Identification of transcriptional regulatory networks specific to pilocytic astrocytoma.

BACKGROUND: Pilocytic Astrocytomas (PAs) are common low-grade central nervous system malignancies for which few recurrent and specific genetic alterations have been identified. In an effort to better understand the molecular biology underlying the pathogenesis of these pediatric brain tumors, we performed higher-order transcriptional network analysis of a large gene expression dataset to identify gene regulatory pathways that are specific to this tumor type, relative to other, more aggressive glial or histologically distinct brain tumours. METHODS: RNA derived from frozen human PA tumours was subjected to microarray-based gene expression profiling, using Affymetrix U133Plus2 GeneChip microarrays. This data set was compared to similar data sets previously generated from non-malignant human brain tissue and other brain tumour types, after appropriate normalization. RESULTS: In this study, we examined gene expression in 66 PA tumors compared to 15 non-malignant cortical brain tissues, and identified 792 genes that demonstrated consistent differential expression between independent sets of PA and non-malignant specimens. From this entire 792 gene set, we used the previously described PAP tool to assemble a core transcriptional regulatory network composed of 6 transcription factor genes (TFs) and 24 target genes, for a total of 55 interactions. A similar analysis of oligodendroglioma and glioblastoma multiforme (GBM) gene expression data sets identified distinct, but overlapping, networks. Most importantly, comparison of each of the brain tumor type-specific networks revealed a network unique to PA that included repressed expression of ONECUT2, a gene frequently methylated in other tumor types, and 13 other uniquely predicted TF-gene interactions. CONCLUSIONS: These results suggest specific transcriptional pathways that may operate to create the unique molecular phenotype of PA and thus opportunities for corresponding targeted therapeutic intervention. Moreover, this study also demonstrates how integration of gene expression data with TF-gene and TF-TF interaction data is a powerful approach to generating testable hypotheses to better understand cell-type specific genetic programs relevant to cancer.

Array-based comparative genomic hybridization identifies CDK4 and FOXM1 alterations as independent predictors of survival in malignant peripheral nerve sheath tumor.

PURPOSE: Malignant peripheral nerve sheath tumors (MPNST) are highly aggressive sarcomas with variable patient survival and few known prognostically relevant genomic biomarkers. To identify survival-associated genomic biomarkers, we performed high-resolution array-based comparative genomic hybridization (aCGH) on a large set of MPNSTs. EXPERIMENTAL DESIGN: Candidate gene alterations identified by aCGH in 38 MPNSTs were validated at the DNA, RNA, and protein levels on these same tumors and an independent set of 87 MPNST specimens. RESULTS: aCGH revealed highly complex copy number alterations, including both previously reported and completely novel loci. Four regions of copy number gain were associated with poor patient survival. Candidate genes in these regions include SOX5 (12p12.1), NOL1 and MLF2 (12p13.31), FOXM1 and FKBP1 (12p13.33), and CDK4 and TSPAN31 (12q14.1). Alterations of these candidate genes and several others of interest (ERBB2, MYC and TP53) were confirmed by at least 1 complementary methodology, including DNA and mRNA quantitative real-time PCR, mRNA expression profiling, and tissue microarray-based fluorescence in situ hybridization and immunohistochemistry. Multivariate analysis showed that CDK4 gain/amplification and increased FOXM1 protein expression were the most significant independent predictors for poor survival in MPNST patients (P < 0.05). CONCLUSIONS: Our study provides new and independently confirmed candidate genes that could serve as genomic biomarkers for overall survival in MPNST patients.

In vitro transformation of primary human CD34+ cells by AML fusion oncogenes: early gene expression profiling reveals possible drug target in AML.

Different fusion oncogenes in acute myeloid leukemia (AML) have distinct clinical and laboratory features suggesting different modes of malignant transformation. Here we compare the in vitro effects of representatives of 4 major groups of AML fusion oncogenes on primary human CD34+ cells. As expected from their clinical similarities, MLL-AF9 and NUP98-HOXA9 had very similar effects in vitro. They both caused erythroid hyperplasia and a clear block in erythroid and myeloid maturation. On the other hand, AML1-ETO and PML-RARA had only modest effects on myeloid and erythroid differentiation. All oncogenes except PML-RARA caused a dramatic increase in long-term proliferation and self-renewal. Gene expression profiling revealed two distinct temporal patterns of gene deregulation. Gene deregulation by MLL-AF9 and NUP98-HOXA9 peaked 3 days after transduction. In contrast, the vast majority of gene deregulation by AML1-ETO and PML-RARA occurred within 6 hours, followed by a dramatic drop in the numbers of deregulated genes. Interestingly, the p53 inhibitor MDM2 was upregulated by AML1-ETO at 6 hours. Nutlin-3, an inhibitor of the interaction between MDM2 and p53, specifically inhibited the proliferation and self-renewal of primary human CD34+ cells transduced with AML1-ETO, suggesting that MDM2 upregulation plays a role in cell transformation by AML1-ETO. These data show that differences among AML fusion oncogenes can be recapitulated in vitro using primary human CD34+ cells and that early gene expression profiling in these cells can reveal potential drug targets in AML.

A white-box approach to microarray probe response characterization: the BaFL pipeline.

BACKGROUND: Microarrays depend on appropriate probe design to deliver the promise of accurate genome-wide measurement. Probe design, ideally, produces a unique probe-target match with homogeneous duplex stability over the complete set of probes. Much of microarray pre-processing is concerned with adjusting for non-ideal probes that do not report target concentration accurately. Cross-hybridizing probes (non-unique), probe composition and structure, as well as platform effects such as instrument limitations, have been shown to affect the interpretation of signal. Data cleansing pipelines seldom filter specifically for these constraints, relying instead on general statistical tests to remove the most variable probes from the samples in a study. This adjusts probes contributing to ProbeSet (gene) values in a study-specific manner. We refer to the complete set of factors as biologically applied filter levels (BaFL) and have assembled an analysis pipeline for managing them consistently. The pipeline and associated experiments reported here examine the outcome of comprehensively excluding probes affected by known factors on inter-experiment target behavior consistency. RESULTS: We present here a 'white box' probe filtering and intensity transformation protocol that incorporates currently understood factors affecting probe and target interactions; the method has been tested on data from the Affymetrix human GeneChip HG-U95Av2, using two independent datasets from studies of a complex lung adenocarcinoma phenotype. The protocol incorporates probe-specific effects from SNPs, cross-hybridization and low heteroduplex affinity, as well as effects from scanner sensitivity, sample batches, and includes simple statistical tests for identifying unresolved biological factors leading to sample variability. Subsequent to filtering for these factors, the consistency and reliability of the remaining measurements is shown to be markedly improved. CONCLUSIONS: The data cleansing protocol yields reproducible estimates of a given probe or ProbeSet's (gene's) relative expression that translates across datasets, allowing for credible cross-experiment comparisons. We provide supporting evidence for the validity of removing several large classes of probes, and for our approaches for removing outlying samples. The resulting expression profiles demonstrate consistency across the two independent datasets. Finally, we demonstrate that, given an appropriate sampling pool, the method enhances the t-test's statistical power to discriminate significantly different means over sample classes.

Acquired copy number alterations in adult acute myeloid leukemia genomes.

Cytogenetic analysis of acute myeloid leukemia (AML) cells has accelerated the identification of genes important for AML pathogenesis. To complement cytogenetic studies and to identify genes altered in AML genomes, we performed genome-wide copy number analysis with paired normal and tumor DNA obtained from 86 adult patients with de novo AML using 1.85 million feature SNP arrays. Acquired copy number alterations (CNAs) were confirmed using an ultra-dense array comparative genomic hybridization platform. A total of 201 somatic CNAs were found in the 86 AML genomes (mean, 2.34 CNAs per genome), with French-American-British system M6 and M7 genomes containing the most changes (10-29 CNAs per genome). Twenty-four percent of AML patients with normal cytogenetics had CNA, whereas 40% of patients with an abnormal karyotype had additional CNA detected by SNP array, and several CNA regions were recurrent. The mRNA expression levels of 57 genes were significantly altered in 27 of 50 recurrent CNA regions <5 megabases in size. A total of 8 uniparental disomy (UPD) segments were identified in the 86 genomes; 6 of 8 UPD calls occurred in samples with a normal karyotype. Collectively, 34 of 86 AML genomes (40%) contained alterations not found with cytogenetics, and 98% of these regions contained genes. Of 86 genomes, 43 (50%) had no CNA or UPD at this level of resolution. In this study of 86 adult AML genomes, the use of an unbiased high-resolution genomic screen identified many genes not previously implicated in AML that may be relevant for pathogenesis, along with many known oncogenes and tumor suppressor genes.