Michael N. Gould

EBSeq: An empirical Bayes hierarchical model for inference in RNA-seq experiments

MOTIVATION Messenger RNA expression is important in normal development and differentiation, as well as in manifestation of disease. RNA-seq experiments allow for the identification of differentially expressed (DE) genes and their corresponding isoforms on a genome-wide scale. However, statistical methods are required to ensure that accurate identifications are made. A number of methods exist for identifying DE genes, but far fewer are available for identifying DE isoforms. When isoform DE is of interest, investigators often apply gene-level (count-based) methods directly to estimates of isoform counts. Doing so is not recommended. In short, estimating isoform expression is relatively straightforward for some groups of isoforms, but more challenging for others. This results in estimation uncertainty that varies across isoform groups. Count-based methods were not designed to accommodate this varying uncertainty, and consequently, application of them for isoform inference results in reduced power for some classes of isoforms and increased false discoveries for others. RESULTS Taking advantage of the merits of empirical Bayesian methods, we have developed EBSeq for identifying DE isoforms in an RNA-seq experiment comparing two or more biological conditions. Results demonstrate substantially improved power and performance of EBSeq for identifying DE isoforms. EBSeq also proves to be a robust approach for identifying DE genes. AVAILABILITY AND IMPLEMENTATION An R package containing examples and sample datasets is available at http://www.biostat.wisc.edu/kendzior/EBSEQ/. SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.

Progress and prospects in rat genetics: a community view

The rat is an important system for modeling human disease. Four years ago, the rich 150-year history of rat research was transformed by the sequencing of the rat genome, ushering in an era of exceptional opportunity for identifying genes and pathways underlying disease phenotypes. Genome-wide association studies in human populations have recently provided a direct approach for finding robust genetic associations in common diseases, but identifying the precise genes and their mechanisms of action remains problematic. In the context of significant progress in rat genomic resources over the past decade, we outline achievements in rat gene discovery to date, show how these findings have been translated to human disease, and document an increasing pace of discovery of new disease genes, pathways and mechanisms. Finally, we present a set of principles that justify continuing and strengthening genetic studies in the rat model, and further development of genomic infrastructure for rat research.

Production of knockout rats using ENU mutagenesis and a yeast-basedscreening assay

The rat is a widely used model in biomedical research and is often the preferred rodent model in many areas of physiological and pathobiological research. Although many genetic tools are available for the rat, methods to produce gene-disrupted knockout rats are greatly needed. In this study, we developed protocols for creating N-ethyl-N-nitrosourea (ENU)-induced germline mutations in several rat strains. F1 preweanling pups from mutagenized Sprague Dawley (SD) male rats were then screened for functional mutations in Brca1 and Brca2 using a yeast gap-repair, ADE2-reporter truncation assay. We produced knockout rats for each of these two breast cancer suppressor genes.

Neutrophil gelatinase-associated lipocalin (NGAL) is a predictor of poor prognosis in human primary breast cancer

Neutrophil gelatinase-associated lipocalin (NGAL) is a small, secreted glycoprotein with proposed functions in cell proliferation, survival and morphogenesis. NGAL is expressed in a variety of tumor types including breast carcinomas, but it is not known whether NGAL contributes directly to breast cancer progression. This study examines the relationship between NGAL expression in breast carcinomas and established clinical prognostic markers as well as clinical outcome. Using immunohistochemistry in tissue microarrays containing well characterized tumor samples from 207 breast cancer patients, NGAL was detected in 68 breast carcinomas in a cytoplasmic location. NGAL expression correlated strongly with negative steroid receptor status, HER-2/neu overexpression, poor histologic grade, the presence of lymph node metastases and a high Ki-67 proliferation index. In univariate survival analysis, NGAL expression was associated with decreased disease-specific survival and decreased disease-free survival in the entire cohort. In multivariate analysis, NGAL remained an independent prognostic marker for disease-free survival. In a subset of patients with estrogen receptor positive tumors, NGAL was significantly associated with decreased disease-free survival. The results show that NGAL expression is a predictor of poor prognosis in primary human breast cancer and suggest that NGAL detection may provide information for risk assessment and identify a subset of patients requiring more aggressive adjuvant therapy.

Structure-activity relationships among monoterpene inhibitors of protein isoprenylation and cell proliferation

The monoterpene d-limonene inhibits the post-translational isoprenylation of p21ras and other small G proteins, a mechanism that may contribute to its efficacy in the chemoprevention and therapy of chemically induced rodent cancers. In the present study, the relative abilities of 26 limonene-like monoterpenes to inhibit protein isoprenylation and cell proliferation were determined. Many monoterpenes were found to be more potent than limonene as inhibitors of small G protein isoprenylation and cell proliferation. The relative potency of limonene-derived monoterpenes was found to be: monohydroxyl = ester = aldehyde > thiol > acid = diol = epoxide > triol = unsubstituted. All monoterpenes that inhibited protein isoprenylation did so in a selective manner, such that 21-26 kDa proteins were preferentially affected. Perillyl alcohol, one of the most potent terpenes, reduced 21-26 kDa protein isoprenylation to 50% of the control level at a concentration of 1 mM, but had no effect on the isoprenylation of 67, 47 or 17 kDa proteins. In particular, p21ras farnesylation was inhibited 40% by 1 mM perillyl alcohol. At the same concentration, perillyl alcohol completely inhibited the proliferation of human HT-29 colon carcinoma cells. The structure-activity relationships observed among the monoterpene isoprenylation inhibitors support a role for small G proteins in cell proliferation, and suggest that many limonene-derived monoterpenes warrant further investigation as antitumor agents.

Mammary carcinoma regression induced by perillyl alcohol, a hydroxylated analog of limonene

The monoterpene perillyl alcohol has been shown to induce the regression of 81% of small mammary carcinomas and up to 75% of advanced mammary carcinomas initiated by 7,12-dimethylbenz(a)anthracene (DMBA) in the Wistar-Furth rat. Dietary perillyl alcohol was greater than 5 times more potent than the monoterpene limonene at inducing tumor regression. Perilly alcohol is rapidly metabolized in the rat, as is limonene. Rats chronically fed perillyl alcohol had the same circulating plasma metabolites as rats fed limonene; however, the levels of these metabolites found in the plasma were higher for perillyl alcohol-fed rats. For example, rats given a 2% perillyl alcohol diet for 10 weeks had plasma levels of terpene metabolites of 0.82 mM whereas those fed a 10% limonene diet for the same period had blood levels of 0.27 mM. It thus appears that the increased potency of perillyl alcohol over limonene in causing tumor regression may be dee at least in part to differences in the pharmacokinetics of these two monoterpenes. We feel that perillyl alcohol is a good candidate for clinical testing of anticancer efficacy in humans.

Heterogeneous expression of the lipocalin NGAL in primary breast cancers

We have previously shown that neu oncogene‐initiated rat mammary carcinomas uniquely over‐express neu‐related lipocalin (NRL), a member of the calycin protein superfamily. Here, we characterize the putative human homolog of NRL, neutrophil gelatinase‐associated lipocalin (NGAL). ngal gene expression was found at moderate levels in only 2 of 17 human tissues examined, breast and lung. When breast cancers were examined for NGAL mRNA and protein levels, they were found to exhibit heterogeneous expression. NGAL levels varied in these tumors from undetectable to exceeding those in normal breast parenchyma. Immuno‐histochemical analysis confirmed the presence of NGAL within breast carcinoma cells but detected only low levels of this protein in normal ductal epithelium. In contrast, large amounts of the protein were localized to the lumen of normal breast ducts in the vicinity of NGAL‐expressing tumors. Interestingly, unlike NRL in rat mammary carcinomas, no significant association between NGAL expression and HER‐2/neu activation was found in human breast tumors. In contrast, a significant correlation between NGAL expression in breast cancer was found with several other markers of poor prognosis, including estrogen and progesterone receptor‐negative status and high proliferation (S‐phase fraction). NGAL levels were stratified as high or low in breast cancers from a cohort of node‐positive patients with known outcome. No significant association between NGAL expression and disease‐free or overall survival was observed. Int. J. Cancer (Pred. Oncol.) 79:565–572, 1998.

A target-selected Apc-mutant rat kindred enhances the modeling of familial human colon cancer

Progress toward the understanding and management of human colon cancer can be significantly advanced if appropriate experimental platforms become available. We have investigated whether a rat model carrying a knockout allele in the gatekeeper gene Adenomatous polyposis coli (Apc) recapitulates familial colon cancer of the human more closely than existing murine models. We have established a mutagen-induced nonsense allele of the rat Apc gene on an inbred F344/NTac (F344) genetic background. Carriers of this mutant allele develop multiple neoplasms with a distribution between the colon and small intestine that closely simulates that found in human familial adenomatous polyposis patients. To distinguish this phenotype from the predominantly small intestinal phenotype found in most Apc-mutant mouse strains, this strain has been designated the polyposis in the rat colon (Pirc) kindred. The Pirc rat kindred provides several unique and favorable features for the study of colon cancer. Tumor-bearing Pirc rats can live at least 17 months, carrying a significant colonic tumor burden. These tumors can be imaged both by micro computed tomography scanning and by classical endoscopy, enabling longitudinal studies of tumor genotype and phenotype as a function of response to chemopreventive and therapeutic regimes. The metacentric character of the rat karyotype, like that of the human and unlike the acrocentric mouse, has enabled us to demonstrate that the loss of the wild-type Apc allele in tumors does not involve chromosome loss. We believe that the Pirc rat kindred can address many of the current gaps in the modeling of human colon cancer.

Identification of metabolites of the antitumor agentd-limonene capable of inhibiting protein isoprenylation and cell growth

SummaryLimonene has been shown to be an effective, nontoxic chemopreventive and chemotherapeutic agent in chemically induced rat mammary-cancer models. The present study characterized circulating metabolites of limonene in female rats and determined their effects on cell growth. Metabolism of limonene was analyzed in plasma extracts by gas chromatography. Rapid conversion of limonene to two major metabolites was detected. These metabolites comprised more than 80% of the circulating limonene-derived material at 1 h after administration and thereafter, whereas limonene itself accounted for only 15%. The metabolites were characterized by mass spectroscopy and infrared spectroscopy. The probable structures were synthesized, and identities were confirmed by comparison of retention times and mass spectra. The two major circulating metabolites of limonene were found to be perillic acid and dihydroperillic acid. We have previously reported that limonene, perillic acid, and dihydroperillic acid inhibit the posttranslational isoprenylation of p21ras and other 21- to 26-kDa cell-growth-associated proteins in NIH3T3 cells and in mammary epithelial cells. In the present study, perillic acid was found to inhibit cell growth in a dose-dependent manner. Thus, perillic acid and dihydroperillic acid, the two major circulating metabolites of limonene in the rat, are more potent inhibitors of protein isoprenylation than is limonene, and perillic acid is also a more potent inhibitor of cell growth. These data raise the possibility that the antitumor effects of limonene in vivo may be mediated via perillic acid and, perhaps, other metabolites.

Cytochromes CYP1A1 and CYP1B1 in the rat mammary gland: Cell-specific expression and regulation by polycyclic aromatic hydrocarbons and hormones

Cultured rat mammary cells express both CYP1A1 and CYP1B1 in response to polycyclic aromatic hydrocarbons (PAH) and 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) in a cell type-specific manner. The expression of each P450 was determined functionally (regioselective PAH metabolism), as apoprotein (immunoblots) and as mRNA (Northern hybridization). The epithelial rat mammary cells (RMEC) expressed CYP1A1, however only after PAH or TCDD treatment. CYP1B1 protein was scarcely detected in these induced RMEC but was surprisingly active as a participant in 7,12-dimethylbenz[a]anthracene (DMBA) metabolism shown through selective antibody inhibition (40% of total activity). CYP1B1 was selectively expressed in the stromal fibroblast population of rat mammary cells to the exclusion of CYP1A1. In the rat mammary fibroblasts (RMF), CYP1B1 protein and associated activity were each present at low levels constitutively and were highly induced by benz[a]anthracene (BA) to a greater extent than by TCDD (12- versus 6-fold). However, BA (10 microM) and TCDD (10 nM) stimulated the 5.2-kb CYP1B1-specific mRNA equally. These increases are consistent with the involvement of the aryl hydrocarbon (Ah) receptor in the transcription of the CYP1B1 gene and with the additional stabilization of CYP1B1 protein by BA, previously observed in embryo fibroblasts. Exactly this regulation of CYP1B1-dependent activity was seen in RMEC suggesting that this arises from exceptionally active CYP1B1 in a small proportion (5%) of residual RMF. The constitutive expression and PAH inducibility of CYP1B1 and CYP1A1 proteins in RMF and RMEC, respectively, were each substantially decreased (approximately 75%) by a hormonal mixture (17 beta-estradiol (0.2 microM) progesterone (1.5 microM) cortisol (1.5 microM) and prolactin (5 micrograms/ml)). Progesterone and cortisol, added singly to RMF suppressed CYP1B1 protein expression (approximately 80%) in both untreated and BA-induced cells, while cortisol also suppressed the 5.2-kb CYP1B1 mRNA. In contrast, 17 beta-estradiol stimulated constitutive expression of CYP1B1 protein (50-75%) and mRNA level (2- to 3-fold), but did not affect CYP1B1 expression in BA-treated RMF. The expression of CYP1A1 and CYP1B1 is therefore highly cell specific even though each is regulated through the Ah receptor. Each P450 exhibits a surprisingly similar pattern of hormonal regulation even though expressed in different cell types.