Selection bias in gene extraction on the basis of microarray gene-expression data

In the context of cancer diagnosis and treatment, we consider the problem of constructing an accurate prediction rule on the basis of a relatively small number of tumor tissue samples of known type containing the expression data on very many (possibly thousands) genes. Recently, results have been presented in the literature suggesting that it is possible to construct a prediction rule from only a few genes such that it has a negligible prediction error rate. However, in these results the test error or the leave-one-out cross-validated error is calculated without allowance for the selection bias. There is no allowance because the rule is either tested on tissue samples that were used in the first instance to select the genes being used in the rule or because the cross-validation of the rule is not external to the selection process; that is, gene selection is not performed in training the rule at each stage of the cross-validation process. We describe how in practice the selection bias can be assessed and corrected for by either performing a cross-validation or applying the bootstrap external to the selection process. We recommend using 10-fold rather than leave-one-out cross-validation, and concerning the bootstrap, we suggest using the so-called. 632+ bootstrap error estimate designed to handle overfitted prediction rules. Using two published data sets, we demonstrate that when correction is made for the selection bias, the cross-validated error is no longer zero for a subset of only a few genes.

On the simultaneous use of clinical and microarray expression data in the cluster analysis of tissue samples

This paper considers a model-based approach to the clustering of tissue samples of a very large number of genes from microarray experiments. It is a nonstandard problem in parametric cluster analysis because the dimension of the feature space (the number of genes) is typically much greater than the number of tissues. Frequently in practice, there are also clinical data available on those cases on which the tissue samples have been obtained. Here we investigate how to use the clinical data in conjunction with the microarray gene expression data to cluster the tissue samples. We propose two mixture model-based approaches in which the number of components in the mixture model corresponds to the number of clusters to be imposed on the tissue samples. One approach specifies the components of the mixture model to be the conditional distributions of the microarray data given the clinical data with the mixing proportions also conditioned on the latter data. Another takes the components of the mixture model to represent the joint distributions of the clinical and microarray data. The approaches are demonstrated on some breast cancer data, as studied recently in van't Veer et al. (2002).

Model-based clustering in gene expression microarrays: An application to breast cancer data

In microarray studies, the application of clustering techniques is often used to derive meaningful insights into the data. In the past, hierarchical methods have been the primary clustering tool employed to perform this task. The hierarchical algorithms have been mainly applied heuristically to these cluster analysis problems. Further, a major limitation of these methods is their inability to determine the number of clusters. Thus there is a need for a model-based approach to these. clustering problems. To this end, McLachlan et al. [7] developed a mixture model-based algorithm (EMMIX-GENE) for the clustering of tissue samples. To further investigate the EMMIX-GENE procedure as a model-based -approach, we present a case study involving the application of EMMIX-GENE to the breast cancer data as studied recently in van 't Veer et al. [10]. Our analysis considers the problem of clustering the tissue samples on the basis of the genes which is a non-standard problem because the number of genes greatly exceed the number of tissue samples. We demonstrate how EMMIX-GENE can be useful in reducing the initial set of genes down to a more computationally manageable size. The results from this analysis also emphasise the difficulty associated with the task of separating two tissue groups on the basis of a particular subset of genes. These results also shed light on why supervised methods have such a high misallocation error rate for the breast cancer data.

Using biplots to interpret gene expression patterns in plants

Sum: Plant biologists in fields of ecology, evolution, genetics and breeding frequently use multivariate methods. This paper illustrates Principal Component Analysis (PCA) and Gabriel's biplot as applied to microarray expression data from plant pathology experiments. Availability: An example program in the publicly distributed statistical language R is available from the web site (www.tpp.uq.edu.au) and by e-mail from the contact. Contact: scott.chapman@csiro.au.

Low prenatal vitamin D alters morphology, cell density and gene expression in adult rat brain: Correlations with schizophrenia

Statement of the study: Based on data from ecological and analytic epidemiological studies, we have proposed that low prenatal vitamin D is a candidate risk-modifying factor for schizophrenia. Previously, we demonstrated that low prenatal vitamin D adversely affected brain development in neonatal rats (Eyles et al, 2003). Here we examine the impact of both prenatal and early life hypovitaminosis D on various outcomes in the adult rat brain. Methods: Female Sprague-Dawley rats were made vitamin D deficient via the use of a special diet (Dyets CA) and lighting conditions that excluded UVB radiation. Animals were kept under these conditions for 6 weeks then mated with males kept under normal conditions. Vitamin deplete dams were kept under these conditions during pregnancy. Offspring from two test groups were examined. Offspring were either reared with dams repleted with vitamin D at birth or remained under deplete conditions till weaning. Both test groups were weaned under normal vitamin D conditions and remained so till testing at adulthood. We compared the brains of adult offspring kept under both test conditions with animals from control environments. Summary of results: We found a significant persistent dose-related increase in lateral ventricle volume and alterations in anterior cingulate and prefrontal cortical cell densities (consistent with the known prodifferentiation properties of this steroid). In both test groups we observed a reduced expression of NGF as well as a down-regulation of transcripts coding for GABAA alpha 4 receptor and two neuronal structural elements; MAP2 and Neurofilament L. Conclusion: These findings provide further evidence that vitamin D is involved in brain development. An increase in prefrontal cortical cell density, a reduction neuronal structural elements and persistent ventriculomegaly are all common anatomical findings in the brains of patients with schizophrenia. The specific reduction in transcripts for neuronal structural proteins but not GFAP is also in accordance with the proposal that frontal cortical architecture in schizophrenia reflects a reduction in connectivity rather than a reduction in glial processes(Goldman-Rakic and Selemon, 1997). These findings confirm the biological plausibility of early life hypovitaminosis D as a risk factor for schizophrenia.

Growth factors and cytokines modulate gene expression of cell-surface proteoglycans in human periodontal ligament cells

Cell-surface proteoglycans have been known to be involved in many functions including interactions with components of the extracellular microenvironment, and act as co-receptors which bind and modify the action of various growth factors and cytokines. The purpose of this study was to determine the regulation by growth factors and cytokines on cell-surface proteoglycan gene expression in cultured human periodontal ligament (PDL) cells. Subconfluent, quiescent PDL cells were treated with various concentrations of serum, bFGF, PDGF-BB, TGF-beta1, IL-1 beta, and IFN-gamma. RT-PCR technique was used, complemented with Northern blot for syndecan-1, to examine the effects of these agents on the mRNA expression of five cell-surface proteoglycans (syndecan-1, syndecan-2, syndecan-4, glypican and betaglycan). Syndecan-1 mRNA levels increased in response to serum, bFGF and PDGF-BB, but decreased in response to TGF-beta1, IL-1 beta and IFN-gamma. In contrast, syndecan-2 mRNA levels were upregulated by TCF-beta1 and IL-1 beta stimulation, but remained unchanged with the other agents. Betaglycan gene expression decreased in response to serum, but was upregulated by TCF-beta1 and unchanged by the other stimulants. Additionally, syndecan-4 and glypican were not significantly altered in response to the regulator molecules studied, with the exception that glypican is decreased in response to IFN-gamma. These data demonstrate that the gene expression of the five cell-surface proteoglycans studied is differentially regulated in PDL cells lending support to the nation of distinct functions for these cell-surface proteoglycans. (C) 2001 Wiley-Liss, inc.

Alterations in gene expression and activity during squamous cell carcinoma development

This study focuses on characterizing the genetic and biological alterations associated with squamous cell carcinoma development. Normal human epidermal keratinocytes (HEKs), cells isolated from a preneoplastic lesion (IEC-1), and two neoplastic cell lines, SCC-25 and COLD-16, were grown as raft cultures, and their gene expression profiles were screened using cDNA arrays. Our data indicated that the expression levels of at least 37 genes were significantly (P less than or equal to 0.05; 1.9% of genes screened) altered in neoplastic cells compared with normal cells. Of these genes, 10 genes were up-regulated and 27 genes were down-regulated in the neoplastic cells. In addition, 51% of the genes altered in the neoplastic cells were already altered in the preneoplastic IEC-1 cells. Immunohistochemical staining of patient tumors was used to verify the cDNA array analysis. Our analysis indicated that alterations in genes associated with extracellular matrix production and apoptosis are disrupted in preneoplastic cells, whereas later stages of neoplasia are associated with alterations in gene expression for genes involved in DNA repair or epidermal growth factor (EGF) receptor/mitogen-activated protein kinase kinase (MAPKK)/MAPK/activator protein-1 (AP-1) signaling. Subsequent functional analysis of the alterations in expression of the EGF receptor/MAPKK/MAPK/AP-1 genes suggested they did not contribute to the neoplastic phenotype.

A mixture model-based approach to the clustering of microarray expression data

Motivation: This paper introduces the software EMMIX-GENE that has been developed for the specific purpose of a model-based approach to the clustering of microarray expression data, in particular, of tissue samples on a very large number of genes. The latter is a nonstandard problem in parametric cluster analysis because the dimension of the feature space (the number of genes) is typically much greater than the number of tissues. A feasible approach is provided by first selecting a subset of the genes relevant for the clustering of the tissue samples by fitting mixtures of t distributions to rank the genes in order of increasing size of the likelihood ratio statistic for the test of one versus two components in the mixture model. The imposition of a threshold on the likelihood ratio statistic used in conjunction with a threshold on the size of a cluster allows the selection of a relevant set of genes. However, even this reduced set of genes will usually be too large for a normal mixture model to be fitted directly to the tissues, and so the use of mixtures of factor analyzers is exploited to reduce effectively the dimension of the feature space of genes. Results: The usefulness of the EMMIX-GENE approach for the clustering of tissue samples is demonstrated on two well-known data sets on colon and leukaemia tissues. For both data sets, relevant subsets of the genes are able to be selected that reveal interesting clusterings of the tissues that are either consistent with the external classification of the tissues or with background and biological knowledge of these sets.

Tissue-specific gene expression in soybean (Glycine max) detected by cDNA microarray analysis

We have constructed cDNA microarrays for soybean (Glycine max L. Merrill), containing approximately 4,100 Unigene ESTs derived from axenic roots, to evaluate their application and utility for functional genomics of organ differentiation in legumes. We assessed microarray technology by conducting studies to evaluate the accuracy of microarray data and have found them to be both reliable and reproducible in repeat hybridisations. Several ESTs showed high levels (>50 fold) of differential expression in either root or shoot tissue of soybean. A small number of physiologically interesting, and differentially expressed sequences found by microarray analysis were verified by both quantitative real-time RT-PCR and Northern blot analysis. There was a linear correlation (r(2) = 0.99, over 5 orders of magnitude) between microarray and quantitative real-time RT-PCR data. Microarray analysis of soybean has enormous potential not only for the discovery of new genes involved in tissue differentiation and function, but also to study the expression of previously characterised genes, gene networks and gene interactions in wild-type, mutant or transgenic; plants.

Effects of A1 adenosine receptor overexpression on normoxic and post-ischemic gene expression

Objectives: To identify potential molecular genetic determinants of cardiovascular ischemic tolerance in wild-type and transgenic hearts overexpressing A(1) adenosine receptors (A(1)ARs). Methods: cDNA microarrays were used to explore expression of 1824 genes ill wild-type hearts and ischemia-tolerant mouse hearts overexpressing A(1)ARs. Results: Overexpression of A(1)ARs reduced post-ischemic contractile dysfunction, limited arrhythmogenesis, and reduced necrosis by similar to80% in hearts subjected to 30 min global ischemia 60 mill reperfusion. Cardioprotection was abrogated by acute A(1)AR antagonism, and only a small number (19) of genes were modified by A(1)AR overexpression in normoxic hearts. Ischemia-reperfusion significantly altered expression of 75 genes in wild-type hearts (14 induced, 61 down-regulated), including genes for metabolic enzymes, structural/motility proteins, cell signaling proteins, defense/growth proteins, and regulators of transcription and translation. A(1)AR overexpression reversed the majority of gene down-regulation whereas gene induction was generally unaltered. Additionally, genes involved in cell defence, signaling and gene expression were selectively modified by ischemia in transgenic hearts (33 induced, 10 down-regulated), possibly contributing to the protected phenotype. Real-time PCR verified changes in nine selected genes, revealing concordance with array data. Transcription of the A(1)AR gene was also modestly reduced post-ischemia, consistent with impaired functional sensitivity to A(1)AR stimulation Conclusions: Data are presented regarding the early post-ischemic gene profile of intact heart. Reduced A(1)AR transcription is observed which may contribute to poor outcome from ischemia. A(1)AR overexpression selectively modifies post-ischemic gene expression, potentially contributing to ischemic-tolerance. (C) 2003 European Society of Cardiology. Published by Elsevier Science B.V. All rights reserved.

Engineering mRNA translation initiation to enhance transient gene expression in Chinese hamster ovary cells

To increase transient expression of recombinant proteins in Chinese hamster ovary cells, we have engineered their protein synthetic capacity by directed manipulation of mRNA translation initiation. To control this process we constructed a nonphosphorylatable Ser51Ala site-directed mutant of eIF2, a subunit of the trimeric eIF2 complex that is implicated in regulation of the global rate of mRNA translation initiation in eukaryotic cells. Phosphorylation of eIF2 by protein kinases inhibits eIF2 activity and is known to increase as cells perceive a range of stress conditions. Using single-and dual-gene plasmids introduced into CHO cells by electroporation, we found that transient expression of the eIF2 Ser51Ala mutant with firefly luciferase resulted in a 3-fold increase in reporter activity, relative to cells transfected with reporter only. This effect was maintained in transfected cells for at least 48 h after transfection. Expression of the wild-type eIF2 protein had no such effect. Elevated luciferase activity was associated with a reduction in the level of eIF2 phosphorylation in cells transfected with the mutant eIF2 construct. Transfection of CHO cells with the luciferase-only construct resulted in a marked decrease in the global rate of protein synthesis in the whole cell population 6 h post-transfection. However, expression of the mutant Ser51Ala or wild-type eIF2 proteins restored the rate of protein synthesis in transfected cells to a level equivalent to or exceeding that of control cells. Associated with this, entry of plasmid DNA into cells during electroporation was visualized by confocal microscopy using a rhodamine-labeled plasmid construct expressing green fluorescent protein. Six hours after transfection, plasmid DNA was present in all cells, albeit to a variable extent. These data suggest that entry of naked DNA into the cell itself functions to inhibit protein synthesis by signaling mechanisms affecting control of mRNA translation by eIF2. This work therefore forms the basis of a rational strategy to generically up-regulate transient expression of recombinant proteins by simultaneous host cell engineering.

Genetic study of alcoholism and novel gene expression in the alcoholic brain

Alcohol dependence may result from neuroadaptation involving alteration of gene expression after long-term alcohol exposure. The systematic study of gene expression profiles of the human alcoholic brain was initiated using the method of polymerase chain reaction (PCR)-differential display and was followed by DNA microarray. To date, more than 100 alcohol-responsive genes have been identified from the frontal cortex, motor cortex and nucleus accumbens of the human brain. These genes have a wide range of functions in the brain and indicate diverse actions of alcohol on neuronal function. This review discusses the current information on the genetic basis of alcoholism and the induction and characterization of these alcohol-responsive genes.

Correlating gene expression with larval competence and the effect of age and parentage on metamorphosis in the tropical abalone Haliotis asinina

The non-geniculate crustose coralline alga (CCA) Mastophora pacifica can induce the metamorphosis of competent Haliotis asinina (Vetigastropoda) larvae. The ability to respond to this natural cue varies considerably with larval age, with a higher proportion of older larvae (e.g. 90 h) able to metamorphose in response to M. pacifica than younger larvae (e.g. 66 h). Here we document the variation in time to acquisition of competence within a larval age class. For example, after 18 h of exposure to M. pacifica, approximately 15 and 36% of 84 and 90-h-old H. asinina larvae had initiated metamorphosis, respectively. This age-dependent response to M. pacifica is also observed when different aged larvae are exposed to CCA for varying periods. A higher proportion of older larvae require shorter periods of exposure to CCA than younger larvae in order to initiate metamorphosis. In this experiment, as in the previous, a small proportion of young larvae were able to respond to brief periods of CCA exposure, suggesting that they had developed the same state of competency as the majority of their older counterparts. Comparisons of the proportions of larvae undergoing metamorphosis between families reveals that parentage also has a significant (P < 0.05) affect on whether an individual will initiate metamorphosis at a given age. These familial differences are more pronounced when younger, largely pre-competent larvae (i.e. 66 h old) are exposed to M. pacifica, with proportions of larvae undergoing metamorphosis differing by as much as 10 fold between families. As these data suggest that variation in the rate of development of the competent state has a genetic basis, and as a first step towards identifying the molecular basis to this variation, we have identified numerous genes that are differentially expressed later in larval development using a differential display approach. Spatial expression analysis of these genes suggests that they may be directly involved in the acquisition of competence, or may play a functional role in the postlarva following metamorphosis.