Medulloblastoma arises in the cerebellum and is the most common malignant brain tumour of childhood, however its molecular basis is not well understood. To assess the role of aberrant epigenetic events in medulloblastoma and identify critical genes in its development, we profiled the promoter methylation status of 11 candidate tumour-suppressor genes (TSGs; p14ARF, p15INK4b, p16INK4a, CASP8, HIC1, EDNRB, TIMP3, TP73, TSLC1, RIZ1 and RASSF1A) in medulloblastoma cell lines, primary tumours and the normal cerebellum. Gene-specific TSG methylation was a significant feature of both medulloblastomas and the cerebellum. Extensive hypermethylation of RASSF1A was detected frequently in medulloblastomas but not in the normal cerebellum (41/44 primary tumours versus 0/5 normal cerebella). In contrast, complete methylation of HIC1 and CASP8 in a subset of primary tumours (17/44 and 14/39) occurred against a consistent background of partial methylation in the normal cerebellum. These data therefore indicate that extensive methylation of RASSF1A, HIC1 and CASP8 are tumour-specific events in medulloblastoma. Moreover, methylation of these genes in medulloblastoma cell lines was associated with their epigenetic transcriptional silencing and methylation-dependent re-expression following treatment with the DNA methyltransferase inhibitor, 5-aza-2′-deoxycytidine. The remaining genes studied showed either low frequency methylation (p14ARF, p16INK4a, RIZ1; <7% of cases), no evidence of methylation (p15INK4b, TIMP3, TP73, TSLC1), or comparable patterns of methylation in the normal cerebellum (EDNRB), suggesting that their hypermethylation does not play a major role in medulloblastoma. Our data demonstrate that tumour-specific hypermethylation affects only a subset of genes, and does not support the existence of a concordant methylation phenotype in this disease. We conclude that epigenetic TSG inactivation is a significant feature of medulloblastoma, and identify RASSF1A, HIC1 and CASP8 as potentially critical genes in its pathogenesis. Furthermore, methylation observed in the normal cerebellum emphasises the requirement for appropriate control tissues when assessing the tumour-specificity of TSG hypermethylation.

Abbreviations: 5-aza CdR, 5-aza-2′-deoxycytidine; COBRA, combined bisulphite and restriction analysis; MSP, methylation-specific PCR; TSG, tumour-suppressor gene

Introduction

Medulloblastoma, the most common malignant brain tumour of childhood, is a primitive neuro-ectodermal tumour arising in the cerebellum. The aggressive clinical behaviour of the tumour and the cognitive and endocrinological long-term side effects of current therapies make both the development of prognostic indicators for disease stratification and the identification of new therapeutic targets a major goal. Current understanding of the molecular biology of medulloblastoma is limited. Cytogenetic studies have described consistent chromosomal aberrations, however molecular genetic studies have identified specific genetic abnormalities in only a small proportion of tumours (reviewed in refs 1,2).

Hypermethylation of promoter-associated CpG islands leading to transcriptional silencing has emerged as an important mechanism of epigenetic inactivation of tumour suppressor genes (TSGs) in cancer development (reviewed in ref. 3). Using a genome-scanning approach, aberrant patterns of CpG island methylation have been described in medulloblastoma (4,5), although few gene-specific events have been identified to date. Recently, we have demonstrated bi-allelic epigenetic inactivation by promoter methylation of the tumour suppressor gene RASSF1A (ras-association domain family protein 1, isoform A) in medulloblastoma (6), further highlighting the potential significance of epigenetic TSG inactivation in the development of this tumour.

To examine more widely the prevalence and role of aberrant promoter methylation events in medulloblastoma and to identify further tumour-specific events in its pathogenesis, we have determined the methylation status of the promoter-associated CpG islands of a series of known or candidate TSGs in primary medulloblastoma tumours and the normal cerebellum. In addition to RASSF1A, three groups of candidate genes were selected for analysis: (i) genes which have been reported previously to show evidence of methylation in medulloblastoma [CASP8 (Caspase 8, cysteine-aspartic acid protease 8), HIC1 (hypermethylated in cancer 1) and p16INK4a] (ii) genes which are epigenetically inactivated in other brain tumours [p14ARF p15INK4b, TIMP3 (tissue inhibitor of metalloprotease 3) and TP73] and (iii) genes which exhibit methylation in several different cancer types, suggesting frequent involvement in tumour development [EDNRB (endothelin B receptor), TSLC1 (tumour suppressor gene in human lung cancer) and RIZ1 (retinoblastoma protein interacting zinc finger gene)] This parallel examination of multiple promoter-associated CpG islands has allowed the compilation of an extensive profile of gene-specific methylation events in medulloblastoma. Importantly, the analysis of normal cerebellar tissue has permitted the identification of tumour-specific hypermethylation of a subset of genes, which we show is associated with their transcriptional silencing. This highlights the potential importance of these epigenetic events in medulloblastoma pathogenesis. Finally, we make a preliminary assessment of the clinico-pathological significance of tumour-specific methylation in medulloblastoma.

Material and methods

 Cell lines and patient material

Eleven medulloblastoma cell lines (DAOY, D283 Med, MHH-MED-1, MHH-MED-8A, D341 Med, D384 Med, D425 Med, D458 Med, UW402 and UW228) were studied. D425 Med and D458 Med were derived from the primary and a metastatic tumour from a single patient, respectively; UW228-2 and UW228-3 were derived from the same tumour but exhibit phenotypic differences (21), all other cell lines were derived independently. All cells were grown under recommended culture conditions, and cell line identity was confirmed prior to use by karyotyping (data not shown). Cell line DNA was extracted using the Qiagen DNeasy kit (Qiagen, Crawley, UK).

A cohort of 44 primary medulloblastomas were analyzed, representative of all the major histopathological subtypes (25 classic, seven large cell/anaplastic and 12 nodular/desmoplastic tumours), and adult and paediatric patients (10 infants <3 years, 30 children 3–16 years and four adults >16 years). Clinical and pathological data were centrally reviewed for this study. The presence of metastases was detected on a pre-operative MRI scan. DNA was extracted from frozen tissues using standard methods and from formalin-fixed, paraffin-embedded tissue using a Nucleon hard tissue kit (Amersham Biosciences, Little Chalfont, UK). Normal cerebellar control DNA consisted of post-mortem material from two infants (new born and 25 months) and three adults (60, 67 and 68 years) who had died of non-neoplastic conditions. Normal peripheral blood DNA consisted of a pool of 20 newborn cord blood DNAs. Local Ethical Committee and Institutional Review Board approval has been obtained for the collection, storage and biological study of all material.

Analysis of promoter methylation status

Bisulphite treatment of DNA was carried out using a CpG genome DNA modification kit (Serologicals, Livingston, UK) according to the manufacturer’s instructions. The methylation status of the CpG islands of p14ARF, p15INK4b, p16INK4a, CASP8, HIC1, EDNRB, TIMP3 and RIZ1 was determined by methylation-specific PCR (MSP) (22) using previously published primer sets (13,16,19,22 - B24#B24–25). The methylation status of the CpG islands of RASSF1A, TP73 and TSLC1 was determined by direct sequencing of the reverse strand and estimation of the relative peak heights of PCR products generated from bisulphite treated DNA using previously published primers (14,17,26). The unmethylated control for MSP consisted of pooled normal neonatal cord blood DNA (see above), the methylated control was the same sample methylated in vitro by SssI methylase (New England Biolabs, Hitchin, UK).

Thirty nanograms of bisulphite treated DNA was used per reaction, the PCR reactions were carried out using previously published conditions. MSP PCR products were separated on a 2.5% agarose gel in 1x TBE [0.09 M Tris–Borate, 0.002 M EDTA (pH 9)]. Products analysed by sequencing were directly sequenced with a CEQ DTCS kit (Beckman Coulter, High Wycombe, UK) using the antisense primer to obtain the reverse sequence. Sequenced products were analyzed on a CEQ 2000XL DNA analysis system (Beckman Coulter), and the methylation status at each CpG residue determined by assessment of the relative peak intensities.

COBRA (combined bisulphite and restriction analysis) of the CASP8 promoter was performed by overnight digestion of a PCR product generated using a non-differential primer set CASP8SQ2 (9) with Taq1 (MBI Fermentas, Hanover, MD) at 65°C, which cuts once within the 455 bp PCR product generating products of 396 and 59 bp. The 455 bp uncut product and the 396 bp digested product were separated on a 4% Nusieve 3:1 agarose gel in 1x TBE.

5-Aza-2′-deoxycytidine (5-aza CdR) treatment and RT–PCR

Four cell lines (D425 Med, MED-8A, DAOY and D283 Med) were grown in the presence or absence of the demethylating agent 5-Aza CdR (5 µM) for 4 days. Medium was renewed daily. HeLa cells were also grown as a positive control as these expressed all four transcripts. RNA was extracted from 107 cells using an RNeasy kit (Qiagen). RNA concentration was determined by spectrophotometry, and agarose gel electrophoresis was used to confirm consistent RNA integrity and quantity from each cell line. One microgram of total RNA was used to synthesize cDNA using a reverse transcription system (Promega, Southampton, UK). cDNA was synthesized using random primers and oligo-d(T) primers in separate reactions, followed by pooling. Equivalent amounts of this cDNA was used for PCR amplification of the RASSF1A, CASP8, HIC1 and ß-actin transcripts. The primers for detecting RASSF1A expression were as described previously (26) from exons 2aß and exon4, generating a 242 bp product. CASP8 primers were a previously published set from exons 1 and 3 generating a 379 bp product (9). HIC1 primers were designed for exon 1B 5′-GCGCGGCGGGGGCTGAGAC and exon 2 5′-GCCCTTGGTGCGCTGGTTGTTGAG, product size 244 bp, to amplify the major transcript (28). ß-Actin, a housekeeping gene, was used as a control for RNA integrity using primers BA67 and BA68 (29). Annealing temperatures for RASSF1A, CASP8, HIC1 and ß-actin were 60, 57, 66 and 60°C respectively. ß-Actin PCRs were amplified for 25 cycles, all other reactions were amplified for 35 cycles. PCR products were electrophoresed on a 2.5% agarose gel in 1x TBE.

Results:

Methylation status of CpG islands in medulloblastoma cell lines

In order to highlight genes that are potentially methylated in medulloblastoma tumours, the methylation status of the 11 candidate genes (RASSF1A, CASP8, HIC1, p16INK4a, p14ARF, p15INK4b, TIMP3, TP73, EDNRB, TSLC1 and RIZ1) was initially investigated in a panel of 11 medulloblastoma cell lines by MSP and bisulphite sequencing, using previously validated primer sets and conditions (13,14,16,17,19,22- B23#B23 - B24#B2426). A summary of these results is presented in Table I. All medulloblastoma cell lines showed evidence of methylation of multiple genes, with between two and six of the 11 CpG islands examined methylated in each cell line.