time, with the same retroviral preparations, and similar levels of TC45 and TC45-D182A were expressed in the pooled colonies (data not shown). These results indicate that TC45 and TC45-D182A may inhibit the proliferation of ⌬EGFR-U87MG cells.
To further analyze the effects of TC45 and TC45-D182A on cellular proliferation, at least three hygromycin B-resistant clones from vector control, TC45, or TC45-D182A U87MG and ⌬EGFR-U87MG populations were isolated and characterized. U87MG and ⌬EGFR-U87MG cells express relatively high amounts of the endogenous 48-kDa TCPTP variant but low levels of endogenous TC45 (Fig. 2B). The U87MG and ⌬EGFR-U87MG stable clones overexpressed similar amounts (ϳ5-fold) of either TC45 or TC45-D182A (Fig. 2B). Indeed, the level of TC45 overexpression was similar to the amount of endogenous 48-kDa TCPTP in these cells (Fig. 2B) and was similar to the endogenous levels of TC45 in other cell types such as HepG2 hepatoma cells (data not shown). The results presented hereon are from one clone for each of the vector control, TC45 or TC45-D182A U87MG, or ⌬EGFR-U87MG cell lines; similar results were obtained for at least two other clones for each of the cell lines generated. We found that expression of either TC45 or the TC45-D182A substrate-trapping mutant significantly inhibited the proliferation of ⌬EGFR-U87MG cells as compared with parental or vector control cells (Fig. 2C). In contrast, neither TC45 nor TC45-D182A had any significant effect on the proliferation of U87MG cells not expressing ⌬EGFR (Fig. 2C). These results are consistent with the observations made after retroviral infection ( Fig. 2A). Moreover, consistent with an effect on cellular proliferation we found that the expression of TC45 or TC45-D182A altered markedly the cell cycle progression of the ⌬EGFR-U87MG cells (Fig. 3). Expression of either TC45 or TC45-D182A in ⌬EGFR-U87MG cells led to a reduction of S and G 2 /M cell populations and a concomitant increase in G o /G 1 populations as compared with parental or vector control cells (Fig. 3). The results indicate that the expression of either TC45 or TC45-D182A leads to a delay in G 1 /S cell cycle progression. Previous studies have shown that the pharmacological inhibition of ⌬EGFR can suppress the proliferation of ⌬EGFR-U87MG cells (22). In this study we find that the treatment of ⌬EGFR-U87MG cells with AG1478, a pharmacological inhibitor of ⌬EGFR, also results in decreased S and G 2 /M and increased G o /G 1 cell populations (Fig. 3). As such, TC45 and TC45-D182A may exert their selective effects on the proliferation of ⌬EGFR-U87MG cells by acting specifically on ⌬EGFR or ⌬EGFR-induced signaling cascades to cause a delay in G 1 /S progression.
Effect of TC45 on ⌬EGFR-mediated Signaling-We examined the tyrosine phosphorylation status of ⌬EGFR in the lysates of TC45 and TC45-D182A ⌬EGFR-U87MG stable cell lines. Consistent with our co-transfection experiments in 293 cells (Fig. 1), we found that TC45 almost completely dephosphorylated ⌬EGFR in the ⌬EGFR-U87MG cells (Fig. 4A). Similar results were obtained when cells were lysed directly in Laemmli sample buffer (to prevent any post-lysis dephosphorylation) (data not shown) indicating that the dephosphorylation was occurring in a cellular context. In contrast, the TC45-D182A substrate-trapping mutant formed a complex with the tyrosine-phosphorylated ⌬EGFR and protected it from dephosphorylation (Fig. 4A), as we found in the 293 cells (Fig. 1) and as we have reported previously for the wild type EGF receptor (14,15). These results indicate that ⌬EGFR is a substrate for TC45 and that the wild type and mutant phosphatases have the potential to regulate cellular proliferation via the modulation of ⌬EGFR signaling. Previously we have shown that transiently expressed TC45 and TC45-D182A act on endogenous wild type EGF receptor in COS1 cells to inhibit the recruitment of the PI3K p85 regulatory subunit and consequent activation of PI3K and downstream Akt (14). The wild type TC45 dephosphorylates the EGF receptor to inhibit recruitment, whereas TC45-D182A forms a complex with the tyrosine-phosphorylated EGF receptor and inhibits recruitment in a competitive manner (14). We assessed the status of PI3K recruitment and downstream activation of Akt in the TC45 and TC45-D182A ⌬EGFR-U87MG stable cell lines. Effects on PI3K recruitment were assessed by monitoring for the presence of p85 in ⌬EGFR immunoprecipitates. We found that TC45 inhibited almost completely the association of p85 with ⌬EGFR, whereas the TC45-D182A mutant had no significant effect on p85 recruitment (Fig. 4B). The activation of protein kinase Akt was then monitored in cell lysates using antibodies specific for Akt phosphorylated on Ser-473. Consistent with the effects on PI3K recruitment, we found that TC45 but not TC45-D182A inhibited significantly the activation of Akt (Fig. 4C). These results indicate that TC45 inhibits the ⌬EGFR-mediated activation of the PI3K/Akt pathway. The inability of TC45-D182A to do the same suggests that the mutant may not be expressed at high enough levels to compete for the recruitment of PI3K to ⌬EGFR.
Nevertheless, the ability of TC45 to inhibit the ⌬EGFRmediated activation of PI3K/Akt is not likely to account for its effects on the proliferation of ⌬EGFR-U87MG cells because both TC45 and TC45-D182A inhibited proliferation (Fig. 2), but only TC45 inhibited PI3K/Akt (Fig. 4). The MAPK pathway is central to the proliferation of cells downstream of growth factor receptors, cytokine receptors, and integrins (23,24), and in many tumor cells, the activation of the MAPK ERK2 is essential for cellular proliferation (25). We examined whether ⌬EGFR-mediated ERK2 activation was necessary for the proliferation of ⌬EGFR-U87MG cells (Fig. 5). First, we found that pharmacological inhibition of ⌬EGFR with AG1478 ablated ⌬EGFR PTK activity and consequent autophosphorylation and diminished the activation of ERK2 (Fig. 5A). These results indicate that the activation of ERK2 in ⌬EGFR-U87MG cells is dependent on ⌬EGFR activity. Second, we found that pharmacological inhibition of ERK2 with the MEK inhibitor PD98059 inhibited significantly cellular proliferation (Fig. 5B). These results demonstrate that the ⌬EGFR-mediated activation of ERK2 is necessary for the proliferation of ⌬EGFR-U87MG cells (Fig. 5C). We determined whether the wild type and mutant phosphatases exerted their effects on ⌬EGFR-U87MG proliferation by inhibiting the ⌬EGFR-mediated activation of ERK2. We assessed the activation state of ERK2 in lysates of TC45and TC45-D182A-expressing ⌬EGFR-U87MG cells using both ERK2 phosphorylation specific antibodies and by monitoring the shift in ERK2 electrophoretic mobility that occurs upon phosphorylation and activation. We found that both TC45 and TC45-D182A inhibited significantly the activation of ERK2 (Fig. 5C). In contrast, TC45 and TC45-D182A had no effect on the activation of ERK2 in U87MG parental cells (data not shown). These results indicate that TC45 and TC45-D182A may act on ⌬EGFR to inhibit the downstream activation of ERK2 and the consequent proliferation of ⌬EGFR-U87MG cells.
Effect of TC45 on Anchorage-independent Growth-We next determined the effects of TC45 and TC45-D182A on the anchorage-independent growth of ⌬EGFR-U87MG cells by ascertaining the ability of our stable cell lines to form colonies in soft agar (Fig. 6). Although no significant difference was observed between ⌬EGFR-U87MG parental and pWZL vector control cells (data not shown), the expression of TC45 inhibited completely the formation of colonies in soft agar (Fig. 6). However, TC45 had no effect on the formation of colonies by U87MG cells that do not express ⌬EGFR (data not shown). In contrast to the effects of TC45 in ⌬EGFR-U87MG cells, expression of TC45-D182A did not appear to have any significant effect on the number of colonies formed, although the colonies were smaller than either ⌬EGFR-U87MG parental or vector control cells (Fig. 6). These results indicate that TC45 but not TC45-D182A ⌬EGFR-U87MG cells were collected in lysis buffer containing 5 mM iodoacetic acid, and both ⌬EGFR protein and phosphotyrosine (pTyr) content were analyzed by immunoblot analysis. B, ⌬EGFR was immunoprecipitated from pWZL, TC45, and TC45D ⌬EGFR-U87MG cell lysates, and the association of p85 PI3K subunit was determined by immunoblot analysis. C, lysates from pWZL, TC45, or TC45D ⌬EGFR-U87MG cells were resolved by SDS-PAGE and immunoblotted with antibodies specific for Akt phosphorylated on Ser-473 (P-Akt). P-Akt immunoblots were stripped and reprobed with antibodies specific for Akt.
FIG. 5. TC45 suppresses ⌬EGFR-mediated ERK2 signaling. A, randomly growing ⌬EGFR-U87MG cells were incubated with the EGF receptor AG1478 pharmacological inhibitor for 16 h. Cell lysates were resolved by SDS-PAGE and immunoblotted with antibodies specific for ⌬EGFR protein, phosphotyrosine (pTyr), ERK2, or the phosphorylated and activated ERK2 (P-ERK2). B, ⌬EGFR-U87MG cells (2 ϫ 10 2 cell/ well) were seeded in 96-well plates and incubated for 24 h. The cells were then either left untreated or incubated with 50 M PD98059 for 3 days. Cellular proliferation at day 0 and day 3 was measured using an MTT assay, and the phosphorylation state of ERK2 at day three was determined using P-ERK2 antibodies. C, lysates from randomly growing ⌬EGFR-U87MG and TC45 or TC45D ⌬EGFR-U87MG cells were resolved by SDS-PAGE and immunoblotted with antibodies specific for the phosphorylated and activated ERK2 (P-ERK2) or ERK2 protein (arrows indicate the mobility shift of the phosphorylated and activated ERK2).
inhibits the anchorage independence of ⌬EGFR-U87MG cells and correlates with the differential effects of TC45 and TC45-D182A on the ⌬EGFR-mediated activation of PI3K. As such, these results suggest that the ⌬EGFR-mediated activation of PI3K may be necessary for the anchorage-independent growth of ⌬EGFR-U87MG cells.
Effect of TC45 on Tumorigenicity and the Survival of Mice-⌬EGFR enhances the tumorigenicity of U87MG glioblastoma cells in vivo and ⌬EGFR PTK activity, autophosphorylation, and signaling are required for this growth advantage (10,11).
Our studies indicate that the MAPK ERK2 and PI3K may be required differentially for the proliferation and anchorage independence of ⌬EGFR-expressing U87MG glioblastoma cells in vitro. To examine the relative importance of these pathways in vivo, we assessed the tumorigenicity of our TC45-and TC45-D182A-expressing ⌬EGFR-U87MG cell lines. The glioblastoma cells were injected stereotactically into the brains of nude mice. We found that the expression of either TC45 or TC45-D182A in the ⌬EGFR-U87MG cells prolonged similarly the survival of mice implanted with these tumor cells as compared with parental or vector control ⌬EGFR cells. Consistent with these results we found that TC45-and TC45-D182A-expressing tumors were smaller than those generated by either ⌬EGFR-U87MG parental or vector control cells (Fig. 7). These results indicate that TC45 has the potential to inhibit the growth of ⌬EGFR-expressing glioblastomas in vivo. DISCUSSION Despite the fact that the ⌬EGFR oncogene is expressed in roughly half of all de novo GBMs, little is known about the signal transduction processes that ⌬EGFR utilizes to increase cellular proliferation and enhance tumor growth (1). Constitutive ⌬EGFR PTK activity and tyrosine phosphorylation-dependent signaling are necessary for the ⌬EGFR-mediated growth of glioblastoma cells in vivo (10,11). As such, the dephosphorylation of ⌬EGFR by tyrosine-specific phosphatases would negate the contribution of ⌬EGFR to gliomagenesis. In this study we have demonstrated that TC45 dephosphorylates and suppresses the tumorigenic potential of ⌬EGFR in glioblastoma cells.
⌬EGFR autophosphorylates on at least five tyrosines in the intracellular C-terminal portion of the receptor (10,11). As for many other receptor PTKs, including the wild type EGF receptor, these phosphotyrosines act as docking sites for the recruitment of Src homology 2 and phosphotyrosine binding domaincontaining proteins for the activation of downstream signaling cascades (26). Whereas the wild type TC45 dephosphorylated ⌬EGFR in human glioblastoma cells, the TC45-D182A mutant formed a complex with the tyrosine-phosphorylated receptor. We found that both TC45 and TC45-D182A inhibited the activation of ERK2 in ⌬EGFR-expressing U87MG cells, but neither had any effect on the activation of ERK2 in U87MG cells not expressing ⌬EGFR. ⌬EGFR activity was necessary for the ac-tivation of ERK2, and this was essential for the proliferation of ⌬EGFR-U87MG cells. These results indicate that TC45 and the D182A mutant may exert their selective effects on ERK2 because of their specific recognition of ⌬EGFR as a cellular substrate. Previous studies have shown that in COS1 cells, TC45 does not inhibit the EGF-induced activation of ERK2 (14). This apparent discrepancy may occur because different cell types were used for the respective studies. Alternatively, these differences may result from the wild type EGF receptor and ⌬EGFR utilizing different mechanisms for the activation of ERK2. The manner by which ERK2 is activated may depend on the extent of EGF receptor activation (27), and although constitutive, the level of ⌬EGFR autophosphorylation is significantly lower than that observed for EGF-stimulated wild type receptor (9,28). Nevertheless, consistent with their effects on ERK2, we found that TC45 and TC45-D182A suppressed significantly the proliferation of ⌬EGFR-expressing U87MG glioblastoma cells and caused a delay in G 1 /S cell cycle progression. Previous studies have demonstrated that ⌬EGFR activity is necessary for the proliferation of ⌬EGFR-U87MG cells in vitro (22). We report that the pharmacological inhibition of ⌬EGFR causes a delay in G 1 /S cell cycle progression similar to the effects of TC45 and TC45-D182A expression. Previous studies have also shown that the expression of ⌬EGFR in U87MG cells enhances Ras activation, which is upstream of the MAPK ERK2, and G 1 /S progression as measured by BrdUrd incorporation (11). As such ⌬EGFR would appear to mediate the progression of cells through G 1 /S via the activation of ERK2. Thus, we propose that TC45 may inhibit selectively the G 1 /S progression and proliferation of ⌬EGFR-expressing U87MG cells by dephosphorylating and maintaining the ⌬EGFR in an inactive state and suppressing the downstream activation of ERK2. However, we cannot exclude the possibility that at least part of the suppression of the ⌬EGFR-mediated activation of ERK2 may occur downstream of ⌬EGFR. We have shown previously that TC45 can recognize the adaptor protein p52 Shc as a cellular substrate (15). In ⌬EGFR-U87MG cells the ⌬EGFR protein associates constitutively with the adaptor proteins Shc and Grb2, which can allow for the recruitment of Ras and the activation of MAPK signaling cascades (11). We have yet to determine whether p52 Shc is necessary for the ⌬EGFR-mediated activation of ERK2 or whether p52 Shc may serve as a substrate for TC45 in the ⌬EGFR-U87MG cells.
The PI3K/Akt signal transduction pathway has been implicated in many receptor PTK-mediated processes including the regulation of cellular proliferation, migration and survival (29,30). We found that in addition to inhibiting ERK2 activation, TC45 also inhibited the recruitment of the p85 regulatory subunit of PI3K and the downstream activation of the protein kinase Akt in ⌬EGFR-U87MG cells. As in the case of wild type EGF receptor signaling (14 -16), we propose that TC45 acts upstream of PI3K on ⌬EGFR to inhibit PI3K recruitment and the concomitant activation of PI3K and Akt. In contrast, the TC45-D182A substrate-trapping mutant had no effect on the recruitment of PI3K or the activation of Akt. That the D182A mutant was unable to inhibit PI3K recruitment can most likely be attributed to substrate trapping by TC45-D182A being less efficient than dephosphorylation. Since TC45 and TC45-D182A suppressed equally the proliferation of ⌬EGFR-U87MG cells, these results suggest that the proliferation of ⌬EGFR-U87MG cells may be independent of ⌬EGFR-mediated PI3K/Akt signaling.
Although ⌬EGFR-mediated PI3K/Akt signaling may not be essential for proliferation, our studies indicate that it may be required for anchorage-independent growth in soft agar. We found that expression of TC45 inhibited completely the formation of ⌬EGFR-U87MG colonies in soft agar. In contrast to TC45, we found that the D182A mutant had no effect on the number of colonies formed by the ⌬EGFR-expressing cells. Given that TC45-D182A had no effect on PI3K/Akt activation, these results suggest that the inhibition of PI3K/Akt may be responsible for the dramatic effect of TC45 on anchorage-independent growth. Consistent with this conclusion, others have shown that PI3K signaling is necessary for the anchorage independent growth of glioblastoma cells (20) and that ⌬EGFRmediated PI3K signaling in NIH3T3 cells is necessary for anchorage independence (7). However, it is important to note that the colonies formed by the TC45-D182A-expressing cells were smaller than those of vector control ⌬EGFR-U87MG cells, indicating that ERK2 signaling and cellular proliferation may also contribute to the growth of cells in soft agar.
Our studies indicate that TC45 and TC45-D182A differentially regulate ⌬EGFR-mediated signaling processes in glioblastoma cells. Whereas TC45 inhibited both ERK2 activation and PI3K/Akt signaling, the TC45-D182A mutant suppressed only ERK2. Ultimately, the real contribution of a signaling pathway to the tumorigenic potential of ⌬EGFR can only be assessed in vivo. We found that in intracerebral xenografts of our stable ⌬EGFR-U87MG cell lines, TC45 and TC45-D182A suppressed similarly the growth of tumors and prolonged similarly the survival of mice. Taken together with our in vitro studies, these results indicate that the ⌬EGFR-mediated activation of ERK2 and the consequent cellular proliferation contribute to the growth of glioblastomas in vivo.
Holland et al. (31) have shown recently that constitutively active Ras, and hence the activation of MAPK signaling cas-cades, together with constitutively active Akt can induce, in mice, high grade gliomas with features of GBMs. Although our studies do not preclude a role for the PI3K/Akt pathway in tumorigenicity, they certainly indicate that the inhibition of ⌬EGFR-mediated PI3K, in addition to ERK2, does not add considerably to the suppression of tumorigenicity. As such, our findings suggest that the inhibition of ERK2 may be sufficient to impede the growth of ⌬EGFR-expressing GBMs, and rational GBM therapies may include inhibitors of ERK2. Moreover, agents that stimulate the expression, activation, or constitutive cytoplasmic localization of TC45 may also provide an alternative strategy for the treatment of ⌬EGFR-expressing GBMs.