Importance of replication in microarray gene expression studies: Statistical methods and evidence from repetitive cDNA hybridizations

We present statistical methods for analyzing replicated cDNA microarray expression data and report the results of a controlled experiment. The study was conducted to investigate inherent variability in gene expression data and the extent to which replication in an experiment produces more consistent and reliable findings. We introduce a statistical model to describe the probability that mRNA is contained in the target sample tissue, converted to probe, and ultimately detected on the slide. We also introduce a method to analyze the combined data from all replicates. Of the 288 genes considered in this controlled experiment, 32 would be expected to produce strong hybridization signals because of the known presence of repetitive sequences within them. Results based on individual replicates, however, show that there are 55, 36, and 58 highly expressed genes in replicates 1, 2, and 3, respectively. On the other hand, an analysis by using the combined data from all 3 replicates reveals that only 2 of the 288 genes are incorrectly classified as expressed. Our experiment shows that any single microarray output is subject to substantial variability. By pooling data from replicates, we can provide a more reliable analysis of gene expression data. Therefore, we conclude that designing experiments with replications will greatly reduce misclassification rates. We recommend that at least three replicates be used in designing experiments by using cDNA microarrays, particularly when gene expression data from single specimens are being analyzed.

DNase I-hypersensitive sites I and II of the human growth hormone locus control region are a major developmental activator of somatotrope gene expression

High-level expression of the human growth hormone (hGH) gene is limited to somatotrope and lactosomatotrope cells of the anterior pituitary. We previously identified a locus control region (LCR) for the hGH gene composed of four tissue-specific DNase I-hypersensitive sites (HS) located between −14.6 kb and −32 kb 5′ to the hGH transcription start site that is responsible for establishing a physiologically regulated chromatin domain for hGH transgene expression in mouse pituitary. In the present study we demonstrated that the LCR mediates somatotrope and lactosomatotrope restriction on an otherwise weakly and diffusely expressed hGH transgene. The subregion of the LCR containing the two pituitary-specific HS, HSI and HSII (−14.6 to −16.2 kb relative to the hGH promoter and denoted HSI,II), was found to be sufficient for mediating somatotrope and lactosomatotrope restriction, for appropriately timed induction of hGH transgene expression between embryonic days 15.5 and 16.5, and for selective extinction of hGH expression in mature lactotropes. When studied by cell transfection, the HSI,II fragment selectively enhanced transcription in a presomatotrope-derived cell line, although at levels (2- to 3-fold) well below that seen in vivo. The LCR activity of the HSI,II element was therefore localized by scoring transgene expression in fetal founder pituitaries at embryonic day 18.5. The data from these studies indicated that a 404-bp segment of the HSI,II region encodes a critical subset of LCR functions, including the establishment of a productive chromatin environment, cell-specific restriction and enhancement of expression, and appropriately timed induction of the hGH transgene during embryonic development.

Modeling stochastic gene expression: Implications for haploinsufficiency

There is increasing recognition that stochastic processes regulate highly predictable patterns of gene expression in developing organisms, but the implications of stochastic gene expression for understanding haploinsufficiency remain largely unexplored. We have used simulations of stochastic gene expression to illustrate that gene copy number and expression deactivation rates are important variables in achieving predictable outcomes. In gene expression systems with non-zero expression deactivation rates, diploid systems had a higher probability of uninterrupted gene expression than haploid systems and were more successful at maintaining gene product above a very low threshold. Systems with relatively rapid expression deactivation rates (unstable gene expression) had more predictable responses to a gradient of inducer than systems with slow or zero expression deactivation rates (stable gene expression), and diploid systems were more predictable than haploid, with or without dosage compensation. We suggest that null mutations of a single allele in a diploid organism could decrease the probability of gene expression and present the hypothesis that some haploinsufficiency syndromes might result from an increased susceptibility to stochastic delays of gene initiation or interruptions of gene expression.

Altered gene expression in the host brain caused by a trematode parasite: Neuropeptide genes are preferentially affected during parasitosis

Schistosome parasites adjust the physiology and behavior of their intermediate molluscan hosts to their own benefit. Previous studies demonstrated effects of the avian-schistosome Trichobilharzia ocellata on peptidergic centers in the brain of the intermediate snail host Lymnaea stagnalis. In particular, electrophysiological properties and peptide release of growth- and reproduction-controlling neuroendocrine neurons were affected. We now have examined the possibility that the expression of genes that control physiology and behavior of the host might be altered during parasitosis. A cDNA library of the brain of parasitized Lymnaea was constructed and differentially screened by using mRNA from the brain of both parasitized and nonparasitized snails. This screening yielded a number of clones, including previously identified cDNAs as well as novel neuronal transcripts, which appear to be differentially regulated. The majority of these transcripts encode neuropeptides. Reverse Northern blot analysis confirmed that neuropeptide gene expression is indeed affected in parasitized animals. Moreover, the expression profiles of 10 transcripts tested showed a differential, parasitic stage-specific regulation. Changes in expression could in many cases already be observed between 1.5 and 5 hr postinfection, suggesting that changes in gene expression are a direct effect of parasitosis. We suggest that direct regulation of neuropeptide gene expression is a strategy of parasites to induce physiological and behavioral changes in the host.

Dynamic modeling of gene expression data

We describe the time evolution of gene expression levels by using a time translational matrix to predict future expression levels of genes based on their expression levels at some initial time. We deduce the time translational matrix for previously published DNA microarray gene expression data sets by modeling them within a linear framework by using the characteristic modes obtained by singular value decomposition. The resulting time translation matrix provides a measure of the relationships among the modes and governs their time evolution. We show that a truncated matrix linking just a few modes is a good approximation of the full time translation matrix. This finding suggests that the number of essential connections among the genes is small.

Mechanism for fetal globin gene expression: Role of the soluble guanylate cyclase–cGMP-dependent protein kinase pathway

Despite considerable concerns with pharmacological stimulation of fetal hemoglobin (Hb F) as a therapeutic option for the β-globin disorders, the molecular basis of action of Hb F-inducing agents remains unclear. Here we show that an intracellular pathway including soluble guanylate cyclase (sGC) and cGMP-dependent protein kinase (PKG) plays a role in induced expression of the γ-globin gene. sGC, an obligate heterodimer of α- and β-subunits, participates in a variety of physiological processes by converting GTP to cGMP. Northern blot analyses with erythroid cell lines expressing different β-like globin genes showed that, whereas the β-subunit is expressed at similar levels, high-level expression of the α-subunit is preferentially observed in erythroid cells expressing γ-globin but not those expressing β-globin. Also, the levels of expression of the γ-globin gene correlate to those of the α-subunit. sGC activators or cGMP analogs increased expression of the γ-globin gene in erythroleukemic cells as well as in primary erythroblasts from normal subjects and patients with β-thalassemia. Nuclear run-off assays showed that the sGC activator protoporphyrin IX stimulates transcription of the γ-globin gene. Furthermore, increased expression of the γ-globin gene by well known Hb F-inducers such as hemin and butyrate was abolished by inhibiting sGC or PKG activity. Taken together, these results strongly suggest that the sGC–PKG pathway constitutes a mechanism that regulates expression of the γ-globin gene. Further characterization of this pathway should permit us to develop new therapeutics for the β-globin disorders.

The effects of aging on gene expression in the hypothalamus and cortex of mice

A better understanding of the molecular effects of aging in the brain may help to reveal important aspects of organismal aging, as well as processes that lead to age-related brain dysfunction. In this study, we have examined differences in gene expression in the hypothalamus and cortex of young and aged mice by using high-density oligonucleotide arrays. A number of key genes involved in neuronal structure and signaling are differentially expressed in both the aged hypothalamus and cortex, including synaptotagmin I, cAMP-dependent protein kinase C β, apolipoprotein E, protein phosphatase 2A, and prostaglandin D. Misregulation of these proteins may contribute to age-related memory deficits and neurodegenerative diseases. In addition, many proteases that play essential roles in regulating neuropeptide metabolism, amyloid precursor protein processing, and neuronal apoptosis are up-regulated in the aged brain and likely contribute significantly to brain aging. Finally, a subset of these genes whose expression is affected by aging are oppositely affected by exposure of mice to an enriched environment, suggesting that these genes may play important roles in learning and memory.

Hypoxia-induced gene expression profiling in the euryoxic fish Gillichthys mirabilis

Hypoxia is important in both biomedical and environmental contexts and necessitates rapid adaptive changes in metabolic organization. Mammals, as air breathers, have a limited capacity to withstand sustained exposure to hypoxia. By contrast, some aquatic animals, such as certain fishes, are routinely exposed and resistant to severe environmental hypoxia. Understanding the changes in gene expression in fishes exposed to hypoxic stress could reveal novel mechanisms of tolerance that may shed new light on hypoxia and ischemia in higher vertebrates. Using cDNA microarrays, we have studied gene expression in a hypoxia-tolerant burrow-dwelling goby fish, Gillichthys mirabilis. We show that a coherent picture of a complex transcriptional response can be generated for a nonmodel organism for which sequence data were unavailable. We demonstrate that: (i) although certain shifts in gene expression mirror changes in mammals, novel genes are differentially expressed in fish; and (ii) tissue-specific patterns of expression reflect the different metabolic roles of tissues during hypoxia.

Plastid redox state and sugars: Interactive regulators of nuclear-encoded photosynthetic gene expression

Feedback regulation of photosynthesis by carbon metabolites has long been recognized, but the underlying cellular mechanisms that control this process remain unclear. By using an Arabidopsis cell culture, we show that a block in photosynthetic electron flux prevents the increase in transcript levels of chlorophyll a/b-binding protein and the small subunit of Rubisco that typically occurs when intracellular sugar levels are depleted. In contrast, the expression of the nitrate reductase gene, which is induced by sugars, is not affected. These findings were confirmed in planta by using Arabidopsis carrying the firefly luciferase reporter gene fused to the plastocyanin and chlorophyll a/b-binding protein 2 gene promoters. Transcription from both promoters increases on carbohydrate depletion. Blocking photosynthetic electron transport with 3-(3′, 4′-dichlorophenyl)-1,1′-dimethylurea prevents this increase in transcription. We conclude that plastid-derived redox signaling can override the sugar-regulated expression of nuclear-encoded photosynthetic genes. In the sugar-response mutant, sucrose uncoupled 6 (sun6), plastocyanin-firefly luciferase transcription actually increases in response to exogenous sucrose rather than decreasing as in the wild type. Interestingly, plastid-derived redox signals do not influence this defective pattern of sugar-regulated gene expression in the sun6 mutant. A model, which invokes a positive inducer originating from the photosynthetic electron transport chain, is proposed to explain the nature of the plastid-derived signal.

Regulation of c-maf gene expression by Pax6 in cultured cells

c-Maf is a bZip transcription factor expressed in developmental and cellular differentiation processes. Recently, a c-maf knockout mouse model, showing abnormal lens development, has been reported. In order to study the regulation mechanisms of c-maf gene expression during the differentiation process we have cloned and functionally characterized the rat c-maf (maf-2) gene. The rat c-maf gene is an intronless gene, covering a length of 3.5 kb. Transient transfection analysis of the 5′-flanking region of the c-maf gene using luciferase as the reporter gene shows that Pax6, a master transcription factor for lens development, strongly activates the c-maf promoter construct. Endogenous c-maf is also activated by the Pax6 expression vector. Electrophoresis mobility shift assay and DNase I footprinting analysis show that at least three Pax6-binding sites are located in the 5′-flanking and 5′-non-coding regions of the rat c-maf gene. The c-maf gene was also markedly activated by its own product, c-Maf, through the MARE (Maf recognition element), suggesting that a positive autoregulatory mechanism controls this gene. In situ hybridization histochemical detection of Pax6 and c-Maf in the E14 lens showed that both mRNAs are expressed in the lens equator where lens epithelial cells are differentiating to lens fiber cells. These results suggest that a Pax6/c-Maf transcription factor cascade is working in lens development.

Saccharomyces Genome Database provides tools to survey gene expression and functional analysis data

Upon the completion of the Saccharomyces cerevisiae genomic sequence in 1996 [Goffeau,A. et al. (1997) Nature, 387, 5], several creative and ambitious projects have been initiated to explore the functions of gene products or gene expression on a genome-wide scale. To help researchers take advantage of these projects, the Saccharomyces Genome Database (SGD) has created two new tools, Function Junction and Expression Connection. Together, the tools form a central resource for querying multiple large-scale analysis projects for data about individual genes. Function Junction provides information from diverse projects that shed light on the role a gene product plays in the cell, while Expression Connection delivers information produced by the ever-increasing number of microarray projects. WWW access to SGD is available at genome-www.stanford.edu/Saccharomyces/.

The TRANSFAC system on gene expression regulation

The TRANSFAC database on transcription factors and their DNA-binding sites and profiles (http://www.gene-regulation.de/) has been quantitatively extended and supplemented by a number of modules. These modules give information about pathologically relevant mutations in regulatory regions and transcription factor genes (PathoDB), scaffold/matrix attached regions (S/MARt DB), signal transduction (TRANSPATH) and gene expression sources (CYTOMER). Altogether, these distinct database modules constitute the TRANSFAC system. They are accompanied by a number of program routines for identifying potential transcription factor binding sites or for localizing individual components in the regulatory network of a cell.

The Mouse Gene Expression Database (GXD)

The Gene Expression Database (GXD) is a community resource of gene expression information for the laboratory mouse. By combining the different types of expression data, GXD aims to provide increasingly complete information about the expression profiles of genes in different mouse strains and mutants, thus enabling valuable insights into the molecular networks that underlie normal development and disease. GXD is integrated with the Mouse Genome Database (MGD). Extensive interconnections with sequence databases and with databases from other species, and the development and use of shared controlled vocabularies extend GXD’s utility for the analysis of gene expression information. GXD is accessible through the Mouse Genome Informatics web site at http://www.informatic s.jax.org/ or directly at http://www.informatics.jax.org/me nus/expression_menu.shtml.

RECODE: a database of frameshifting, bypassing and codon redefinition utilized for gene expression

The RECODE database is a compilation of ‘programmed’ translational recoding events taken from the scientific literature and personal communications. The database deals with programmed ribosomal frameshifting, codon redefinition and translational bypass occurring in a variety of organisms. The entries for each event include the sequences of the corresponding genes, their encoded proteins for both the normal and alternate decoding, the types of the recoding events involved, trans-factors and cis-elements that influence recoding. The database is freely available at http://recode.genetics.utah.edu/.