Phosphatidylinositol 3′-Kinase/AKT Signaling Is Activated in Medulloblastoma Cell Proliferation and Is Associated with Reduced Expression of PTEN

Purpose: Medulloblastomas represent the most frequent malignant brain tumors of childhood. They are supposed to originate from cerebellar neural precursor cells. Recently, it has been shown that Sonic Hedgehog–induced formation of medulloblastoma in an animal model is significantly enhanced by activation of the phosphatidylinositol 3′-kinase (PI3K) signaling pathway.

Experimental Design: To examine a role for PI3K/AKT signaling in the molecular pathogenesis of human medulloblastoma, we did an immunohistochemical study of the expression of Ser⁴⁷³-phosphorylated (p)-AKT protein in 22 medulloblastoma samples: All samples displayed p-AKT expression. To investigate if an activated PI3K/AKT pathway is required for medulloblastoma cell growth, we treated five human medulloblastoma cell lines with increasing concentrations of the PI3K inhibitor LY294002 and analyzed cellular proliferation and apoptosis. The antiproliferative effect could be antagonized by overexpressing constitutively active AKT. As the activation of PI3K/AKT signaling may be associated with alterations of the PTEN gene located at 10q23.3, a chromosomal region subject to frequent allelic losses in medulloblastoma, we screened PTEN for mutations and mRNA expression.

Results: Proliferation of all of the medulloblastoma cell lines was dependent on PI3K/AKT signaling, whereas apoptosis was not prominently affected. Allelic loss was detected in 16% of the cases. One medulloblastoma cell line was found to carry a truncating mutation in the PTEN coding sequence. Even more important, PTEN mRNA and protein levels were found to be significantly lower in medulloblastomas compared with normal cerebellar tissue of different developmental stages. Reduction of PTEN expression was found to be associated with PTEN promoter hypermethylation in 50% of the tumor samples.

Conclusions: We conclude that activation of the PI3K/AKT pathway constitutes an important step in the molecular pathogenesis of medulloblastoma and that dysregulation of PTEN may play a significant role in this context.

Medulloblastomas represent the most frequent malignant brain tumors of childhood with an incidence of five cases per 1 million children (1). For a subgroup of medulloblastoma, the desmoplastic variant, an activation of the Shh/Ptch signaling pathway has been shown to be a major molecular pathogenic event, occurring through inactivating mutations of the receptor gene PTCH, activating mutations of its signaling partners, or merely at the transcriptional level (2–4). Animals hemideficient in patched (ptc^+/−) develop medulloblastoma and rhabdomyosarcoma in 15% to 30% depending on the genetic background (5). In view of the fact that the majority of ptc^+/− mutants do not develop tumors, an activation of Hedgehog signaling alone is obviously not sufficient for medulloblastoma induction (6). Very recently, it has been shown that in mice, activated Akt is able to significantly augment the incidence of Shh-induced medulloblastoma (7).

AKT/protein kinase B is a 57 kDa Ser/Thr kinase and the cellular homologue of the viral oncoprotein v-Akt. It represents a central mediator involved in the signal transduction of different growth-controlling pathways that involve phosphatidylinositol 3′-kinases (PI3K). Upon stimulation of different growth factor receptors (e.g., the insulin-like growth factor-I receptor or the epidermal growth factor receptor), PI3K catalyzes the generation of phosphatidylinositol-3,4,5-triphosphate from phosphatidylinositol-4,5-triphosphate. Phosphatidylinositol-3,4,5-triphosphate acts as a potent second messenger to recruit AKT/protein kinase B to the plasma membrane where it is activated by phosphorylation through 3-phosphoinositol-dependent protein kinases. Subsequently, AKT/PKB itself phosphorylates diverse growth-controlling effectors, such as mammalian target of rapamycin, glycogen synthase kinase-3β, or MDM2 (for reviews, see refs. 8, 9). The intracellular level of phosphatidylinositol-3,4,5-triphosphate is tightly regulated by the lipid and protein phosphatase PTEN/MMAC, which dephosphorylates primarily its 3′-inositol position. PTEN represents a putative tumor suppressor gene in medulloblastoma because loss of PTEN function would contribute to an overactivation of the PI3K/AKT signaling pathway, thereby generating a biological context similar to a mouse model published recently (7). In addition, medulloblastoma show frequent losses of heterozygosity of chromosome 10q, where PTEN is located (10q23.31; refs. 10, 11).

The present study was carried out to determine if AKT activation occurs in human medulloblastoma, if PI3K/AKT activation is essential for the proliferation of human medulloblastoma cells, and if this activation is associated with mutations or a reduced mRNA expression of PTEN. A panel of 22 medulloblastoma samples was analyzed for the expression of Ser⁴⁷³-phosphorylated (p)-AKT protein. We then treated five medulloblastoma cell lines with the PI3K inhibitor LY294002 and determined its effect on cellular proliferation and apoptosis. To elucidate the putative role of PTEN for AKT activation in medulloblastoma, a series of 72 human medulloblastoma biopsies and 10 cell lines was screened for PTEN allelic loss and mutations and PTEN mRNA expression levels were evaluated in 24 medulloblastoma biopsies and 11 fetal and 5 postnatal cerebella. Finally, we analyzed PTEN promoter methylation in 10 tumors and 2 fetal cerebella by methylation-specific PCR. Our data indicate that AKT activation plays an essential role in medulloblastoma and that it is associated with genetic and epigenetic alterations of the PTEN tumor suppressor gene.

Patients, tumors, and cell lines. We examined a total of 114 medulloblastoma samples, including 85 classic medulloblastoma, 27 desmoplastic medulloblastoma, and 2 medullomyoblastomas. All tumors were classified according to the revised WHO classification of brain tumors. The majority of the patients were enrolled in the German Society of Pediatric Hematology and Oncology multicenter treatment study for pediatric malignant brain tumors (HIT). DNA was available in a subgroup of cases from peripheral blood. The analyzed cell lines were DAOY, D283-MED, D425-MED, D341-MED, D556-MED, MHH-MED-1, MHH-MED-2, MHH-MED-3, MHH-MED-4 (12–16), MEB-MED-8A, MEB-MED-8S, and 1580WÜ.⁵

DAOY, D283-MED, and D341-MED cells were purchased from American Type Culture Collection (London, United Kingdom). Additionally, 12 fetal, 1 childhood, and 4 adult cerebellar samples were included in the analysis. Fetal cerebella <20 weeks of gestational age were obtained from electively aborted fetuses with the consent of the mother under a fetal tissue collection protocol approved by the ethics committee of the hospital. Fetal tissues >20 weeks were obtained at autopsy. The postnatal control samples were prepared from tissue adjacent to vascular malformations. Patients gave their informed consent to use the material that was removed during an operation. Histologic examination of the fetal and postnatal samples did not show signs of neurologic disorders or cancer. Due to technical reasons, for immunohistochemical experiments addressing developmental issues, we used cerebella from 3- and 15-day-old F₁ offspring of C57BI6 mice that were prepared, fixed, and embedded in paraffin.

DNA extraction. DNA was isolated from peripheral blood and tumor tissue by a standard proteinase K/SDS digestion followed by phenol/chloroform extraction. Tissue fragments selected for DNA extraction were checked by frozen sectioning to ensure that they consisted either of tumor or normal cerebellar tissue. Only fragments with a tumor cell content of at least 80% were included.

Analysis of losses of heterozygosity of the chromosomal region 10q23. Loss of heterozygosity (LOH) analysis was done by microsatellite analysis in 54 samples for which constitutional genomic DNA was available. Markers included D10S215, D10S541, and AFMaa086wg9. Primer sequences are available from the genome database.⁶

PCR was carried out in a final volume of 10 μL containing 100 ng DNA, 5 pmol of each primer, 10 mmol/L Tris-HCl (pH 8.3), 50 mmol/L KCl, 1.0 to 1.5 mmol/L MgCl₂, 200 mmol/L of each deoxynucleotide, and 0.25 unit of Taq polymerase (Life Technologies, Karlsruhe, Germany) with annealing temperatures of 56°C to 58°C. Products were separated by denaturating PAGE and visualized by silver staining as described before (17).

Single-strand conformational polymorphism analysis of the PTEN gene and DNA sequencing. Sixty-two medulloblastomas and 10 cell lines were included in the mutational analysis of the PTEN gene. Ten to 50 ng genomic DNA were used in each PCR reaction for amplification of 12 fragments spanning the coding region of the gene. Primer sequences and detailed PCR conditions are listed in Table 1. The reactions were all done using a Uno Thermoblock Cycler (Biometra, Göttingen, Germany) in a final volume of 10 μL using reagents as described above. Mutational analysis using the single-strand conformation polymorphism method was done as described (18). PCR products showing aberrantly migrating bands were subjected to repeated electrophoresis, bands were excised and eluted, and the DNA was reamplified. The resulting PCR products were purified using the QIAquick PCR purification kit (Qiagen, Hilden, Germany). Sequencing was done with the Applied Biosystems PRISM Dye Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems, Foster City, CA) and a GeneAmp PCR system 9600 (Perkin-Elmer, Wellesley, MA), using 20 ng of PCR product as a template. The sequencing reaction products were analyzed using an Applied Biosystems 377 sequencer (Applied Biosystems).

RNA extraction, cDNA preparation, and semiquantitative reverse transcription-PCR. Twenty-four human medulloblastoma biopsies (13 of classic and 11 of desmoplastic histology) and nine medulloblastoma cell lines were analyzed for PTEN mRNA expression. Because medulloblastoma are composed of undifferentiated neural cells, 12 fetal human cerebellar samples (15-30 weeks of gestational age) were included in the study as control tissues. Additionally, 4 adult (47-67 years) and 1 pediatric (1 year) cerebellar samples were analyzed. To verify tissue composition, tumor tissue samples were examined by frozen section before RNA extraction. Only fragments with a tumor cell content of at least 80% were subjected to RNA extraction. Total cellular RNA was extracted with the Trizol reagent (Life Technologies), according to the instructions of the manufacturer. To remove any contaminating DNA, the samples were digested with DNase I (Promega). Subsequently, DNase was removed by an additional round of Trizol extraction. Reverse transcription was done using the Superscript II Preamplification System (Life Technologies) with random hexamers as primers in a final volume of 20 μL, using the protocol provided by the manufacturer. Semiquantitative analysis of PTEN mRNA expression was done with a duplex reverse transcription-PCR for β₂-microglobulin (5′-TGTCTTTCAGCAAGGACTGG-3′ and 5′-GATGCTGCTTACATGTCTCG-3′) and PTEN (5′-TTCCATCCTGCAGAAGAAGC-3′ and 5′-ACGCCTTCAAGTCTTTCTGC-3′) using primers spanning intronic sequences to exclude signals from contaminating genomic DNA. Reverse primers were 5′-labeled with fluorescent dyes (5-FAM; MWG Biotech, Munich, Germany). Fragment sizes were 148 bp for β₂-microglobulin and 216 bp for PTEN. PCR reactions were carried out in a final volume of 10 μL containing 100 ng of cDNA, 5 pmol of each primer, 10 mmol/L Tris-HCl (pH 8.3), 50 mmol/L KCl, 1.5 mmol/L MgCl₂, 200 mmol/L of each deoxynucleotide, and 0.25 unit of Taq polymerase (Life Technologies). The PCR program consisted of an initial denaturation at 94°C for 5 minutes followed by 33 cycles of 94°C for 35 seconds, 61°C for 40 seconds, and 72°C for 40 seconds, and a final extension step at 72°C for 10 minutes. The number of cycles had been determined to be in the exponential phase by kinetic experiments using 20 to 38 cycles. The resulting PCR products were loaded onto 4.5% denaturating acrylamide gels and reaction products were analyzed on a semiautomated DNA sequencer (ABI 377) using the Genescan software (ABI, Darmstadt, Germany). Each PCR was carried out at least in triplicate. Relative expression levels were calculated as ratios of the signal intensities of PTEN and β₂-microglobulin products.

Methylation-specific PCR. Bisulfite treatment of genomic DNA of 10 medulloblastoma and 2 fetal cerebellar samples of which sufficient material was available was done essentially as described before (19). Methylation-specific PCR for the PTEN promoter region was done with primers 5′-TTTTTTTTCGGTTTTTCGAGGC-3′ and 5′-CAATCGCGTCCCAACGCCG-3′ (methylated sequence) and 5′-TTTTGAGGTGTTTGGGTTTTTGGT-3′ and 5′-ACACAATCACATCCCAACACCA-3′ (unmethylated sequence) amplifying 134 and 124 bp fragments located in the PTEN 5′ region (20). Human lymphocytic DNA artificially methylated with SssI Methylase (New England Biolabs, Frankfurt, Germany) and untreated lymphocytic DNA were used as controls. Products were analyzed on 2% agarose gels stained with ethidium bromide.

Immunohistochemistry. Paraffin sections were cut at 4 μm, mounted on positively charged slides (Superfrost, Menzel, Braunschweig, Germany), and air-dried overnight at 37°C. Before staining, the sections were deparaffinized in xylene and rehydrated in a graded alcohol sequence. Subsequently, sections were boiled for 10 minutes at 600 W in citrate buffer solution (pH 6.0) in a microwave oven. Then, to block endogenous peroxidase activity, the slides were incubated in 1% hydrogen peroxide diluted in methanol for 30 minutes. Finally, they were rehydrated in distilled water, followed by buffer. For p-AKT (Ser⁴⁷³) staining, PBS (used for PTEN) was substituted by TBS. The slides were then incubated in a blocking solution (PBS or TBS with 5% nonfat dry milk and 2% normal swine serum) for 60 minutes at room temperature. This was followed by a 15-minute incubation with avidin-biotin blocking solutions (avidin-biotin blocking kit; Vector Laboratories, Inc., Burlington, Ontario, Canada). The solution was removed from the slides using a filter paper and the polyclonal anti-PTEN (1:50, Zymed Laboratories, San Francisco, CA) and anti-p-AKT (Ser⁴⁷³; 1:25, Cell Signaling Technology, Frankfurt, Germany) antibodies were added to the samples overnight at 4°C (PTEN) or 22°C (p-AKT). After removing unbound antibody by several rinses with PBS or TBS containing 0.1% Triton X-100, the bound antibody was detected using the avidin-biotin complex method (DAKO, Glostrup, Denmark) and visualized by diaminobenzidine tetrahydrochloride. Slides were lightly counterstained with hematoxylin. Finally, the sections were dehydrated in graded alcohols and mounted in Corbit-Balsam (Hecht, Kiel, Germany). As positive controls, prostatic tissue containing carcinoma and normal parenchyma (PTEN) and breast cancer biopsies (p-AKT) were used. Immunostaining against Ki-67 was carried out using the MIB-1 antibody (DAKO), the horseradish peroxidase kit and diaminobenzidine tetrahydrochloride (Zytomed, Berlin, Germany).

Culture and treatment of human medulloblastoma cell lines, [³H]thymidine uptake assays. Medulloblastoma cell lines were cultured, as described before, under serum-supplemented conditions (15). For [³H]thymidine uptake assays, cells were grown in 96-well dishes (Nunc, Wiesbaden, Germany) in a volume of 100 μL and a concentration of 1 × 10⁵/mL (except 5 × 10⁴/mL for D283-MED and D425-MED). Treatments were done with increasing amounts of LY294002 (Cell Signaling Technology). DMSO concentrations were always <0.1% and appropriate controls were included. Assays were done for 48 hours using a minimum of five replicates. For [³H]thymidine uptake assays, 12.5 μL of medium containing [³H]thymidine (3 μCi/mL, Amersham, Freiburg, Germany) were added for the last 12 hours of treatment. Cells were collected using a cell harvester (Packard Filtermate 196) and [³H]thymidine uptake was determined in a microplate scintillation and luminescence counter (Topcount NXT, Packard), using a specific software (Topcount NXT software, version 1.6). [³H]thymidine uptake was normalized to the basal uptake in controls.

Analysis of apoptosis (4′,6-diamidino-2-phenylindole staining). Cells were grown under the serum-supplemented conditions as described above in a volume of 3 mL in six-well culture dishes (Nunc). They were treated for 48 hours with individual doses of LY294002 (GI₅₀ and GI₇₅) or DMSO as a control. Cells were harvested, washed in PBS, fixed with 3.7% paraformaldehyde for 10 minutes at room temperature, and washed again in PBS. They were then treated with 1 μg/mL 4′,6-diamidino-2-phenylindole (Sigma, Tantkirchen, Germany) for 10 minutes, washed twice with PBS, and mounted on appropriate slides using Fluoromount-G medium (Southern Biotechnology Associates, Inc.; Birmingham, AL). Stained nuclei were visualized and photographed using a Leica DMLB fluorescence microscope. For each assay, 300 cells were analyzed in triplicates. Apoptotic cells were morphologically defined by chromatin condensation and fragmentation.

Expression vectors, cell transfection, and cell sorting. The constitutively active myristoylated AKT construct and its control empty vector (pUSE) were obtained from Upstate (Lake Placid, NY). Twenty-four hours after seeding, D283-MED cells were cotransfected with the expression construct or control vector together with the pmaxGFP vector (Amaxa, Cologne, Germany) using FuGene6, following the instructions of the manufacturer (Roche, Basel, Switzerland). After another 24 hours, the cells were sorted on a fluorescence-activated cell sorting DiVa cellsorter (BD Biosciences, Heidelberg, Germany) according to their green fluorescent protein fluorescence intensity. Propidium iodide was added to the cell suspension immediately before sorting at a final concentration of 0.5 μg/mL to exclude dead cells. The standard 488 nm argon laser line was used for excitation of green fluorescent protein and propidium iodide. Green fluorescent protein fluorescence was recorded on a 520/20 bandpass filter and a 630/22 bandpass filter was used for monitoring propidium iodide uptake. Untransfected cells served as controls to determine levels of autofluorescence. Sorted cells were collected in bovine serum albumin–precoated test tubes. After sorting, cells were cultured in 96-well plates, as described above, in medium supplemented with 1% Penstrep (Invitrogen, Carlsbad, CA). Myristoylated AKT expressing cells and control cells were treated with 14.6 μmol/L LY294002 (corresponding to the GI₅₀ value). To analyze cellular proliferation, [³H]thymidine uptake assays were done as described above.

Western blot analysis. Cells were cultured as described above in a volume of 3 mL in six-well culture dishes (Nunc). Treatments with 5 and 15 μmol/L LY294002 or control DMSO were done for 1 hour. Lysis, electrophoresis, and staining were done as described before (21) using the p-(Ser⁴⁷³)-AKT and AKT polyclonal antibodies (Cell Signaling Technologies, Frankfurt, Germany) following the instructions of the manufacturer.

Medulloblastomas exhibit a significant expression of AKT phosphorylated at Ser⁴⁷³. In the immunohistochemical analysis of p-AKT (Ser⁴⁷³) expression, all 22 medulloblastoma samples showed detectable levels of p-AKT. Five cases (two classic medulloblastoma and three desmoplastic medulloblastoma) showed strong, 12 cases (8 classic medulloblastoma, 4 desmoplastic medulloblastoma) moderate, and 5 cases (all classic medulloblastoma) weak expression. Two representative cases are shown in Fig. 1 together with staining for the Ki-67 (MIB-1) proliferation marker. Proliferative activity was particularly prominent in areas with AKT phosphorylation. p-AKT expression was not present in adjacent normal cerebellar tissue (Fig. 1). However, a murine cerebellar sample of postnatal day 3 displayed significant p-AKT staining in the external granule layer that contains neuronal precursors that display a high proliferative activity. In a 15-day-old murine cerebellum, p-AKT was not detectable anymore (data not shown).

Proliferation of human medulloblastoma cells depends on an activated PI3K/AKT signaling pathway. Five medulloblastoma cell lines were cultured with increasing concentrations of the PI3K inhibitor LY294002. All cell lines showed a significant drug-induced inhibition of [³H]thymidine uptake (Fig. 2). The 50% (GI₅₀) and 75% (GI₇₅) growth inhibition levels were calculated by nonlinear regression using Prism 4 for Macintosh, version 4.0a. Mean GI₅₀ ± SD of all cell lines was 15.7 ± 1.1 μmol/L; mean GI₇₅ ± SD was 28.9 ± 5.8 μmol/L. Reduction of [³H]thymidine uptake was associated with a dose-dependent decrease in the levels of phosphorylated (Ser⁴⁷³)-AKT (Fig. 2). The cell lines 1580WÜ and D425-MED, which tolerated slightly higher doses of LY294002, showed lower levels of basal AKT phosphorylation. To assess the effects of PI3K/AKT inhibition on apoptosis, we did 4′,6-diamidino-2-phenylindole staining. The experiments revealed that LY294002 leads to a significant increase in the apoptotic rate of MHH-MED-1 (t test GI₅₀, P < 0.05; GI₇₅, P < 0.01), 1580WÜ (t test GI₅₀, P < 0.05; GI₇₅, P < 0.001), and D425-MED (t test, GI₅₀: not significant; GI₇₅, P < 0.05). However, the overall apoptotic rate [mean at GI₅₀, 2.69 ± 0.43 %; max (D425-MED), 4.99 ± 0.88%] seemed rather moderate compared with the changes in proliferation rates (Fig. 2). To find out if an activated AKT signal can counteract the inhibitory effects of LY294002, we overexpressed myristoylated AKT in D283-MED cells. In the control experiment, treatment with 14.6 μmol/L LY294002 (corresponding to the GI₅₀ calculated from the experiment shown in Fig. 2) led to a decrease in proliferative activity to 41%. [³H]thymidine uptake in cells expressing constitutively active AKT and treated with the same dose was significantly higher (63%; t test, P < 0.01; Fig. 3).

Medulloblastomas display LOH of the chromosomal region 10q23.31. Fifty-four medulloblastoma samples were included in the analysis of the allelic status of chromosomal region 10q23.1. Informative patterns were obtained in 37 of 51 samples analyzed at D10S215, 13 of 53 analyzed at AFMa086wg9, and 25 of 52 analyzed at D10S541. LOH was detected in 5.4%, 15.4%, and 16% of the informative samples, respectively. All medulloblastomas with LOH showed the classic, nondesmoplastic histology. Altogether, four samples with LOH 10q23.1 were identified. Representative data are shown in Fig. 4.

PTEN is mutated in the medulloblastoma cell line MHH-MED-1. Sixty-two medulloblastoma samples and 10 medulloblastoma cell lines were included in a mutational analysis of the PTEN coding region. One sample (D324) showed an intronic polymorphism (C→A) in intron 7 (data not shown). The medulloblastoma cell line MHH-MED-1 was identified as carrying a truncating mutation in exon 8, codon 274 (Fig. 4).

PTEN mRNA expression is significantly reduced in medulloblastomas compared with normal fetal and adult cerebellum. Twenty-four human medulloblastoma biopsies, 10 medulloblastoma cell lines, 12 fetal cerebellar samples, and 4 postnatal cerebellar samples were analyzed for PTEN mRNA expression (Fig. 5). Statistical analysis, using the Mann-Whitney test, revealed that medulloblastoma (median, 0.54), independent of their histologic subtype, express PTEN mRNA at a level significantly lower than fetal (median, 2.01; P < 0.0001) and postnatal cerebellum (median, 1.24; P = 0.035) and the entire set of normal cerebellar tissues (median, 1.51; P < 0.0001). There was no significant difference between classic medulloblastoma and desmoplastic medulloblastoma (Fig. 5). In addition, PTEN mRNA expression was analyzed in nine medulloblastoma cell lines, including four of those used in the in vitro study (not D283-MED). In all of these cell lines, PTEN was found to be expressed at detectable levels corresponding roughly to expression levels found in the tumor samples (data not shown). Immunostaining for PTEN on 11 tumor samples revealed that, in contrast to normal cerebellum, PTEN protein is not detectable in medulloblastoma by immunohistochemistry (Fig. 5).

Transcriptional reduction of PTEN is associated with hypermethylation of its promoter. Five of 10 medulloblastoma samples (50%) analyzed for PTEN promoter methylation by methylation-specific PCR displayed a distinct methylated band, whereas five tumor samples did not. Those samples that were also included in the mRNA expression analysis and that displayed PTEN promoter methylation showed very low PTEN mRNA levels: D1106 (mean ± SE, 0.41 ± 0.10), D1103 (0.42 ± 0.04), D1198 (0.51 ± 0.06), and D963 (0.51 ± 0.09; Fig. 6). D1165 expressed higher levels of PTEN mRNA (1.28 ± 0.4) and was found to be unmethylated in the PTEN 5′ region; however, D967 expressed low levels of PTEN (0.41 ± 0.09) without being methylated. The two fetal cerebellar samples analyzed were found to be unmethylated in the selected 5′ promoter fragment and expressed high levels of PTEN mRNA: R1628 (1.35 ± 0.13) and R1585 (1.21).

Although medulloblastoma represent the most frequent malignant brain tumor in childhood, their molecular pathogenesis is only partially understood. In the present study, we show that an activated PI3K/AKT signaling pathway constitutes a common event in medulloblastoma. Furthermore, we show that PI3K signaling pathway activity is important for medulloblastoma cell proliferation and that this activation is associated with a significantly reduced mRNA and protein expression of the PTEN tumor suppressor gene product. We provide evidence that epigenetic mechanisms may be involved in PTEN inactivation in medulloblastoma.

Medulloblastomas represent a heterogenous group of highly malignant embryonal brain tumors. Histopathologically, a classic and a desmoplastic subtype of medulloblastoma can be distinguished. The molecular phenotype of the desmoplastic subtype, representing ∼25% of medulloblastoma, has recently been well characterized. This group of tumors is associated with inactivating mutations of the PTCH tumor suppressor gene, activating mutations of its interaction partners, and an activation of the Hedgehog-Patched signaling pathway (2–4, 22, 23). Recent data have shown that PI3K/AKT signaling enhances the effects of Hedgehog in these tumors and their cerebellar precursors, and that it may represent an indispensable precondition for the molecular realization of the Hedgehog signal (21, 24, 25). In agreement with this finding, retroviral co-overexpression of Shh and activated Akt in cerebellar neural progenitors of the mouse newborn cerebellum significantly increased the incidence of tumors compared with the incidence of tumors induced by Shh alone (7).

On the other hand, less information is available on the more frequent medulloblastoma variant, the classic subtype. Two lines of evidence argue in favor of a role for PI3K/AKT signaling in these tumors. First, several growth factor receptors involving the PI3K/AKT cascade in their signal transduction are expressed in medulloblastoma, e.g., the insulin-like growth factor-I receptor (26, 27), TrkB (28), PDGFRB (29, 30), CXCR4 (31, 32), ErbB2 (33), and c-KIT (34). Second, medulloblastoma have been reported to display, in a considerable fraction, allelic losses in the chromosomal region 10q23 where the tumor suppressor gene PTEN, a negative regulator of the PI3K/AKT signaling pathway, is located (35).

We therefore screened a panel of 22 medulloblastoma samples for expression of the activated form of AKT, p-(Ser⁴⁷³)-AKT. Our finding of AKT phosphorylation in all samples, independent of the histopathologic subtype, is in good agreement with previous reports (7, 36, 37), which found increased expression of Ser⁴⁷³-phosphorylated AKT in a significant subset of medulloblastoma or PNET samples. Our data imply that PI3K/AKT pathway activation—because it represents a common phenomenon—might play an important role in medulloblastoma. We, therefore, wanted to analyze the functional relevance of PI3K/AKT signaling in medulloblastoma proliferation. Our data show that the growth of all medulloblastoma cell lines tested does depend on an activated PI3K/AKT pathway. 4′,6-Diamidino-2-phenylindole staining excluded that the effects of LY294002 on [³H]thymidine uptake are predominantly due to an induction of apoptosis. Because an overexpression of constitutively activated AKT could partially rescue D283-MED cells from LY294002-induced growth inhibition, we conclude that AKT is an essential component in the PI3K-dependent cascade activated in medulloblastoma. However, as myristoylated AKT did not provide a complete compensation of LY294002-induced antiproliferative effects, further downstream targets of PI3K seem to be activated in medulloblastoma. Pharmacologic inhibition of the PI3K/AKT pathway may represent a promising therapeutic approach. Recently, Castillo et al. (38) reported that phosphatidylinositol ether lipid analogues inhibited AKT activity by interacting with the phosphoinosite-binding site in the PH domain of AKT.

As medulloblastomas are known to display allelic losses of the chromosomal region 10q, where the PTEN tumor suppressor gene is located, we wanted to know if this gene is altered in medulloblastoma. Inactivation of PTEN constitutes a possible mechanism that might be responsible for AKT activation. Our finding of allelic losses of the chromosomal region 10q23 in 16% of medulloblastoma corresponds roughly to previous data (39–41). Interestingly, Reardon et al. (42) showed that losses of 10q are associated with metastatic disease at diagnosis. In our single-strand conformation polymorphism analysis of a series of 62 medulloblastoma samples and 10 medulloblastoma cell lines, only a single mutation in PTEN was detected. It was found in MHH-MED-1, a cell line established from the spinal fluid metastases of a relapsed tumor in a 10-year-old boy (15). Unfortunately, no tumor tissues from this patient were available for analysis. To our knowledge, only a small panel of 22 medulloblastoma samples has been studied for PTEN mutations previously. The only mutation detected occurred in a recurrent tumor that also carried a p53 mutation (43). Thus, it seems that the frequency of PTEN mutations is low in medulloblastoma and that mutations may be associated with a more aggressive or relapsing phenotype of the tumor.

Alternatively, PTEN could be inactivated at the transcriptional level, potentially involving epigenetic mechanisms. Indeed, mRNA expression analysis showed that PTEN is expressed at significantly lower levels in the tumor samples compared with cerebellar control tissues. However, persistent detection of PTEN mRNA in all medulloblastoma cell lines analyzed argues against a high incidence of homozygous deletion of PTEN, as suggested in a very recent report (44).

As epigenetic inactivation of PTEN has recently been reported for other tumors, including breast cancer (45), pancreatic cancer, colorectal cancer, and glioblastomas (46–48), we analyzed a set of 10 randomly chosen medulloblastoma and 2 fetal cerebellar samples of which sufficient DNA was available by methylation-specific PCR. We found hypermethylation of the PTEN promoter in 50% of the tumors and, in a significant subset of these cases, PTEN mRNA expression was reduced to 20% to 25% compared with the median of fetal cerebellum. This result was further confirmed by the absence of PTEN protein with immunostaining. Thus, mechanisms beyond the transcriptional level may be involved in the regulation of PTEN. Indeed, a recent report shows that mouse PTEN mRNA contains an alternative 5′ untranslated sequence with an inhibitory effect on its translation (49). Predominant transcription of homologous human sequences may contribute to reduced PTEN protein levels.

In summary, our findings provide evidence that PI3K/AKT signaling may play an important role in human medulloblastoma, in particular in context with further signaling pathways involved in its pathogenesis, e.g., Hedgehog and Wnt signaling. The data also indicate that PTEN may be involved in activation of the PI3K/AKT pathway in medulloblastoma, possibly representing an alternative or additional molecular step to abundant membrane receptor activation or mutations in PI3KCA (50). Additional studies will be required to further characterize the role of the PI3K/AKT/PTEN system in medulloblastoma.

We thank Andreas Dolf and Peter Wurst for excellent technical assistance, U. Klatt for photographic work, and Drs. Darell Bigner and Greg Riggins (Duke Comprehensive Cancer Center, Duke University Medical Center, Durham, NC) for kindly providing the D425-MED and D556-MED cells.