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.

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.

Visualization of biochemical networks in living cells

Functional annotation of novel genes can be achieved by detection of interactions of their encoded proteins with known proteins followed by assays to validate that the gene participates in a specific cellular function. We report an experimental strategy that allows for detection of protein interactions and functional assays with a single reporter system. Interactions among biochemical network component proteins are detected and probed with stimulators and inhibitors of the network. In addition, the cellular location of the interacting proteins is determined. We used this strategy to map a signal transduction network that controls initiation of translation in eukaryotes. We analyzed 35 different pairs of full-length proteins and identified 14 interactions, of which five have not been observed previously, suggesting that the organization of the pathway is more ramified and integrated than previously shown. Our results demonstrate the feasibility of using this strategy in efforts of genomewide functional annotation.

The nu gene acts cell-autonomously and is required for differentiation of thymic epithelial progenitors.

The nude mutation (nu) causes athymia and hairlessness, but the molecular mechanisms by which it acts have not been determined. To address the role of nu in thymogenesis, we investigated whether all or part of the nude thymic epithelium could be rescued by the presence of wild-type cells in nude <--> wild-type chimeric mice. Detailed immunohistochemical analyses revealed that nude-derived cells could persist in the chimeric thymus but could not contribute to cortical or medullary epithelial networks. Nude-derived cells, present in few clusters in the medulla, expressed markers of a rare subpopulation of adult medullary epithelium. The thymic epithelial rudiment of nude mice strongly expressed these same markers, which may therefore define committed immature thymic epithelial precursor cells. To our knowledge, these data provide the first evidence that the nu gene product acts cell-autonomously and is necessary for the development of all major subpopulations of mature thymic epithelium. We propose that nu acts to regulate growth and/or differentiation, but not determination, of thymic epithelial progenitors.

The Drosophila melanogaster dodo (dod) gene, conserved in humans, is functionally interchangeable with the ESS1 cell division gene of Saccharomyces cerevisiae.

We have sequenced the region of DNA adjacent to and including the flightless (fli) gene of Drosophila melanogaster and molecularly characterized four transcription units within it, which we have named tweety (twe), flightless (fli), dodo (dod), and penguin (pen). We have performed deletion and transgenic analysis to determine the consequences of the quadruple gene removal. Only the flightless gene is vital to the organism; the simultaneous absence of the other three allows the overriding majority of individuals to develop to adulthood and to fly normally. These gene deletion results are evaluated in the context of the redundancy and degeneracy inherent in many genetic networks. Our cDNA analyses and data-base searches reveal that the predicted dodo protein has homologs in other eukaryotes and that it is made up of two different domains. The first, designated WW, is involved in protein-protein interactions and is found in functionally diverse proteins including human dystrophin. The second is involved in accelerating protein folding and unfolding and is found in Escherichia coli in a new family of peptidylprolyl cis-trans isomerases (PPIases; EC 5.2.1.8). In eukaryotes, PPIases occur in the nucleus and the cytoplasm and can form stable associations with transcription factors, receptors, and kinases. Given this particular combination of domains, the dodo protein may well participate in a multisubunit complex involved in the folding and activation of signaling molecules. When we expressed the dodo gene product in Saccharomyces cerevisiae, it rescued the lethal phenotype of the ESS1 cell division gene.

Transient Activation of Meox1 Is an Early Component of the Gene Regulatory Network Downstream of Hoxa2

Hox genes encode transcription factors that regulate morphogenesis in all animals with bilateral symmetry. Although Hox genes have been extensively studied, their molecular function is not clear in vertebrates, and only a limited number of genes regulated by Hox transcription factors have been identified. Hoxa2 is required for correct development of the second branchial arch, its major domain of expression. We now show that Meox1 is genetically downstream from Hoxa2 and is a direct target. Meox1 expression is downregulated in the second arch of Hoxa2 mouse mutant embryos. In chromatin immunoprecipitation (ChIP), Hoxa2 binds to the Meox1 proximal promoter. Two highly conserved binding sites contained in this sequence are required for Hoxa2-dependent activation of the Meox1 promoter. Remarkably, in the absence of Meox1 and its close homolog Meox2, the second branchial arch develops abnormally and two of the three skeletal elements patterned by Hoxa2 are malformed. Finally, we show that Meox1 can specifically bind the DNA sequences recognized by Hoxa2 on its functional target genes. These results provide new insight into the Hoxa2 regulatory network that controls branchial arch identity.

Statistical Hurdle Models for Single Cell Gene Expression: Differential Expression and Graphical Modeling

Thesis (Ph.D.)--University of Washington, 2016-06

Prediction of subcellular localization using sequence-biased recurrent networks

Motivation: Targeting peptides direct nascent proteins to their specific subcellular compartment. Knowledge of targeting signals enables informed drug design and reliable annotation of gene products. However, due to the low similarity of such sequences and the dynamical nature of the sorting process, the computational prediction of subcellular localization of proteins is challenging. Results: We contrast the use of feed forward models as employed by the popular TargetP/SignalP predictors with a sequence-biased recurrent network model. The models are evaluated in terms of performance at the residue level and at the sequence level, and demonstrate that recurrent networks improve the overall prediction performance. Compared to the original results reported for TargetP, an ensemble of the tested models increases the accuracy by 6 and 5% on non-plant and plant data, respectively.

LPS regulates proinflammatory gene expression in macrophages by altering histone deacetylase expression

Bacterial LPS triggers dramatic changes in gene expression in macrophages. We show here that LPS regulated several members of the histone deacetylase (HDAC) family at the mRNA level in murine bone marrow-derived macrophages (BMM). LPS transiently repressed, then induced a number of HDACs (Hdac-4, 5, 7) in BMM, whereas Hdac-1 mRNA was induced more rapidly. Treatment of BMM with trichostatin A (TSA), an inhibitor of HDACs, enhanced LPS-induced expression of the Cox-2, Cxcl2, and Ifit2 genes. In the case of Cox-2, this effect was also apparent at the promoter level. Overexpression of Hdac-8 in RAW264 murine macrophages blocked the ability of LPS to induce Cox-2 mRNA. Another class of LPS-inducible genes, which included Ccl2, Ccl7, and Edn1, was suppressed by TSA, an effect most likely mediated by PU.1 degradation. Hence, HDACs act as potent and selective negative regulators of proinflammatory gene expression and act to prevent excessive inflammatory responses in macrophages.

Stochastic delay differential equations for genetic regulatory networks

Time delay is an important aspect in the modelling of genetic regulation due to slow biochemical reactions such as gene transcription and translation, and protein diffusion between the cytosol and nucleus. In this paper we introduce a general mathematical formalism via stochastic delay differential equations for describing time delays in genetic regulatory networks. Based on recent developments with the delay stochastic simulation algorithm, the delay chemical masterequation and the delay reaction rate equation are developed for describing biological reactions with time delay, which leads to stochastic delay differential equations derived from the Langevin approach. Two simple genetic regulatory networks are used to study the impact of' intrinsic noise on the system dynamics where there are delays. (c) 2006 Elsevier B.V. All rights reserved.

RNA: Networks & imaging

The past few years have brought about a fundamental change in our understanding and definition of the RNA world and its role in the functional and regulatory architecture of the cell. The discovery of small RNAs that regulate many aspects of differentiation and development have joined the already known non-coding RNAs that are involved in chromosome dosage compensation, imprinting, and other functions to become key players in regulating the flow of genetic information. It is also evident that there are tens or even hundreds of thousands of other non-coding RNAs that are transcribed from the mammalian genome, as well as many other yet-to-be-discovered small regulatory RNAs. In the recent symposium RNA: Networks & Imaging held in Heidelberg, the dual roles of RNA as a messenger and a regulator in the flow of genetic information were discussed and new molecular genetic and imaging methods to study RNA presented.

Molecular classification of cutaneous malignant melanoma by gene expression profiling

The most common human cancers are malignant neoplasms of the skin(1,2). Incidence of cutaneous melanoma is rising especially steeply, with minimal progress in non-surgical treatment of advanced disease(3,4). Despite significant effort to identify independent predictors of melanoma outcome, no accepted histopathological, molecular or immunohistochemical marker defines subsets of this neoplasm(2,3). Accordingly, though melanoma is thought to present with different 'taxonomic' forms, these are considered part of a continuous spectrum rather than discrete entities(2). Here we report the discovery of a subset of melanomas identified by mathematical analysis of gene expression in a series of samples. Remarkably, many genes underlying the classification of this subset are differentially regulated in invasive melanomas that form primitive tubular networks in vitro, a feature of some highly aggressive metastatic melanomas(5). Global transcript analysis can identify unrecognized subtypes of cutaneous melanoma and predict experimentally verifiable phenotypic characteristics that may be of importance to disease progression.

Evolving noisy oscillatory dynamics in genetic regulatory networks

We introduce a genetic programming (GP) approach for evolving genetic networks that demonstrate desired dynamics when simulated as a discrete stochastic process. Our representation of genetic networks is based on a biochemical reaction model including key elements such as transcription, translation and post-translational modifications. The stochastic, reaction-based GP system is similar but not identical with algorithmic chemistries. We evolved genetic networks with noisy oscillatory dynamics. The results show the practicality of evolving particular dynamics in gene regulatory networks when modelled with intrinsic noise.

Evolving genetic regulatory networks using an artificial genome

Boolean models of genetic regulatory networks (GRNs) have been shown to exhibit many of the characteristic dynamics of real GRNs, with gene expression patterns settling to point attractors or limit cycles, or displaying chaotic behaviour, depending upon the connectivity of the network and the relative proportions of excitatory and inhibitory interactions. This range of behaviours is only apparent, however, when the nodes of the GRN are updated synchronously, a biologically implausible state of affairs. In this paper we demonstrate that evolution can produce GRNs with interesting dynamics under an asynchronous update scheme. We use an Artificial Genome to generate networks which exhibit limit cycle dynamics when updated synchronously, but collapse to a point attractor when updated asynchronously. Using a hill climbing algorithm the networks are then evolved using a fitness function which rewards patterns of gene expression which revisit as many previously seen states as possible. The final networks exhibit “fuzzy limit cycle” dynamics when updated asynchronously.

Multilevel modelling for inference of genetic regulatory networks

Time-course experiments with microarrays are often used to study dynamic biological systems and genetic regulatory networks (GRNs) that model how genes influence each other in cell-level development of organisms. The inference for GRNs provides important insights into the fundamental biological processes such as growth and is useful in disease diagnosis and genomic drug design. Due to the experimental design, multilevel data hierarchies are often present in time-course gene expression data. Most existing methods, however, ignore the dependency of the expression measurements over time and the correlation among gene expression profiles. Such independence assumptions violate regulatory interactions and can result in overlooking certain important subject effects and lead to spurious inference for regulatory networks or mechanisms. In this paper, a multilevel mixed-effects model is adopted to incorporate data hierarchies in the analysis of time-course data, where temporal and subject effects are both assumed to be random. The method starts with the clustering of genes by fitting the mixture model within the multilevel random-effects model framework using the expectation-maximization (EM) algorithm. The network of regulatory interactions is then determined by searching for regulatory control elements (activators and inhibitors) shared by the clusters of co-expressed genes, based on a time-lagged correlation coefficients measurement. The method is applied to two real time-course datasets from the budding yeast (Saccharomyces cerevisiae) genome. It is shown that the proposed method provides clusters of cell-cycle regulated genes that are supported by existing gene function annotations, and hence enables inference on regulatory interactions for the genetic network.