Karol L. Thompson

Analysis of variance components in gene expression data

MOTIVATION A microarray experiment is a multi-step process, and each step is a potential source of variation. There are two major sources of variation: biological variation and technical variation. This study presents a variance-components approach to investigating animal-to-animal, between-array, within-array and day-to-day variations for two data sets. The first data set involved estimation of technical variances for pooled control and pooled treated RNA samples. The variance components included between-array, and two nested within-array variances: between-section (the upper- and lower-sections of the array are replicates) and within-section (two adjacent spots of the same gene are printed within each section). The second experiment was conducted on four different weeks. Each week there were reference and test samples with a dye-flip replicate in two hybridization days. The variance components included week-to-week, animal-to-animal and between-array and within-array variances. RESULTS We applied the linear mixed-effects model to quantify different sources of variation. In the first data set, we found that the between-array variance is greater than the between-section variance, which, in turn, is greater than the within-section variance. In the second data set, for the reference samples, the week-to-week variance is larger than the between-array variance, which, in turn, is slightly larger than the within-array variance. For the test samples, the week-to-week variance has the largest variation. The animal-to-animal variance is slightly larger than the between-array and within-array variances. However, in a gene-by-gene analysis, the animal-to-animal variance is smaller than the between-array variance in four out of five housekeeping genes. In summary, the largest variation observed is the week-to-week effect. Another important source of variability is the animal-to-animal variation. Finally, we describe the use of variance-component estimates to determine optimal numbers of animals, arrays per animal and sections per array in planning microarray experiments.

Sources of variation in baseline gene expression levels from toxicogenomics study control animals across multiple laboratories

BackgroundThe use of gene expression profiling in both clinical and laboratory settings would be enhanced by better characterization of variance due to individual, environmental, and technical factors. Meta-analysis of microarray data from untreated or vehicle-treated animals within the control arm of toxicogenomics studies could yield useful information on baseline fluctuations in gene expression, although control animal data has not been available on a scale and in a form best served for data-mining.ResultsA dataset of control animal microarray expression data was assembled by a working group of the Health and Environmental Sciences Institutes Technical Committee on the Application of Genomics in Mechanism Based Risk Assessment in order to provide a public resource for assessments of variability in baseline gene expression. Data from over 500 Affymetrix microarrays from control rat liver and kidney were collected from 16 different institutions. Thirty-five biological and technical factors were obtained for each animal, describing a wide range of study characteristics, and a subset were evaluated in detail for their contribution to total variability using multivariate statistical and graphical techniques.ConclusionThe study factors that emerged as key sources of variability included gender, organ section, strain, and fasting state. These and other study factors were identified as key descriptors that should be included in the minimal information about a toxicogenomics study needed for interpretation of results by an independent source. Genes that are the most and least variable, gender-selective, or altered by fasting were also identified and functionally categorized. Better characterization of gene expression variability in control animals will aid in the design of toxicogenomics studies and in the interpretation of their results.

An evaluation of the hemizygous transgenic Tg.AC mouse for carcinogenicity testing of pharmaceuticals. II. A genotypic marker that predicts tumorigenic responsiveness.

The Tg.AC transgenic mouse skin paint assay is one of the short-term carcinogenesis models that has been proposed as a replacement for 1 species in the conventional 2-yr bioassay required for safety testing of pharmaceuticals. In our initial efforts to evaluate the sensitivity and specificity of this model for human pharmaceuticals, 61% of the hemizygous Tg.AC mice in the positive control groups were refractory to treatment with 12-O-tetradecanoylphorbol 13-acetate (TPA). Tg.AC mice are reported to carry ≤10 copies of a transgcne consisting of a ζ-globin promoter fused to the v-Ha-ras structural gene with a terminal simian virus 40 (SV40) polyaden-ylation signal. Southern blot hybridization of genomic DNA from 66 tail biopsies using a ζ-globin probe revealed that all of the hemizygous Tg.AC mice screened contained approximately 40 copies of the transgene and that mice unresponsive to TPA had lost a 2-kb BamHI fragment containing ζ-gIobin promoter sequence. By systematic screening of Tg.AC breeder mice for this diagnostic marker of phenotypic responsiveness, it should be possible to selectively enrich the Tg.AC mouse colony to consist exclusively of responders and to guard against further genetic instability.

Improvement in the Reproducibility and Accuracy of DNA Microarray Quantification by Optimizing Hybridization Conditions

BackgroundDNA microarrays, which have been increasingly used to monitor mRNA transcripts at a global level, can provide detailed insight into cellular processes involved in response to drugs and toxins. This is leading to new understandings of signaling networks that operate in the cell, and the molecular basis of diseases. Custom printed oligonucleotide arrays have proven to be an effective way to facilitate the applications of DNA microarray technology. A successful microarray experiment, however, involves many steps: well-designed oligonucleotide probes, printing, RNA extraction and labeling, hybridization, and imaging. Optimization is essential to generate reliable microarray data.ResultsHybridization and washing steps are crucial for a successful microarray experiment. By following the hybridization and washing conditions recommended by an oligonucleotide provider, it was found that the expression ratios were compressed greater than expected and data analysis revealed a high degree of non-specific binding. A series of experiments was conducted using rat mixed tissue RNA reference material (MTRRM) and other RNA samples to optimize the hybridization and washing conditions. The optimized hybridization and washing conditions greatly reduced the non-specific binding and improved the accuracy of spot intensity measurements.ConclusionThe results from the optimized hybridization and washing conditions greatly improved the reproducibility and accuracy of expression ratios. These experiments also suggested the importance of probe designs using better bioinformatics approaches and the need for common reference RNA samples for platform performance evaluation in order to fulfill the potential of DNA microarray technology.

Normalization Methods for Analysis of Microarray Gene-Expression Data

This paper investigates subset normalization to adjust for location biases (e.g., splotches) combined with global normalization for intensity biases (e.g., saturation). A data set from a toxicogenomic experiment using the same control and the same treated sample hybridized to six different microarrays is used to contrast the different normalization methods. Simple t-tests were used to compare two samples for dye effects and for treatment effects. The numbers of genes that reproducibly showed significant p-values for the unnormalized data and normalized data from different methods were evaluated for assessment of different normalization methods. The one-sample t-statistic of the ratio of red to green samples was used to test for dye effects using only control data. For treatment effects, in addition to the one-sample t-test of the ratio of the treated to control samples, the two-sample t-test for testing the difference between treated and control samples was also used to compare the two approaches. The method that combines a subset approach (median or lowessfit) for location adjustment with a global lowess fit for intensity adjustment appears to perform well.

Use of a mixed tissue RNA design for performance assessments on multiple microarray formats

The comparability and reliability of data generated using microarray technology would be enhanced by use of a common set of standards that allow accuracy, reproducibility and dynamic range assessments on multiple formats. We designed and tested a complex biological reagent for performance measurements on three commercial oligonucleotide array formats that differ in probe design and signal measurement methodology. The reagent is a set of two mixtures with different proportions of RNA for each of four rat tissues (brain, liver, kidney and testes). The design provides four known ratio measurements of >200 reference probes, which were chosen for their tissue-selectivity, dynamic range coverage and alignment to the same exemplar transcript sequence across all three platforms. The data generated from testing three biological replicates of the reagent at eight laboratories on three array formats provides a benchmark set for both laboratory and data processing performance assessments. Close agreement with target ratios adjusted for sample complexity was achieved on all platforms and low variance was observed among platforms, replicates and sites. The mixed tissue design produces a reagent with known gene expression changes within a complex sample and can serve as a paradigm for performance standards for microarrays that target other species.

Evaluation of miR-216a and miR-217 as potential biomarkers of acute pancreatic injury in rats and mice

Abstract Mild injury of the exocrine pancreas is often asymptomatic and can be under- or mis-diagnosed. The pancreas-enriched microRNAs miR-216a and miR-217 were evaluated as potential serum biomarkers of exocrine pancreas injury in rodent models of acute pancreatitis induced by caerulein, l-arginine, and pancreatic duct ligation. Both microRNAs showed time- and dose- relevant responses to pancreatic injury and wider dynamic ranges of response than serum amylase or lipase. Pancreas-selective microRNAs were found to be relatively sensitive serum biomarkers of pancreatic injury in rodents with potentially greater specificity than the current standard assays.

An Evaluation of the Hemizygous Transgenic Tg.AC Mouse for Carcinogenicity Testing of Pharmaceuticals. I. Evidence for a Confounding Nonresponder Phenotype

We have completed 2 26-wk studies to evaluate the hemizygous transgenic Tg.AC mouse, which has been proposed as an alternative short term model for testing carcinogenicity. We attempted to evaluate the response to the known rodent carcinogens cyclophosphamide, phenolphthalein, and tamoxifen and to the noncarcinogen chlorpheniramine following topical application. In the first study, a weak response (2/17 animals) was observed to the positive control 12-O-tetradecanoylphorbol 13-acetate (TPA in ethanol, 1.25 μg), and no response was observed to cyclophosphamide, phenolphthalein, or chlorpheniramine, despite evidence for skin penetration. The second study compared 1.25 μg and 6.25 μg of TPA in ethanol and acetone solutions. Tamoxifen was also evaluated in both solvents and orally. No significant response was observed to tamoxifen by skin paint or oral routes. Over 60% of the high dose TPA-treated animals showed no (0 or 1) papilloma response, and 30% of the animals each developed more than 32 papillomas. The heterogenous response to high dose TPA may be related to variability in the responsiveness of hemizygous animals. In light of these findings, further Tg.AC studies should employ homozygous animals, and the underlying cause for heterogeneity in the tumorigenic response of Tg.AC mice should be identified and eliminated.

Workgroup report : Review of genomics data based on experience with mock submissions-view of the CDER pharmacology toxicology nonclinical pharmacogenomics subcommittee

Over the past few years, both the U.S. Food and Drug Administration (FDA) and the pharmaceutical industry have recognized the potential importance of pharmacogenomics and toxicogenomics to drug development. To resolve the uncertainties surrounding the use of microarray technology and the presentation of genomics data for regulatory purposes, several pharmaceutical companies and genomics technology providers have provided the FDA with reports of genomics studies that included supporting toxicology data (e.g., serum chemistry, histopathology). These studies were not associated with any active drug application and were exploratory or hypothesis generating in nature. For training purposes, these reports were reviewed by the Nonclinical Pharmacogenomics Subcommittee consisting of the Center for Drug Evaluation and Research pharmacology and toxicology researchers and reviewers. In this article, we describe some of these submissions and report on our assessment of data content, format, and quality control metrics that were useful for evaluating these nonclinical genomics submissions, specifically in relation to the proposed MIAME/MINTox (minimum information about a microarray experiment/minimum information needed for a toxicology experiment) recommendations. These genomics submissions allowed both researchers and regulators to gain experience in the process of reviewing and analyzing toxicogenomics data. The experience will allow development of recommendations for the submission and review of these data as the state of the science evolves.

Age-related differences in susceptibility to cisplatin-induced renal toxicity

Limited experimental models exist to assess drug toxicity in pediatric populations. We recently reported how a multi‐age rat model could be used for pre‐clinical studies of comparative drug toxicity in pediatric populations. The objective of this study was to expand the utility of this animal model, which previously demonstrated an age‐dependent sensitivity to the classic nephrotoxic compound, gentamicin, to another nephrotoxicant, namely cisplatin (Cis). Sprague–Dawley rats (10, 25, 40 and 80 days old) were injected with a single dose of Cis (0, 1, 3 or 6 mg kg−1 i.p.). Urine samples were collected prior and up to 72 h after treatment in animals that were ≥ 25 days old. Several serum, urinary and ‘omic’ injury biomarkers as well as renal histopathology lesions were evaluated. Statistically significant changes were noted with different injury biomarkers in different age groups. The order of age‐related Cis‐induced nephrotoxicity was different than our previous study with gentamicin: 80 > 40 > 10 > 25 day‐old vs 10 ≥ 80 > 40 > 25‐day‐old rats, respectively. The increased levels of kidney injury molecule‐1 (Kim‐1: urinary protein/tissue mRNA) provided evidence of early Cis‐induced nephrotoxicity in the most sensitive age group (80 days old). Levels of Kim‐1 tissue mRNA and urinary protein were significantly correlated to each other and to the severity of renal histopathology lesions. These data indicate that the multi‐age rat model can be used to demonstrate different age‐related sensitivities to renal injury using mechanistically distinct nephrotoxicants, which is reflected in measurements of a variety of metabolite, gene transcript and protein biomarkers. Published in 2009 by John Wiley & Sons, Ltd.