Interleukin-13 Receptor α2, EphA2, and Fos-Related Antigen 1 as Molecular Denominators of High-Grade Astrocytomas and Specific Targets for Combinatorial Therapy

Materials and Methods

Tissue, cell culture, and antibodies. Human GBM tumor tissue was obtained from the operating room, snap frozen, and stored at −80°C until use. For culture, fresh tissue was minced into small pieces, 5 mL digestion buffer (1 mg/mL collagenase type II, 1 mg/mL collagenase type IV, 2 mg/mL DNase, 4% NuSerum in DMEM) were added, and the mixture was applied to tissue homogenizer, collected, and incubated for 45 min at 37°C with shaking. Cells were washed and established for cell culture in RPMI 1640 (Invitrogen), 10% fetal bovine serum (Sigma), 100 μg/mL sodium pyruvate, 20 μg/mL L-Proline (Sigma), 1× HT supplement consisting of 0.1 μmol/L sodium hypoxanthine and 0.016 μmol/L thymidine, 5 units/mL Penicillin G, and 5 units/mL streptomycin sulfate (Invitrogen). Human GBM xenograft tumors were provided as a generous gift from Dr. C. David James (University of California-San Francisco). Tumors were dissociated and cultured in DMEM (Invitrogen) with 10% fetal bovine serum to obtain cell lines and used for analysis within the first five passages. U-251 MG were obtained from American Type Culture Collection and grown in DMEM with 10% fetal bovine serum and 0.1 mmol/L MEM nonessential amino acids (Sigma). Anti–Fra-1 and donkey anti-goat IgG–horseradish peroxidase (HRP) were obtained from Santa Cruz Biotechnology. Anti-EphA2 B208 antibody was graciously provided by MedImmune, Inc.. Anti-EphA2 D7, anti–β-actin, goat anti-mouse IgG-HRP, and goat anti-rabbit IgG-HRP were obtained from Sigma. Anti–IL-13Rα2 was obtained from R&D Systems. Anti-GFAP was obtained from Dako and was used to stain cells grown on 12-mm coverslips and fixed in 10% formalin followed by incubation with donkey anti-rabbit rhodamine (Jackson Labs.).

Immunohistochemistry. Tissue microarrays were obtained from Cybrdi, Inc.. Histologic designations of tumor grade for each section were verified by a neuropathologist (C.S.). Slides were heated at 65°C, deparaffinized in xylene, and rehydrated. Antigen retrieval was done with 10 mmol/L sodium citrate buffer (pH 6.0) by microwaving twice for 5 min. Endogenous peroxidase activity was quenched by incubating slides for 30 min in peroxide/methanol. EphA2 and Fra-1 staining was done using the SensiTek HRP anti-polyvalent kit (ScyTek Laboratories). Slides were blocked and incubated with primary antibody or PBS overnight at 4°C, followed by incubation with ScyTek biotinylated secondary antibody for 15 min, then ScyTek avidin-HRP for 20 min. IL-13Rα2 staining was done using R&D Systems Cell and Tissue Staining kit according to the manufacturer. Visualization with ScyTek AEC/chromagen was done and allowed to proceed for 3 to 5 min. Slides were counterstained in hematoxylin for 1 min and mounted with Crystal-Mount (Biomedia). Photomicrographs were taken with a 40× magnification lens with a Retiga 4000 camera using ImagePro Plus v5.1. Images were processed with Jasc Paint Shop Pro v6.01. Tissue sections were scored (a) based on the average percentage range of specific positive-stained cells within the entire section and assigned to one of the frequency categories (0-10%, 10-50%, or 50-100% positive-staining cells) and (b) based on the overall specific staining intensity for a given marker throughout the section and assigned a score (0, none; 1, weak; 2, moderate; or 3, strong). All sections were scored blindly by one person using a 20× objective lens. Diagnosis was first confirmed, and then the staining frequency and intensity were validated independently by a neuropathologist (C.S.). Twelve percent of scores were initially different, all within one degree of staining category on the chosen scales. Discrepancies were discussed and resolved, resulting in 8% of scores differing from the original score. Notably, among the specimens in question, two of three was grade II astrocytomas.

Western blotting. Western blotting was done as described previously (8). Membranes were incubated with primary antibody overnight at 4°C and with secondary antibody conjugated with HRP (goat anti-mouse IgG, goat anti-rabbit IgG, or donkey anti-goat IgG) 1 h at room temperature. Detection was done using the enhanced chemiluminescence plus Western Blotting Detection System (GE Healthcare). Membranes were exposed to autoradiographic film X-OMAT AR, and films were scanned at 600× dpi and images compiled using Jasc Paint Shop Pro v 6.0.

Cytotoxicity assay. Human GBM explant cells (5 × 10³) or U-251 MG cells (1 × 10³) were plated per well in 96-well culture plates. IL-13.PE38QQR (made in house; ref. 33) and ephrinA1-PE38QQR (made as described in ref. 20) were diluted in PBS + 0.1% bovine serum albumin, added to each well, and incubated at 37°C for 48 h. Cells treated with cycloheximide served as a positive control for cell death. Cell viability was determined using a 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt/phenazine methosulfate cell proliferation assay (Promega). Cells were incubated with the 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt/phenazine methosulfate dye for 2 to 4 h, and absorbance was measured at 490 nm using a microplate reader. Each concentration of drug was tested in quadruplicate in each assay, and viability of cells treated with drug was calculated as percentage of untreated control cells.

Statistical analyses. All data sets were analyzed by one-way ANOVA followed by Bonferroni's multiple comparison test to determine level of significance between pairs of data sets. P < 0.05 was considered significant.

Results

IL-13Rα2, EphA2, and Fra-1 expression in human astrocytomas and normal brain. To investigate the expression of IL-13Rα2, EphA2, and Fra-1 in astrocytomas, we did immunohistochemistry for each of these three proteins using commercially available tissue microarrays containing pathologically verified tumor tissue from 16 patients with WHO grade II or low-grade astrocytoma, 14 with WHO grade III or anaplastic astrocytoma, and 46 with WHO grade IV GBM, in addition to nine normal brain tissue samples. Specific staining for each protein was blindly scored first with respect to the percentage of positively stained tumor cells per tissue section. Table 1 displays the number and percentage of patients within each histologic subtype having a frequency of 0% to 10%, 10% to 50%, or 50% to 100% of tumor cells per section stained positive for IL-13Rα2, EphA2, and Fra-1. When considering frequency, those specimens with the most tumor cells stained positive (50-100%) were GBM, with 78%, 91%, and 72% of GBM specimens falling into this group for IL-13Rα2, EphA2, and Fra-1, respectively (Table 1; Fig. 1A ). Those tumors with a frequency of 0% to 10% and 10% to 50% of positive-staining cells were more commonly low-grade and anaplastic astrocytomas rather than GBM (Table 1; Fig. 1A). Importantly, the majority of normal brain specimens fell into the 0% to 10% frequency category for all three markers (Table 1: Fig. 1A). Together, these findings suggest that the frequency of expression of IL-13Rα2, EphA2, and Fra-1 within a tumor is likely associated with astrocytoma grade.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Analysis of IL-13Rα2, EphA2, and Fra-1 expression in astrocytomas and normal brain. A, frequency of IL-13Rα2, EphA2, or Fra-1 expression. Tissue sections stained for IL-13Rα2, EphA2, or Fra-1 were analyzed and placed into one of three categories based on the average percentage of positive-staining cells per section: 0-10%, 10-50%, or 50-100%. Columns, percentage of total samples in each histologic category: astrocytomas, low-grade astrocytoma (A), anaplastic astrocytoma (AA), and GBM, and normal brain. B-D, IL-13Rα2, EphA2, and Fra-1 expression with respect to staining intensity. Tissue sections stained for IL-13Rα2, EphA2, or Fra-1 were analyzed based on the overall specific staining intensity of the marker in each section and assigned a score: 0, none; 1, low; 2, moderate; 3, strong. B, expression of IL-13Rα2, EphA2, and Fra-1 depicted as staining intensity versus the percentage of samples in each histologic subtype (normal brain, low-grade astrocytoma, anaplastic astrocytoma, or GBM). C, expression of IL-13Rα2, EphA2, or Fra-1 depicted as a function of histologic subtype versus staining intensity. *P < 0.05, **P < 0.01, ***P < 0.001 versus GBM. D, expression of IL-13Rα2, EphA2, or Fra-1 as a histogram representing the staining intensity of each patient sample with respect to histologic subtype.

A similar pattern was observed with respect to intensity of IL-13Rα2, EphA2, and Fra-1 staining. Each specimen was assigned a numerical score based on the overall specific staining intensity of each marker: 0, none; 1, weak; 2, moderate; 3, strong. The staining intensity with respect to the percentage of samples in each staining category for normal brain, low-grade astrocytoma, anaplastic astrocytoma, and GBM for each marker is depicted in Table 2 and Fig. 1B. GBM comprised 30%, 61%, and 50% of samples with strong staining (intensity 3) for IL-13Rα2, EphA2, and Fra-1, respectively (Table 2; Fig. 1B). In contrast, normal brain never stained strongly for any of the three factors (Table 2; Fig. 1B). IL-13Rα2, EphA2, and Fra-1 staining intensity was significantly higher in GBM when compared with normal brain (P < 0.01, IL-13Rα2; P < 0.001, EphA2; P < 0.05, Fra-1; Fig. 1C). Moreover, all three factors were expressed at a significantly higher level in GBM than low-grade astrocytomas (P < 0.05, IL-13Rα2; P < 0.01, EphA2; P < 0.05, Fra-1), and EphA2 and Fra-1 were more prevalent in GBM than anaplastic astrocytomas (P < 0.01, EphA2; P < 0.01, Fra-1; Fig. 1C). Overall, the intensity of expression of all three markers increased with astrocytoma grade from normal brain to GBM (Fig. 1B-D). The intensity of IL-13Rα2, EphA2, and Fra1 for each individual patient specimen is shown in Fig. 1D, further representing a pattern suggestive of the involvement of all three factors in astrocytoma formation/progression. This increase in expression of IL-13Rα2, EphA2, and Fra-1 as related to tumor grade is shown in representative sections of normal brain, low-grade astrocytoma, anaplastic astrocytoma, and GBM (Fig. 2A ).

• Download figure
• Open in new tab
• Download powerpoint

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Immunohistochemistry and Western blot analysis of IL-13Rα2, EphA2, and Fra-1 expression in astrocytomas, normal brain, and human GBM xenograft tumors. A, photomicrographs of IL-13Rα2, EphA2, and Fra-1 immunostaining in three different representative sections of normal brain, low-grade astrocytomas, anaplastic astrocytomas, and GBM. B, photomicrographs of IL-13Rα2, EphA2, and Fra-1 immunostaining in a representative subset of 4 of the 16 GBM patient specimens described in Table 3. C, Western blot for IL-13Rα2, EphA2, and Fra-1 in human GBM tumor tissue lysates (arbitrarily numbered 1-9). β-Actin served as a control for equal loading of proteins. D, Western blot for IL-13Rα2, EphA2, and Fra-1 in human GBM xenograft tumors and cell lines and EGFR amplification status in the tumors. The numbers of xenograft cell lines correspond to the tumors from which they originated.

Combined expression of EphA2, IL-13Rα2, and Fra-1 in GBM. Despite our finding that IL-13Rα2, EphA2, and Fra-1 individually are overexpressed in a majority of GBM tumors when compared with low-grade and anaplastic astrocytomas and normal brain, none of these markers were strongly overexpressed in 100% of GBM cases, and some were expressed in <50% of cells within a given tumor (Tables 1 and 2; Fig. 1A-D). An expression profile for IL-13Rα2, EphA2, and Fra-1 for a representative of 16 of the GBM samples analyzed depicts both the staining intensity score for each tumor, as well as the frequency of positive-staining tumor cells (Table 3 ). These data revealed that 4 of 16 GBM patients (25%) had tumors that strongly overexpressed all three factors (staining intensity score, 3), albeit at variable frequencies of positive-staining tumor cells/specimen (Table 3). The remaining tumors either had low or moderate expression of one or more of the factors, such as in patient 7 for EphA2 (Table 3). It was evident, however, that at least one of the three markers was, in all cases, present at a level of intensity that we expect to be sufficient for effective therapeutic molecular targeting (moderate or strong staining intensity) and in a majority of tumor cells within the specimen (Table 3). The expression of all three proteins by immunohistochemistry is shown in patients 7, 9, 13, and 14 (Fig. 2B). Overall, 100% of GBM specimens overexpressed at least one of the three targets at a moderate or strong intensity.

Western blotting of human GBM tumor tissue revealed results consistent with the immunohistochemical analysis. IL-13Rα2 was abundantly overexpressed in five of nine specimens (tumors 1, 5, 6, 7, 8; Fig. 2C), EphA2 in five of nine (tumors 1, 3, 4, 5, 9; Fig. 2C), and Fra-1 also in five of nine (tumors 2, 4, 6, 8, 9; Fig. 2C). Notably, those tumors which had low expression of one or more of the proteins abundantly overexpressed at least one of the other two. For example, tumor 4 did not overexpress IL-13Rα2 but displayed abundant EphA2 and Fra-1 (Fig. 2C). We observed a similar pattern for EphA2 in tumors 2, 6, and 8 and for Fra-1 in tumors 3 and 5 (Fig. 2C). Importantly, we found that all GBM tumor specimens analyzed by Western blotting abundantly overexpressed at least one of the three proteins (Fig. 2C), further supporting our findings from immunohistochemistry analysis that the combined expression of IL-13Rα2, EphA2, and Fra-1 in GBM patient tumors is 100%. Moreover, these results strongly support the need for a combinatorial targeted approach to attack tumors in all GBM patients.

IL-13Rα2, EphA2, and Fra-1 expression in human GBM xenograft tumors and cell lines. To further explore IL-13Rα2, EphA2, and Fra-1 expression in GBM, we analyzed explanted human GBM tumors that have been serially passaged in nude mice and established to resemble characteristics of the original tumor (34, 35). In addition, some of these tumors were cultured to obtain cell lines, and whole lysates from these cells were also used for the analysis. Western blotting revealed abundant expression of IL-13Rα2, EphA2, and Fra-1 in all three xenograft cell lines and a majority of the xenograft tumor specimens (Fig. 2D); all three cell lines were found previously to be highly susceptible to IL-13–based cytotoxins (data not shown). IL-13Rα2 was overexpressed in five of seven tumors (tumors 5, 6, 12, 46, and 15; Fig. 2D), EphA2 in six of seven tumors (tumors 6, 12, 46, 15, 8, and 26; Fig. 2D), and Fra-1 in four of seven tumors (tumors 46, 15, 8, and 26; Fig. 2D). Notably, every tumor displayed abundant overexpression of at least one of the three proteins (Fig. 2D), supporting our immunohistochemical staining and results from Western blotting of human GBM tumor tissue. Interestingly, the expression of all three factors varied somewhat when comparing tumor specimens and the corresponding established cell lines (xenograft cell lines and tumors 5, 6, and 12; Fig. 2D). These tumors have been previously characterized for EGFR amplification (5 of 12 tumors analyzed had EGFR amplified and all of the EGFR-positive tumors were used in this assay; Fig. 2D and C), and it has been shown that the EGFR status of the tumors has a tendency to change upon establishment in culture (34), which may also hold true for the markers of interest here. Interestingly, tumor 5, which has no EGFR amplification, has no detectable EphA2 or Fra-1. In contrast, EphA2 and Fra-1 are expressed in tumor 6, which has amplified EGFRvIII, and in tumors 12, 46, 15, 8, and 26, which overexpress wild-type EGFR. Thus, our analysis reveals that EphA2 and Fra-1 expression correlates in general with the expression of EGFR in these tumors, in accordance with previous data revealing EphA2 and Fra-1 as part of a small group of genes whose expression is controlled by EGFRvIII (31). However, they can also be overexpressed independently of EGFRs, indicative of other mechanisms involved in their up-regulation (31).

IL-13Rα2 and EphA2-targeted cytotoxins kill GBM cells expressing varying levels of the molecular markers. To confirm the utility of IL-13Rα2 and EphA2 as combinatorial molecular targets in GBM, we tested the ability of cytotoxins targeted to these plasma membrane receptors to kill primary cells cultured from fresh isolates of a pathologically verified human GBM tumor specimen. IL-13.E13K.PE38QQR is a recombinant IL-13Rα2–targeted cytotoxin consisting of a mutated IL-13 ligand fused to PE38QQR, a derivative of Pseudomonas exotoxin A (33). EphrinA1-PE38QQR is an EphA2-targeted cytotoxin consisting of the ephrinA1 ligand chemically conjugated to PE38QQR (20). Both cytotoxins have been found to have potent and specific killing activity against established GBM cell lines in culture (20, 33). We found that tumor cells derived from BTCOE 4536, a GBM tumor, are killed by IL-13.E13K.PE38QQR and, to a much lesser extent, ephrinA1-PE38QQR (Fig. 3A ). Accordingly, Western blotting showed that these cells highly overexpressed IL-13Rα2 and did not overexpress EphA2 (Fig. 3B). The cells derived from BTCOE 4536 are, indeed, of astrocytic origin, as they expressed the astrocyte marker GFAP (Fig. 3C). Moreover, established human GBM cells were susceptible to killing with the targeted cytotoxins in accordance with their level of target protein expression. U-251 MG cells overexpress both IL-13Rα2 and EphA2 to a similar degree (Fig. 3B) and were thus potently killed by both IL-13.E13K.PE38QQR and ephrinA1-PE38QQR (Fig. 3A). These results support the need for specific combinatorial approach simultaneously targeting multiple markers to have a single therapeutic option suitable for all GBM patients.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Human GBM cells are killed by IL-13Rα2–targeted and EphA2-targeted cytotoxins. A, cell viability of human GBM primary explant cells (BTCOE 4536) and established human GBM cells (U-251 MG) in response to IL13.E13K-PE38QQR or ephrinA1-PE38QQR. B, Western blot of IL-13Rα2 and EphA2 immunoreactivity in BTCOE 4536 and U-251 MG cells. C, immunofluorescent staining for GFAP in BTCOE 4536 cells.

Discussion

IL-13Rα2, EphA2, and Fra-1, three proteins that we have shown previously to be individually abundant in GBM, have now been concomitantly analyzed for their expression in the same specimens of human WHO grades II to IV astrocytomas and normal brain, with an emphasis on combinatorial expression in GBM. We have found that the expression of all of these proteins is significantly associated with GBM compared with low-grade and anaplastic astrocytomas and normal brain. Moreover, IL-13Rα2, EphA2, and Fra-1 are abundant in human GBM xenograft tumors serially passaged in mice, as well as in GBM cells derived from these tumors. The majority of tumors from our analysis of GBM patients overexpress at least two of these proteins in 50% to 100% of tumor cells and at a moderate or strong intensity. Importantly, all of the GBM tumors analyzed overexpressed at least one of the three factors, providing a rationale for combinatorial molecular targeting of these proteins for imaging and therapeutics that will be applicable to nearly all patients with GBM. Thus, IL-13Rα2, EphA2, and Fra-1 are molecular denominators of GBM and are attractive targets for experimental therapies, including immunotherapy.

Heterogeneity is a prominent feature of high-grade astrocytomas (36). This heterogeneity is multifaceted and is considered, for example, with respect to the proliferative capacity of cells within a tumor, invasion and motility, mutations in tumor-suppressor genes and oncogenes, altered gene expression due to factors in the tumor microenvironment, such as hypoxia or neovascularization, and altered expression of plasma membrane receptors. These variables are in some cases related to one another, but do not occur to the same degree in every tumor and rarely occur in a constant manner throughout the tumor. Thus, therapies directed to a single target or used in isolation without other approaches to treatment are unlikely to fully succeed. It was once thought that molecular common denominators of GBM do not exist as they do for some other cancers, such as leukemia and breast cancer. Our previous findings, however, have revealed that molecular targeting of most patients with GBM is not only possible, but due to the resistance of GBM tumors to current standard therapies, is likely necessary to improve the outcome of patients with GBM (6, 37).

IL-13Rα2 has previously been found to be overexpressed in ∼75% of GBM patients (6, 12, 13). Our current analysis is in full agreement with these findings, that 76% of GBM tumors overexpress the receptor at a moderate or strong level. IL-13Rα2 is a very attractive molecular target in GBM, especially due to the fact that the only other normal tissue in which the gene for this plasma membrane receptor is considerably expressed is the testes (7). In fact, a recombinant cytotoxin composed of IL-13 fused to a modified bacterial toxin has been developed, which destroys IL-13Rα2–overexpressing GBM cells and is in phase III clinical trials, but also destroys those cells expressing the shared receptor for IL-13 and IL-4 and is thus nonspecific (14, 15). New cytotoxins with enhanced specificity for IL-13Rα2 have been designed and produced and are expected to enter clinical trials in the near future (33, 37, 38). Other immunotherapy-based approaches include generation of targeted cytolytic T cells expressing a mutated IL-13 chimeric T-cell receptor or zetakine (39) and the identification of IL-13Rα2 peptide analogues capable of inducing a glioma-targeted cytolytic T-cell response when given as a vaccine (40). Additionally, IL-13Rα2–targeted adenoviral constructs have been developed and could potentially be used as gene therapy vectors for the treatment of gliomas (41). It is evident, however, from our data that IL-13Rα2 is not expressed in 100% of patients with GBM nor is it present on every cell within a given tumor. This observation holds true for the other molecular targets we have investigated, which prompted the combinatorial analysis of several proteins together, as well as a rationale for the development of multifaceted, multitargeted molecular therapies.

We have previously shown that the EphA2 receptor tyrosine kinase is a novel molecular marker and target in GBM due to its abundant overexpression in GBM cell lines and patient samples and virtual absence in normal brain (8). Here, we have found that EphA2 is strongly overexpressed in 60% of GBM patient tumors and expressed at a moderate or strong level in 98% of GBM tumors. EphA2 is an attractive target not only due to the fact that it is an internalized plasma membrane receptor specifically expressed on the surface of GBM cells, but also due to its functional role in mediating the malignant properties of GBM and other solid tumors (8, 17, 42, 43). Moreover, a genome-wide screen recently identified EphA2 as being functionally important in GBM and related to patient survival, further validating the receptor as an attractive target in GBM (44). EphA2 is present in tumor cell lines and tissues in a nontyrosine phosphorylated state (17, 43), and this form of the receptor has been shown to be an oncoprotein capable of transforming mammary epithelial cells (17). Decreasing the expression of EphA2 protein results in a decrease in tumor volume in mice (19, 44), and activation and subsequent down-regulation of the receptor by treatment with its endogenous ligand, ephrinA1, suppresses the malignant phenotype of GBM (8) and other tumor cells in vitro (17) and also suppresses tumor growth in vivo (23). Antibody targeting of EphA2 also inhibits the malignant behavior of tumor cells (22). Furthermore, we have observed EphA2 to be expressed in some patients at very high levels in the tumor vasculature, which has also been shown for other solid tumors (45, 46), as well as in GBM tumor cells invading into the normal brain.⁵ This expression of EphA2 in all three GBM compartments provides an additional benefit to therapies targeting this receptor, whether they are based on the ephrinA1 ligand and deliver cytotoxic proteins to cells via EphA2, are designed to restore signaling through the receptor, or decrease its expression to modulate the malignant properties of tumors. In addition, EphA2 has been identified as a novel target for glioma vaccines, and vaccination with EphA2-derived peptides resulted in an immune response that prevented the establishment and growth of both EphA2-positive and EphA2-negative tumors in vivo, supporting the notion that EphA2-targeted therapies may be effective in other tumor compartments that overexpress the receptor, such as the vascular endothelium (47).

Fra-1 differs from IL-13Rα2 and EphA2 in that it is an AP-1 transcription factor rather than a plasma membrane receptor. We have previously shown that it is abundant in gliomas and plays a major role in the oncogenic properties of GBM cells (9, 26). Interestingly, although Fra-1 is a nuclear transcription factor, our analysis revealed that it is, in many cases, also localized to the cytoplasm, an observation that we have made previously when ectopic Fra-1 localized also heavily to cytoplasm (9) and which has been reported in breast cancer as well (48). We have shown that overexpression of Fra-1 in malignant glioma cells causes profound changes in cell morphology and tumorigenicity (9). Fra-1 has also been shown to have an effect on a number of other solid tumors, such as thyroid, prostate, and breast cancer, and the possibility of using Fra-1 as a therapeutic target in these malignancies and others has been explored (25). One therapeutic approach for targeting Fra-1 entails gene therapy to knockdown the expression of this transcription factor, potentially preventing the expression of proteins that contribute to tumor progression, angiogenesis, or invasion. For example, targeting Fra-1 with small interfering RNA inhibits the invasive phenotype of breast cancer cells (49). Interestingly, a Fra-1–based vaccine has been developed, in which an immune response is directed against Fra-1 antigenic peptides displayed by MHC on the surface of tumor cells overexpressing this protein (50). We have previously found that growth factor stimulation of glioma cells results in an increase in Fra-1 and vascular endothelial growth factor-D (26), supporting a possible role for Fra-1 in tumor cells in mediating the transcription of genes involved in angiogenesis, many of which have AP-1 transcription factor binding sites in their promoter regions (49). The protein products of these Fra-1 inducible genes then have the ability to act in the context of the tumor microenvironment to affect malignant processes, such as neovascularization and tumor recurrence and progression.

We have observed that IL-13Rα2, EphA2, and Fra-1 are expressed in GBM xenograft tumors that have been serially passaged in mice and in cell lines established from these tumors. All but one of the tumors we analyzed for expression of these markers have amplification of EGFR or EGFRvIII (34). Despite the fact that EGFR signaling has been implicated in the control of expression of all three of these factors (18, 28, 29) and our finding that Fra-1 and EphA2 were absent from the tumor which lacked EGFR amplification, we did not observe overexpression of all the markers in every tumor in which EGFR was amplified. Moreover, an analysis of EGFR expression on the same specimens as for IL-13Rα2, EphA2, and Fra-1, revealed that 40% of GBM tumors overexpressed the receptor at a moderate or strong intensity (data not shown), which is in accordance with data reported in the literature (27). These observations illustrate that although EGFR may contribute to the control of expression of these molecular targets, the expression level of each marker in a given tumor is a result of a more complex regulatory system involving other pathways and oncogenes/tumor suppressor genes. Importantly for therapeutic considerations, however, at least one of the three targets was expressed in every xenograft tumor.

IL-13Rα2, EphA2, and Fra-1 represent a novel combination of factors that are each expressed at high levels in a subset of GBM patients, but when combined are expressed in virtually all patients with GBM. Importantly, we were unable to readily detect the expression of these proteins in normal brain, which makes them highly suited as targets for molecular diagnostics and therapeutics designed to spare healthy brain tissue from the harms of nonspecific antitumor therapies. A combinatorial approach to novel drug development using these proteins as targets would spare the expensive and time-consuming process of individualized molecular profiling and should make molecularly targeted therapies more economical and feasible (5). Most importantly, this work provides the rationale for development of a therapeutic approach targeting IL-13Rα2, EphA2, and Fra-1, concomitantly to improve the outcome of patients with GBM.