Analysis of oncogenic signaling networks in glioblastoma identifies ASPM as a molecular target

Glioblastoma is the most common primary malignant brain tumor of adults and one of the most lethal of all cancers. Patients with this disease have a median survival of 15 months from the time of diagnosis despite surgery, radiation, and chemotherapy. New treatment approaches are needed. Recent works suggest that glioblastoma patients may benefit from molecularly targeted therapies. Here, we address the compelling need for identification of new molecular targets. Leveraging global gene expression data from two independent sets of clinical tumor samples (n = 55 and n = 65), we identify a gene coexpression module in glioblastoma that is also present in breast cancer and significantly overlaps with the "metasignature" for undifferentiated cancer. Studies in an isogenic model system demonstrate that this module is downstream of the mutant epidermal growth factor receptor, EGFRvIII, and that it can be inhibited by the epidermal growth factor receptor tyrosine kinase inhibitor Erlotinib. We identify ASPM (abnormal spindle-like microcephaly associated) as a key gene within this module and demonstrate its overexpression in glioblastoma relative to normal brain (or body tissues). Finally, we show that ASPM inhibition by siRNA-mediated knockdown inhibits tumor cell proliferation and neural stem cell proliferation, supporting ASPM as a potential molecular target in glioblastoma. Our weighted gene coexpression network analysis provides a blueprint for leveraging genomic data to identify key control networks and molecular targets for glioblastoma, and the principle eluted from our work can be applied to other cancers.

Tumor necrosis factor-α-induced protein 3 down-regulates nuclear factor-κB-mediated drug resistance in vitro and is a favorable clinical prognostic factor in human glioblastomas

1508

Background: Resistance to chemotherapy is major impediment to the successful treatment of human glioblastomas. Methods: We used an integrated resistance model and genomics tools to globally explore molecular factors and cellular pathways mediating resistance to O6-alkylating agents in glioblastoma cells. Results: We identified a transcriptomic signature of 286 genes that predicts a common in vitro and in vivo resistance phenotype to these agents. This signature was significantly enriched for genes with functions in organismal survival (27.5%) and cell death (49.0%), both with P < 0.00001. Modularity was a predominant organizational principle of the signature, with functions being carried out by groups of interacting molecules in overlapping networks. A highly significant network was built around nuclear factor-κB (NF-κB), which included the persistent alterations of various NF-κB pathway elements. Tumor necrosis factor-α-induced protein 3 (TNFAIP3) was identified as a new endogenous regulatory component of a putative cytoplasmic signaling cascade that mediates NF-κB activation in response to DNA damage caused by O6-alkylating agents. Expression of the corresponding zinc finger protein A20 closely mirrored the expression of the TNFAIP3 transcript, and was inversely related to NF-κB activation status in the resistant cells. A prediction model based on the resistance signature enabled the subclassification of an independent, validation cohort of 31 glioblastomas into two outcome groups (P = .037). TNFAIP3 expression was a favorable factor in patient prognosis (P = .028), and was part of an optimized four-gene predictor (TNFAIP3, CD44 antigen, syndecan 1, and F-box protein 32) significantly associated with patient survival (P = .022). Conclusions: Our results offer strong evidence for TNFAIP3 as a key regulator of the cytoplasmic signaling to activate NF-κB en route to O6-alkylating agent resistance in glioblastoma cells. This gene may be an attractive target for therapeutic modulation of glioblastomas.

No significant financial relationships to disclose.

Identification of transforming growth factor-beta1-binding protein overexpression in carmustine-resistant glioma cells by MRNA differential display

BACKGROUND

The authors previously demonstrated the presence of cells in primary human malignant gliomas that intrinsically are resistant to carmustine (BCNU). Numerous studies have identified mechanisms of therapy resistance in these cells; however, the authors' work and that of others suggest that additional mechanisms of resistance exist.

METHODS

The authors identified a glioma cell line that lacks detectable methylguanine methyltransferase expression and does not alter its expression of glutathione-S-transferase-pi in response to BCNU chemotherapy. This cell line was used in mRNA differential display experiments to identify genes involved in what to the authors' knowledge were previously undescribed mechanisms of resistance.

RESULTS

The overexpression of the gene encoding the transforming growth factor latency binding protein was demonstrated in glioma cells selected for resistance to BCNU, compared with their parental unselected cells.

CONCLUSIONS

Transforming growth factor-beta1 has pleiotropic functions in transformed and normal cells. Although activation of TGF-beta1 does not appear to be a causative factor in BCNU resistance in the current study, it may be involved in the growth of these resistant cells.

Expression of a receptor protein tyrosine phosphatase in human glial tumors

We have analyzed expression of a receptor protein tyrosine phosphatase (RPTPzeta/beta) in tissue samples from 23 human gliomas. Using the reverse transcription-polymerase chain reaction (RT-PCR) technique, we assayed for the presence or absence of mRNA transcripts encoding the intact receptor and 2 alternatively spliced forms of RPTPzeta/beta. Transcripts encoding the intact and truncated receptors were expressed in all of the lower grade gliomas (WHO grade 1-3) analyzed, but not in 55% of the grade 4 glioblastomas multiforme (GBM). However, this subset of GBMs did express an alternatively spliced secreted form comprised of only the RPTPzeta/beta extracellular domain. Our data suggests there may be a correlation between the loss of transcripts encoding the receptor forms of RPTPzeta/beta and progression from low to high grade gliomas. This work provides additional evidence for the importance of phosphatase isoform expression in human tumors.

Expression of the genes encoding myelin basic protein and proteolipid protein in human malignant gliomas

Pathological differentiation of oligodendroglioma and mixed oligoastrocytoma from astrocytoma is difficult, relying on morphological characteristics due to the lack of reliable immunohistochemical stains. Oligodendrocytes, the presumed cell of origin of oligodendrogliomas, highly express the genes encoding myelin basic protein (MBP) and proteolipid protein (PLP). We analyzed the expression of these genes to determine whether they might be useful molecular markers of oligodendrocytic tumors. MBP and PLP were highly expressed in all oligodendrogliomas and minimally expressed in glioblastomas multiforme. MBP was highly expressed in mixed oligoastrocytomas, whereas PLP expression was minimal. The association between tumor classification and expression of the MBP and PLP genes was statistically significant. Expression of these genes may serve as a useful molecular marker for some subtypes of human gliomas.

Biological and molecular analysis of a low-grade recurrence of a glioblastoma multiforme

We and others have reported that human malignant gliomas demonstrate intratumor heterogeneity in which many regions may be benign; however, the presence of regions of increased malignancy in these same tumors is generally indicative of poor patient prognosis. These data suggested that tumor progression may be a local phenomenon, resulting in regions that progress to a more malignant type prior to the progression of the entire tumor. Implicit in this premise is the idea that molecular markers of tumor progression may be detectable prior to histological evidence of progression. This report details analyses performed on a primary and recurrent tumor obtained from the same patient in which the primary tumor was of a higher histological grade than the recurrent tumor. Results of molecular, cytogenetic, flow cytometric, and histological analyses of the primary tumor were indicative of a grade 4 glioblastoma multiforme. Standard cytogenetic and flow cytometric analyses demonstrated that the cells were near-diploid with a stem line population of 46,XX normal G-banded karyotypes. In contrast, tissue resected from the recurrent tumor 5 months later was histologically less malignant; however, the molecular, cytogenetic, and flow cytometric analyses of this sample demonstrated the presence of specific genetic abnormalities typically found in more malignant tumors. These data demonstrate that specific molecular and/or genetic changes leading to tumor progression may become detectable in a glioma prior to the appearance of histological features of a higher grade tumor.

Expression of the tumor suppressor gene DCC in human gliomas

Reduced expression and/or allelic loss of the putative tumor suppressor gene DCC has been demonstrated in colorectal, gastric, pancreatic, esophageal, breast, and hematological malignancies. We examined the expression of the DCC gene in 22 tissue samples from human gliomas (glioblastoma multiforme, oligodendroglioma, and mixed oligodendroglioma/astrocytoma). Seven of 8 glioblastomas multiforme (88%) had reduced or absent DCC expression, and 8 of the other 14 tumors underexpressed DCC when compared to normal brain tissue. These results demonstrate that reduced expression of DCC occurs in human malignant gliomas and may be part of a common genetic pathway leading to neoplastic transformation and/or tumor progression.

BCNU-resistant human glioma cells with over-representation of chromosomes 7 and 22 demonstrate increased copy number and expression of platelet-derived growth factor genes

We used standard karyotypic analyses of first-division cells to identify a subpopulation of cells in primary malignant gliomas with over-representation of chromosomes 7 and 22. These cells are a minor subpopulation in the primary tumor but become the dominant population after treatment in vitro of the cells with the chemotherapeutic agent 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU). The selection for a cell with this specific karyotypic abnormality suggests that these chromosomes contain genes important to the growth of BCNU-resistant cells. Southern blot hybridization analyses demonstrate an increased copy number of the genes encoding platelet-derived growth factor (PDGF) A-chain and B-chain, which have been mapped to chromosomes 7 and 22, respectively. Reverse transcription followed by polymerase chain reaction (RT-PCR) analysis demonstrates increased expression of these genes. In addition, these cells secrete a mitogenic factor that stimulates 3H-thymidine uptake in NIH 3T3 cells. This factor is sensitive to anti-PDGF antibodies and beta-mercaptoethanol, but not to anti-EGF antibodies. These data suggest that autocrine and/or paracrine mechanisms occur in human malignant gliomas, and that over-expression of PDGF may play a role in the growth of BCNU-resistant cells in these tumors.

Dissociation of AIDS-related vacuolar myelopathy and productive HIV-1 infection of the spinal cord

Although merging clinically within the spectrum of the AIDS dementia complex, vacuolar myelopathy is a pathologically distinct entity detected in up to 30% of autopsied patients succumbing to the late complications of human immunodeficiency virus type 1 (HIV-1) infection. Using immunohistochemistry and in situ hybridization to detect an HIV-1 core protein and viral mRNA, respectively, in tissue sections, and culture isolation to assess infectious virus in tissue homogenates, we found that vacuolar myelopathy was independent of productive HIV-1 infection of the spinal cord and brain. These results indicate that AIDS-associated vacuolar myelopathy is either not related directly to spinal cord HIV-1 infection or involves nonproductive infection and pathobiological processes distinct from those responsible for the multinucleated-cell inflammatory infiltrates that serve as histopathologic markers of productive CNS HIV-1 infection.

Transcriptional activity of the herpes simplex virus genome during establishment, maintenance, and reactivation of in vitro virus latency

We previously have described a model of in vitro herpes simplex virus (HSV) latency in which latent infection was (i) established with human leukocyte interferon (IFN-alpha) in combination with (E)-5-(2-bromovinyl)-2'-deoxyuridine (BVDU) or 9-[(2-hydroxyethoxy)methyl]guanine (acyclovir); (ii) maintained after termination of combined inhibitor treatment by incubation at 40.5 degrees, and (iii) reactivated by either reducing the incubation temperature to 37 degrees or by superinfecting at the elevated temperature with human cytomegalovirus (HCMV). We now report the use of this system to examine the transcriptional activity of the HSV genome during establishment, maintenance, and reactivation of HSV latency in vitro. Numerous species of virus-specific polyadenylated RNAs were present during the first 3 days of combined BVDU and IFN-alpha treatment of HSV type 1 (HSV-1)-infected human fetus lung fibroblast cells. However, after 7 days of combined inhibitor treatment, only a very small quantity of virus-specific RNA could be detected utilizing the short unique region of the HSV-1 genome as probe. After terminating combined BVDU and IFN-alpha treatment and increasing the temperature from 37 to 40.5 degrees on day 7 after infection, virus-specific RNA was undetectable by RNA blot hybridization analysis; however, a small amount of HSV-specific RNA was detected in 2% of the cells by in situ hybridization. The HSV-1 transcriptional products produced after HCMV superinfection in the presence of selected inhibitors of macromolecular synthesis also were examined and demonstrated that the efficient activation of HSV-1 immediate-early gene transcription required the expression of not only immediate-early HCMV gene product(s), but also at least a subset of early-late gene products.

Incorporation and metabolism of 2'-fluoro-5-substituted arabinosyl pyrimidines and their selective inhibition of viral DNA synthesis in herpes simplex virus type 1 (HSV-1)-infected and mock-infected Vero cells

The incorporation and metabolism of 2'-fluoro-5-substituted arabinosyl pyrimidine analogs, and their selective inhibition of viral DNA synthesis in herpes simplex virus type 1 (HSV-1)-infected and mock-infected Vero cells were studied by HPLC and CsCl isopycnic density gradient analysis of isolated DNAs. The amounts of radiolabeled analogs incorporated as parent compound following 10 microM exposure for 4 h were 10-fold higher in HSV-1-infected vs mock-infected cells for 2'-fluoro-5-difluoromethyl-Ara-U (F2FMAU); 4.3-fold higher for 5-ethyl deoxyuridine (EdU); 2.6-fold higher for 2'-fluoro-5-methyl-Ara-U (FMAU) and 1.7-fold higher for dThd. For 2'-fluoro-5-ethyl-Ara-U (FEAU), 3.0 pmole of unchanged moiety was incorporated per 10(6) HSV-1-infected cells but no incorporation was detected in mock-infected cells. HPLC profiles showed that the percentages of radiolabeled analogs incorporated as parent compound in the DNA extracted from HSV-1-infected cells were 31.0% for F2FMAU, 99.6% for EdU, 83.5% for FEAU and 98.3% for FMAU; from mock-infected cells, they were 63.6% for F2FMAU, 96.7% for EdU, 97.3% for FMAU and no incorporation into DNA for FEAU was detected. CsCl density gradient analyses of isolated DNA showed that viral DNA synthesis was inhibited 98% by 10 microM FEAU, 92% by 10 microM F2FMAU, 90% by 2 microM FMAU and 80% by 50 microM EdU, whereas cellular DNA synthesis was inhibited by 53, 44, 61, 66 and 54%, respectively. We conclude that: (a) FEAU incorporation into host-cell DNA was not detectable but FEAU was selectively incorporated into HSV-infected cells; (b) FMAU and FEAU were metabolically stable; however, F2FMAU was extensively metabolized; (c) FEAU and F2FMAU were among the most selective inhibitors of HSV-1 DNA synthesis while allowing cellular DNA synthesis to continue.

The brain in AIDS: central nervous system HIV-1 infection and AIDS dementia complex

Infection with human immunodeficiency virus type 1 (HIV-1) is frequently complicated in its late stages by the AIDS dementia complex, a neurological syndrome characterized by abnormalities in cognition, motor performance, and behavior. This dementia is due partially or wholly to a direct effect of the virus on the brain rather than to opportunistic infection, but its pathogenesis is not well understood. Productive HIV-1 brain infection is detected only in a subset of patients and is confined largely or exclusively to macrophages, microglia, and derivative multinucleated cells that are formed by virus-induced cell fusion. Absence of cytolytic infection of neurons, oligodentrocytes, and astrocytes has focused attention on the possible role of indirect mechanisms of brain dysfunction related to either virus or cell-coded toxins. Delayed development of the AIDS dementia complex, despite both early exposure of the nervous system to HIV-1 and chronic leptomeningeal infection, indicates that although this virus is "neurotropic," it is relatively nonpathogenic for the brain in the absence of immunosuppression. Within the context of the permissive effect of immunosuppression, genetic changes in HIV-1 may underlie the neuropathological heterogeneity of the AIDS dementia complex and its relatively independent course in relation to the systemic manifestations of AIDS noted in some patients.

Activation of latent herpes simplex virus type 2 infection in vitro requires a (E)-5-(2-bromovinyl)-2'-deoxyuridine-sensitive gene function

Previous studies have shown that herpes simplex virus (HSV) type 2 (HSV-2) can be maintained in a latent state in a limited number of cells by elevating the incubation temperature after treatment of HSV-infected human fetus lung fibroblast cells with metabolic inhibitors. Superinfection with human cytomegalovirus (HCMV) of latently infected cells maintained at the elevated temperature reactivated latent virus. In addition, superinfection with temperature-sensitive mutants indicated that reactivation of latent HSV in vitro did not require the expression of late gene function(s) of the superinfecting virus. We now report the (i) design of an in vitro HSV-2-latency system in which a higher percentage of cells contain a virus genome that can be activated; and (ii) subsequent use of this system to further characterize the virus activation process. Superinfection with a transcription-negative temperature-sensitive mutant of HSV type 1 (HSV-1) did not reactivate HSV-2-replication, suggesting that adsorption and penetration of the superinfecting virus were not sufficient for reactivation of the latent virus. Furthermore, superinfection with HSV-1 in the presence of (E)-5-(2-bromovinyl)-2'-deoxyuridine did not reactivate HSV-2 replication, suggesting that the expression of the immediate-early gene products are not sufficient for HSV-2 reactivation. Collectively, these data suggest that in addition to the expression of immediate-early gene function(s) at least a subset of early HSV-1 gene products are required for reactivation of latent HSV-2 in vitro.

Regulation of glucokinase activity in the domestic fowl

The factors which regulate soluble and particulate glucokinase and hexokinase activity in the liver of domestic chickens has been investigated. Pretreatment with oral administration (via tube feeding) of glucose plus injection of insulin resulted in a significant increase in the activity of soluble (p less than 0.01) and particulate (p less than 0.01) glucokinase activity whereas fasting for 48 hours reduced glucokinase and hexokinase activity (p less than 0.01) in the particulate fraction only. Treatment of fed chickens for 2 weeks with thyroxine (50 micrograms: i.m. daily) plus triiodothyronine (50 micrograms) resulted in a marginal decrease (NS) in soluble glucokinase activity but significantly increased soluble hexokinase (p less than 0.05) activity. Thyroidectomized animals showed a decline in both soluble glucokinase (p less than 0.01) and hexokinase (p less than 0.025) activity. There was no effect of thyroid hormone manipulation on particulate glucokinase activity although there was a significant reduction in particulate hexokinase activity (p less than 0.05) in thyroidectomized birds. These data establish a physiological role for the glucokinase enzyme activity in avian carbohydrate metabolism and suggest that in contrast with the mammal, the particulate fraction is the more physiologically important enzyme.

Prolonged herpes simplex virus latency in vitro after treatment of infected cells with acyclovir and human leukocyte interferon

We previously demonstrated that herpes simplex virus type 1 (HSV-1) can be established in a latent form in vitro by the treatment of HSV-infected human cells with (E)-5-(2-bromovinyl)-2'-deoxyuridine (BVDU) in combination with human leukocyte interferon (IFN-alpha). We now report that the substitution of BVDU with 9-[(2-hydoxyethoxy)methyl]guanine (acyclovir; ACV) during a combined treatment with IFN-alpha inhibited HSV-1 replication and established in vitro virus latency that could be maintained for a longer period after inhibitor removal and a continued incubation at 37 degrees C. By contrast, the treatment of HSV-1-infected cells with combined IFN-alpha and 9-(1,3-dihydroxy-2-propoxymethyl)guanine, a congener of ACV, failed to establish in vitro virus latency. Furthermore, none of these inhibitors used alone was sufficient to establish in vitro virus latency. The use of nucleoside analogs differing from BVDU in their modes of action has enabled us to initiate studies designed to extend in vitro virus latency.

Analysis of the herpes simplex virus genome during in vitro latency in human diploid fibroblasts and rat sensory neurons

We have previously designed in vitro model systems to characterize the herpes simplex virus type 1 (HSV-1) genome during in vitro virus latency. Latency was established by treatment of infected human embryo lung fibroblast (HEL-F) cells or rat fetal neurons with (E)-5-(2-bromovinyl)-2'-deoxyuridine and human leukocyte interferon and was maintained by increasing the incubation temperature after inhibitor removal. Virus was reactivated by reducing the incubation temperature. We have now examined the HSV-1-specific DNA content of latently infected HEL-F cells and rat fetal neurons treated with (E)-5-(2-bromovinyl)-2'-deoxyuridine and human leukocyte interferon and increased temperature. The HEL-F cell population contained, on an average, between 0.25 and 0.5 copies of most, if not all, HSV-1 HindIII and XbaI DNA fragments per haploid cell genome equivalent. In contrast, the latently infected neurons contained, on an average, 8 to 10 copies per haploid cell genome equivalent of most HSV-1 BamHI DNA fragments. There was no detectable alteration in size or molarity of the HSV-1 terminal or junction DNA fragments obtained by HindIII, XbaI, or BamHI digestion of the latently infected neuron or HEL-F cell DNA, as compared with digestion of a reconstruction mixture of purified HSV-1 virion and HEL-F cell DNAs. These data suggest that the predominant form of the HSV-1 genome in either latently infected cell population is nonintegrated, linear, and nonconcatameric.

High efficiency latency and activation of herpes simplex virus in human cells

Herpes simplex virus (HSV) exists in humans in a latent form that can be activated. To characterize the molecular basis of the cell-virus interactions and to analyze the state of the latent HSV genome, an in vitro model system was established. In this system a large fraction of the latently infected cells contain an HSV genome that can be activated. Cell survival was reduced minimally after repression of high multiplicity HSV type 1 (HSV-1) infection of human fibroblast cells with (E)-5-(2-bromovinyl)-2'-deoxyuridine in combination with human leukocyte interferon (IFN-alpha). A minimum of 1 to 3 percent of the surviving cells contained an HSV genome that could be activated either by human cytomegalovirus superinfection or reduction in incubation temperature.

The effect of high hydrostatic pressure on eukaryotic protein synthesis

The pressure response of two eukaryotic protein synthesizing systems has been characterized. The rabbit reticulocyte system has been tested, both in vivo and in vitro, using endogenous polysomes and polyuridylic acid (poly U). In addition, the poly U-directed polyphenylalanine synthesizing system obtained from wheat germ was utilized. The effect of pressure on eukaryotic protein synthesis has been found to be basically similar to that observed in prokaryotic systems, although the response of the eukaryotic protein synthesizing system is somewhat more complex signifying a greater influence of overlapping reactions. Magnesium was found to affect eukaryotic systems in much the same way as has been reported for prokaryotic systems, i.e., increasing the Mg2+ concentration in a protein synthesizing system increases the barotolerance exhibited by the system. Under conditions of high Mg2+ concentration, however, extreme (up to 160%) stimulation of protein synthesis at lower pressure levels was observed in the eukaryotic systems. Such high stimulation is not apparent in prokaryotic systems. The poly U-directed wheat germ system exhibited the most barotolerant polypeptide synthesis ever seen in our laboratory. This extreme barotolerance was only slightly decreased when the system was tested at reduced concentrations of magnesium.