Evaluation of the Prognostic Value of CD44 in Glioblastoma Multiforme

Materials and Methods

Patient population. This study was approved and performed according to the guidelines of the Institutional Review Board of Chang Gung Memorial Hospital (approval #94-182). Written, informed consent was obtained from all patients. The patient population consisted of 73 adults whose gliomas or supratentorial GBM were verified histologically between May 2003 and April 2006 (Table I). Histopathological examinations were performed by a neuropathologist according to WHO criteria. Tumors showing prominent microvascular proliferation and/or necrosis, in addition to high cellularity, marked nuclear atypia, and remarkable mitotic activity were diagnosed as GBM.

Region-specific specimen collection. Deep-seated tumors were removed using an intraoperative navigation system (BrainLAB, Feldkirchen, Germany) that minimized invasiveness and maximized patient safety and accurate tumor resection. Brain tissue samples were collected from the resection zone, categorized as peripheral normal brain, tumor marginal tissue or tumor core (according to the navigators' assessment; Figure 1a), and stored in liquid nitrogen.

Preparation of fluorescently labeled cDNA probes. Twenty μg of total RNA extracted from tumor samples were mixed with 2 μg of oligo-dT primer in a total volume of 13 μl, heated to 70°C for 10 min, and cooled on ice. To this mixture was added 1.5 μl of 20× nucleotide cocktail (10 μM each 2′-deoxyadenosine 5′-triphosphate, 2′-deoxycytidine 5′-triphosphate, 2′-deoxyguanosine 5′-triphosphate, and 6 μM 2′-deoxythymidine 5′-triphosphate), 3 μl of either Cy3-2′-deoxyuridine 5′-triphosphate (dUTP) or Cy5-dUTP (GE Healthcare Bio-Sciences, Piscataway, NJ, USA), 6 μl of 5× first-strand buffer, 3 μl of 0.1 M DL-Dithiothreitol, 0.1 μl of RNAguard (GE Healthcare), and 2 μl of 200 units/ml Superscript II reverse transcriptase (Invitrogen, Taipei City, Taiwan, ROC). After 2 h at 42°C, RNA strands were degraded by adding 5 μl of 0.5 N NaOH and incubation for 10 min at 70°C. Samples were neutralized by adding 7.5 μl of 1 M Tris-HCl (pH 7.5). The separately synthesized Cy3- and Cy5-labeled targets were combined and mixed with 20 μg of human Cot-1 DNA (Invitrogen) in a final volume of 500 μl. Target probes were purified by centrifugation using a Microcon YM-30 filtration unit (Millipore, Billerica, MA, USA), and concentrated to a final volume of 5 μl.

DNA microarray hybridization. Whole human genome (GE Healthcare) and a customized oncogene and kinase cDNA microarray consisting of 15,488 probes (National Cheng Kung University Microarray Center, Tainan, Taiwan, ROC) were used in this study. Four sets of brain tissue were analyzed: three from glioblastoma tissues and one from trauma tissue (i.e. control). A total of 2.5 μl of 10 μg/ml poly(A) RNA (Roche Applied Science, Indianapolis, IN, USA), 0.5 μl of 10% SDS, 3 μl of 20× PM (0.1% BSA, 1% SDS), 15 μl of formamide, and 3 μl of 20× SSC (1.5 M NaCl, 150 mM sodium citrate, pH 7.0) were added to each concentrated target, and the volume adjusted to 30 μl with distilled water. Target mixtures were denatured for 2 min at 100°C, incubated in the dark for 20-30 min at room temperature, and then placed on the array under a 24 × 40-mm cover glass. Hybridizations were performed for 16 h at 42°C in a humidified chamber. After hybridization, arrays were washed at room temperature in 2xSSC with 0.l% SDS for 10 min, 0.1× SSC and 0.l% SDS for 10 min, and 0.1× SSC for 5 min. All data were calibrated against commercial human RNA (Clontech, Mountain View, CA, USA).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

MRI of representative tumor, indicating sampling areas (periphery, margin and core).

Scanning and data processing. Washed arrays were scanned using a GSI Lumonics ScanArray 4000 (Packard BioChip Technologies, Billerica, MA, USA). The data (16-bit TIFF images) were analyzed using QuantArray software (PerkinElmer, Waltham, MA, USA), converting the signal intensity of each spot into a text format. The background (i.e. the average intensity of blank spots) was subtracted. Cut-off values for each experiment were set at 5-9-fold of the lowest signal intensity, leaving ~3,000 data points with relatively high signal intensities. For each sample, data points affected by experimental artifacts were eliminated by log transforming the Cy3/Cy5 ratios and subtracting the median of all log (Cy3/Cy5) values from each log (Cy3/Cy5) value. For each gene, variances among the samples were calculated. Genes with more than three missing data values were excluded from further analyses. Potential protein—protein interactions between candidate gene products were analyzed using MetaCore™ software (GeneGo, St. Joseph, MI, USA).

Real-time polymerase chain reaction. The following primers and probe for CD44 were used (Applied Biosystems, Foster City, CA, USA): forward: 5′-CAACTCCATCTGTGCAGCAAA-3′; reverse: 5′-GTAACCTCCTGAAGTGCTCGTC-3′; probe: 5′-CATATTGC TTCAATGCTTCAG CTCCACCTG-3′. Other probes included CD63 antigen (CD63; Hs00156390_m1), fibronectin 1 (FN1; Hs01549940_m1), growth arrest and DNA-damage-inducible, alpha (GADD45A; Hs00355052_g1) and secreted phosphoprotein 1 (SPP1; Hs00959010_m1). Eukaryotic 18S rRNA (Hs99999901_s1) was used as an endogenous control. Real-time PCR cycles were performed as follows: 50°C for 2 min; 95°C for 15 min; forty cycles at 95°C for 15 s and 60°C for 1 min. Experiments were performed in triplicate. Gene expression levels were calculated by the ΔΔ Ct method and normalized against the 18S control.

Immunoblotting analysis. Brain tumor samples were washed twice in ice-cold PBS and lysed on ice in ice-cold T-PER tissue protein extraction reagent (Pierce, Rockford, IL, USA) containing protease inhibitor cocktail (Sigma, St. Louis, MO, USA). Lysates were cleared by centrifugation, and total protein concentrations were determined by Bradford assay (Bio-Rad, Hercules, CA, USA). Protein samples (10 μg/lane) were separated on 12% polyacrylamide gels by SDS—PAGE and transferred to polyvinylidene difluoride membranes (Millipore). Blots were blocked overnight in TBS (20 mM Tris-HCl, 150 mM NaCl, 0.1% Tween-20, 0.5 μM EDTA, pH 7.4) containing 5% nonfat dry milk, incubated for 2 h with anti-human CD44 antibodies (1:50,000; R&D Systems, Abingdon, UK), and then incubated with horseradish peroxidase-conjugated goat anti-mouse IgG (1:20,000; PerkinElmer) for 1 h. Human CD44 was detected using the Western Lightning kit (PerkinElmer) according to the manufacturer's instructions.

Immunocytochemistry. Tissue sections from peripheral, marginal and tumor core regions were deparaffinized, treated with 3% H₂O₂ for 10 min at room temperature, and then microwaved in 0.01 M citrate buffer (pH 6.0) to retrieve antigenicity. The sections were blocked with 1% BSA in PBS for 20 min at room temperature and incubated overnight with a goat anti-human CD44 monoclonal antibody (1:100 dilution in PBS containing 1% BSA). Samples were washed four times with PBS and incubated with goat anti-mouse IgG (PerkinElmer). Immunocomplexes were visualized by the LSAB 2 HRP kit (Dako, Carpinteria, CA, USA) using 3,3′-diaminobenzidine tetrachloride as a substrate. Sections were counterstained lightly with hematoxylin, dehydrated with a graded alcohol series, cleared with xylene, and mounted with coverslips.

Statistical analyses. All data were analyzed using SPSS statistical software, v. 13.0. Gene expression in different tumor regions was analyzed using nonparametric methods for paired samples. Differences between low and high gene expression groups were evaluated using the log rank test; the Spearman correlation coefficient was used to estimate the significance of the correlation. Kaplan—Meier survival curves were used for univariate survival analyses. Cox regression was also used in univariate and multivariate models with adjustments for age.