Adenovirus-mediated overexpression of p15INK4B inhibits human glioma cell growth, induces replicative senescence, and inhibits telomerase activity similarly to p16INK4A.

The genes encoding the cyclin-dependent kinase inhibitors p16INK4A (CDKN2A) and p15INK4B (CDKN2B) are frequently homozygously deleted in a variety of tumor cell lines and primary tumors, including glioblastomas in which 40-50% of primary tumors display homozygous deletions of these two loci. Although the role of p16 as a tumor suppressor has been well documented, it has remained less well studied whether p15 plays a similar growth-suppressing role. Here, we have used replication-defective recombinant adenoviruses to compare the effects of expressing wild-type p16 and p15 in glioma cell lines. After infection, high levels of p16 and p15 were observed in two human glioma cell lines (U251 MG and U373 MG). Both inhibitors were found in complex with CDK4 and CDK6. Expression of p16 and p15 had indistinguishable effects on U251 MG, which has homozygous deletion of CDKN2A and CDKN2B, but a wild-type retinoblastoma (RB) gene. Cells were growth-arrested, showed no increased apoptosis, and displayed a markedly altered cellular morphology and repression of telomerase activity. Transduced cells became enlarged and flattened and expressed senescence-associated beta-galactosidase, thus fulfilling criteria for replicative senescence. In contrast, the growth and morphology of U373 MG, which expresses p16 and p15 endogenously, but undetectable levels of RB protein, were not affected by exogenous overexpression of either inhibitor. Thus, we conclude that overexpression of p15 has a similar ability to inhibit cell proliferation, to cause replicative senescence, and to inhibit telomerase activity as p16 in glioma cells with an intact RB protein pathway.

Cryopreservation of viable human glioblastoma xenografts.

Human tumour xenografts maintained in nude mice are a valuable research tool. The passaging and maintenance of human tumour xenografts in immune-deficient animals are expensive and labour-intensive. This study presents a protocol that permits long-term cryopreservation of viable glioblastoma xenograft tissue pieces in liquid nitrogen. Twenty different human glioblastoma xenografts that have been successfully transplanted and repeatedly passaged in nude mice were cryopreserved to validate the method. Different passages were cryopreserved for up to 40 months. On retransplantation of the tumours, all cases except one grew successfully. In order to ensure that the individual tumours did grow (not induced mouse tumours) restriction fragment length polymorphism analysis with variable number of tandem repeats (VNTR) probes was carried out. The method permits the long-term storage of viable, retransplantable glioblastoma xenografts. It minimizes animal use for the maintenance of xenografts and permits cryo-back-up of valuable tumours, thus markedly reducing the cost and increasing the accessibility of human tumour xenografts as a research tool in biology and genetics.

Deregulation of the p14ARF/MDM2/p53 pathway is a prerequisite for human astrocytic gliomas with G1-S transition control gene abnormalities.

Deregulation of G1-S transition control in cell cycle is one of the important mechanisms in the development of human tumors including astrocytic gliomas. We have previously reported that approximately two-thirds of glioblastomas (GBs) had abnormalities of G1-S transition control either by mutation/homozygous deletion of RB1 or CDKN2A p16INK4A), or amplification of CDK4 (K. Ichimura et al., Oncogene, 13: 1065-1072, 1996). However, abnormalities of G1-S transition control genes may induce p53-dependent apoptosis in cells. Recent investigations suggest that p14ARF is induced in response to abnormal cell cycle entry and results in p53 accumulation by inhibiting MDM2-mediated transactivational silencing and degradation of p53. To investigate the roles of the G1-S transition control system and the p14ARF/MDM2/p53 pathway in the development of astrocytic gliomas, we examined abnormalities of genes involved in these regulatory pathways in a total of 190 primary human astrocytic gliomas of different malignancy grades [136 GBs, 39 anaplastic astrocytomas (AAs) and 15 astrocytomas (As)]. Sixty-seven percent of GBs (91/136) and 21% of AAs (8/39) had abnormalities of the G1-S control system either by mutation/homozygous deletion of RB1, CDKN2A or CDKN2B, or amplification of CDK4. Seventy-six percent of GBs (103 of 136), 72% of AAs (28 of 39), and 67% of As (10 of 15) had deregulated p53 pathway either by mutation of TP53, amplification of MDM2, or homozygous deletion/mutation of p14ARF. When all of the data were combined and compared, 96% of GBs (87 of 91) and 88% of AAs (7 of 8) with abnormal G1-S transition control also had deregulated p53 pathway. Thus, we demonstrate that deregulation of the G1-S transition control system was almost always accompanied by inactivation of the p53 pathway, clearly illustrating the cooperative roles of these two systems in the development/progression of primary human astrocytic gliomas.

The complexity of the 7p12 amplicon in human astrocytic gliomas: detailed mapping of 246 tumors.

Approximately 30%-35% of human glioblastomas have epidermal growth factor receptor (EGFR) gene amplification. Amplicons containing the EGFR gene frequently include several unidentified adjacent genes. Amplification of adjacent genes in the absence of EGFR amplification has been documented. To define the region in detail, we produced a YAC contig map, determining the orientation of the EGFR gene and the general order of 11 STS and EST markers. Seventy-six tumors with amplification of the region were found in a series of 246 human astrocytic gliomas. The amplicons showed amplification of contiguous or noncontiguous loci both telomeric and centromeric to the EGFR gene. Six percent (12/190) of the glioblastomas and 1 anaplastic astrocytomas had amplicons excluding the EGFR gene. These amplicons commonly contain amplified loci telomeric to the EGFR gene. Some of the amplified loci were found to be consistently overexpressed. The findings suggest that there may be other gene target(s) for amplification in the 7p12 region in addition to EGFR.

Mutational profile of the PTEN gene in primary human astrocytic tumors and cultivated xenografts.

The genetic abnormality most frequently identified in glioblastomas is loss of alleles on chromosome 10. We have performed a comprehensive study of the PTEN tumor suppressor gene on 10q23, including loss of heterozygosity (LOH) analysis, multiplex PCR, mutation analysis, and reverse transcription PCR (RT-PCR). In total, 151 glioblastomas, 41 anaplastic astrocytomas, 15 astrocytomas, and 13 glioma cell lines were analyzed as well as 23 xenografts derived from primary glioblastomas, which allows a comparison of the PTEN gene status in primary tumors versus xenografts. Homozygous deletions were found in 7% of the glioblastomas and 40% showed mutation of a single retained allele. This mutation frequency is higher than reported previously. The large number of mutations identified allows the presentation of a mutational profile along the coding sequence. The majority of mutations appear to affect conserved residues or structurally conserved regions. PTEN alterations were selected for in xenografts, and there is evidence that they may even facilitate establishment of xenografts. No alterations were found in astrocytomas and only 5% of anaplastic astrocytomas had mutations. Thus, loss of wild type PTEN represents one of the major abnormalities associated with astrocytic tumor progression to glioblastoma and provides a strong selective growth advantage when cultivating glioblastoma tissue in xenografts. No correlation with EGFR amplification was evident.

Acquired rearrangement of an amplified epidermal growth factor receptor (EGFR) gene in a human glioblastoma xenograft.

Amplification of the epidermal growth factor receptor (EGFR) occurs in about 40% of human glioblastomas. In half of these cases, rearrangements of the amplified gene result in aberrant transcripts and proteins. The most frequent rearrangement affects the external domain of the receptor and results in nonbinding of ligand and constitutive activity. Less frequent rearrangements involve changes resulting in the loss of cytoplasmic amino acid sequences necessary for downregulation of the receptor following ligand binding. Here we report the development and selection for a rearranged amplified EGF receptor, which lacks cytoplasmic amino acid sequences in a human glioblastoma xenograft. An identical aberration has previously been reported in glioblastoma tissue. The patient tumor material, as well as the first passages of the xenograft showed amplification of the EGFR gene, but no evidence of gene rearrangement or an aberrant transcript. Interphase FISH data show the amplified gene on double minutes. Between passages 3 and 16, the growth rate of the xenograft almost doubled, the rearranged amplicon became dominant, as did the aberrant transcript, indicating selection under these conditions.

Distinct patterns of deletion on 10p and 10q suggest involvement of multiple tumor suppressor genes in the development of astrocytic gliomas of different malignancy grades.

Deletions of chromosome 10 are the most frequent genetic abnormality in glioblastomas. Several commonly deleted regions have been proposed; however, they are not coincident. We have deletion mapped chromosome 10 in 198 astrocytic gliomas using 53 microsatellite markers. Two commonly deleted regions on 10p were identified, one of which lies between D10S594 and D10S559 and the other between D10S1713 and D10S189. Most of 10q deletions were large and included a region distal to D10S554. Four glioblastomas of 122 had patterns suggestive of homozygous deletions at D10S541, a locus close and distally located to the PTEN/MMAC1 gene. Losses of alleles were found not only in glioblastomas (93%) but also in anaplastic astrocytomas (66%) and in astrocytomas (35%). Most glioblastomas lost one entire chromosome 10, while astrocytomas preferentially lost only 10p. The data suggest that a number of tumor suppressor genes on chromosome 10, in addition to PTEN/MMAC1, may be associated with astrocytic glioma development.

Infrequent methylation of CDKN2A(MTS1/p16) and rare mutation of both CDKN2A and CDKN2B(MTS2/p15) in primary astrocytic tumours.

In a series of 46 glioblastomas, 16 anaplastic astrocytomas and eight astrocytomas, all tumours retaining one or both alleles of CDKN2A (48 tumours) and CDKN2B (49 tumours) were subjected to sequence analysis (entire coding region and splice acceptor and donor sites). One glioblastoma with hemizygous deletion of CDKN2A showed a missense mutation in exon 2 (codon 83) that would result in the substitution of tyrosine for histidine in the protein. None of the tumours retaining alleles of CDKN2B showed mutations of this gene. Glioblastomas with retention of both alleles of CDKN2A (14 tumours) and CDKN2B (16 tumours) expressed transcripts for these genes. In contrast, 7/13 glioblastomas with hemizygous deletions of CDKN2A and 8/11 glioblastomas with hemizygous deletions of CDKN2B showed no or weak expression. Anaplastic astrocytomas and astrocytomas showed a considerable variation in the expression of both genes, regardless of whether they retained one or two copies of the genes. The methylation status of the 5' CpG island of the CDKN2A gene was studied in all 15 tumours retaining only one allele of CDKN2A as well as in the six tumours showing no significant expression of transcript despite their retaining both CDKN2A alleles. Three tumours (one of each malignancy grade studied) were found to have partially methylated the 5' CpG island of CDKN2A. It appears that in human astrocytic gliomas point mutations of the CDKN2A and CDKN2B genes are uncommon and hypermethylation of the 5' CpG region of CDKN2A does not appear to be a major mechanism for inhibiting transcription of this gene.

Human glioblastomas with no alterations of the CDKN2A (p16INK4A, MTS1) and CDK4 genes have frequent mutations of the retinoblastoma gene.

A series of 195 human gliomas were studied as to the status of their CDKN2A, CDK4 and RB1 genes. Among 120 glioblastomas, 40% had no wild-type CDKN2A gene, 12% amplified the CDK4 gene, and 14% had no wild-type RBI gene. With two exceptions, each tumour had only one of these abnormalities. Thus the majority of the glioblastomas (64%) had distinct genetic aberrations which would obviously disrupt the control of transition from G1 to the S-phase of the cell cycle. A further 30% had loss of one allele of the CDKN2A and/or RBI genes. Only seven (6%) glioblastomas had no abnormalities of these genes. Anaplastic astrocytomas showed similar changes to the glioblastomas but at lower frequencies-34% showing no aberrations of the genes analysed. The astrocytomas showed solely loss of one allele of the RBI gene in 28% of tumours, with retention of one wild-type copy. In the glioblastomas with no alterations of CDKN2A, CDK4 or RB1, several other genes (CCND1, CCND2, CCND3, CDK6, E2F, CDK7, MYC and MYCN) whose products take part in cell cycle regulation were examined. No abnormalities were detected. Thus some aberration of the CDKN2A, CDK4 and RB1 genes appears to be almost obligatory in glioblastomas.