In silico two-hybrid system for the selection of physically interacting protein pairs

Deciphering the interaction links between proteins has become one of the main tasks of experimental and bioinformatic methodologies. Reconstruction of complex networks of interactions in simple cellular systems by integrating predicted interaction networks with available experimental data is becoming one of the most demanding needs in the postgenomic era. On the basis of the study of correlated mutations in multiple sequence alignments, we propose a new method (in silico two-hybrid, i2h) that directly addresses the detection of physically interacting protein pairs and identifies the most likely sequence regions involved in the interactions. We have applied the system to several test sets, showing that it can discriminate between true and false interactions in a significant number of cases. We have also analyzed a large collection of E. coli protein pairs as a first step toward the virtual reconstruction of its complete interaction network.     

Modelling autocatalytic networks with artificial microbiology

Cells can usefully be equated to autocatalytic networks that increase in mass and then divide. To begin to model relationships between autocatalytic networks and cell division, we have written a program of artificial chemistry that simulates a cell fed by monomers. These monomers are symbols that can be assembled into linear (non-branched) polymers to give different lengths. A reaction is catalysed by a particular polymer or 'enzyme' that may itself be a reactant of that reaction (autocatalysis). These reactions are only studied within the confines of the 'cell' or 'reaction chamber'. There is a flux of material through the cell and eventually the mass of polymers reaches a threshold at which we analyse the cell. Our results indicate a similarity between the connectivity of the reaction network and that of real metabolic networks. Developing the model will entail attributing increased probabilities of reactions to polymers that are colocalised to evaluate the consequences of the dynamics of large assemblies of diverse molecules (hyperstructures) and of cell division.     

Networks in phylogenetic analysis: new tools for population biology

Phylogenetic analysis has changed greatly in the past decade, including the more widespread appreciation of the idea that evolutionary histories are not always tree-like, and may, thus, be best represented as reticulated networks rather than as strictly dichotomous trees. Reconstructing such histories in the absence of a bifurcating speciation process is even more difficult than the usual procedure, and a range of alternative strategies have been developed. There seem to be two basic uses for a network model of evolution: the display of real but unobservable evolutionary events (i.e. a hypothesis of the true phylogenetic history), and the display of character conflict within the data itself (i.e. a summary of the data). These two general approaches are briefly reviewed here, and the strengths and weaknesses of the different implementations are compared and contrasted. Each network methodology seems to have limitations in terms of how it responds to increasing complexity (e.g. conflict) in the data, and therefore each is likely to be more appropriate for one of the two uses than for the other. Several examples using parasitological data sets illustrate the uses of networks within the context of population biology.     

The Na+/H+ antiporter of the thermohalophilic bacterium Rhodothermus marinus

In the thermohalophilic bacterium Rhodothermus marinus, the NADH:quinone oxidoreductase (complex I) is encoded by two single genes and two operons, one of which contains the genes for five complex I subunits, nqo10-nqo14, a pterin carbinolamine dehydratase, and a putative single subunit Na+/H+ antiporter. Here we report that the latter encodes indeed a functional Na+/H+ antiporter, which is able to confer resistance to Na+, but not to Li+ to an Escherichia coli strain defective in Na+/H+ antiporters. In addition, an extensive amino acid sequence comparison with several single subunit Na+/H+ antiporters from different groups, namely NhaA, NhaB, NhaC, and NhaD, suggests that this might be the first member of a new type of Na+/H+ antiporters, which we propose to call NhaE.     

Rescue of PINK1 Protein Null-specific Mitochondrial Complex IV Deficits by Ginsenoside Re Activation of Nitric Oxide Signaling

PINK1, linked to familial Parkinson''s disease, is known to affect mitochondrial function. Here we identified a novel regulatory role of PINK1 in the maintenance of complex IV activity and characterized a novel mechanism by which NO signaling restored complex IV deficiency in PINK1 null dopaminergic neuronal cells. In PINK1 null cells, levels of specific chaperones, including Hsp60, leucine-rich pentatricopeptide repeat-containing (LRPPRC), and Hsp90, were severely decreased. LRPPRC and Hsp90 were found to act upstream of Hsp60 to regulate complex IV activity. Specifically, knockdown of Hsp60 resulted in a decrease in complex IV activity, whereas antagonistic inhibition of Hsp90 by 17-(allylamino) geldanamycin decreased both Hsp60 and complex IV activity. In contrast, overexpression of the PINK1-interacting factor LRPPRC augmented complex IV activity by up-regulating Hsp60. A similar recovery of complex IV activity was also induced by coexpression of Hsp90 and Hsp60. Drug screening identified ginsenoside Re as a compound capable of reversing the deficit in complex IV activity in PINK1 null cells through specific increases of LRPPRC, Hsp90, and Hsp60 levels. The pharmacological effects of ginsenoside Re could be reversed by treatment of the pan-NOS inhibitor l-NG-Nitroarginine Methyl Ester (l-NAME) and could also be reproduced by low-level NO treatment. These results suggest that PINK1 regulates complex IV activity via interactions with upstream regulators of Hsp60, such as LRPPRC and Hsp90. Furthermore, they demonstrate that treatment with ginsenoside Re enhances functioning of the defective PINK1-Hsp90/LRPPRC-Hsp60-complex IV signaling axis in PINK1 null neurons by restoring NO levels, providing potential for new therapeutics targeting mitochondrial dysfunction in Parkinson''s disease.     

Analyzing steady states of dynamics of bio-molecules from the structure of regulatory networks

Regulatory relations between biological molecules constitute complex network systems and realize diverse biological functions through the dynamics of molecular activities. However, we currently have very little understanding of the relationship between the structure of a regulatory network and its dynamical properties. In this paper we introduce a new method, named “linkage logic” to analyze the dynamics of network systems. By this method, we can restrict possible steady states of a given complex network system from the knowledge of regulatory linkages alone. The regulatory linkage simply specifies the list of variables that affect the dynamics of each variable. We formalize two aspects of the linkage logic: the “Principle of Compatibility” determines the upper limit of the diversity of possible steady states of the dynamics realized by a given network; the “Principle of Dependency” determines the possible combinations of states of the system. By combining these two aspects, (i) for a given network, we can identify a cluster of nodes that gives an alternative representation of the steady states of the whole system, (ii) we can reduce a given complex network into a simpler one without loss of the ability to generate the diversity of steady states, (iii) we can examine the consistency between the structure of network and observed set of steady states, and (iv) sometimes we can predict unknown states or unknown regulations from an observed set of steady states alone. We illustrate the method by several applications to an experimentally determined regulatory network for biological functions.     

Structure-guided Mutation of the Conserved G3-box Glycine in Rheb Generates a Constitutively Activated Regulator of Mammalian Target of Rapamycin (mTOR)

Constitutively activated variants of small GTPases, which provide valuable functional probes of their role in cellular signaling pathways, can often be generated by mutating the canonical catalytic residue (e.g. Ras Q61L) to impair GTP hydrolysis. However, this general approach is ineffective for a substantial fraction of the small GTPase family in which this residue is not conserved (e.g. Rap) or not catalytic (e.g. Rheb). Using a novel engineering approach, we have manipulated nucleotide binding through structure-guided substitutions of an ultraconserved glycine residue in the G3-box motif (DXXG). Substitution of Rheb Gly-63 with alanine impaired both intrinsic and TSC2 GTPase-activating protein (GAP)-mediated GTP hydrolysis by displacing the hydrolytic water molecule, whereas introduction of a bulkier valine side chain selectively blocked GTP binding by steric occlusion of the γ-phosphate. Rheb G63A stimulated phosphorylation of the mTORC1 substrate p70S6 kinase more strongly than wild-type, thus offering a new tool for mammalian target of rapamycin (mTOR) signaling.     

Deterministic convergence of chaos injection-based gradient method for training feedforward neural networks

It has been shown that, by adding a chaotic sequence to the weight update during the training of neural networks, the chaos injection-based gradient method (CIBGM) is superior to the standard backpropagation algorithm. This paper presents the theoretical convergence analysis of CIBGM for training feedforward neural networks. We consider both the case of batch learning as well as the case of online learning. Under mild conditions, we prove the weak convergence, i.e., the training error tends to a constant and the gradient of the error function tends to zero. Moreover, the strong convergence of CIBGM is also obtained with the help of an extra condition. The theoretical results are substantiated by a simulation example.