A linear programming approach for estimating the structure of a sparse linear genetic network from transcript profiling data

Background: A genetic network can be represented as a directed graph in which a node corresponds to a gene and a directed edge specifies the direction of influence of one gene on another. The reconstruction of such networks from transcript profiling data remains an important yet challenging endeavor. A transcript profile specifies the abundances of many genes in a biological sample of interest. Prevailing strategies for learning the structure of a genetic network from high-dimensional transcript profiling data assume sparsity and linearity. Many methods consider relatively small directed graphs, inferring graphs with up to a few hundred nodes. This work examines large undirected graphs representations of genetic networks, graphs with many thousands of nodes where an undirected edge between two nodes does not indicate the direction of influence, and the problem of estimating the structure of such a sparse linear genetic network (SLGN) from transcript profiling data. Results: The structure learning task is cast as a sparse linear regression problem which is then posed as a LASSO (l1-constrained fitting) problem and solved finally by formulating a Linear Program (LP). A bound on the Generalization Error of this approach is given in terms of the Leave-One-Out Error. The accuracy and utility of LP-SLGNs is assessed quantitatively and qualitatively using simulated and real data. The Dialogue for Reverse Engineering Assessments and Methods (DREAM) initiative provides gold standard data sets and evaluation metrics that enable and facilitate the comparison of algorithms for deducing the structure of networks. The structures of LP-SLGNs estimated from the INSILICO1, INSILICO2 and INSILICO3 simulated DREAM2 data sets are comparable to those proposed by the first and/or second ranked teams in the DREAM2 competition. The structures of LP-SLGNs estimated from two published Saccharomyces cerevisae cell cycle transcript profiling data sets capture known regulatory associations. In each S. cerevisiae LP-SLGN, the number of nodes with a particular degree follows an approximate power law suggesting that its degree distributions is similar to that observed in real-world networks. Inspection of these LP-SLGNs suggests biological hypotheses amenable to experimental verification. Conclusion: A statistically robust and computationally efficient LP-based method for estimating the topology of a large sparse undirected graph from high-dimensional data yields representations of genetic networks that are biologically plausible and useful abstractions of the structures of real genetic networks. Analysis of the statistical and topological properties of learned LP-SLGNs may have practical value; for example, genes with high random walk betweenness, a measure of the centrality of a node in a graph, are good candidates for intervention studies and hence integrated computational – experimental investigations designed to infer more realistic and sophisticated probabilistic directed graphical model representations of genetic networks. The LP-based solutions of the sparse linear regression problem described here may provide a method for learning the structure of transcription factor networks from transcript profiling and transcription factor binding motif data.

Fine-tuning of the signalling network controlling morphogenesis and stem cell development in teeth

Tooth development is regulated by sequential and reciprocal interactions between epithelium and mesenchyme. The molecular mechanisms underlying this regulation are conserved and most of the participating molecules belong to several signalling families. Research focusing on mouse teeth has uncovered many aspects of tooth development, including molecular and evolutionary specifi cs, and in addition offered a valuable system to analyse the regulation of epithelial stem cells. In mice the spatial and temporal regulation of cell differentiation and the mechanisms of patterning during development can be analysed both in vivo and in vitro. Follistatin (Fst), a negative regulator of TGFβ superfamily signalling, is an important inhibitor during embryonic development. We showed the necessity of modulation of TGFβ signalling by Fst in three different regulatory steps during tooth development. First we showed that tinkering with the level of TGFβ signalling by Fst may cause variation in the molar cusp patterning and crown morphogenesis. Second, our results indicated that in the continuously growing mouse incisors asymmetric expression of Fst is responsible for the labial-lingual patterning of ameloblast differentiation and enamel formation. Two TGFβ superfamily signals, BMP and Activin, are required for proper ameloblast differentiation and Fst modulates their effects. Third, we identifi ed a complex signalling network regulating the maintenance and proliferation of epithelial stem cells in the incisor, and showed that Fst is an essential modulator of this regulation. FGF3 in cooperation with FGF10 stimulates proliferation of epithelial stem cells and transit amplifying cells in the labial cervical loop. BMP4 represses Fgf3 expression whereas Activin inhibits the repressive effect of BMP4 on the labial side. Thus, Fst inhibits Activin rather than BMP4 in the cervical loop area and limits the proliferation of lingual epithelium, thereby causing the asymmetric maintenance and proliferation of epithelial stem cells. In addition, we detected Lgr5, a Wnt target gene and an epithelial stem cell marker in the intestine, in the putative epithelial stem cells of the incisor, suggesting that Lgr5 is a marker of incisor stem cells but is not regulated by Wnt/β-catenin signalling in the incisor. Thus the epithelial stem cells in the incisor may not be directly regulated by Wnt/β-catenin signalling. In conclusion, we showed in the mouse incisors that modulating the balance between inductive and inhibitory signals constitutes a key mechanism regulating the epithelial stem cells and ameloblast differentiation. Furthermore, we found additional support for the location of the putative epithelial stem cells and for the stemness of these cells. In the mouse molar we showed the necessity of fi ne-tuning the signalling in the regulation of the crown morphogenesis, and that altering the levels of an inhibitor can cause variation in the crown patterning.

NETGEM: Network Embedded Temporal GEnerative Model for gene expression data

Background: Temporal analysis of gene expression data has been limited to identifying genes whose expression varies with time and/or correlation between genes that have similar temporal profiles. Often, the methods do not consider the underlying network constraints that connect the genes. It is becoming increasingly evident that interactions change substantially with time. Thus far, there is no systematic method to relate the temporal changes in gene expression to the dynamics of interactions between them. Information on interaction dynamics would open up possibilities for discovering new mechanisms of regulation by providing valuable insight into identifying time-sensitive interactions as well as permit studies on the effect of a genetic perturbation. Results: We present NETGEM, a tractable model rooted in Markov dynamics, for analyzing the dynamics of the interactions between proteins based on the dynamics of the expression changes of the genes that encode them. The model treats the interaction strengths as random variables which are modulated by suitable priors. This approach is necessitated by the extremely small sample size of the datasets, relative to the number of interactions. The model is amenable to a linear time algorithm for efficient inference. Using temporal gene expression data, NETGEM was successful in identifying (i) temporal interactions and determining their strength, (ii) functional categories of the actively interacting partners and (iii) dynamics of interactions in perturbed networks. Conclusions: NETGEM represents an optimal trade-off between model complexity and data requirement. It was able to deduce actively interacting genes and functional categories from temporal gene expression data. It permits inference by incorporating the information available in perturbed networks. Given that the inputs to NETGEM are only the network and the temporal variation of the nodes, this algorithm promises to have widespread applications, beyond biological systems. The source code for NETGEM is available from https://github.com/vjethava/NETGEM

Isolation of the murine interleukin 2 gene and characterization of its regulatory architecture

Interleukin 2 (IL2) is the primary growth hormone used by mature T cells and this lymphokine plays an important role in the magnification of cell-mediated immune responses. Under normal circumstances its expression is limited to antigen-activated type 1 helper T cells (T_H1) and the ability to transcribe this gene is often regarded as evidence for commitment to this developmental lineage. There is, however, abundant evidence than many non-T_H1 T cells, under appropriate conditions, possess the ability to express this gene. Of paramount interest in the study of T-cell development is the mechanisms by which differentiating thymocytes are endowed with particular combinations of cell surface proteins and response repertoires. For example, why do most helper T cells express the CD4 differentiation antigen?

As a first step in understanding these developmental processes the gene encoding IL2 was isolated from a mouse genomic library by probing with a conspecific IL2 cDNA. The sequence of the 5' flanking region from + 1 to -2800 was determined and compared to the previously reported human sequence. Extensive identity exists between +1 and -580 (86%) and sites previously shown to be crucial for the proper expression of the human gene are well conserved in both sequence location in the mouse counterpart.

Transient expression assays were used to evaluate the contribution of various genomic sequences to high-level gene expression mediated by a cloned IL2 promoter fragment. Differing lengths of 5' flanking DNA, all terminating in the 5' untranslated region, were linked to a reporter gene, bacterial chloramphenicol acetyltransferase (CAT) and enzyme activity was measured after introduction into IL2-producing cell lines. No CAT was ever detected without stimulation of the recipient cells. A cloned promoter fragment containing only 321 bp of upstream DNA was expressed well in both Jurkat and EL4.El cells. Addition of intragenic or downstream DNA to these 5' IL2-CAT constructs showed that no obvious regulatory regions resided there. However, increasing the extent of 5' DNA from -321 to -2800 revealed several positive and negative regulatory elements. One negative region that was well characterized resided between -750 and -1000 and consisted almost exclusively of alternating purine and pyrimidines. There is no sequence resembling this in the human gene now, but there is evidence that there may have once been.

No region, when deleted, could relax either the stringent induction-dependence on cell-type specificity displayed by this promoter. Reagents that modulated endogenous IL2 expression, such as cAMP, cyclosporin A, and IL1, affected expression of the 5' IL2-CAT constructs also. For a given reagent, expression from all expressible constructs was suppressed or enhanced to the same extent. This suggests that these modulators affect IL2 expression through perturbation of a central inductive signal rather than by summation of the effects of discrete, independently regulated, negative and positive transcription factors.

Some biocybernetic aspects in gene expression and regulatory control

In eucaryotes, gene expression and control is a complex nonlinear process, where there are many control mechanisms and ways, both physic, chemical and informational control. By the exploration from the angle of biocybernetics, the authors suggest that gene expression is a co-control process. In this process, physic, chemical and informational feedback controls are associated and influential each other, and are cross and co-functional. The physic, chemical and informational control ways composed an order non-linear feedback control system in eucaryotes.

A dependent partition-valued process for multitask clustering and time evolving network modelling

The fundamental aim of clustering algorithms is to partition data points. We consider tasks where the discovered partition is allowed to vary with some covariate such as space or time. One approach would be to use fragmentation-coagulation processes, but these, being Markov processes, are restricted to linear or tree structured covariate spaces. We define a partition-valued process on an arbitrary covariate space using Gaussian processes. We use the process to construct a multitask clustering model which partitions datapoints in a similar way across multiple data sources, and a time series model of network data which allows cluster assignments to vary over time. We describe sampling algorithms for inference and apply our method to defining cancer subtypes based on different types of cellular characteristics, finding regulatory modules from gene expression data from multiple human populations, and discovering time varying community structure in a social network.

Gene structures and promoter characteristics of interferon regulatory factor 1 (IRF-1), IRF-2 and IRF-7 from snakehead Channa argus

Three interferon regulatory factor (IRF) genes, CaIRF-1, CaIRF-2 and CaIRF-7, and their promoters of snakehead (Channa argus) were cloned and characterized. The CaIRF-1 gene consists of ten exons, spans 4.3 kb and encodes a putative peptide of 299 aa. The CaIRF-2 gene consists of nine exons, spans 8 kb and encodes a putative peptide of 328 aa. The gene organizations of CaIRF-1 and CaIRF-2 are very similar to that of human IRF-1 and IRF-2 except more compact. Comparison of exon-intron organization of the two genes indicated a common evolutionary structure, notably within the exons encoding the DNA binding domain (DBD) of the two factors. The CaIRF-7 gene spans 4.1 kb and encodes a putative peptide of 437 aa. However, the gene organization of CaIRF-7 consisting of ten exons is different to human IRF-7a gene which has an intron in 5' UTR. Three CaIRFs share homology in N-terminal encompassing the DBD that contains a characteristic repeat of tryptophan residues. The promoters of CaIRF-1 and CaIRF-2 genes contain the conserved sites for NF-kappa B and Sp1. The gamma-IFN activation sites (GAS) were found in the promoters of CaIRF-1 and CaIRF-7. The promoter of CaIRF-7 contains conserved interferon stimulating response element (ISRE) which is characteristic of IFN-induced gene promoter, and suggests that there also exist intracellular amplifier circuit in fish IFN signal pathway. Moreover, the element GAAANN oriented in both directions is repeated in CaIRF promoter regions, which confers to further inducibility by IFN. The constitutive expression of CaIRF genes were found to increase obviously in response to induction by the known IFN-inducer poly I:C. (c) 2008 Published by Elsevier Ltd.

Characterization of muscle-regulatory gene, MyoD, from flounder (Paralichthys olivaceus) and analysis of its expression patterns during embryogenesis

Specification and differentiation of skeletal muscle cells are driven by the activity of genes encoding members of the myogenic regulatory factors (MRFs). In vertebrates, the MRF family includes MyoD, Myf5, myogenin, and MRF4. The MRFs are capable of converting a variety of nonmuscle cells into myoblasts and myotubes. To better understand their roles in fish muscle development, we isolated the MyoD gene from flounder (Paralichthys olivaceus) and analyzed its structure and patterns of expression. Sequence analysis showed that flounder MyoD shared a structure similar to that of vertebrate MRFs with three exons and two introns, and its protein contained a highly conserved basic helix-loop-helix domain (bHLH). Comparison of sequences revealed that flounder MyoD was highly conserved with other fish MyoD genes. Sequence alignment and phylogenetic analysis indicated that flounder MyoD, seabream (Sparus aurata) MyoD1, takifugu (Takifugu rubripes) MyoD, and tilapia (Oreochromis aureus) MyoD were more likely to be homologous genes. Flounder MyoD expression was first detected as two rows of presomitic cells in the segmental plate. From somitogenesis, MyoD transcripts were present in the adaxial cells that give rise to slow muscles and the lateral somitic cells that give rise to fast muscles. After 30 somites formed, MyoD expression decreased in the somites except the caudal somites, coincident with somite maturation. In the hatching stage, MyoD was expressed in other muscle cells and caudal somites. It was detected only in muscle in the growing fish.

Punctuated evolution and transitional hybrid network in an ancestral cell cycle of fungi.

Although cell cycle control is an ancient, conserved, and essential process, some core animal and fungal cell cycle regulators share no more sequence identity than non-homologous proteins. Here, we show that evolution along the fungal lineage was punctuated by the early acquisition and entrainment of the SBF transcription factor through horizontal gene transfer. Cell cycle evolution in the fungal ancestor then proceeded through a hybrid network containing both SBF and its ancestral animal counterpart E2F, which is still maintained in many basal fungi. We hypothesize that a virally-derived SBF may have initially hijacked cell cycle control by activating transcription via the cis-regulatory elements targeted by the ancestral cell cycle regulator E2F, much like extant viral oncogenes. Consistent with this hypothesis, we show that SBF can regulate promoters with E2F binding sites in budding yeast.

Characterization of the 5' upstream regulatory region of the human myostatin gene: regulation of myostatin gene transcription by dexamethasone in vitro

We cloned and characterized a 3.3-kb fragment containing the 5'-regulatory region of the human myostatin gene. The promoter sequence contains putative muscle growth response elements for glucocorticoid, androgen, thyroid hormone, myogenic differentiation factor 1, myocyte enhancer factor 2, peroxisome proliferator-activated receptor, and nuclear factor-kappaB. To identify sites important for myostatin's gene transcription and regulation, eight deletion constructs were placed in C(2)C(12) and L6 skeletal muscle cells. Transcriptional activity of the constructs was found to be significantly higher in myotubes compared with that of myoblasts. To investigate whether glucocorticoids regulate myostatin gene expression, we incubated both cell lines with dexamethasone. On both occasions, dexamethasone dose dependently increased both the promoter's transcriptional activity and the endogenous myostatin expression. The effects of dexamethasone were blocked when the cells were coincubated with the glucocorticoid receptor antagonist RU-486. These findings suggest that glucocorticoids upregulate myostatin expression by inducing gene transcription, possibly through a glucocorticoid receptor-mediated pathway. We speculate that glucocorticoid-associated muscle atrophy might be due in part to the upregulation of myostatin expression.

Association between polymorphism in regulatory region of gene encoding tumour necrosis factor alpha and risk of Alzheimer's disease and vascular dementia: a case-control study

BACKGROUND: Deposition of beta-amyloid in the brains of patients with Alzheimer's disease is thought to precede a chain of events that leads to an inflammatory response by the brain. We postulated that genetic variation in the regulatory region of the gene for the proinflammatory cytokine tumour necrosis factor alpha (TNF-alpha) leads to increased risk of Alzheimer's disease and vascular dementia. METHODS: A polymorphism in the regulatory region of the TNF-alpha gene was analysed in a case-control study. The polymorphism (C-850T) was typed in 242 patients with sporadic Alzheimer's disease, 81 patients with vascular dementia, 61 stroke patients without dementia, and 235 normal controls. These groups of individuals were also genotyped for the apolipoprotein E polymorphism, and the vascular dementia and stroke groups were typed at the HLA-DR locus. FINDINGS: The distribution of TNF-alpha genotypes in the vascular dementia group differed significantly from that in the stroke and normal control groups, giving an odds ratio of 2.51 (95% CI 1.49-4.21) for the development of vascular dementia for individuals with a CT or TT genotype. Logistic regression analysis indicated that the possession of the T allele significantly increased the risk of Alzheimer's disease associated with carriage of the apolipoprotein E epsilon4 allele (odds ratio 2.73 [1.68-4.44] for those with apolipoprotein E epsilon4 but no TNF-alpha T, vs 4.62 [2.38-8.96] for those with apolipoprotein E epsilon4 and TNF-alpha T; p=0.03). INTERPRETATION: Possession of the TNF-alpha T allele significantly increases the risk of vascular dementia, and increases the risk of Alzheimer's disease associated with apolipoprotein E. Although further research is needed, these findings suggest a potential role for anti-inflammatory therapy in vascular dementia and Alzheimer's disease, and perhaps especially in patients who have had a stroke.