|Home | About | Journals | Submit | Contact Us | Français|
Human malignant glioma cell lines and adult brain tumors overexpress high levels of interleukin-13 receptor α2 chain (IL-13Rα2). Because the IL-13Rα2 chain is an important target for cancer therapy and prognosis for patients with brainstem glioma (BSG) remains dismal, we investigated the expression of this receptor in specimens of diffusely infiltrative pediatric BSG relative to normal brain tissue. Twenty-eight BSG specimens and 15 normal brain specimens were investigated for IL-13Rα2 protein expression by immunohistochemical analysis (IHC) using two different antibodies in two different laboratories. Highly sensitive Q-dot–based IHC and in situ hybridization (ISH) assays were also developed to identify IL-13Rα2 protein and RNA in these specimens. The results were evaluated independently in two laboratories in a blinded fashion. By Q-dot IHC or a standard IHC assay, 17 of 28 (61%) tumor specimens showed modest to strong staining for IL-13Rα2, while 15 normal brain tissue samples showed weak expression for IL-13Rα2 protein. Significant interrater agreement between the two laboratories was seen in the assessment of IL-13Rα2 intensity. High-level IL-13Rα2 RNA expression was detected in tumor samples by Q-dot ISH, but only weak RNA expression was observed in normal brain. Significant agreement between ISH and IHC assays was observed (simple kappa [κ] estimate = 0.358, weighted κ = 0.89, p = 0.001). IL-13Rα2 protein and mRNA are expressed to significantly higher levels in BSG than in normal brain tissue. Both IHC and ISH represent robust methods to detect expression of the IL-13Rα2 receptor in BSG that could represent an important new drug target for treatment of this disease.
Malignant astrocytoma/glioblastoma multiforme (GBM) is the third leading cause of cancer-related death among children and adolescents in the United States.1 In contrast to adult brain tumors, which occur typically in the cerebral hemispheres, pediatric tumors are localized predominantly in the posterior fossa and brainstem.2 Brainstem glioma (BSG) accounts for ~20% of all childhood brain tumors. Despite numerous attempts to improve treatment of this disease, more than 90% of children with BSG develop diffusely infiltrative tumors and succumb to disease within 2 years of diagnosis.3,4 The intimate relation of BSG to vital control centers renders extensive surgical resection of the tumor impossible, and conventional chemotherapy and radio-therapy are ineffective in the treatment of this tumor. Therefore, new treatments of BSG are urgently needed.
Several proteins expressed by brain tumor cells that may serve as targets of new treatments have been identified. These include cell surface expression of fibroblast growth factor receptor-1b,2 epidermal growth factor receptor,5,6 and tranferrin receptor.7 We have demonstrated abundant expression of receptors for interleukin 4 (IL-4) and IL-13 on adult and pediatric brain tumors and meningiomas.8–12 In contrast, normal human brain expresses barely detectable levels of IL-4R or IL-13R.8,10 The differential expression of certain receptors between normal and malignant brain tissue may identify important biologic processes in cancer development. Further, proteins that are expressed selectively on cancer cells could be used as targets of therapeutic agents that deliver toxic agents specifically to malignant cells.
The expression and structure of IL-4 and IL-13 receptors on glioma and other human tumors have been studied extensively. IL-4 receptor complex exists in two different types. Type I IL-4 receptors are composed of IL-4Rα (also known as IL-4Rβ) and IL-2R gamma common (γc) subunits, whereas type II receptors have IL-4Rα and IL-13Rα1 subunits.13,14 IL-13R receptors also exist in at least two different types. Type I IL-13R is composed of IL-13Rα1 (also known as IL-13Rα′), IL-13Rα2 (also known as IL-13Rα), and IL-4Rα chains, whereas type II IL-13R consists of IL-4Rα and IL-13Rα1 chains.15,16 The role of IL-2 receptor common γc chain in the formation of IL-13R complex is not clear. It has been shown that the introduction of IL-2R γc can decrease IL-13 and IL-4 binding and interferes in the functioning of both receptors in cells that normally do not express this chain.15,17–19 These and other studies have shown that IL-4Rα and IL-13α1 chains are shared between IL-4R and IL-13R complexes. Furthermore, both chains are required for signal transduction through type II IL-4R and both type I and II IL-13R.15 IL-4 and IL-13 mediate signal transduction through JAK/STAT pathways. They phosphorylate and activate different JAK kinases, but both cytokines phosphorylate and activate the same STAT6 protein.20,21 In contrast to IL-4Rα and IL-13Rα1, the IL-13Rα2 chain does not seem to signal through the STAT6 pathway; it inhibits signaling through the STAT6 pathway by both IL-13R and IL-4R.13,22–24 Most recently, we have reported that IL-13 could signal through IL-13Rα2 in a STAT6- independent, AP-1–dependent manner to induce activation of the TGFβ1 promoter resulting in inflammation and fibrosis in animals.25
To target IL-13R, we have developed an IL-13 cytotoxin that consists of IL-13 and a mutated form of Pseudomonas exotoxin (PE). Recombinant IL-13PE cytotoxin is found to be highly cytotoxic to IL-13R–positive renal cell carcinoma (RCC) cells and other IL-13Rα2–positive cancer cells derived from malignant glioma, AIDS-associated Kaposi’s sarcoma (AIDS-KS), squamous cell carcinoma of the head and neck (SCCHN), ovarian carcinoma, and prostate carcinoma.26–31 In various animal models of human cancer (e.g., glioblastoma, AIDS-KS, SCCHN, and ovarian carcinoma), IL-13PE has shown remarkable antitumor effects in vivo.28,31–35 On the basis of these and other preclinical studies, several phase I/II clinical trials were initiated at various medical centers to determine safety and tolerability of IL-13PE in patients with malignant brain tumors.36–47 All these studies have demonstrated that intratumoral and peritumoral infusion of IL13-PE is well tolerated and seems to show clinical benefit to patients.48 Based on these results, a phase III clinical study was initiated in which two to three catheters were placed peritumorally and IL-13PE at a concentration of 0.5 μg/ml was infused.38 This multicenter trial was completed for patient accrual in December 2005. The results for safety and overall survival when compared with standard Gliadel treatment are being carefully evaluated. The preliminary findings suggest that there was no difference in survival in patients treated with these two therapies. The data analysis is underway to examine the group of patients with clinical benefit and correlation with IL-13Rα2 expression. In addition, limitations for drug delivery to infiltrating tumors are being evaluated.
Since IL-13R targeting by IL-13PE may provide clinical benefit to patients with IL-13R–positive tumors, here we have examined whether BSGs express IL-13R. We have demonstrated previously that primary explants of human malignant gliomas and pediatric brain tumors overexpress IL-13R in more than 72% of specimens.49,50 The information on IL-13R expression in BSGs will be useful to further determine whether these are potential targets for IL-13PE. Therefore, both IL-13 receptor α2 mRNA and protein expression were examined in 28 cases of pediatric BSG and in 15 normal brain specimens. Our data show that IL-13R is expressed selectively to high levels in BSG and can be detected reliably using either immunohistochemical analysis (IHC) or in situ hybridization (ISH).
Tumor and normal brain samples were obtained after securing approval from the U.S. Food and Drug Administration (FDA) Research Involving Human Subjects Committee (RIHSC) and the St. Jude Children’s Research Hospital (SJCRH) Institutional Review Board. Formalin- fixed tumor sections were collected from 28 children (≤ 17 years old) with a radiological or histological (biopsy performed at time of diagnosis or postmortem) diagnosis of intrinsic BSG recruited under the Pediatric Brain Tumor Consortium (PBTC) study PBTC-N06. The histologic grading and pathology review of these samples has been described previously. Sixteen patients were female, and 12 were male. The mean age at diagnosis was 6.9 years (range, 2–14 years).6
Fifteen normal brain specimens including nine pediatric and six adult brain samples were obtained from the National Cancer Institute–supported Co-operative Human Tissue Network (CHTN) and commercially available human tissue arrays (Cybrdi, Frederick, MD, USA; Biomax, Gaithersburg, MD, USA).
IL-13Rα2 levels in BSG and normal tissue sections were determined using two different antibodies at two different institutions. Polyclonal chicken antibody against IL-13Rα2 protein (Genway, San Diego, CA, USA) was used for IHC assay at SJCRH, and goat polyclonal antibody against IL-13Rα2 (R&D, Minneapolis, MN, USA) was used at the Tumor Vaccines and Biotechnology Branch (TVBB), Center for Biologics Evaluation and Research (CBER), FDA (Bethesda, MD, USA). At both institutions, 5-μ paraffin sections were deparaffinized, treated with 100%, 75%, and 50% 200-proof ethyl alcohol prepared in RNase-free water, and microwaved for 15 min in antigen retrieval reagent to unmask the antigen. Auto-fluorescence in tissue sections was minimized by sodium borohydride treatment, and the sections were blocked for 2 h in block buffer consisting of 5% rabbit serum and 1% biotin-free bovine serum albumin in 1× phosphate-buffered saline (PBS). At TVBB, tissue sections were then incubated in primary antibody at a concentration of 0.5 μg/ml for 16 h at 4° C, washed, and incubated with biotinylated rabbit antigoat antibody. These sections were then reacted with Q-dot 655 streptavidin (0.5 μg/ml) for 45 min, washed, and incubated further with biotinylated antistreptavidin antibody (1 μg/ml) for 45 min to amplify the fluorescent signals of the immunostaining. In the final step of the assay, the samples were incubated with Q-dot 655 streptavidin (0.5 μg/ml) for 45 min at room temperature. After washing with PBS three times, the sections were mounted with 90% glycerol and viewed in a Nikon fluorescence microscope using Q-dot 655 filters (Chroma, Rockingham, VT, USA). For control, the samples were immunostained with isotype control goat immunoglobulin G (IgG) and processed through all steps of the protocol. At SJCRH 3,3′-diaminobenzidine/peroxidase assay was used to localize antibody binding in IHC as described previously.51
The tumor sections were evaluated and graded for IL-13Rα2 immunostaining twice at different time points independently at SJCRH and at TVBB (B.H.J. and F.V.) in a blinded fashion. Normal brain sections were evaluated at TVBB. The percentage of IL-13Rα2–positive fields in tissue sections was counted by viewing the tumor or tissue section under the same magnification (×200). Intensity of immunostaining was also recorded on a semiquantitative scale (<1+, 1+, 2+, and 3+). After staining and positive field assessment, codes were deidentified, and results were compared between the two institutions.
ISH assay was performed in BSG and normal brain specimens for the expression of IL-13Rα2 RNA using Q-dot 525–labeled probe hybridization technique (Joshi et al., unpublished results). For ISH, we used the same tumor specimens previously used for IHC assay after stripping off the antigen–primary antibody–secondary antibody complex by incubating with Restore Western Stripping buffer (Pierce, Rockford, IL, USA) for 2 h at room temperature. The specimens were washed with 1× PBS and incubated in 25 mM glycine buffer (pH 2.5) for 30 min to inactivate residual Q-dot–based fluorescence on the specimens. The samples were washed with 1× PBS and permeabilized by incubating with 5 μg/ml proteinase K (Sigma-Aldrich, St. Louis, MO, USA) for 15 min at room temperature. The DNA in the sections was destroyed by incubating with 5 U/ml of DNase for 6 h at room temperature. The sections were washed three times and hybridized with an in vitro–transcribed biotinylated antisense riboprobe for IL-13Rα2; RNA was dissolved in 2× hybridization buffer (4× saline sodium citrate [SSC], 0.2 M sodium phosphate [pH 6.50], 2× Denhardt’s solution, and 0.1mg/ml sodium azide) and 20% dextran sulfate in deionized formamide. An in vitro transcribed biotinylated sense riboprobe for IL-13Rα2 was used as a negative control. The slides were heated to 65° C for 5 min, hybridized for 16 h at 42° C, and subsequently washed with 0.2× SSC three times and 1× PBS two times. The sections were then incubated with 0.5 μg/ml streptividin–Q-dot 525 for 45 min, washed three times with PBS, and reacted with biotinylated anti-streptavidin antibody and with 0.5 μg/ml streptividin–Q-dot 525 for 45 min for amplification of the hybridized signals. The slides were washed three times with PBS, dried, mounted with 90% glycerol, and viewed under a Nikon fluorescence microscope using Q-dot 525 filters. The fluorescence microscopic images were digitized and analyzed using IP lab software (Scanalytics, Alexandria, VA, USA). The tissue sections were evaluated and graded for IL-13Rα2 hybridized fluorescence intensity twice at different time points by two authors (B.H.J. and F.V.) in a blinded fashion.
The results of the IHC and ISH analyses were analyzed at the Operations and Biostatistics Center of PBTC and also at the Biostatistics Branch, Office of Biostatistics and Epidemiology, CBER, FDA. To investigate agreement between the SJCRH and the FDA labs and between ISH and IHC assays, simple kappa (κ) and weighted κ statistics were estimated, and 95% confidence intervals and p values are provided. Fisher’s exact test was used to investigate the association between two dichotomous factors. The Cochran-Armitage trend test was used to detect trend differences between the levels of a dichotomous variable and an ordinal variable, the Cochran-Mantel-Haenszel test for correlations between two ordinal categorical variables, and Spearman’s rank correlation to investigate rank correlations between two continuous variables.
At FDA/CBER, 28 BSG tumor specimens were analyzed for IL-13Rα2 expression by highly sensitive Q-dot–based IHC analysis (Table 1). This IHC technique is highly sensitive for the detection of low fluorescence signals because of the amplification of signal through biotin-streptavidin complex on the tissue section. Overall, 17 of 28 (61%) BSG specimens expressed IL-13Rα2 with varying degrees of immunostaining that ranged from weak to strong staining (1+, 2+, and 3+). Representative BSG-positive specimens are shown in Fig. 1A. Five samples showed 1+, 3 samples 2+, 9 samples 3+, and 11 samples <1+ staining intensity. IHC staining was observed at both the cell surface and intracellularly (Fig. 1A; results not shown). Tissue sections stained with isotype IgG showed no or weak staining. In contrast to tumors, eight pediatric and six adult normal brain tissue samples demonstrated weak staining for IL-13Rα2 chain (<1+). However, sample of brain tissue taken from a pediatric patient with epilepsy showed 1+ fluorescent intensity (Table 2, Fig. 1B).
IL-13Rα2 expression by IHC performed at SJCRH also identified IL-13Rα2 expression in 17 of 28 tumors. The mean percentage of IL-13Rα2 tumor cell expression in the 17 immunopositive BSG sections was 80% (range, 60%–100%). Significant agreement was seen in the IHC score recorded by the two different methods and the two different research sites (simple κ = 0.284; weighted κ = 0.424; exact p = 0.025). Interestingly, the percentage of IL-13R–positive tumor samples recorded in the two different sites for each tumor also significantly rank-correlated with each other (Spearman’s rank correlation coefficient, rS = 0.55; p = 0.0023). Normal brain sections showed few positive cells (Table 2).
ISH analysis of the same 28 BSG specimens and 15 normal brain specimens employed for IHC was performed after stripping sections of primary and secondary antibodies. Nineteen (68%) showed >1+ staining, while nine BSG samples showed <1+ staining. Among >1+ intensity samples, five showed 1+, five showed 2+, four showed 3+, and five showed 4+ staining intensity. Fig. 2A shows examples of ISH staining of four randomly selected BSG specimens displaying <1+ to 3+ ISH intensity. In contrast, no staining was seen in sections in which an in vitro–transcribed biotinylated sense control riboprobe to IL-13Rα2 was used. Consistent with IHC, only one normal brain sample showed 1+ ISH staining intensity, while 14 of 15 tested normal brain specimens showed <1+ staining intensity (Table 2, Fig. 2B).
We also analyzed the results to determine whether any relationship existed between IL-13Rα2 expression and clinical variables. As shown in Table 1, the immunofluorescence intensity of immunostaining for IL-13Rα2 expression had no correlation with the clinical variables (e.g., age, sex, or different grades of tumor), even though a majority of specimens overexpressed this receptor chain at protein and mRNA levels. Although this conclusion was made at both SJCRH and FDA/CBER laboratories, it is important to note that the sample size is rather small. It is difficult to conclude whether IL-13Rα2 expression correlates with tumor types. Additional samples need to be analyzed for a definite conclusion.
As shown in Table 3, 61%–68% of BSG specimens had at least >1+ IL-13Rα2 intensity measured by both IHC and ISH analyses. Agreement between ISH and IHC assays for IL-13Rα2 intensities was significantly correlated (p < 0.0001); simple κ estimate = 0.358, and weighted κ = 0.89 (p = 0.001). Further, analysis of these data suggested that percent positive fields for IL-13Rα2 intensities obtained from IHC and ISH techniques was also highly rank-correlated (rS = 0.88, p < 0.0001). A similar relationship between ISH and IHC was observed in normal brain specimens, suggesting that ISH and IHC intensities of IL-13Rα2 expression correlated significantly (p < 0.001).
We demonstrate that diffusely infiltrative pediatric BSG expresses readily detectable levels of IL-13Rα2 RNA and protein. In contrast, this receptor was not detected in the majority of normal pediatric and adult brain samples. Importantly, RNA and protein IL-13Rα2 expression levels were significantly correlated in our studies, and the results of two different IHC studies conducted in two separate research facilities were also in close agreement. Our results also suggest heterogeneity in IL-13Rα2 expression, particularly in high-grade BSG tumors. More than 50% of grade 4 tumors had low levels of IL-13α2 expression. Since the sample size is small, additional samples are needed to confirm this conclusion. Together, these data indicate that IL-13Rα2 is expressed to relatively high levels in pediatric BSG and can be readily and reliably detected using either IHC or ISH.
These data are important because IL-13Rα2 is being developed as a potential drug target for pediatric glioma, and reliable detection of this protein will be crucial for the conduct of appropriate clinical trials. We have previously demonstrated that the IL-13Rα2 chain is over-expressed in adult GBM primary culture cells and cell lines.49,50,52 We have also reported that IL-13Rα2 is over-expressed in 83% of pediatric brain tumor specimens of non-BSG.50 These pediatric brain tumors express type I IL-13R, which is composed of the IL-13Rα1, IL-13Rα2, and IL-4Rα chains. Thus, the current study extends these previous observations to show that IL-13Rα2 is also expressed in BSG.
It is important to note that we were able to successfully perform ISH assays on the same tissue sections employed for IHC analysis following stripping of antigen- antibody immune complexes. This technology may be used to study protein and RNA expression and perhaps multiple gene products in BSG when often small biopsy specimens are available.
The significance of IL-13Rα2 expression in BSG or any other tumor tissue is not clear. IL-13Rα2 has been shown to be a major IL-13 binding component of the IL-13R complex.16 After binding to IL-13, the IL-13Rα2 chain internalizes inside the cell for biological activity.53 Whether IL-13Rα2 signals in BSG cells is not known. We have also considered the role of IL-13Rα2 in tumorigenicity and metastasis. In that regard, the IL-13Rα2 gene may undergo mutations or rearrangements causing transformation of normal cells. Although our initial studies have found no mutations in the IL-13Rα2 gene in cell cultures derived from adult glioma samples, it is still possible that this gene undergoes unknown changes.54 Mutation analysis of IL-13Rα2 in BSG has not been performed.
Previous studies have demonstrated that the IL-13Rα2 chain can sensitize human glioma cells to a chimeric fusion protein comprising IL-13 and a mutated form of Pseudomonas exotoxin (termed IL-13–PE38QQR). This cytotoxin is highly cytotoxic to IL-13R–positive malignancies, including brain tumors in vitro and in vivo.29,32,55–57 Based on these observations, it is predicted that IL-13 cytotoxin will also be active in pediatric patients with BSG. However, because of heterogeneity in receptor expression, it is noted that IL-13Rα2 expression could be a critical factor in BSG patients for evaluating response to IL-13 immunotoxin-based therapy. Also, as not all pediatric BSG cells are positive for the IL-13Rα2 chain, it is possible that IL-13 cytotoxin will not eliminate all malignant cells. Therefore, it is noteworthy and we have reported that plasmid-mediated gene transfer of IL-13Rα2 sensitizes tumors to IL-13 cytotoxin in vitro and in vivo.34,58 IL-13 cytotoxin eliminated the majority of malignant cells in in vivo animal models after IL-13Rα2 plasmid injection. Tumor regression was attributed not only to cytotoxic effect of IL-13 cytotoxin but also to infiltration by innate immune cells. Therefore, it is possible that in the clinical setting, IL-13 cytotoxin will be effective in eliminating most tumor cells, even if they do not uniformly express the IL-13Rα2 chain.58,59 Thus, direct intratumoral administration of IL-13 cytotoxin may provide a useful strategy for patients in whom surgical resection cannot be achieved. Because IL-13 cytotoxin has been shown to be well tolerated when administered directly to adult glioma or to brain tissue adjoining the tumor resection cavity37,48,49,56,60 and in the brainstem in monkeys,61 this agent may have utility in the treatment of pediatric patients with BSG. Therefore, additional preclinical and clinical studies need to be performed to determine the utility of IL-13R targeting for BSG therapy.
This work was supported in part by NIH grant U01 CA81457 for the Pediatric Brain Tumor Consortium (PBTC) and the American Lebanese Syrian Associated Charities. We thank Drs. Brent McCright and Robert Aksamit, both from CBER, for their help in reading the manuscript and providing critical comments. We also thank Dr. James Boyett of PBTC, SJCRH for his help in statistical analysis. The views presented in this article do not necessarily reflect those of the FDA. These studies were conducted as part of collaboration between the FDA and NeoPharm, Inc., under a Cooperative Research and Development Agreement (CRADA).