Long-range regulatory interactions at the 4q25 atrial fibrillation risk locus involve PITX2c and ENPEP.

BACKGROUND: Recent genome-wide association studies have uncovered genomic loci that underlie an increased risk for atrial fibrillation, the major cardiac arrhythmia in humans. The most significant locus is located in a gene desert at 4q25, approximately 170 kilobases upstream of PITX2, which codes for a transcription factor involved in embryonic left-right asymmetry and cardiac development. However, how this genomic region functionally and structurally relates to PITX2 and atrial fibrillation is unknown. RESULTS: To characterise its function, we tested genomic fragments from 4q25 for transcriptional activity in a mouse atrial cardiomyocyte cell line and in transgenic mouse embryos, identifying a non-tissue-specific potentiator regulatory element. Chromosome conformation capture revealed that this region physically interacts with the promoter of the cardiac specific isoform of Pitx2. Surprisingly, this regulatory region also interacts with the promoter of the next neighbouring gene, Enpep, which we show to be expressed in regions of the developing mouse heart essential for cardiac electrical activity. CONCLUSIONS: Our data suggest that de-regulation of both PITX2 and ENPEP could contribute to an increased risk of atrial fibrillation in carriers of disease-associated variants, and show the challenges that we face in the functional analysis of genome-wide disease associations.

An evolutionarily conserved three-dimensional structure in the vertebrate Irx clusters facilitates enhancer sharing and coregulation.

Developmental gene clusters are paradigms for the study of gene regulation; however, the mechanisms that mediate phenomena such as coregulation and enhancer sharing remain largely elusive. Here we address this issue by analysing the vertebrate Irx clusters. We first present a deep enhancer screen of a 2-Mbp span covering the IrxA cluster. Using chromosome conformation capture, we show that enhancer sharing is widespread within the cluster, explaining its evolutionarily conserved organization. We also identify a three-dimensional architecture, probably formed through interactions with CCCTC-binding factor, which is present within both Irx clusters of mouse, Xenopus and zebrafish. This architecture brings the promoters of the first two genes together in the same chromatin landscape. We propose that this unique and evolutionarily conserved genomic architecture of the vertebrate Irx clusters is essential for the coregulation of the first two genes and simultaneously maintains the third gene in a partially independent regulatory landscape.

Allelic variation at the 8q23.3 colorectal cancer risk locus functions as a cis-acting regulator of EIF3H.

Common genetic variation at human 8q23.3 is significantly associated with colorectal cancer (CRC) risk. To elucidate the basis of this association we compared the frequency of common variants at 8q23.3 in 1,964 CRC cases and 2,081 healthy controls. Reporter gene studies showed that the single nucleotide polymorphism rs16888589 acts as an allele-specific transcriptional repressor. Chromosome conformation capture (3C) analysis demonstrated that the genomic region harboring rs16888589 interacts with the promoter of gene for eukaryotic translation initiation factor 3, subunit H (EIF3H). We show that increased expression of EIF3H gene increases CRC growth and invasiveness thereby providing a biological mechanism for the 8q23.3 association. These data provide evidence for a functional basis for the non-coding risk variant rs16888589 at 8q23.3 and provides novel insight into the etiological basis of CRC.

Genome-wide profiling of p63 DNA-binding sites identifies an element that regulates gene expression during limb development in the 7q21 SHFM1 locus.

Heterozygous mutations in p63 are associated with split hand/foot malformations (SHFM), orofacial clefting, and ectodermal abnormalities. Elucidation of the p63 gene network that includes target genes and regulatory elements may reveal new genes for other malformation disorders. We performed genome-wide DNA-binding profiling by chromatin immunoprecipitation (ChIP), followed by deep sequencing (ChIP-seq) in primary human keratinocytes, and identified potential target genes and regulatory elements controlled by p63. We show that p63 binds to an enhancer element in the SHFM1 locus on chromosome 7q and that this element controls expression of DLX6 and possibly DLX5, both of which are important for limb development. A unique micro-deletion including this enhancer element, but not the DLX5/DLX6 genes, was identified in a patient with SHFM. Our study strongly indicates disruption of a non-coding cis-regulatory element located more than 250 kb from the DLX5/DLX6 genes as a novel disease mechanism in SHFM1. These data provide a proof-of-concept that the catalogue of p63 binding sites identified in this study may be of relevance to the studies of SHFM and other congenital malformations that resemble the p63-associated phenotypes.

Long-range gene regulation links genomic type 2 diabetes and obesity risk regions to HHEX, SOX4, and IRX3.

Genome-wide association studies identified noncoding SNPs associated with type 2 diabetes and obesity in linkage disequilibrium (LD) blocks encompassing HHEX-IDE and introns of CDKAL1 and FTO [Sladek R, et al. (2007) Nature 445:881-885; Steinthorsdottir V, et al. (2007) Nat. Genet 39:770-775; Frayling TM, et al. (2007) Science 316:889-894]. We show that these LD blocks contain highly conserved noncoding elements and overlap with the genomic regulatory blocks of the transcription factor genes HHEX, SOX4, and IRX3. We report that human highly conserved noncoding elements in LD with the risk SNPs drive expression in endoderm or pancreas in transgenic mice and zebrafish. Both HHEX and SOX4 have recently been implicated in pancreas development and the regulation of insulin secretion, but IRX3 had no prior association with pancreatic function or development. Knockdown of its orthologue in zebrafish, irx3a, increased the number of pancreatic ghrelin-producing epsilon cells and decreased the number of insulin-producing beta-cells and glucagon-producing alpha-cells, thereby suggesting a direct link of pancreatic IRX3 function to both obesity and type 2 diabetes.

Understanding the regulatory genome.

The sequencing of the whole genome of multiple species provides us with the instruction book of how to build an organism and make it work, plus a detailed history of how diversity was generated during evolution. Unfortunately, we still understand only a small fraction, which is locating where genes are and deciphering the proteins they code for. The next step is to understand how the correct amount of gene products are produced in space and time to obtain a fully functioning organism, from the egg to the adult. This is what is known as the regulatory genome, a term coined by Eric H. Davidson. In this review, we examine what we know about gene regulation from a genomic point of view, revise the current in silico, in vitro and in vivo methodological approaches to study transcriptional regulation, and point to the power of phylogenetic footprinting as a guide to regulatory element discovery. The advantages and limitations of each approach are considered, with the emerging view that only large-scale studies and data-crunching will give us insight into the language of genomic regulatory systems, and allow the discovery of regulatory codes in the genome.

Gene dosage and mutational analyses of EGFR in oligodendrogliomas.

We have studied amplification/gene-dosage and sequence variations of the EGFR gene in 41 oligodendroglial tumours graded according to the WHO classification (21 oligodendrogliomas grade II, 13 oligodendrogliomas grade III and 6 oligoastrocytomas grade II-III), using multiplex ligation-dependent probe amplification (MLPA), real-time quantitative PCR, and PCR/SSCP techniques. To determine gene-dose we studied exons 11 (extracellular domain) and 25 (intracellular domain) in the EGFR gene. Overdose (1- to 5-fold increase) was present in exon 11 in 21 of 41 samples (52.5% of cases) and in exon 25, in 7 of 41 samples (17.5% of cases). Gene amplification > 5-fold increase) was present in exon 11, in 17 of 41 samples (42.5% of cases), and in exon 25 in 6 of 41 samples (15% of cases). Three tumours (two grade II oligodendrogliglioma, one mixed oligoastrocytoma) displayed high level amplifications: > 100 gene copies were identified by both real-time quantitative PCR and MLPA analyses. Gene sequence alterations were identified by PCR/SSCP and sequencing in four cases: two missense mutations: G1051A (Ala351Thr) and G2216A (Arg739Hys); one nonsense mutation: C2934T (Asp978Asp); and an 18 bp deletion in position 2423-2441 of E19. These changes were present only in tumoral DNA, not in the corresponding constitutional patients' DNA. We also found four previously unidentified polymorphic variants: G2025A (Ala675Ala), C2233T (Leu745Leu), C2895T (Treo965Treo) and C3168T (Asp1056Asp), and three previously described polymorphic changes: E12+22 Tright curved arrow A, G1748A (Arg583Lys) and A2547T (Glu849Glu). Our findings demonstrate that mutations and amplification/overdose in the EGFR gene are present in low-grade oligodendroglial tumours, and may contribute to the development of these brain neoplasms.

Real-time quantitative PCR analysis of regions involved in gene amplification reveals gene overdose in low-grade astrocytic gliomas.

We have studied gene amplification of genes located in 1q32 (GAC1, ELF3, MDM4, and ren1), 4q11 (PDGFR-alpha), and in 12q13-14 (MDM2 and CDK4) using quantitative real-time PCR in a group of 86 tumors consisting of 44 WHO grade IV glioblastomas (GBM) (34 primary and 10 secondary tumors), 21 WHO grade III anaplastic astrocytomas (AA), and 21 WHO grade II astrocytomas (AII). Gene amplification was present in 56 of the 86 samples (65%) in at least 1 gene in our series. GAC1 (51%) and MDM4 (27%) were the most frequently amplified genes within the 1q32 amplicon, and their higher amplification frequency was statistically significant (P<0.05, chi) in the low-grade astrocytomas. Concordant co-amplification was determined for ELF3 and ren1 or ren1 and MDM4 in the grade III-IV tumors. MDM2 amplification was significantly more frequent in primary GBM (16%) than was in secondary GBM (0%). The present study shows that gene amplification in the studied regions is already present in low-grade astrocytic tumors and that amplification of some genes may represent another molecular marker to differentiate primary from secondary GBM.

No evidence of INI1hSNF5 (SMARCB1) and PARVG point mutations in oligodendroglial neoplasms.

Allelic losses of chromosome 22 found in oligodendrogliomas suggest that at least one tumor suppressor gene on chromosome 22 is inactivated during the multistep process of tumorigenesis in this glial tumor. INI1hSNF5 (HUGO symbol: SMARCB1), located at 22q11, encodes a component of the ATP-dependent chromatin remodeling hSWI-SNF complex; it is a tumor suppressor gene that is mutated in several malignant tumors. The PARVG gene, located at 22q13, has been found to exhibit reduced expression in some cancer lines. Both genes are thus candidate tumor suppressors, potentially involved in the pathogenesis of gliomas. We performed mutation analyses of INI1hSNF5 and PARVG in a series of 40 oligodendrogliomas, but only sequence polymorphic variations were identified. Accordingly, INI1hSNF5 and PARVG do not seem to be the tumor suppressor genes involved in oligodendroglioma development and progression.

Real-time quantitative PCR analysis of gene dosages reveals gene amplification in low-grade oligodendrogliomas.

Proto-oncogene amplification is an important alteration that is present in about 45% to 50% of high-grade human gliomas. We studied this mechanism in 8 genes (cyclin-dependent kinase-4 [CDK4], MDM2, MDM4, renin-angiotensin system-1, ELF3, GAC1, human epidermal growth factor receptor-2, and platelet-derived growth factor receptor-A gene) in a series of 40 oligodendrogliomas (World Health Organization (WHO) grade II, 21; WHO grade III, 13; and WHO grade II-III oligoastrocytomas, 6) using real-time quantitative polymerase chain reaction. Amplification of at least 1 of these genes was detected in 58% of samples (23/40). By histopathologic grade, 67% of grade II oligodendrogliomas (14/21), 46% of grade III anaplastic oligodendrogliomas (6/13), and 50% of mixed oligoastrocytomas (3/6) were positive for amplification of at least 1 gene. CDK4, MDM2, and GAC1 were the most frequently involved genes (12/40 [30%], 12/40 [30%], and 13/40 [33%], respectively). Our findings demonstrate gene amplification in low-grade samples indicating that it is an important alteration in the early steps of oligodendroglioma development and, therefore, might be considered a molecular mechanism leading to malignant progression toward anaplastic forms.

Mutational study of the 1p located genes p18ink4c, Patched-2, RIZ1 and KIF1B in oligodendrogliomas.

Loss of 1p heterozygosity is one of the most characteristic events in oligodendrogliomas. Several genes located in this region have been previously studied to find the target gene implicated in the development of this tumor without success. Patched-2, RIZ1 and KIF1B are novel oncosuppressor genes located at 1p and involved in different kinds of tumors. We have studied these genes and p18(ink4c) using PCR/SSCP methods to detect sequence variations in a series of 40 oligodendrogliomas in which the allelic status at 1p was analyzed. Polymorphisms or no sequence changes were detected in all four genes analyzed. None of the genes analyzed seem to be the target-gene mapped at 1p involved by mutation in oligodendroglioma development.

Genetic and epigenetic alteration of the NF2 gene in sporadic meningiomas.

The role of the NF2 gene in the development of meningiomas has recently been documented; inactivating mutations plus allelic loss at 22q, the site of this gene (at 22q12), have been identified in both sporadic and neurofibromatosis type 2-associated tumors. Although epigenetic inactivation through aberrant CpG island methylation of the NF2 5' flanking region has been documented in schwannoma (another NF2-associated neoplasm), data on participation of this epigenetic modification in meningiomas are not yet widely available. Using methylation-specific PCR (MSP) plus sequencing, we assessed the presence of aberrant promoter NF2 methylation in a series of 88 meningiomas (61 grade I, 24 grade II, and 3 grade III), in which the allelic constitution at 22q and the NF2 mutational status also were determined by RFLP/microsatellite and PCR-SSCP analyses. Chromosome 22 allelic loss, NF2 gene mutation, and aberrant NF2 promoter methylation were detected in 49%, 24%, and 26% of cases, respectively. Aberrant NF2 methylation with loss of heterozygosity (LOH) at 22q was found in five cases, and aberrant methylation with NF2 mutation in another; LOH 22q and the mutation were found in 16 samples. The aberrant methylation of the NF2 gene also was the sole alteration in 15 samples, most of which were from grade I tumors. These results indicate that aberrant NF2 hypermethylation may participate in the development of a significant proportion of sporadic meningiomas, primarily those of grade I.

Molecular analysis of the erbB gene family calmodulin-binding and calmodulin-like domains in astrocytic gliomas.

Primarily involved in cell proliferation and differentiation processes, the plasma membrane-bound ErbB tyrosine kinase receptor family is formed by four members: erbB1/EGFR, erbB2/HER2/Neu, erbB3/HER3 and erbB4/HER4. Calmodulin (CaM) is a Ca2+-binding protein involved in the regulation of multiple intracellular processes that binds directly to EGFR in the presence of Ca2+, inhibiting its tyrosine kinase activity. Two main regions in the receptor have been implicated in this relationship: the calmodulin-binding domain (CaM-BD) and the calmodulin-like domain (CaM-LD); their sequences are highly conserved in other members of this family of receptors. The presence of mutations, amplification and/or overexpression and genomic rearrangement of these domains was investigated for all four erbB family genes in a series of 89 glial tumors, including 44 WHO grade IV glioblastomas, 21 WHO grade III anaplastic astrocytomas, and 24 WHO grade II astrocytomas. Gene alterations were only found in the regions of interest in EGFR. One glioblastoma showed an in frame tandem duplication of the intracellular region including CaM-LD (exons 18-25). CaM-BD gene overdose was evidenced in 18 tumors that showed EGFR amplification in other domains. Over-expression of CaM-BD and CaM-LD was detected in 6 and 17 cases, respectively, of the 19 tumors in which this study was performed. The other three genes coding for the ErbB receptors did not present point mutations, or rearrangements, and only a very low amplification rate was found for erbB2 (1 case) and erbB3 (4 cases). No overexpression of erbB2, erbB3 or erbB4 was detected. These findings suggest that EGFR is the main erbB gene family member non-randomly involved in malignant glioma development, and that the two domains under study, due to their high conservation and wide separation in the EGFR sequence, are good marker regions for evaluating EGFR/erbB1 gene amplification, as well as for analysing the presence of transcripts corresponding to truncated cytosolic forms of the receptor in these tumors.

Hypermethylation of the DNA repair gene MGMT: association with TP53 G:C to A:T transitions in a series of 469 nervous system tumors.

O6-methylguanine-DNA methyltransferase (MGMT) plays a major role in repairing DNA damage from alkylating agents. By removing alkyl groups from the O6-position in guanine, MGMT can prevent G:C to A:T transition mutations, a type of variation frequently involving TP53 mutations in brain tumors. Promoter hypermethylation of CpG islands in tumor-related genes can lead to their transcriptional inactivation, and this epigenetic mechanism has been shown to participate in MGMT silencing in some cancers, including those affecting the nervous system. Accordingly, a link between both genetic and epigenetic anomalies may exist in these neoplasms. To determine the relevance of defective MGMT function due to aberrant methylation in relation to the presence of TP53 mutations, we studied 469 nervous system tumors (including all major histological subtypes) for MGMT promoter methylation and TP53 mutations at exons 5-8. Overall, aberrant methylation occurred in 38% of the samples (180/469), with values higher than 50% in the more malignant forms such as glioblastomas and anaplastic gliomas including those with astrocytic, oligodendroglial and ependymal differentiation. In contrast, the non-glial tumors displayed an overall aberrant MGMT promoter methylation of 26%, even though this group includes highly malignant tumors such as neuroblastomas, medulloblastomas and brain metastases. Overall, TP53 mutations were found in 25% of the methylated MGMT tumors (45/180), whereas only 10% of the unmethylated MGMT tumors (30/289) showed TP53 changes (P < 0.001). G:C to A:T changes occurred at CpG sites in 9% of methylated tumors, and in 0.7% of the unmethylated samples. This type of transition at non-CpG dinucleotides was also more frequent in the tumors with aberrant MGMT methylation (5%) than the unmethylated tumors (0.7%). These data suggest that MGMT silencing as a result of promoter hypermethylation may lead to G:C to A:T transition mutations in the TP53 gene of some histological nervous system tumor subtypes.

DNA methylation of multiple promoter-associated CpG islands in meningiomas: relationship with the allelic status at 1p and 22q.

The purpose of this research was to examine the DNA methylation profile of meningiomas. Accordingly, we examined the DNA methylation status of ten tumor-related genes (RB1, p16(INK4a), p73, MGMT, ER, DAPK, TIMP-3, p14(ARF), THBS1, and Caspase-8) in 98 meningiomas (68 grade I; 27 grade II; and 3 grade III samples) using methylation-specific PCR and sequencing. The most frequently methylated genes were THBS1 (30%), TIMP-3 (24%), p16(INK4a) (17%), MGMT (16%), p73 (15%), ER (15%), and p14(ARF) (13%), whereas methylation was relatively rare in the other genes (<10%). Methylation occurred in at least one gene in 77.5% of the cases and in three or more genes in 25.5%. Methylation was tumor specific since it was absent in the controls: two non-neoplastic meningeal samples and two non-neoplastic brain samples. The frequency of aberrant gene methylation in grade I versus grade II-III tumors showed some differences for TIMP-3, THBS1, MGMT, p16(INK4a) and p73; these differences reached statistical significance for TIMP-3: 18% in grade I versus 40% in grade II-III (P < 0.02). Our previous loss of heterozygosity studies provided the allelic constitution at 1p and 22q for 60 of the 98 meningiomas included in this report. The level of aberrant promoter methylation increased in tumors (30 samples) displaying 1p loss (either isolated or as concurrent deletion at 1p/22q; P = 0.014). These meningiomas primarily accumulated the epigenetic changes of THBS1 (14/30; 47%; P < 0.005), TIMP-3 (12/30; 40%; P < 0.05), p73 (10/30; 26%; P < 0.02) and p14(ARF) /p16(INK4a)(7/30 each one; 23%; not significant). Our findings indicate that aberrant DNA methylation of promoter-associated CpG islands in meningiomas contributes to the development of these tumors.

Deletion and aberrant CpG island methylation of Caspase 8 gene in medulloblastoma.

Aberrant methylation of promoter CpG islands in human genes is an alternative genetic inactivation mechanism that contributes to the development of human tumors. Nevertheless, few studies have analyzed methylation in medulloblastomas. We determined the frequency of aberrant CpG island methylation for Caspase 8 (CASP8) in a group of 24 medulloblastomas arising in 8 adult and 16 pediatric patients. Complete methylation of CASP8 was found in 15 tumors (62%) and one case displayed hemimethylation. Three samples amplified neither of the two primer sets for methylated or unmethylated alleles, suggesting that genomic deletion occurred in the 5' flanking region of CASP8. Our findings suggest that methylation commonly contributes to CASP8 silencing in medulloblastomas and that homozygous deletion or severe sequence changes involving the promoter region may be another mechanism leading to CASP8 inactivation in this neoplasm.

Aberrant CpG island methylation of multiple genes in ependymal tumors.

Aberrant methylation of promoter CpG islands in human genes represents an alternative mechanism for genetic inactivation, and contributes to the development of human tumors. Nevertheless, thus far, few reports have analyzed methylation in ependymomas. We determined the frequency of aberrant CpG island methylation of several tumor-associated genes: p16(INK4a), RB1, MGMT, DAPK, TIMP3, THBS1, TP73, NF2 and Caspase 8 in a group of 27 ependymomas, consisting of 22 WHO grade II samples and five anaplastic WHO grade III tumors. The respective methylation indices (number of genes methylated/total genes analyzed) for both tumor groups was 0.195 and 0.198. Overall methylation rates greater than 20% were detected in MGMT, TIMP3, THBS1 and TP73. NF2 and Caspase 8 each presented hypermethylation in less than 10% of cases, and the cell-cycle regulators RB1/p16(INK4a) were hypermethylated in 4% and 18% of the samples, respectively, mostly affecting the low-grade forms. Our findings suggest that methylation commonly contributes to the inactivation of cancer-related genes in ependymomas.

Promoter methylation status of multiple genes in brain metastases of solid tumors.

The aberrant methylation of the CpG island promoter regions acquired by tumor cells is one mechanism for loss of gene function. The high methylation rate for RB1 and death-associated protein-kinase gene (DAP-kinase) (60 and 90%, respectively) previously found in brain metastases suggests this mechanism could be non-randomly associated to tumor progression and metastasis. Thus, in addition to these two genes, we determined the methylation status of the genes p16INK4a, glutathione S-transferase P1 (GSTP1), O6-methylguanine DNA methyltransferase (MGMT), thrombospondin-1 (THBS1), p14ARF, TP53, p73, and tissue inhibitor of metalloproteinase 3 (TIMP-3), in 18 brain metastases of solid tumors, with methylation specific PCR. The metastases were derived from malignant melanoma (three cases), lung carcinoma (six cases), breast carcinoma (three cases), ovarian carcinoma (two cases) and one each from colon, kidney, bladder and undifferentiated carcinoma. We detected methylation levels in the tumor samples of 83% in p16INK4a, 72% in DAP-kinase, 56% in THBS1, 50% in RB1, 39% in MGMT, 33% in GSTP1 and p14ARF each, 22% in p73 and TIMP-3 each, and 11% in TP53. The methylation index (number of genes methylated/number of genes tested) varied between 0.1 and 0.6, with an average of 0.42, indicating that a high grade of gene methylation accumulates parallel to the tumor metastasis process. Our data suggest an important role for gene methylation in the development of brain metastases, primarily involving epigenetic silencing of DAP-kinase, THBS1 and the cell-cycle regulators RB1/p16INK4a.

CpG island methylation in sporadic and neurofibromatis type 2-associated schwannomas.

PURPOSE: The purpose of this research was to examine the DNA methylation profile of schwannomas. EXPERIMENTAL DESIGN: We examined the DNA methylation status of 12 tumor-related genes (NF2, RB1, p14(ARF), p16(INK4a), p73, TIMP-3, MGMT, DAPK, THBS1, caspase-8, TP53, and GSTP1) in 44 sporadic and/or NF2-associated schwannomas using methylation-specific PCR. RESULTS: The most frequently methylated genes were THBS1 (36%), p73 (27%), MGMT (20%), NF2 (18%), and TIMP-3 (18%). The RB1/p16INK4a gene pair displayed aberrant methylayed alleles in 15% of cases, whereas methylation was relatively rare in the other genes (<5%). Methylation was tumor specific because it was absent in two nonneoplastic nerve sheath samples and two nonneoplastic brain samples studied as controls. CONCLUSIONS: Our findings indicate that aberrant methylation seems to be a mechanism for NF2 gene inactivation, considered an early step in schwannoma tumorigenesis, and as well, aberrant hypermethylation of other tumor-related genes might represent secondary events that also contribute to the development of these tumors.

Aberrant promoter methylation of multiple genes in oligodendrogliomas and ependymomas.

Promoter hypermethylation represents a primary mechanism in the inactivation of tumor suppressor genes during tumorigenesis. To determine the frequency and timing of hypermethylation during carcinogenesis of nonastrocytic tumors, we analyzed promoter methylation status of 10 tumor-associated genes in a series of 41 oligodendrogliomas (22 World Health Organization [WHO] grade II; 13 WHO grade III; 6 WHO grade II-III oligoastrocytomas) and 7 WHO grade II-III ependymomas, as well as 2 nonneoplastic brain samples, by a methylation-specific polymerase chain reaction. Aberrant CpG island methylation was detected in 9 of 10 genes analyzed, and all but one sample displayed anomalies in at least one gene. The frequencies of hypermethylation for the 10 genes were as follows, in oligodendrogliomas and ependymomas, respectively: 80% and 28% for MGMT; 70% and 28% for GSTP1; 66% and 57% for DAPK; 44% and 28% for TP14(ARF); 39% and 0% for THBS1; 24% and 28% for TIMP3; 24% and 14% for TP73; 22% and 0% for TP16(INK4A); 3% and 14% for RB1; and 0% in both neoplasms for TP53. No methylation of these genes was detected in normal brain tissue samples. We conclude that a high frequency of aberrant methylation of the 5' CpG island of the MGMT, GSTP1, TP14(ARF), THBS1, TIMP3, and TP73 genes is observed in nonastrocytic neoplasms. This aberration seems to occur early in the carcinogenesis process (it is already present in the low-grade forms), although in some instances (DAPK, THBS1, and TP73) it appears also associated with the genesis of anaplastic forms.