Knowledge-based analysis of microarray gene expression data by using support vector machines

We introduce a method of functionally classifying genes by using gene expression data from DNA microarray hybridization experiments. The method is based on the theory of support vector machines (SVMs). SVMs are considered a supervised computer learning method because they exploit prior knowledge of gene function to identify unknown genes of similar function from expression data. SVMs avoid several problems associated with unsupervised clustering methods, such as hierarchical clustering and self-organizing maps. SVMs have many mathematical features that make them attractive for gene expression analysis, including their flexibility in choosing a similarity function, sparseness of solution when dealing with large data sets, the ability to handle large feature spaces, and the ability to identify outliers. We test several SVMs that use different similarity metrics, as well as some other supervised learning methods, and find that the SVMs best identify sets of genes with a common function using expression data. Finally, we use SVMs to predict functional roles for uncharacterized yeast ORFs based on their expression data.

Generation of longer cDNA fragments from serial analysis of gene expression tags for gene identification

We have developed a technique called the generation of longer cDNA fragments from serial analysis of gene expression (SAGE) tags for gene identification (GLGI), to convert SAGE tags of 10 bases into their corresponding 3′ cDNA fragments covering hundred bases. A primer containing the 10-base SAGE tag is used as the sense primer, and a single base anchored oligo(dT) primer is used as an antisense primer in PCR, together with Pfu DNA polymerase. By using this approach, a cDNA fragment extending from the SAGE tag toward the 3′ end of the corresponding sequence can be generated. Application of the GLGI technique can solve two critical issues in applying the SAGE technique: one is that a longer fragment corresponding to a SAGE tag, which has no match in databases, can be generated for further studies; the other is that the specific fragment corresponding to a SAGE tag can be identified from multiple sequences that match the same SAGE tag. The development of the GLGI method provides several potential applications. First, it provides a strategy for even wider application of the SAGE technique for quantitative analysis of global gene expression. Second, a combined application of SAGE/GLGI can be used to complete the catalogue of the expressed genes in human and in other eukaryotic species. Third, it can be used to identify the 3′ cDNA sequence from any exon within a gene. It can also be used to confirm the reality of exons predicted by bioinformatic tools in genomic sequences. Fourth, a combined application of SAGE/GLGI can be applied to define the 3′ boundary of expressed genes in the genomic sequences in human and in other eukaryotic genomes.

Gene expression, synthesis, and secretion of interleukin 18 and interleukin 1β are differentially regulated in human blood mononuclear cells and mouse spleen cells

Interleukin (IL)-18, formerly called interferon γ (IFN-γ)-inducing factor, is biologically and structurally related to IL-1β. A comparison of gene expression, synthesis, and processing of IL-18 with that of IL-1β was made in human peripheral blood mononuclear cells (PBMCs) and in human whole blood. Similar to IL-1β, the precursor for IL-18 requires processing by caspase 1. In PBMCs, mature but not precursor IL-18 induces IFN-γ; in whole human blood stimulated with endotoxin, inhibition of caspase 1 reduces IFN-γ production by an IL-1β-independent mechanism. Unlike the precursor for IL-1β, precursor for IL-18 was expressed constitutively in PBMCs and in fresh whole blood from healthy human donors. Western blotting of endotoxin-stimulated PBMCs revealed processed IL-1β in the supernatants via an caspase 1-dependent pathway. However, in the same supernatants, only unprocessed precursor IL-18 was found. Unexpectedly, precursor IL-18 was found in freshly obtained PBMCs and constitutive IL-18 gene expression was present in whole blood of healthy donors, whereas constitutive IL-1β gene expression is absent. Similar to human PBMCs, mouse spleen cells also constitutively contained the preformed precursor for IL-18 and expressed steady-state IL-18 mRNA, but there was no IL-1β protein and no spontaneous gene expression for IL-1β in these same preparations. We conclude that although IL-18 and IL-1β are likely members of the same family, constitutive gene expression, synthesis, and processing are different for the two cytokines.

Imaging adenoviral-directed reporter gene expression in living animals with positron emission tomography

We are developing quantitative assays to repeatedly and noninvasively image expression of reporter genes in living animals, using positron emission tomography (PET). We synthesized positron-emitting 8-[18F]fluoroganciclovir (FGCV) and demonstrated that this compound is a substrate for the herpes simplex virus 1 thymidine kinase enzyme (HSV1-TK). Using positron-emitting FGCV as a PET reporter probe, we imaged adenovirus-directed hepatic expression of the HSV1-tk reporter gene in living mice. There is a significant positive correlation between the percent injected dose of FGCV retained per gram of liver and the levels of hepatic HSV1-tk reporter gene expression (r2 > 0.80). Over a similar range of HSV1-tk expression in vivo, the percent injected dose retained per gram of liver was 0–23% for ganciclovir and 0–3% for FGCV. Repeated, noninvasive, and quantitative imaging of PET reporter gene expression should be a valuable tool for studies of human gene therapy, of organ/cell transplantation, and of both environmental and behavioral modulation of gene expression in transgenic mice.

A binding site for homeodomain and Pax proteins is necessary for L1 cell adhesion molecule gene expression by Pax-6 and bone morphogenetic proteins

The cell adhesion molecule L1 regulates axonal guidance and fasciculation during development. We previously identified the regulatory region of the L1 gene and showed that it was sufficient for establishing the neural pattern of L1 expression in transgenic mice. In the present study, we characterize a DNA element within this region called the HPD that contains binding motifs for both homeodomain and Pax proteins and responds to signals from bone morphogenetic proteins (BMPs). An ATTA sequence within the core of the HPD was required for binding to the homeodomain protein Barx2 while a separate paired domain recognition motif was necessary for binding to Pax-6. In cellular transfection experiments, L1-luciferase reporter constructs containing the HPD were activated an average of 4-fold by Pax-6 in N2A cells and 5-fold by BMP-2 and BMP-4 in Ng108 cells. Both of these responses were eliminated on deletion of the HPD from L1 constructs. In transgenic mice, deletion of the HPD from an L1-lacZ reporter resulted in a loss of β-galactosidase expression in the telencephalon and mesencephalon. Collectively, our experiments indicate that the HPD regulates L1 expression in neural tissues via homeodomain and Pax proteins and is likely to be a target of BMP signaling during development.

Overexpression of a Shaker-type potassium channel in mammalian central nervous system dysregulates native potassium channel gene expression

The nervous system maintains a delicate balance between excitation and inhibition, partly through the complex interplay between voltage-gated sodium and potassium ion channels. Because K+ channel blockade or gene deletion causes hyperexcitability, it is generally assumed that increases in K+ channel gene expression should reduce neuronal network excitability. We have tested this hypothesis by creating a transgenic mouse that expresses a Shaker-type K+ channel gene. Paradoxically, we find that addition of the extra K+ channel gene results in a hyperexcitable rather than a hypoexcitable phenotype. The presence of the transgene leads to a complex deregulation of endogenous Shaker genes in the adult central nervous system as well as an increase in network excitability that includes spontaneous cortical spike and wave discharges and a lower threshold for epileptiform bursting in isolated hippocampal slices. These data suggest that an increase in K+ channel gene dosage leads to dysregulation of normal K+ channel gene expression, and it may underlie a mechanism contributing to the pathogenesis of human aneuploidies such as Down syndrome.

Male gametic cell-specific gene expression in flowering plants

The role of the male gamete—the sperm cell—in the process of fertilization is to recognize, adhere to, and fuse with the female gamete. These highly specialized functions are expected to be controlled by activation of a unique set of genes. However, male gametic cells traditionally have been regarded as transcriptionally quiescent because of highly condensed chromatin and a very reduced amount of cytoplasm. Here, we provide evidence for male gamete-specific gene expression in flowering plants. We identified and characterized a gene, LGC1, which was shown to be expressed exclusively in the male gametic cells. The gene product of LGC1 was localized at the surface of male gametic cells, suggesting a possible role in sperm–egg interactions. These findings represent an important step toward defining the molecular mechanisms of male gamete development and the cellular processes involved in fertilization of flowering plants.

Toward functional genomics in bacteria: Analysis of gene expression in Escherichia coli from a bacterial artificial chromosome library of Bacillus cereus

As the study of microbes moves into the era of functional genomics, there is an increasing need for molecular tools for analysis of a wide diversity of microorganisms. Currently, biological study of many prokaryotes of agricultural, medical, and fundamental scientific interest is limited by the lack of adequate genetic tools. We report the application of the bacterial artificial chromosome (BAC) vector to prokaryotic biology as a powerful approach to address this need. We constructed a BAC library in Escherichia coli from genomic DNA of the Gram-positive bacterium Bacillus cereus. This library provides 5.75-fold coverage of the B. cereus genome, with an average insert size of 98 kb. To determine the extent of heterologous expression of B. cereus genes in the library, we screened it for expression of several B. cereus activities in the E. coli host. Clones expressing 6 of 10 activities tested were identified in the library, namely, ampicillin resistance, zwittermicin A resistance, esculin hydrolysis, hemolysis, orange pigment production, and lecithinase activity. We analyzed selected BAC clones genetically to identify rapidly specific B. cereus loci. These results suggest that BAC libraries will provide a powerful approach for studying gene expression from diverse prokaryotes.

Phosphate is a specific signal for induction of osteopontin gene expression

Osteopontin is a phosphorylated glycoprotein secreted to the mineralizing extracellular matrix by osteoblasts during bone development. It is believed to facilitate the attachment of osteoblasts and osteoclasts to the extracellular matrix, allowing them to perform their respective functions during osteogenesis. Several other functions have been suggested for this protein, and its up-regulation is associated with various disease states related to calcification, including arterial plaque formation and the formation of kidney stones. Although expression of this gene has been demonstrated in multiple tissues, its regulation is not well understood. Our previous studies on the roles of the retinoblastoma protein (pRB) and p300/CBP in the regulation of osteoblast differentiation revealed a link between osteopontin induction and the synthesis of alkaline phosphatase. In this paper, we describe results specifically linking induction of osteopontin to the enzymatic activity of alkaline phosphatase in the medium, which results in the generation of free phosphate. This elevation of free phosphate in the medium is sufficient to signal induction of osteopontin RNA and protein. The strong and specific induction of osteopontin in direct response to increased phosphate levels provides a mechanism to explain how expression of this product is normally regulated in bone and suggests how it may become up-regulated in damaged tissue.

Fundamental patterns underlying gene expression profiles: Simplicity from complexity

Analysis of previously published sets of DNA microarray gene expression data by singular value decomposition has uncovered underlying patterns or “characteristic modes” in their temporal profiles. These patterns contribute unequally to the structure of the expression profiles. Moreover, the essential features of a given set of expression profiles are captured using just a small number of characteristic modes. This leads to the striking conclusion that the transcriptional response of a genome is orchestrated in a few fundamental patterns of gene expression change. These patterns are both simple and robust, dominating the alterations in expression of genes throughout the genome. Moreover, the characteristic modes of gene expression change in response to environmental perturbations are similar in such distant organisms as yeast and human cells. This analysis reveals simple regularities in the seemingly complex transcriptional transitions of diverse cells to new states, and these provide insights into the operation of the underlying genetic networks.

LXRα functions as a cAMP-responsive transcriptional regulator of gene expression

LXRα is a member of a nuclear receptor superfamily that regulates transcription. LXRα forms a heterodimer with RXRα, another member of this family, to regulate the expression of cholesterol 7α-hydroxylase by means of binding to the DR4-type cis-element. Here, we describe a function for LXRα as a cAMP-responsive regulator of renin and c-myc gene transcriptions by the interaction with a specific cis-acting DNA element, CNRE (an overlapping cAMP response element and a negative response element). Our previous studies showed that renin gene expression is regulated by cAMP, at least partly, through the CNRE sequence in its 5′-flanking region. This sequence is also found in c-myc and several other genes. Based on our cloning results using the yeast one-hybrid system, we discovered that the mouse homologue of human LXRα binds to the CNRE and demonstrated that it binds as a monomer. To define the function of LXRα on gene expression, we transfected the renin-producing renal As4.1 cells with LXRα expression plasmid. Overexpression of LXRα in As4.1 cells confers cAMP inducibility to reporter constructs containing the renin CNRE. After stable transfection of LXRα, As4.1 cells show a cAMP-inducible up-regulation of renin mRNA expression. In parallel experiments, we demonstrated that LXRα can also bind to the homologous CNRE in the c-myc promoter. cAMP promotes transcription through c-myc/CNRE:LXRα interaction in LXRα transiently transfected cells and increases c-myc mRNA expression in stably transfected cells. Identification of LXRα as a cAMP-responsive nuclear modulator of renin and c-myc expression not only has cardiovascular significance but may have generalized implication in the regulation of gene transcription.

Retinoic acid induces sodium/iodide symporter gene expression and radioiodide uptake in the MCF-7 breast cancer cell line

The sodium/iodide symporter (NIS) stimulates iodide uptake in normal lactating breast, but is not known to be active in nonlactating breast or breast cancer. We studied NIS gene regulation and iodide uptake in MCF-7 cells, an estrogen receptor (ER)-positive human breast cancer cell line. All-trans retinoic acid (tRA) treatment stimulated iodide uptake in a time- and dose-dependent fashion up to ≈9.4-fold above baseline. Stimulation with selective retinoid compounds indicated that the induction of iodide uptake was mediated by retinoic acid receptor. Treatment with tRA markedly stimulated NIS mRNA and immunoreactive protein (≈68 kDa). tRA stimulated NIS gene transcription ≈4-fold, as shown by nuclear run-on assay. No induction of iodide uptake was observed with RA treatment of an ER-negative human breast cancer cell line, MDA-MB 231, or a normal human breast cell line, MCF-12A. The iodide efflux rate of tRA-treated MCF-7 cells was slow (t1/2 = 24 min), compared with that in FRTL-5 thyroid cells (t1/2 = 3.9 min), favoring iodide retention in MCF-7 cells. An in vitro clonogenic assay demonstrated selective cytotoxicity with 131I after tRA stimulation of MCF-7 cells. tRA up-regulates NIS gene expression and iodide uptake in an ER-positive breast cancer cell line. Stimulation of radioiodide uptake after systemic retinoid treatment may be useful for diagnosis and treatment of some differentiated breast cancers.

Functional interaction between c-Jun and promoter factor Sp1 in epidermal growth factor-induced gene expression of human 12(S)-lipoxygenase

The functional role of the interaction between c-Jun and simian virus 40 promoter factor 1 (Sp1) in epidermal growth factor (EGF)-induced expression of 12(S)-lipoxygenase gene in human epidermoid carcinoma A431 cells was studied. Coimmunoprecipitation experiments indicated that EGF stimulated interaction between c-Jun and Sp1 in a time-dependent manner. Overexpression of Ha-ras and c-Jun also enhanced the amount of c-Jun binding to Sp1. In addition, the c-Jun dominant negative mutant TAM-67 not only inhibited the coimmunoprecipitated c-Jun binding to Sp1 in a dose-dependent manner in cells overexpressing c-Jun but also reduced promoter activity of the 12(S)-lipoxygenase gene induced by c-Jun overexpression. Treatment of cells with EGF increased the interaction between the Sp1 oligonucleotide and nuclear c-Jun/Sp1 in a time-dependent manner. Furthermore, EGF activated the chimeric promoter consisting of 10 tandem GAL4-binding sites, which replaced the three Sp1-binding sites in the 12(S)lipoxygenase promoter only when coexpressed with GAL4-c-Jun () fusion proteins. These results indicate that the direct interaction between c-Jun and Sp1 induced by EGF cooperatively activated expression of the 12(S)-lipoxygenase gene, and that Sp1 may serve at least in part as a carrier bringing c-Jun to the promoter, thus transactivating the transcriptional activity of 12(S)-lipoxygenase gene.

Identification of diverse nerve growth factor-regulated genes by serial analysis of gene expression (SAGE) profiling

Neurotrophic factors such as nerve growth factor (NGF) promote a wide variety of responses in neurons, including differentiation, survival, plasticity, and repair. Such actions often require changes in gene expression. To identify the regulated genes and thereby to more fully understand the NGF mechanism, we carried out serial analysis of gene expression (SAGE) profiling of transcripts derived from rat PC12 cells before and after NGF-promoted neuronal differentiation. Multiple criteria supported the reliability of the profile. Approximately 157,000 SAGE tags were analyzed, representing at least 21,000 unique transcripts. Of these, nearly 800 were regulated by 6-fold or more in response to NGF. Approximately 150 of the regulated transcripts have been matched to named genes, the majority of which were not previously known to be NGF-responsive. Functional categorization of the regulated genes provides insight into the complex, integrated mechanism by which NGF promotes its multiple actions. It is anticipated that as genomic sequence information accrues the data derived here will continue to provide information about neurotrophic factor mechanisms.

Regional and strain-specific gene expression mapping in the adult mouse brain

To determine the genetic causes and molecular mechanisms responsible for neurobehavioral differences in mice, we used highly parallel gene expression profiling to detect genes that are differentially expressed between the 129SvEv and C57BL/6 mouse strains at baseline and in response to seizure. In addition, we identified genes that are differentially expressed in specific brain regions. We found that approximately 1% of expressed genes are differentially expressed between strains in at least one region of the brain and that the gene expression response to seizure is significantly different between the two inbred strains. The results lead to the identification of differences in gene expression that may account for distinct phenotypes in inbred strains and the unique functions of specific brain regions.