¹Genome Science Division, Research Center for Advanced Science and Technology, University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8904, Japan

²Department of Neurosurgery, Faculty of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan

³Department of Neurosurgery, Saitama Medical School, 38-2 Morohongo, Moroyama-machi, Iruma-gun, Saitama 350-0495, Japan

⁴Department of Neurosurgery, School of Medicine, Teikyo University, 2-11-2 Kaga, Itabashi-ku, Tokyo 173-8606, Japan

⁵CREST (Core Research for Evolutional Science and Technology), Japan Science and Technology Corporation, 4-1-8 Honcho, Kawaguchi, Saitama 332-0012, Japan

Correspondence to: K Ueki, Department of Neurosurgery, Faculty of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan. E-mail: kueki-tky@umin.ac.jp or H Aburatani, Genome Science Division, Research Center for Advanced Science and Technology, University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8904, Japan. E-mail: haburata-tky@umin.ac.jp

Oligodendrogliomas frequently, but not always show sensitivity to chemotherapy and recent studies demonstrated that allelic loss of chromosome 1p is highly associated with this chemosensitivity. To gain insight into the molecular mechanism of such difference, we examined comprehensive gene expression profiles of 11 oligodendroglial tumors, six with and five without 1pLOH (loss of heterozygosity), and two normal brain tissues using the oligonucleotide microarray (GeneChip). Statistically significant numbers of genes were expressed differentially between the two genetic subsets. Clustering analysis separated the tumor subsets well. The tumors with 1pLOH had similar expression profiles to the normal brain for those differentially expressed genes. Many genes showing higher expression in tumors with 1pLOH were presumed to have functions in nervous tissues. Notably, the majority of the 123 genes showing significant expression reduction in tumors with 1pLOH were either on chromosome 1 (50%) or on 19 (10%), and the average expression reduction ratio was about 50% (0.54±0.13) possibly reflecting the chromosomal deletion. Thus, the biological difference between the genetic subsets of oligodendroglioma was indeed reflected to gene expression profile, which provided baseline information for further studies to elucidate the mechanism of chemosensitivity in gliomas.

Oncogene (2002) 21, 3961-3968 doi:10.1038/sj.onc.1205495

oligodendroglioma; oligonucleotide microarray; loss of heterozygosity

Introduction

Oligodendrogliomas are a major type of gliomas which constitute approximately 5% of all primary brain tumors or 10 to 25% of all intracranial gliomas (Kleihues and Cavenee, 2000). One of the important recent findings in neuro-oncology was that those oligodendrogliomas frequently showed remarkable sensitivity to chemotherapy, especially to a regimen using procarbazine, CCNU and vincristine (PCV therapy) (Cairncross and Macdonald, 1988). However, the response rate to PCV therapy remains 60-80%, and 20-30% of tumors are resistant to chemotherapy and have worse prognosis. Therefore, within this histologically indistinguishable entity, there apparently exist subgroups showing different biological behavior. Recent molecular genetic studies on oligodendrogliomas revealed that allelic loss of chromosome 1p, which is found in 60-80% of oligodendrogliomas and often accompanied with allelic loss of 19q (Smith et al., 1999), was highly associated with the treatment responsiveness and also with a better prognosis (Cairncross et al., 1998; Ino et al., 2001). Thus, it is now being recognized that loss of chromosome 1p is a marker separating oligodendrogliomas into subgroups showing different biological behavior. In addition to its important clinical implications, understanding of the underlying molecular mechanisms of such a difference may lead to a new treatment strategy for all gliomas. Unfortunately, the putative tumor suppressor genes at chromosomes 1p and 19q, obvious keys to investigate the molecular biologic features of the tumor cells, are yet to be identified despite vigorous investigations. Several attractive candidates on chromosome 1p include TP73 (1p36.3) and CDKN2C (1p32), but neither has been shown to be altered in the majority of oligodendrogliomas (Husemann et al., 1999; Mai et al., 1998). Although 1p loss is also found in many other neoplasms including neuroblastomas, the search for the suppressor gene in such neoplasms has not been successful either (Ohira et al., 2000; Schwab et al., 1996). To gain insight into the molecular basis of the biological difference among oligodendrogliomas, we turned to recently developed oligonucleotide microarray technology. By analysing comprehensive gene expressions, several studies have now shown that the expression profiles correlated well with the histology and clinical grades in human neoplasms including gliomas (Golub et al., 1999; Huang et al., 2000; Watson et al., 2001). Therefore, we performed a comparative study of the gene expression profiles between the genetic subgroups of oligodendrogliomas based on the 1p status.

Results

Genetic alterations in oligodendroglial tumor samples

Of 40 oligodendroglial tumors we could collect, we selected six cases with 1pLOH (loss of heterozygosity) and five cases without 1pLOH from which we could obtain good quality RNA evaluable with the GeneChip system (Affymetrix, Santa Clara, CA, USA). Histological diagnoses and the results of molecular genetic analysis are summarized in Table 1. There were seven oligodendrogliomas, one oligoastrocytoma and three anaplastic oligodendrogliomas. In all six tumors with 1pLOH, all of the informative 1p markers showed LOH, indicating that deletion involved the whole arm of chromosome 1p (data not shown). Of the six cases with 1pLOH, five cases also had 19qLOH and one case was non-informative on examined 19q markers. None of the six tumors with 1pLOH had TP53 mutation, and three of the five tumors without 1pLOH had TP53 mutation. No case had 10qLOH.

The statistical analysis of genes differentially expressed by 1p status

To select genes that were expressed differentially by 1p status, we used prediction value (P-value) in neighborhood analysis, which was recently described as useful for extracting genes expressed uniformly high in one group and low in the other (Golub et al., 1999). We listed a total of 209 genes that had an absolute P-value of more than one, of which 86 genes showed higher expression and 123 genes showed lower expression in tumors with 1pLOH. These numbers of the genes were significantly higher than expected in random grouping tested by 1000 times permutation test (P<0.01), indicating that these two subgroups indeed have significantly different gene expression profiles. When Mann-Whitney test with cut-off P-values of 0.05 and 0.01 were used, 288 and 123 genes were detected as differentially expressed by the 1p status, and those numbers were higher than the expected numbers in permutation test which were 115 and 33 in median, respectively. Of the 209 genes selected by prediction value, more than 90% (192 genes) were also included in the 288 genes selected by a P-value of 0.05 by Mann-Whitney test, indicating the consistency of those two methods in selecting differentially expressed genes. We used the 209 genes for further analysis.

Clustering analysis was performed to classify all 13 samples using Pearson correlation with these extracted 209 genes (Figure 1). The tumor subsets were separated well and the normal brain samples were clustered into the same group with the tumors with 1pLOH. Among the five tumors without 1pLOH, expression profiles were not markedly different between the tumors with and without TP53 mutation in this clustering analysis.

Genes showing higher expression in tumors with 1pLOH

Of the 86 genes selected by P-value, 24 genes whose mean average difference had more than threefold difference between the two groups were listed in Table 2. The average differences of those genes in normal brain RNA were close to those in tumors with 1pLOH as expected by the clustering analysis. Based on the UniGene on National Center for Biotechnology Information (NCBI), 14 of the 24 genes were predominantly expressed in brain or neural tissues (KIAA0985, RGS7, human clone 23695, INA, KIAA0750, MYT1L, human clone 23560, PTPRN, SLC1A2, HAPIP, SNCB, SNAP25, L1CAM and OLFM1), and were likely to have some function in the nervous system. In the normal brain samples, genes that are predominantly expressed in glial cells such as glial fibrillary acidic protein (GFAP) and myelin basic protein (MBP) were also well expressed, indicating that these samples contained many glial cells.

Genes showing lower expression in tumors with 1pLOH and their chromosomal location

Of the 123 genes selected by P-value, 61 genes (50%) were mapped to chromosome 1 (58 to 1p, 1 to 1q, and 2 to 1p or 1q) and 12 genes (10%) were mapped to chromosome 19 (11 to 19q, and 1 to 19p), while 50 genes (41%) were mapped to other chromosomes. When we focused on top 30 genes that had an absolute P-value of more than 1.5, 83% (25 genes) were mapped to chromosome 1 or 19. Relative expressions of the 73 genes, 61 on chromosome 1 and 12 on chromosome 19, in tumors with 1pLOH compared to tumors without 1pLOH were 0.54±0.13 in average. Of the 123 genes, 16 genes whose mean average difference had more than threefold difference between the two groups were listed in Table 3 (whole list of the selected genes would be available on request).

The validation using semi-quantitative RT-PCR

Of the 24 higher expressed and 16 lower expressed genes in 1pLOH tumors, semi-quantitative RT-PCR was performed on nine known genes whose differences were more than fourfold and also more than 40 in mean average difference between the two groups (indicated by * in Tables 2 and 3). The results of RT-PCR corresponded well to the GeneChip data (Figure 2). The additional tumor samples showed similar expression pattern to the same 1p status cases examined by GeneChip, although there were two cases (case 18 and 19) which showed exceptional expression pattern. Those two cases neither had allelic loss on 1p/19q nor had TP53 mutation. The case 19 was rather similar to the 1pLOH tumors. Case 18 showed lower expression in some of the genes that had higher expression in other tumors without 1pLOH.

Expression of genes on chromosome 1p

The relative expressions of the genes on chromosome 1p (n=158) in tumors with 1pLOH against tumors without 1pLOH were arranged on the chromosome map to see their relationship with chromosomal loci (Figure 3). Genes showing lower expression in tumors with 1pLOH were distributed over the whole chromosome 1p arm. There also were many genes whose expressions were not decreased in 1pLOH tumors, which were also found in various chromosomal loci.

Discussion

Using the oligonucleotide microarray technology, we could identify genes that were differentially expressed between the subgroups of oligodendroglioma by the 1p status. Results of semi-quantitative RT-PCR performed on some of the identified genes were concordant with the chip analysis data, confirming the fidelity of the system in general. Additional oligodendrogliomas studied by RT-PCR showed similar expression pattern to the GeneChip cases according to their 1p status. Of the five tumors without 1pLOH and without TP53 mutation, however, one additional case was rather similar to the tumors with 1pLOH and another additional case also showed some inconsistency. Such variations of gene expression pattern suggest heterogeneity in tumors without 1pLOH and that more than two subgroups may exist in oligodendroglial tumors, as reported recently (Ino et al., 2001). The numbers of cases analysed in our study were still limited, and a larger-scale study would enable detailed classification of oligodendroglial tumors based on gene expression profiles. Nonetheless, our data clearly showed that oligodendrogliomas of different genetic subsets indeed had distinct gene expression pattern, and could identify many differentially expressed genes.

Five of 10 cases without 1pLOH had TP53 mutation, and the expression patterns of the genes examined by RT-PCR were not significantly different between tumors with and without TP53 mutation. There was no apparent difference in the expression pattern of the 209 genes among five tumors without 1pLOH examined by GeneChip, regardless of the TP53 mutation status (data not shown). Therefore, the listed genes were most likely extracted by the status of 1pLOH than that of TP53.

Of the genes showing higher expression in tumors with 1pLOH (see Table 2), we noticed several genes that were expressed predominantly in normal brain or neural tissue, indicating that those genes are functional in the normal nervous system. Contamination of normal tissue cells was not likely, because allelic losses observed on the microsatellite analysis were almost complete in all cases, indicating that the examined tissues consisted mostly of tumor cells. For example, MYT1L encodes a zinc finger protein which plays a role in the development of neurons in the central nervous system (Kim et al., 1997), and PTPRN, which had especially similar expression pattern to MYT1L in the RT-PCR analysis, is implicated in neuroendocrine secretory processes. SNCB plays a role in neuronal plasticity, SLC1A2 is a glial high affinity glutamate transporter, and HAPIP is also abundantly expressed in the neural tissues. L1CAM is an axonal glycoprotein involved in neuronal migration and differentiation (Kenwrick et al., 2000). In combination with the results of clustering, these data may suggest that tumors without 1pLOH are more distant from normal brain, possibly reflecting their differentiation status.

Analysis on the differentially expressed genes provided potentially interesting information on their chromosomal locations. Of the top 123 genes whose expressions were most significantly decreased in tumors with 1pLOH, nearly 60% were located either on chromosome 1 or chromosome 19, with the ratio of expression levels to tumors without 1pLOH around 50%. It was reported that nearly all oligodendrogliomas with 1p and 19q LOH lose the entire arm of 1p and 19q (Bigner et al., 1999; Nigro et al., 2001; Smith et al., 1999), which was also confirmed by our microsatellite analysis on 1p. Therefore, reduced expression of genes in a wide range of 1p is likely to be a consequence of losing one copy of each gene. On the other hand, some genes on chromosome 1p had higher expressions in tumors with 1pLOH, suggesting that the expression regulations of those genes were not simply dependent on the copy number. In a few genes such as COL11A1 and RBBP4, the relative expressions were remarkably low probably because of their overexpression in tumors without 1pLOH, rather than their expression reduction in 1p losing tumors (see Table 3). Despite the rather comprehensive expression analysis, we still could not pinpoint a particular gene that would affect the chemosensitivity of oligodendrogliomas. None of the genes previously suggested to be related with chemosensitivity, such as O⁶-methylguanine-DNA methyltransferase (MGMT), multidrug resistance 1 (MDR1), multidrug resistance-associated protein (MRP), glutathione S-transpherase pi, metallothionein and topoisomerase II (Nutt et al., 2000; Tanaka et al., 2000), were detected as differentially expressed genes in our study. On the other hand, we showed that significant numbers of genes were differentially expressed between oligodendroglioma subsets, including expression reduction of numerous genes in the chromosome 1p. An interesting question is whether about 50% reduction of such numerous genes in the same chromosomal region could have any biological effect on tumorigenesis or chemosensitivity. Recent studies showed that loss of one copy of a gene and subsequent reduction of its expression level is possibly related to tumorigenesis, a phenomenon called haplo-insufficiency (Fero et al., 1998; Gutmann et al., 1999). Whether similar phenomenon underlies the biological features of oligodendroglioma remains to be investigated.

From a technological point of view, it could be an important observation that the oligonucleotide microarray may be quantitative enough to detect expression reduction caused by the allelic loss in numerous genes. cDNA microarray and serial analysis of gene expression (SAGE) have been tried with some success to detect increase or decrease of expressions of certain genes which were altered by gene amplifications or deletions (Caron et al., 2001; Pollack et al., 1999). Our data indicated that oligonucleotide microarray would be a good system to identify such genes.

In summary, we showed that genetic subsets in oligodendrogliomas by 1p status were reflected in gene expression profile. Some of the interesting genes differentially expressed included genes implicated in the function of nervous tissues, genes on chromosome 1p and 19q. Molecular mechanism of chemosensitivity and chemoresistance may well be represented by those differentially expressed genes, and our data would serve as good baseline data for the future studies to solve that clinically important question.

Materials and methods

Sample preparation

Tumor samples obtained at surgery were snap frozen in liquid nitrogen and stored at -80°C until use. Histological diagnosis was made on formalin-fixed paraffin-embedded tissues processed separately. To minimize the notorious variability of the histological diagnosis in oligodendroglial tumors, the histology slides were reviewed by four independent neuropathologists to make consensus diagnoses following the WHO classification (Kleihues and Cavenee, 2000). Paired blood samples were obtained after written informed consents, and were subjected to DNA extraction for the microsatellite analysis. Of 40 oligodendroglial tumors, six tumors with 1pLOH and five without 1pLOH were selected for expression profiling using GeneChip system (Affymetrix). Total RNA from normal whole brain was purchased from two different providers (Clontech, Palo Alto, CA, USA and Life Technologies, Inc., Rockville, MD, USA), which were used to see the expression profile of the normal neurons and glial cells.

Genetic analysis

LOH assay on chromosomes 1p, 19q and 10q to detect allelic losses were performed using Genetic Analyzer 310 (Applied Biosystems, Foster City, CA, USA) as previously described. The following microsatellite markers located at the commonly deleted in gliomas were used: D1S244, D1S2734, and D1S402 for 1p (1p36), D19S112, D19S596, D19S412 and D19S219 for 19q (19q13), D10S1744, D10S1680 and D10S583 for 10q (10q22-23) (Ueki et al., 2000). For tumors with 1pLOH, four additional 1p markers were further examined to see the range of the deletion: D1S1166 (1p13), D1S495 (1p22), D1S2835 (1p32) and D1S2657 (1p34). The SSCP assay for exons 5 to 8 of TP53 was performed using previously published primer pairs (Fults et al., 1992), again using Genetic Analyzer 310. Exons showing migration shift were PCR amplified again and were directly sequenced using BigDye Terminator Kit (Applied Biosystems) following the manufacturer's protocol. Established comparative multiplex PCR assays were used to detect homozygous deletion of CDKN2A (Ueki et al., 1996). For RNA extraction, the frozen tumor sample was homogenized in Isogen (Nippon Gene, Osaka, Japan) and total RNA was isolated following manufacturer's instructions.

Gene Chip experiment

Five g of total RNA from each sample were used to synthesize biotin-labeled cRNA, which was then hybridized to the high-density oligonucleotide array (GeneChip Human U95A array; Affymetrix) following the previously published protocol with minor modifications (Ishii et al., 2000). Arrays contain probe sets for approximately 12 626 human genes and ESTs, which were selected from Build 95 of the UniGene Database (derived form GenBank 113, dbEST/10-02-99). After washing, arrays were stained with streptavidin-phycoerythrin (Molecular Probes, Inc., Eugene, OR, USA) and analysed by a Hewlett-Packard Scanner to collect primary data. The GeneChip 3.3 software (Affymetrix) was used to calculate the average difference for each gene probe on the array, which was shown as an intensity value of gene expression defined by Affymetrix using their algorithm. The average difference has been shown to quantitatively reflect the abundance of a particular mRNA molecule in a population (Ishii et al., 2000; Lockhart et al., 1996). To allow comparison among multiple arrays, the average differences were normalized for each array by assigning the average of overall average difference values to be 100. A value of two was assigned to every average difference below two. Of the total 12 626 probe sets represented on the array, control probes and genes scored as absent (not detected) by the expression algorithm in GeneChip software (Affymetrix) or less than 100 in all 13 samples were excluded from the analysis because of low confidence of scarcely expressed genes, and 5668 probe sets were left.

Selection of differentially expressed genes

For the selection of differentially expressed genes by 1p status, we used prediction value (P-value) which reflects the difference between two groups, given by (1-2)/( 1+ 2) when (1, 1) and (2, 2) denote the means and standard deviations of the log of the expression level of gene for the sample in group 1 and group 2, respectively (Golub et al., 1999). Pre-filtering was applied to select probe sets whose maximum and minimum average difference among 11 tumor samples differed by more than 100, and had more than two-fold difference. For the remaining 3875 probe sets, the prediction values were calculated. We also used Mann-Whitney test, which measures whether the distribution of gene expression level between two groups is overlapped.

Clustering analysis

The expression patterns of samples were statistically analysed using GeneSpring 4.0 software (Silicon Genetics, Redwood City, CA, USA). Average differences were converted into logarithm, and hierarchical clustering was carried out using Pearson correlation coefficient of 0.8 (Eisen et al., 1998).

Semi-quantitative RT-PCR

Semi-quantitative RT-PCR was performed using 13 samples used for GeneChip analysis and additional nine oligodendroglial tumors. Of the nine additional cases, four cases had combined 1p and 19q LOH, while five cases had neither genetic alteration. Two of the five additional cases without 1pLOH had TP53 mutation. cDNA was synthesized with oligo-dT primer from 2 g total RNA, using SuperScript Preamplification System (Life Technologies, Inc.). The concentration of the cDNA was equalized using the GAPDH gene expression as a control. PCR was then performed with 2 l of cDNA for 31-37 cycles, consisted of 30 s of denaturing at 94°C, 30 s of annealing at 63-70°C and 1 min of extension at 72°C. The primer sets used are listed in Table 4. PCR products were separated by electrophoresis on 1.5% agarose gels and were visualized with ethidium bromide staining. Numbers of PCR cycles were optimized to ensure product intensity within the linear phase of amplification. For each primer set, the amplicon was sequenced after subcloned into pGEM-T Easy vector (Promega, Madison, WI, USA) to confirm that the correct target gene was amplified.

Identification of gene location

Chromosomal loci of the genes were identified using the locus information from the web sites of GenBank, UniGene and LocusLink on NCBI, by referring to the corresponding GenBank accession number of each probe set.

Detailed chromosomal locations of 950 genes mapped on 1p were obtained from the web site of Map Viewer (Homo sapiens build 26) on NCBI, in which the gene locations are shown by distances from the telomere of the short arm. These 950 genes were matched to the probe sets on Human U95A array by referring to the LocusLink ID, UniGene ID and GenBank accession number, which identified 502 probe sets represented on U95A array. Genes with expressions (average difference) scored as absent or less than 100 in average of 11 tumors were excluded because of low confidence in evaluating genes with low expression. For the 158 genes remaining, we calculated relative expressions by dividing the mean expressions in tumors with 1pLOH by those in tumors without 1pLOH, which were then arranged on the chromosome map.

Acknowledgements

We thank Drs Yoichi Nakazato, Takanori Hirose, Nobuaki Funada and Junko Hirato for reviewing the histology. We also thank Xijin Ge and Makoto Kano for their help in statistical analysis and Hiroko Meguro for her technical assistance. This work was supported by Grants-in-Aid from the Ministry of Education, Culture, Sports, Science and Technology, Japan (No. 11671357 and No. 13470184). This study was carried out as a part of The Technology Development for Analysis of Protein Expression and Interaction in Bioconsortia on R&D of New Industrial Science and Technology Frontiers which was performed by Industrial Science, Technology and Environmental Policy Bureau, Ministry of Economy, Trade and Industry, and entrusted by New Energy Development Organization (NEDO).

Bigner SH, Matthews MR, Rasheed BK, Wiltshire RN, Friedman HS, Friedman AH, Stenzel TT, Dawes DM, McLendon RE, Bigner DD. (1999). Am. J. Pathol., 155: 375-386. MEDLINE

Cairncross JG, Macdonald DR. (1988). Ann. Neurol., 23: 360-364. MEDLINE

Cairncross JG, Ueki K, Zlatescu MC, Lisle DK, Finkelstein DM, Hammond RR, Silver JS, Stark PC, Macdonald DR, Ino Y, Ramsay DA, Louis DN. (1998). J. Natl. Cancer Inst., 90: 1473-1479. MEDLINE

Caron H, van Schaik B, van der Mee M, Baas F, Riggins G, van Sluis P, Hermus MC, van Asperen R, Boon K, Voute PA, Heisterkamp S, van Kampen A, Versteeg R. (2001). Science, 291: 1289-1292. Article MEDLINE

Eisen MB, Spellman PT, Brown PO, Botstein D. (1998). Proc. Natl. Acad. Sci. USA, 95: 14863-14868. Article MEDLINE

Fero ML, Randel E, Gurley KE, Roberts JM, Kemp CJ. (1998). Nature, 396: 177-180. Article MEDLINE

Fults D, Brockmeyer D, Tullous MW, Pedone CA, Cawthon RM. (1992). Cancer Res., 52: 674-679. MEDLINE

Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri MA, Bloomfield CD, Lander ES. (1999). Science, 286: 531-537. Article MEDLINE

Gutmann DH, Loehr A, Zhang Y, Kim J, Henkemeyer M, Cashen A. (1999). Oncogene, 18: 4450-4459. MEDLINE

Huang H, Colella S, Kurrer M, Yonekawa Y, Kleihues P, Ohgaki H. (2000). Cancer Res., 60: 6868-6874. MEDLINE

Husemann K, Wolter M, Buschges R, Bostrom J, Sabel M, Reifenberger G. (1999). J. Neuropathol. Exp. Neurol., 58: 1041-1050. MEDLINE

Ino Y, Betensky RA, Zlatescu MC, Sasaki H, Macdonald DR, Stemmer-Rachamimov AO, Ramsay DA, Cairncross JG, Louis DN. (2001). Clin. Cancer Res., 7: 839-845. MEDLINE

Ishii M, Hashimoto S, Tsutsumi S, Wada Y, Matsushima K, Kodama T, Aburatani H. (2000). Genomics, 68: 136-143. Article MEDLINE

Kenwrick S, Watkins A, de Angelis E. (2000). Hum. Mol. Genet., 9: 879-886. MEDLINE

Kim JG, Armstrong RC, v Agoston D, Robinsky A, Wiese C, Nagle J, Hudson LD. (1997). J. Neurosci. Res., 50: 272-290. Article MEDLINE

Kleihues P, Cavenee WK. ed (2000). Pathology and Genetics of Tumours of the Nervous System. IARC Press: Lyon.

Lockhart DJ, Dong H, Byrne MC, Follettie MT, Gallo MV, Chee MS, Mittmann M, Wang C, Kobayashi M, Horton H, Brown EL. (1996). Nat. Biotechnol., 14: 1675-1680. MEDLINE

Mai M, Huang H, Reed C, Qian C, Smith JS, Alderete B, Jenkins R, Smith DI, Liu W. (1998). Genomics, 51: 359-363. Article MEDLINE

Nigro JM, Takahashi MA, Ginzinger DG, Law M, Passe S, Jenkins RB, Aldape K. (2001). Am. J. Pathol., 158: 1253-1262. MEDLINE

Nutt CL, Noble M, Chambers AF, Cairncross JG. (2000). Cancer Res., 60: 4812-4818. MEDLINE

Ohira M, Kageyama H, Mihara M, Furuta S, Machida T, Shishikura T, Takayasu H, Islam A, Nakamura Y, Takahashi M, Tomioka N, Sakiyama S, Kaneko Y, Toyoda A, Hattori M, Sakaki Y, Ohki M, Horii A, Soeda E, Inazawa J, Seki N, Kuma H, Nozawa I, Nakagawara A. (2000). Oncogene, 19: 4302-4307. MEDLINE

Pollack JR, Perou CM, Alizadeh AA, Eisen MB, Pergamenschikov A, Williams CF, Jeffrey SS, Botstein D, Brown PO. (1999). Nat. Genet., 23: 41-46. Article MEDLINE

Schwab M, Praml C, Amler LC. (1996). Genes Chromosomes Cancer, 16: 211-229. MEDLINE

Smith JS, Alderete B, Minn Y, Borell TJ, Perry A, Mohapatra G, Hosek SM, Kimmel D, O'Fallon J, Yates A, Feuerstein BG, Burger PC, Scheithauer BW, Jenkins RB. (1999). Oncogene, 18: 4144-4152. MEDLINE

Tanaka S, Kamitani H, Amin MR, Watanabe T, Oka H, Fujii K, Nagashima T, Hori T. (2000). J. Neurooncol., 46: 157-171. MEDLINE

Ueki K, Ono Y, Henson JW, Efird JT, von Deimling A, Louis DN. (1996). Cancer Res., 56: 150-153. MEDLINE

Ueki K, Nishikawa R, Nakazato Y, Hirose T, Hirato J, Funada N, Fujimaki T, Hojo S, Kubo O, Ide T, Usui M, Ochiai C, Ito S, Takahashi H, Mukasa A, Asai A, Kirino T. (2002). Clin. Cancer Res., 8: 196-201. MEDLINE

Watson MA, Perry A, Budhjara V, Hicks C, Shannon WD, Rich KM. (2001). Cancer Res., 61: 1825-1829. MEDLINE

Figure 1 Hierarchical clustering of 11 oligodendroglial tumors and two normal brain samples using the 209 genes selected by P-value. Each column represents a gene and each row represents a sample. Red indicates increased expression, and blue indicates decreased gene expression. Expression of each gene is normalized to its median in this figure. The dendrogram indicates the degree of similarity between their expression profiles. Normal brain samples were clustered into the same group with the tumors with 1pLOH. LOH: loss of heterozygosity

Figure 2 Results of semi-quantitative RT-PCR. Eleven tumors (# 1-11) used in the GeneChip experiments (printed in boldface), nine additional tumors (# 12-20), and two normal brain tissues (# 21, 22) were analysed. The upper six genes and the middle three genes showed higher and lower expression in 1pLOH tumors in the GeneChip experiment, respectively. GAPDH was used as a control

Figure 3 The relative expressions of the genes on chromosome 1p. Horizontal-axis represents the location of the examined 158 genes demonstrated from telomere (left) to centromere (right). Vertical-axis represents relative expressions with error bar, dividing the mean expressions (average difference) of tumors with 1pLOH by those of tumors without 1pLOH. The value beyond 1.7 is not shown in this figure

Table 1 Summary of oligodendroglial tumors used in GeneChip experiments

Table 2 List of genes showing higher expression in tumors with 1pLOH

Table 3 List of genes showing lower expression in tumors with 1pLOH

Table 4 Primer pairs used in RT-PCR