Jennifer Walling

Epigenetic-Mediated Dysfunction of the Bone Morphogenetic Protein Pathway Inhibits Differentiation of Glioblastoma-Initiating Cells

Despite similarities between tumor-initiating cells with stem-like properties (TICs) and normal neural stem cells, we hypothesized that there may be differences in their differentiation potentials. We now demonstrate that both bone morphogenetic protein (BMP)-mediated and ciliary neurotrophic factor (CNTF)-mediated Jak/STAT-dependent astroglial differentiation is impaired due to EZH2-dependent epigenetic silencing of BMP receptor 1B (BMPR1B) in a subset of glioblastoma TICs. Forced expression of BMPR1B either by transgene expression or demethylation of the promoter restores their differentiation capabilities and induces loss of their tumorigenicity. We propose that deregulation of the BMP developmental pathway in a subset of glioblastoma TICs contributes to their tumorigenicity both by desensitizing TICs to normal differentiation cues and by converting otherwise cytostatic signals to proproliferative signals.

Unsupervised Analysis of Transcriptomic Profiles Reveals Six Glioma Subtypes

Gliomas are the most common type of primary brain tumors in adults and a significant cause of cancer-related mortality. Defining glioma subtypes based on objective genetic and molecular signatures may allow for a more rational, patient-specific approach to therapy in the future. Classifications based on gene expression data have been attempted in the past with varying success and with only some concordance between studies, possibly due to inherent bias that can be introduced through the use of analytic methodologies that make a priori selection of genes before classification. To overcome this potential source of bias, we have applied two unsupervised machine learning methods to genome-wide gene expression profiles of 159 gliomas, thereby establishing a robust glioma classification model relying only on the molecular data. The model predicts for two major groups of gliomas (oligodendroglioma-rich and glioblastoma-rich groups) separable into six hierarchically nested subtypes. We then identified six sets of classifiers that can be used to assign any given glioma to the corresponding subtype and validated these classifiers using both internal (189 additional independent samples) and two external data sets (341 patients). Application of the classification system to the external glioma data sets allowed us to identify previously unrecognized prognostic groups within previously published data and within The Cancer Genome Atlas glioblastoma samples and the different biological pathways associated with the different glioma subtypes offering a potential clue to the pathogenesis and possibly therapeutic targets for tumors within each subtype.

Genomic changes and gene expression profiles reveal that established glioma cell lines are poorly representative of primary human gliomas.

Genetic aberrations, such as gene amplification, deletions, and loss of heterozygosity, are hallmarks of cancer and are thought to be major contributors to the neoplastic process. Established cancer cell lines have been the primary in vitro and in vivo models for cancer for more than 2 decades; however, few such cell lines have been extensively characterized at the genomic level. Here, we present a high-resolution genome-wide chromosomal alteration and gene expression analyses of five of the most commonly used glioma cell lines and compare the findings with those observed in 83 primary human gliomas. Although genomic alterations known to occur in primary tumors were identified in the cell lines, we also observed several novel recurrent aberrations in the glioma cell lines that are not frequently represented in primary tumors. Additionally, a global gene expression cluster distinct from primary tumors was identified in the glioma cell lines. Our results indicate that established cell lines are generally a poor representation of primary tumor biology, presenting a host of genomic and gene expression changes not observed in primary tissues, although some discrete features of glioma biology were conserved in the established cell lines. Refined maps of genetic alterations and transcriptional divergence from the original tumor type, such as the one presented here, may help serve as a guideline for a more biologically rational and clinically relevant selection of the most appropriate glioma model for a given experiment. (Mol Cancer Res 2008;6(1):21–30)

A purine scaffold Hsp90 inhibitor destabilizes BCL-6 and has specific antitumor activity in BCL-6–dependent B cell lymphomas

We report that heat shock protein 90 (Hsp90) inhibitors selectively kill diffuse large B cell lymphomas (DLBCLs) that depend on the BCL-6 transcriptional repressor. We found that endogenous Hsp90 interacts with BCL-6 in DLBCL cells and can stabilize BCL-6 mRNA and protein. Hsp90 formed a complex with BCL-6 at its target promoters, and Hsp90 inhibitors derepressed BCL-6 target genes. A stable mutant of BCL-6 rescued DLBCL cells from Hsp90 inhibitor–induced apoptosis. BCL-6 and Hsp90 were almost invariantly coexpressed in the nuclei of primary DLBCL cells, suggesting that their interaction is relevant in this disease. We examined the pharmacokinetics, toxicity and efficacy of PU-H71, a recently developed purine-derived Hsp90 inhibitor. PU-H71 preferentially accumulated in lymphomas compared to normal tissues and selectively suppressed BCL-6–dependent DLBCLs in vivo, inducing reactivation of key BCL-6 target genes and apoptosis. PU-H71 also induced cell death in primary human DLBCL specimens.

High-resolution Global Genomic Survey of 178 Gliomas Reveals Novel Regions of Copy Number Alteration and Allelic Imbalances

Primary brain tumors are the fourth leading cause of cancer mortality in adults under the age of 54 years and the leading cause of cancer mortality in children in the United States. Therapy for the most common type of primary brain tumors, gliomas, remains suboptimal. The development of new and more effective treatments will likely require a better understanding of the biology of these tumors. Here, we show that use of the high-density 100K single-nucleotide polymorphism arrays in a large number of primary tumor samples allows for a much higher resolution survey of the glioma genome than has been previously reported in any tumor type. We not only confirmed alterations in genomic areas previously reported to be affected in gliomas, but we also refined the location of those sites and uncovered multiple, previously unknown regions that are affected by copy number alterations (amplifications, homozygous and heterozygous deletions) as well as allelic imbalances (loss of heterozygosity/gene conversions). The wealth of genomic data produced may allow for the development of a more rational molecular classification of gliomas and serve as an important starting point in the search for new molecular therapeutic targets.

Recurrent epimutation of SDHC in gastrointestinal stromal tumors

Methylation of the SDH gene explains the loss of SDH gene expression in SDH wild-type gastrointestinal stromal tumors. All Roads Lead to Loss of Expression Gastrointestinal stromal tumors are the most common mesenchymal tumors in the gastrointestinal tract, and they can occur in isolation or as part of a constellation of cancers known as Carney triad. A subtype of this cancer, characterized by lack of expression in a gene called SDH, is not well understood and lacks a specific treatment, and this is the type that most commonly occurs in children. Now, Killian et al. have identified methylation of the SDH gene in patients with SDH-deficient gastrointestinal stromal tumors who lack mutations in the SDH gene. This finding provides a common link explaining the pathogenesis of these SDH-deficient tumors, including many of the ones associated with Carney triad. Succinate dehydrogenase (SDH) is a conserved effector of cellular metabolism and energy production, and loss of SDH function is a driver mechanism in several cancers. SDH-deficient gastrointestinal stromal tumors (dSDH GISTs) collectively manifest similar phenotypes, including hypermethylated epigenomic signatures, tendency to occur in pediatric patients, and lack of KIT/PDGFRA mutations. dSDH GISTs often harbor deleterious mutations in SDH subunit genes (SDHA, SDHB, SDHC, and SDHD, termed SDHx), but some are SDHx wild type (WT). To further elucidate mechanisms of SDH deactivation in SDHx-WT GIST, we performed targeted exome sequencing on 59 dSDH GISTs to identify 43 SDHx-mutant and 16 SDHx-WT cases. Genome-wide DNA methylation and expression profiling exposed SDHC promoter–specific CpG island hypermethylation and gene silencing in SDHx-WT dSDH GISTs [15 of 16 cases (94%)]. Six of 15 SDHC-epimutant GISTs occurred in the setting of the multitumor syndrome Carney triad. We observed neither SDHB promoter hypermethylation nor large deletions on chromosome 1q in any SDHx-WT cases. Deep genome sequencing of a 130-kbp (kilo–base pair) window around SDHC revealed no recognizable sequence anomalies in SDHC-epimutant tumors. More than 2000 benign and tumor reference tissues, including stem cells and malignancies with a hypermethylator epigenotype, exhibit solely a non-epimutant SDHC promoter. Mosaic constitutional SDHC promoter hypermethylation in blood and saliva from patients with SDHC-epimutant GIST implicates a postzygotic mechanism in the establishment and maintenance of SDHC epimutation. The discovery of SDHC epimutation provides a unifying explanation for the pathogenesis of dSDH GIST, whereby loss of SDH function stems from either SDHx mutation or SDHC epimutation.

Prediction of Associations between microRNAs and Gene Expression in Glioma Biology

Despite progress in the determination of miR interactions, their regulatory role in cancer is only beginning to be unraveled. Utilizing gene expression data from 27 glioblastoma samples we found that the mere knowledge of physical interactions between specific mRNAs and miRs can be used to determine associated regulatory interactions, allowing us to identify 626 associated interactions, involving 128 miRs that putatively modulate the expression of 246 mRNAs. Experimentally determining the expression of miRs, we found an over-representation of over(under)-expressed miRs with various predicted mRNA target sequences. Such significantly associated miRs that putatively bind over-expressed genes strongly tend to have binding sites nearby the 3′UTR of the corresponding mRNAs, suggesting that the presence of the miRs near the translation stop site may be a factor in their regulatory ability. Our analysis predicted a significant association between miR-128 and the protein kinase WEE1, which we subsequently validated experimentally by showing that the over-expression of the naturally under-expressed miR-128 in glioma cells resulted in the inhibition of WEE1 in glioblastoma cells.

Predicting in vitro drug sensitivity using Random Forests

MOTIVATION Panels of cell lines such as the NCI-60 have long been used to test drug candidates for their ability to inhibit proliferation. Predictive models of in vitro drug sensitivity have previously been constructed using gene expression signatures generated from gene expression microarrays. These statistical models allow the prediction of drug response for cell lines not in the original NCI-60. We improve on existing techniques by developing a novel multistep algorithm that builds regression models of drug response using Random Forest, an ensemble approach based on classification and regression trees (CART). RESULTS This method proved successful in predicting drug response for both a panel of 19 Breast Cancer and 7 Glioma cell lines, outperformed other methods based on differential gene expression, and has general utility for any application that seeks to relate gene expression data to a continuous output variable. IMPLEMENTATION Software was written in the R language and will be available together with associated gene expression and drug response data as the package ivDrug at http://r-forge.r-project.org.

Histone Demethylase Jumonji D3 (JMJD3) as a Tumor Suppressor by Regulating p53 Protein Nuclear Stabilization

Histone methylation regulates normal stem cell fate decisions through a coordinated interplay between histone methyltransferases and demethylases at lineage specific genes. Malignant transformation is associated with aberrant accumulation of repressive histone modifications, such as polycomb mediated histone 3 lysine 27 (H3K27me3) resulting in a histone methylation mediated block to differentiation. The relevance, however, of histone demethylases in cancer remains less clear. We report that JMJD3, a H3K27me3 demethylase, is induced during differentiation of glioblastoma stem cells (GSCs), where it promotes a differentiation-like phenotype via chromatin dependent (INK4A/ARF locus activation) and chromatin independent (nuclear p53 protein stabilization) mechanisms. Our findings indicate that deregulation of JMJD3 may contribute to gliomagenesis via inhibition of the p53 pathway resulting in a block to terminal differentiation.

A phase I trial of enzastaurin in patients with recurrent gliomas.

Purpose: Enzastaurin is a selective inhibitor of protein kinase C β. Prior phase I studies did not show increased drug exposures with escalating once daily administration. Limits from gastrointestinal absorption may be overcome by twice daily dosing, potentially improving antitumor effects. Experimental Design: We conducted a phase I dose escalation study in 26 patients with recurrent malignant glioma, stratified by use of enzyme-inducing antiepileptic drugs, to investigate whether divided twice daily dosing results in higher exposures compared with once daily dosing. Phosphorylated glycogen synthase 3 β was analyzed as a potential biomarker of enzastaurin activity. Results: Enzastaurin was poorly tolerated at all dose levels evaluated (500, 800, and 1,000 mg total daily), with thrombocytopenia and prolonged QTc as dose-limiting toxicities. The average drug concentration of enzastaurin under steady-state conditions was doubled by twice daily dosing compared with daily dosing [1.990; 90% confidence interval (CI), 1.450-2.730]. Additionally, geometric mean ratios doubled with 800 versus 500 mg dosing for both daily (2.687; 90% CI, 1.232-5.860) and twice daily regimens (1.852; 90% CI, 0.799-4.292). Two patients achieved long-term benefit (over 150 weeks progression free). Conclusions: Higher and more frequent dosing of enzastaurin resulted in improved drug exposure but with unacceptable toxicity at the doses tested. Phosphorylated glycogen synthase 3 β may be a useful biomarker of the biological activity of enzastaurin. Enzastaurin has activity in a subset of malignant glioma patients and warrants continued study in combination with other agents using a maximal once daily dose of 500 mg.