We showed previously that CCAAT/enhancer binding protein alpha (C/EBP alpha), a tissue-specific transcription factor, is a candidate tumor suppressor in lung cancer. In the present study, we have performed a transcriptional profiling study of C/EBP alpha target genes using an inducible cell line system. This study led to the identification of hepatocyte nuclear factor 3beta (HNF3 beta), a transcription factor known to play a role in airway differentiation, as a downstream target of C/EBP alpha. We found down-regulation of HNF3 beta expression in a large proportion of lung cancer cell lines examined and identified two novel mutants of HNF3 beta, as well as hypermethylation of the HNF3 beta promoter. We also developed a tetracycline-inducible cell line model to study the cellular consequences of HNF3 beta expression. Conditional expression of HNF3 beta led to significant growth reduction, proliferation arrest, apoptosis, and loss of clonogenic ability, suggesting additionally that HNF3 beta is a novel tumor suppressor in lung cancer. This is the first study to show genetic abnormalities of lung-specific differentiation pathways in the development of lung cancer.

Loss of PTEN function leads to activation of phosphoinositide 3-kinase (PI3K) signaling and Akt. Clinical trials are now testing whether mammalian target of rapamycin (mTOR) inhibition is useful in treating PTEN-null cancers. Here, we report that mTOR inhibition induced apoptosis of epithelial cells and the complete reversal of a neoplastic phenotype in the prostate of mice expressing human AKT1 in the ventral prostate. Induction of cell death required the mitochondrial pathway, as prostate-specific coexpression of BCL2 blocked apoptosis. Thus, there is an mTOR-dependent survival signal required downstream of Akt. Bcl2 expression, however, only partially restored intraluminal cell growth in the setting of mTOR inhibition. Expression profiling showed that Hif-1 alpha targets, including genes encoding most glycolytic enzymes, constituted the dominant transcriptional response to AKT activation and mTOR inhibition. These data suggest that the expansion of AKT-driven prostate epithelial cells requires mTOR-dependent survival signaling and activation of HIF-1 alpha, and that clinical resistance to mTOR inhibitors may emerge through BCL2 expression and/or upregulation of HIF-1 alpha activity.

Gene expression analysis of human leukemias has provided insight into disease classification and mechanisms of oncogenesis. Its success is particularly evident for acute leukemias with rearrangement of the mixed lineage leukemia (MLL) gene on chromosome 11q23. Unlike most other recurrent translocations, MLL rearrangements are found in leukemias classified as acute myelogenous leukemia (AML) or acute lymphoblastic leukemia (ALL). In addition, MLL-rearranged leukemias often express both myeloid- and lymphoid-associated genes. These unusual characteristics have generated much interest in the cell of origin and the mechanism of transformation by MLL rearrangements. Here we review insights gained from characterization of MLL-rearranged human leukemias by genome-wide expression profiling and compare these to data from model systems.

We recently found that MLL-rearranged acute lymphoblastic leukemias (MLL) have a unique gene expression profile including high level expression of the receptor tyrosine kinase FLT3. We hypothesized that FLT3 might be a therapeutic target in MLL and found that 5 of 30 MLLs contain mutations in the activation loop of FLT3 that result in constitutive activation. Three are a newly described deletion of I836 and the others are D835 mutations. The recently described FLT3 inhibitor PKC412 proved cytotoxic to Ba/F3 cells dependent upon activated FLT3 containing either mutation. PKC412 is also differentially cytotoxic to leukemia cells with MLL translocations and FLT3 that is activated by either overexpression of the wild-type receptor or mutation. Finally, we developed a mouse model of MLL and used bioluminescent imaging to determine that PKC412 is active against MLL in vivo.

Here we describe how patterns of gene expression in human tumors have been deconvoluted to reveal a mechanism of action for the cyclin D1 oncogene. Computational analysis of the expression patterns of thousands of genes across hundreds of tumor specimens suggested that a transcription factor, C/EBPbeta/Nf-Il6, participates in the consequences of cyclin D1 overexpression. Functional analyses confirmed the involvement of C/EBPbeta in the regulation of genes affected by cyclin D1 and established this protein as an indispensable effector of a potentially important facet of cyclin D1 biology. This work demonstrates that tumor gene expression databases can be used to study the function of a human oncogene in situ.

To determine whether Akt activation was sufficient for the transformation of normal prostate epithelial cells, murine prostate restricted Akt kinase activity was generated in transgenic mice (MPAKT mice). Akt expression led to p70S6K activation, prostatic intraepithelial neoplasia (PIN), and bladder obstruction. mRNA expression profiles from MPAKT ventral prostate revealed similarities to human cancer and an angiogenic signature that included three angiogenin family members, one of which was found elevated in the plasma of men with prostate cancer. Thus, the MPAKT model may be useful in studying the role of Akt in prostate epithelial cell transformation and in the discovery of molecular markers relevant to human disease.

DNA microarrays can be used to identify gene expression changes characteristic of human disease. This is challenging, however, when relevant differences are subtle at the level of individual genes. We introduce an analytical strategy, Gene Set Enrichment Analysis, designed to detect modest but coordinate changes in the expression of groups of functionally related genes. Using this approach, we identify a set of genes involved in oxidative phosphorylation whose expression is coordinately decreased in human diabetic muscle. Expression of these genes is high at sites of insulin-mediated glucose disposal, activated by PGC-1alpha and correlated with total-body aerobic capacity. Our results associate this gene set with clinically important variation in human metabolism and illustrate the value of pathway relationships in the analysis of genomic profiling experiments.

A statistical methodology for estimating dataset size requirements for classifying microarray data using learning curves is introduced. The goal is to use existing classification results to estimate dataset size requirements for future classification experiments and to evaluate the gain in accuracy and significance of classifiers built with additional data. The method is based on fitting inverse power-law models to construct empirical learning curves. It also includes a permutation test procedure to assess the statistical significance of classification performance for a given dataset size. This procedure is applied to several molecular classification problems representing a broad spectrum of levels of complexity.

In modern clinical neuro-oncology, histopathological diagnosis affects therapeutic decisions and prognostic estimation more than any other variable. Among high-grade gliomas, histologically classic glioblastomas and anaplastic oligodendrogliomas follow markedly different clinical courses. Unfortunately, many malignant gliomas are diagnostically challenging; these nonclassic lesions are difficult to classify by histological features, generating considerable interobserver variability and limited diagnostic reproducibility. The resulting tentative pathological diagnoses create significant clinical confusion. We investigated whether gene expression profiling, coupled with class prediction methodology, could be used to classify high-grade gliomas in a manner more objective, explicit, and consistent than standard pathology. Microarray analysis was used to determine the expression of approximately 12000 genes in a set of 50 gliomas, 28 glioblastomas and 22 anaplastic oligodendrogliomas. Supervised learning approaches were used to build a two-class prediction model based on a subset of 14 glioblastomas and 7 anaplastic oligodendrogliomas with classic histology. A 20-feature k-nearest neighbor model correctly classified 18 of the 21 classic cases in leave-one-out cross-validation when compared with pathological diagnoses. This model was then used to predict the classification of clinically common, histologically nonclassic samples. When tumors were classified according to pathology, the survival of patients with nonclassic glioblastoma and nonclassic anaplastic oligodendroglioma was not significantly different (P = 0.19). However, class distinctions according to the model were significantly associated with survival outcome (P = 0.05). This class prediction model was capable of classifying high-grade, nonclassic glial tumors objectively and reproducibly. Moreover, the model provided a more accurate predictor of prognosis in these nonclassic lesions than did pathological classification. These data suggest that class prediction models, based on defined molecular profiles, classify diagnostically challenging malignant gliomas in a manner that better correlates with clinical outcome than does standard pathology.