Establishment of the DU.528 human lymphohemopoietic stem cell line.

We have established the DU.528 cell line from the pretreatment leukemia cells of a patient who underwent a T lymphoblastic-to-promyelocytic phenotype conversion during treatment with the adenosine deaminase inhibitor, deoxycoformycin. The cell line and clones obtained from it by limiting dilution have the same karyotype previously found in the patient's pretreatment T lymphoblasts and post-deoxycoformycin treatment promyelocytes. DU.528 cells in continuous culture for greater than 2 yr display a predominant undifferentiated T lymphoblastoid phenotype. These cells spontaneously generate progeny of at least three lineages, T lymphoid, granulocytic/monocytic, and erythroid. The surface marker most consistently expressed by DU.528 cells in the undifferentiated state is the 3A1 antigen, which has been found on prothymocytes in the embryonic thymus. Some undifferentiated DU.528 cells also expressed the IL-2 receptor, but no other T cell differentiation antigens. Exposure of DU.528 cells to a variety of agents induced myeloid maturation; adenosine and deoxyadenosine, in the presence of deoxycoformycin, induced expression of myeloid differentiation antigens. Our results suggest that DU.528 is a lymphohematopoietic stem cell line and support the hypothesis that differentiation of pluripotent stem cells may be altered by genetic deficiency of adenosine deaminase. DU.528 cells may provide a useful model for examining factors that regulate stem cell proliferation and differentiation.

Identification of an amplified, highly expressed gene in a human glioma.

A gene, termed gli, was identified that is amplified more than 50-fold in a malignant glioma. The gene is expressed at high levels in the original tumor and its derived cell line and is located at chromosome 12 position (q13 to q14.3). The gli gene is a member of a select group of cellular genes that are genetically altered in primary human tumors.

PTEN, a putative protein tyrosine phosphatase gene mutated in human brain, breast, and prostate cancer.

Mapping of homozygous deletions on human chromosome 10q23 has led to the isolation of a candidate tumor suppressor gene, PTEN, that appears to be mutated at considerable frequency in human cancers. In preliminary screens, mutations of PTEN were detected in 31% (13/42) of glioblastoma cell lines and xenografts, 100% (4/4) of prostate cancer cell lines, 6% (4/65) of breast cancer cell lines and xenografts, and 17% (3/18) of primary glioblastomas. The predicted PTEN product has a protein tyrosine phosphatase domain and extensive homology to tensin, a protein that interacts with actin filaments at focal adhesions. These homologies suggest that PTEN may suppress tumor cell growth by antagonizing protein tyrosine kinases and may regulate tumor cell invasion and metastasis through interactions at focal adhesions.

The GLI-Kruppel family of human genes.

Previous characterization of GLI, a gene found to be amplified and expressed in a subset of human brain tumors, revealed the presence of five tandem zinc fingers related to those of Krüppel (Kr), a Drosophila segmentation gene of the gap class. We have used the GLI cDNA as a molecular probe to isolate related sequences from the human genome. Partial characterization of six related loci, including sequence determination, expression studies, and chromosome localization, revealed that each locus could encode a separate finger protein. The predicted proteins all had similar H-C links, i.e., a conserved stretch of 9 amino acids connecting the C-terminal histidine of one finger to the N-terminal cysteine of the next. On the basis of amino acid sequence and intron-exon organization, the genes could be placed into one of two subgroups: the GLI subgroup (with the consensus finger amino acid sequence [Y/F]XCX3GCX3[F/Y]X5LX2HX3-4H[T/S]GEKP) or the Kr subgroup (with the consensus finger amino acid sequence [Y/F]XCX2CX3FX5LX2HXRXHTGEKP). Unlike GLI or Kr, most of the newly isolated genes were expressed in many adult tissues. The predicted proteins probably control the expression of other genes and, by analogy with Kr and GLI, may be important in human development, tissue-specific differentiation, or neoplasia.

Demonstration of complex antigenic heterogeneity in a human glioma cell line and eight derived clones by specific monoclonal antibodies.

We have investigated the antigenic heterogeneity of human glioma cells and its correlation with other parameters of tumor cell heterogeneity (karyotype, 2':3' cyclic nucleotide 3'-phosphohydrolase expression, in vitro morphology) using the established human glioma cell line D-54 MG and eight single-cell-derived clones. The panel of antibodies used included 3 previously described heterologous sera raised against human gliomas and lamb oligodendroglia and 10 monoclonal antibodies with demonstrated reactivity for tumors of neuroectodermal origin, human fetal tissue, or human Thy-1. Antigen expression was determined by cell surface radioimmunoassay and peroxidase-antiperoxidase immunohistology. The use of a monoclonal antibody panel composed of ten reagents of varied specificity resulted in the demonstration of highly variable and complex antigenic patterns on the cell surfaces of cloned subpopulations of the human glioma cell line D-54 MG. Only one antigen, human Thy-1, was present on the parent line and all clones; the remaining nine antigens exhibited a distribution unrelated to other predictive parameters of genotypic or phenotypic heterogeneity such as karyotype, 2':3' cyclic nucleotide 3'-phosphohydrolase expression, or in vitro morphology. With the exception of clones 3 and 4, which shared a common antigen profile but exhibited distinctly different in vitro morphological patterns, the detected antigenic profile of each clone was distinct, with the proportion of expressed antigens ranging from 2 of 10 (clone 2) to 10 of 10 (clone 1). The demonstration of distinct, selectively maintained cell subpopulations within a human glioma cell line has direct implications for immunotherapeutic designs.

Comparison of intravenous versus intracarotid therapy with 1,3-bis(2-chloroethyl)-1-nitrosourea in a rat brain tumor model.

Currently numerous clinical trials are in progress utilizing intracarotid (i.c.) 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU) for the treatment of malignant gliomas based upon the proposed focal nature of these tumors and the assumption that the i.c. route delivers higher levels of drug to the tumor. To date, however, increased efficacy in an animal model has not been clearly demonstrated for the i.c. delivery of BCNU. We have evaluated the dose-response curve for the i.v. and i.c. administration of BCNU in a commonly utilized experimental brain tumor model, the 9L rat gliosarcoma. An initial toxicity trial utilizing the i.p. 10% lethal dose (LD10) of BCNU by the i.v. and i.c. routes failed to demonstrate any significance in toxicity between the two routes. Tumor-bearing animals were then treated on Day 15-16 after tumor inoculations with 1, 10, 25, 50, 75, and 100% of the LD10 dose by either the i.v. or the i.c. route. Both i.v. and i.c. BCNU gave maximum survival increases at 75-100% LD10 doses, and there was no therapeutic advantage seen from i.c. delivery. However, at 50% of the LD10 dose (6.65 mg/kg), triplicate experiments demonstrated that the i.c. but not the i.v. dose maintained maximum efficacy equivalent to 100% of the LD10 given either i.v. or i.c. When the dose was reduced to 25% of the LD10 dose (3.33 mg/kg), two of three experiments showed efficacy of the i.c. delivery of this lower drug dosage to be equivalent to 100% of the LD10 given i.v. or i.c. The i.v. dosage resulted in a significant reduction in survival in all three trials. At 10% of the LD10 dose (1.30 mg/kg), neither the i.v. nor the i.c. administration retained equivalent efficacy to 100% of the LD10. However, in one of two trials, the i.c. groups had statistically better survival than controls, while in neither experiment was any advantage over controls seen in the i.v. treated groups. At 1% of the LD10 dose, neither the i.v. nor the i.c. route demonstrated any therapeutic efficacy. From our data, the advantage of the i.c. delivery of BCNU in the intracranial 9L rat gliosarcoma appears to be in the fact that significantly lower dosages than those given i.v. may be utilized to achieve equivalent survival with potentially less systemic toxicity.

Effect of epidermal growth factor on glioma cell growth, migration, and invasion in vitro.

Effects of epidermal growth factor (EGF) and an antibody (Ab-528) reactive against the binding site for EGF on human EGF receptors were studied on multicellular tumor spheroids obtained from three human glioma cell lines with high (D-37 MG), medium (D-247 MG), and low (D-263 MG) levels of EGF receptor expression. The D-247 MG and D-263 MG spheroids grew slowly or not at all in the absence of EGF, while in the presence of EGF they were growth stimulated. Tumor cell migration, as measured by the spread of cells from spheroids on a plastic substratum, was increased by the addition of EGF for all three cell lines. Stimulation of migration could be blocked by a subsequent addition of Ab-528 to the medium at a concentration of 50 micrograms/ml. Invasiveness of glioma cell spheroids into fetal rat brain aggregates was related to EGF receptor expression; the two lines with medium to high receptor expression (D-247 MG and D-37 MG) were invasive, while the line with low EGF receptor expression (D-263 MG) was noninvasive, as assessed by an in vitro coculture assay. In the D-247 MG cell line, morphometry revealed EGF-enhanced invasiveness of the tumor cells. The addition of the Ab-528 to EGF-treated cocultures reduced invasion in both D-247 MG and D-37 MG cell lines. Antibody Ab-528 alone did not affect glioma cell growth or migration but did inhibit invasiveness. The present study suggests that, in brain tumors with an increased number of normal-sized Mr 170,000 EGF receptors, EGF or an EGF-like ligand such as transforming growth factor-alpha may selectively facilitate expansive tumor growth and tumor cell invasion. This effect may in part be blocked or retarded by specific antibodies to the EGF receptor.

Amplification of the c-myc gene in human medulloblastoma cell lines and xenografts.

Cultured cell lines and xenografts derived from 7 human medulloblastomas were evaluated for amplification of the c-myc, N-myc, epidermal growth factor receptor, and gli genes by Southern blot analysis. Karyotypes of the original biopsies and early passaged cells demonstrated double minute chromosomes in 4 of the 7 cases. All 7 samples (3 cell lines and 4 xenografts) from the 4 tumors with double minute chromosomes contained amplification of the c-myc gene. Cell lines and xenografts derived from the 3 biopsies without double minute chromosomes failed to demonstrate amplification of the 4 genes which were tested, but a rearrangement of the c-myc gene occurred in 1 of the 3 tumors. These observations demonstrate that the c-myc gene is often amplified and/or rearranged in human medulloblastomas and suggest that amplification of this gene provides a growth advantage for medulloblastoma cells in vitro and in vivo.

Infrequent p53 gene mutations in medulloblastomas.

Cytogenetic and molecular studies of medulloblastomas have demonstrated frequent loss of sequences from the short arm of chromosome 17, possibly implicating loss or inactivation of the p53 tumor suppressor gene. We amplified exons 5 through 8 of the p53 gene by the polymerase chain reaction technique. These segments, which encompass the regions usually mutated in human tumors, were sequenced to search for p53 mutations in 12 medulloblastoma tumors, 8 xenografts, and 3 permanent cell lines. Mutation of the p53 gene was found in only 1 of 3 cell lines tested and in none of the xenografts or primary tumors studied. Our results suggest that p53 is mutated in an unusual way or that a second tumor suppressor gene on the short arm of chromosome 17 is involved in the pathogenesis of medulloblastoma.

Alterations of the TP53 gene in human gliomas.

Glial tumors of all grades and histological types from 72 adults and 48 children were analyzed for mutations of the TP53 gene, loss of heterozygosity (LOH) for 17p, and accumulation of TP53 protein to determine whether the incidence and type of TP53 alterations differ among tumors of different histological type and between tumors from adults and children. These tumors were also evaluated for LOH for chromosome 10 and for amplification of the epidermal growth factor receptor, C-MYC, N-MYC, GLI, platelet-derived growth factor receptor-alpha, and murine double minute 2 genes to determine the patterns of molecular alterations involved in the progression of these neoplasms. Seventeen of the 120 tumors contained mutations of the TP53 gene. One of the tumors with TP53 gene mutation was from one of the 48 patients less than 18 years of age. Twelve of the 17 tumors with mutations occurred among the 27 patients in the 18-45-year age group, while 4 tumors with mutations were among the 45 patients more than 45 years old. There was also an increased incidence of TP53 mutation in patients with anaplastic astrocytoma histology. However, no significant association between presence of TP53 mutation and patient survival was observed. These studies demonstrate that TP53 gene mutations are a common mechanism for glial cell neoplasms in the 18-45-year age group but are unrelated to progression and advanced histological grade. LOH for chromosome 10 and gene amplification, however, occurring in 82 and 40%, respectively, of glioblastoma multiforme, whether seen alone or along with TP53 gene alterations, are related to advanced histological grade of the tumor. In childhood gliomas, in contrast, TP53 gene alterations, LOH for 17p and 10q, and gene amplification are uncommon in tumors of all grades, suggesting that presently unknown mechanisms are responsible for the genesis and progression of these tumors.

Deletion of p16 and p15 genes in brain tumors.

We have used molecular genetic methods to examine the status of cell cycle-inhibitory genes in human brain tumors. We found that p16 and a neighboring gene, p15, were often homozygously deleted in glioblastoma multiformes but not in medulloblastomas or ependymomas. The deletions occurred in both primary tumors and their derived xenografts, but no intragenic mutations in either of the two genes were found. The p15 gene was expressed in a more widespread pattern in normal tissues than p16, but the products of both genes had similar capacities to bind to cyclin D-dependent kinases 4 and 6. These data suggest that the target of deletion in glioblastoma multiforme includes both p15 and p16 genes. The reason that homozygous deletions, rather than intragenic mutations, are so common in these tumors may be that deletion is a more efficient mechanism for simultaneous inactivation of both genes.

Amplification and expression of the epidermal growth factor receptor gene in human glioma xenografts.

Xenografts from eight malignant human gliomas were established in athymic mice and were used to study amplification and expression of the epidermal growth factor receptor (EGFR) gene. Tissue identity between biopsy and xenografts was confirmed by karyotypic profiles, which showed that each glioma xenograft retained structural abnormalities, including double minute chromosomes, present in the parent glioma. EGFR gene amplification was found in six of the eight glioma biopsies and their corresponding xenografts. Expression of the EGFR gene was measured by Scatchard analysis, affinity reactions, immunoprecipitations, Western immunoblots, and immunocytochemistry; significant expression of the EGFR gene was only detectable in xenografts with EGFR gene amplification. Moreover, five of the six xenografts with EGFR gene amplification demonstrated structural alterations of the EGFR gene, which was associated with low-molecular-weight EGFR proteins. These xenografts represent an excellent tissue source and in vivo model system for characterizing the epidermal growth factor receptor in malignant human gliomas.

Establishment of a melphalan-resistant rhabdomyosarcoma xenograft with cross-resistance to vincristine and enhanced sensitivity following buthionine sulfoximine-mediated glutathione depletion.

A melphalan-resistant human rhabdomyosarcoma xenograft, TE-671 MR, was established in athymic mice by serial melphalan treatment of the parent xenograft, TE-671, at the 10% lethal dosage (LD10); significant resistance was evident after ten passages of the tumor. TE-671 MR demonstrated a doubling time of 3.5 days and a latency period to 1000-mm3 tumors of 27.5 days. The glutathione level of TE-671 MR was 2.36 mumol/g tumor, wet weight, 2-fold higher than the parent line. The glutathione S-transferase activity of TE-671 MR was 117.8 mumol/min/mg protein, essentially unchanged from the parent line. Although TE-671 MR demonstrated cross-resistance to vincristine, dot blot analysis did not reveal an elevated expression of mdr1 mRNA in the resistant line. TE-671 MR demonstrated a 9.7-day growth delay following treatment with melphalan at the LD10 (compared to 20.9 days for the parent line). Treatment with L-buthionine-SR-sulfoximine (BSO) resulted in increased sensitivity to melphalan subsequently administered at 50% of the LD10 (melphalan alone, growth delays of 3.7 and 4.6 days in duplicate trials; melphalan plus BSO, growth delays of 7.2 and 9.8 days). Sensitivity to melphalan equal to that of the parent line TE-671 was not achieved, however. Treatment with BSO did not result in significantly enhanced sensitivity to subsequently administered vincristine (50% of the LD10) (vincristine alone, growth delays of 6.8 and 6.9 days in duplicate trials; vincristine plus BSO, growth delays of 10.9 and 7.5 days). These results suggest that generation of melphalan resistance may be associated with development of cross-resistance to vincristine; this resistance may be associated with (although not necessarily mediated by) glutathione elevation; this resistance may be partially overcome by BSO-mediated depletion of glutathione.

PTEN gene mutations are seen in high-grade but not in low-grade gliomas.

The PTEN gene, located on 10q23, has recently been implicated as a candidate tumor suppressor gene in brain, breast and prostate tumors. In the present study, 123 brain tumors, including various grades and histological types of gliomas occurring in children and adults, were analyzed for PTEN mutations by SSCP assay and sequencing. Mutations in the PTEN gene were found in 13 of 42 adult glioblastomas and 3 of 13 adult anaplastic astrocytomas, whereas none of the 21 low-grade adult gliomas or the 22 childhood gliomas of all grades showed mutations. The single medulloblastoma with a mutation was a recurrent tumor that also possessed a p53 mutation. High-grade adult gliomas with PTEN mutations included cases that also contained gene amplification or p53 gene mutations, as well as cases that did not contain either of these abnormalities. There was no obvious relationship between presence of PTEN mutation and survival; however, there was a tendency for PTEN mutations to occur in older age group patients. This analysis suggest that PTEN gene mutations are restricted to high-grade adult gliomas and that this abnormality is independent of the presence or absence of gene amplification or p53 gene mutation in these tumors.

Characterization of the epidermal growth factor receptor in human glioma cell lines and xenografts.

Both permanent cultured cell lines and athymic mouse xenografts were established from two human glioblastomas. Biopsies from D-245 MG and D-270 MG contained amplified and rearranged epidermal growth factor receptor (EGFR) genes. Although the gene amplification and rearrangement seen originally was maintained in the xenografts, cultured cell lines established from these biopsies lost the amplified rearranged genes in vitro. Analysis of these cell lines and 11 additional permanent human glioma cell lines with normal EGFR gene copy number showed from 2.7 x 10(3) to 4.1 x 10(5) high affinity EGFRs/cell by radioreceptor assay. The RNase A protection assay showed minimal differences in the quantity of EGFR mRNA among the 13 glioma lines, while the D-245 MG and D-270 MG xenografts expressed approximately 10-20 times as much EGFR mRNA as the corresponding cell lines. Immunoprecipitation of EGFR from these lines, including D-245 MG and D-270 MG, demonstrated only the intact Mr 170,000 Da form, while truncated Mr 145,000 Da and 100,000 Da EGFR proteins were immunoprecipitated from the D-270 MG and D-245 MG xenografts, respectively. These studies demonstrate that gliomas with amplification of the EGFR gene are capable of establishing in culture but that the amplified rearranged genes are not maintained. Possible explanations are that the abnormal genes are lost during serial passage or that the cells with amplified rearranged genes only represent a minor subpopulation of cells, which are unable to grow in culture. In either case, these observations suggest that high expression and structural abnormalities of EGFR proteins generated by amplification and rearrangement of the EGFR gene provide a growth advantage for gliomas in vivo but not in vitro.

Specific chromosomal abnormalities in malignant human gliomas.

Karyotypic analysis of 54 malignant human gliomas (5 anaplastic astrocytomas, 43 glioblastoma multiformes, 3 gliosarcomas, 2 giant cell glioblastomas, 1 anaplastic mixed glioma) has demonstrated that 12 tumors contained normal stemlines or only lacked one sex chromosome. The 42 tumors with abnormal karyotypes included 38 tumors which could be completely analyzed. Six of these 38 cases had near-triploid or near-tetraploid stemlines and 32 had near-diploid stemlines. Statistically significant numerical deviations in the near-diploid group were gains of chromosome 7 (26 of 32; P less than 0.001), and losses of chromosome 10 (19 of 32; P less than 0.001). Double minutes occurred in 18 of 32 near diploid tumors. The distribution of structural abnormalities was analyzed statistically by comparing the incidence of breakpoint in each chromosomal arm to the expected value based on chromosomal arm length. This analysis demonstrated that structural abnormalities of 9p and 19q were significant statistically (P less than 0.005 and P = 0.02, respectively). Although chromosome 1, 6p, the centromeric region of chromosome 11, 13q, and 15q were also frequently involved in structural abnormalities, the incidence of these breaks did not reach statistical significance. This demonstration of specific chromosomal abnormalities in near-diploid gliomas provides the basis for the investigation of genes which may be quantitatively or qualitatively altered in these neoplasms.

Activity of intrathecal 4-hydroperoxycyclophosphamide in a nude rat model of human neoplastic meningitis.

Neoplastic meningitis can result from leptomeningeal dissemination of a variety of cancers. We now report the development of animal models of human neoplastic meningitis and activity of intrathecal 4-hydroperoxycyclophosphamide (4-HC) against the human rhabdomyosarcoma cell line TE-671 and the human glioma cell line D-54 MG grown in the subarachnoid space of athymic rats. The injection of 5 x 10(5) TE-671 or D-54 MG cells resulted in leptomeningeal tumor growth from the base of the brain to the cauda equina. Daily weights and neurological examinations revealed progressive neurological deficits and weight loss, with death occurring between Days 21 and 27 for TE-671 and Days 14 and 26 for D-54 MG. 4-HC toxicity in non-tumor-bearing rats was assessed at dose levels of 2.0, 10.0, 15.0, and 20.0 mM, with clinical and histological evidence of neurotoxicity observed at the 2 highest dose levels. Intrathecal treatment with 4-HC on Day 8 following injection of TE-671 resulted in an increase in median survival of 20% (P = 0.04) at 1.0 mM 4-HC and 41% (P less than 0.001) at 2.5 mM 4-HC. Intrathecal treatment with 4-HC (2.5 mM) on Day 5 following injection of D-54 MG resulted in an increase in median survival of 23% (P = 0.009). These studies show the usefulness of the athymic rat model of human neoplastic meningitis and demonstrate the efficacy in vivo of intrathecally administered 4-HC against a human glioma and a human rhabdomyosarcoma cell line and the lack of toxicity at therapeutic levels of 4-HC in normal athymic rats.

Oncogene amplification in pediatric brain tumors.

Despite a considerable amount of information concerning chromosomal and molecular abnormalities found in gliomas in adults, relatively little is known regarding these abnormalities in pediatric brain tumors. We have analyzed DNA from 37 primary brain tumors and 4 tumor-derived cell lines for oncogene amplification. Probes utilized represent 11 known oncogenes (erbB1, gli, neu, myc, L-myc, N-myc, H-ras, K-ras, N-ras, sis, and src). Of 20 primary medulloblastomas studied, only one tumor was found to have erbB1 amplification. In contrast, of the 4 medulloblastoma cell lines studied, 1 had c-myc amplification, 1 had erbB1 amplification, and 1 had amplification of N-myc. Twelve glial brain tumors were analyzed, and only 1 case with amplification of the erbB1 oncogene was found. Other tumors studied include 1 meningioma, 2 ependymomas, 1 anaplastic ependymoma, and 1 cerebral primitive neuroectodermal tumor, none of which had oncogene amplification. These results suggest that oncogene amplification is relatively uncommon in primary medulloblastomas, but the frequency and diversity of oncogene amplification is greater in tumors that can be established as cell lines. The lower frequency of erbB1 amplification in glial brain tumors in children compared to adults is consistent with the generally lower grade of glial tumor histology seen in pediatric patients. However, the case with amplification of the erbB1 oncogene represented 1 of 2 cases of glioblastoma multiforme we studied, which suggests that pediatric glioblastoma multiforme may have a similar frequency of erbB1 oncogene amplification to glioblastomas seen in adults. Our results suggest that oncogene amplification is a relatively uncommon mechanism of oncogene activation in pediatric brain tumors, and they provide molecular evidence for heterogeneity in tumors classified as medulloblastomas.

Expression of O6-methylguanine-DNA methyltransferase in six human medulloblastoma cell lines.

Six well characterized human medulloblastoma cell lines (D283 Med, Daoy, D341 Med, D384 Med, D425 Med, and D458 Med) were examined for the expression of O6-methylguanine-DNA methyltransferase (MGMT) by activity and Western and Northern blot analysis. High levels of MGMT activity were present in D283 Med, Daoy, D341 Med, and D384 Med (1.36, 0.80, 1.68, and 1.62 pmol/mg of protein, respectively), but negligible MGMT activity was detected in D425 Med and D458 Med (0.06 and 0.05 pmol/mg of protein, respectively), which were derived separately at different times from the same patient. The presence of MGMT protein and its transcript was demonstrated in D283 Med, Daoy, D341 Med, and D384 Med, but both the protein and the mRNA were undetectable in D425 Med and D458 Med. Nevertheless, all six cell lines contained an apparently unaltered MGMT gene, as determined by Southern blot analysis. The absence of MGMT activity in D425 Med and D458 Med is likely due to the absence of the protein, resulting from a lack of transcription of the MGMT gene. The varying levels of expression of MGMT in medulloblastoma cells found in this study should provide a molecular basis for drug design and selection in chemotherapy of this tumor.

Methylator resistance mediated by mismatch repair deficiency in a glioblastoma multiforme xenograft.

A methylator-resistant human glioblastoma multiforme xenograft, D-245 MG (PR), in athymic nude mice was established by serially treating the parent xenograft D-245 MG with procarbazine. D-245 MG xenografts were sensitive to procarbazine, temozolomide, N-methyl-N-nitrosourea, 1,3-bis(2-chloroethyl)-1-nitrosourea, 9-aminocamptothecin, topotecan, CPT-11, cyclophosphamide, and busulfan. D-245 MG (PR) xenografts were resistant to procarbazine, temozolomide, N-methyl-N-nitrosourea, and busulfan, but they were sensitive to the other agents. Both D-245 MG and D-245 MG (PR) xenografts displayed no O6-alkylguanine-DNA alkyltransferase activity, and their levels of glutathione and glutathione-S-transferase were similar. D-245 MG xenografts expressed the human mismatch repair proteins hMSH2 and hMLH1, whereas D-245 MG (PR) expressed hMLH1 but not hMSH2.