Prediction-based fingerprints of protein-protein interactions

The recognition of protein interaction sites is an important intermediate step toward identification of functionally relevant residues and understanding protein function, facilitating experimental efforts in that regard. Toward that goal, the authors propose a novel representation for the recognition of protein-protein interaction sites that integrates enhanced relative solvent accessibility (RSA) predictions with high resolution structural data. An observation that RSA predictions are biased toward the level of surface exposure consistent with protein complexes led the authors to investigate the difference between the predicted and actual (i.e., observed in an unbound structure) RSA of an amino acid residue as a fingerprint of interaction sites. The authors demonstrate that RSA prediction-based fingerprints of protein interactions significantly improve the discrimination between interacting and noninteracting sites, compared with evolutionary conservation, physicochemical characteristics, structure-derived and other features considered before. On the basis of these observations, the authors developed a new method for the prediction of protein-protein interaction sites, using machine learning approaches to combine the most informative features into the final predictor. For training and validation, the authors used several large sets of protein complexes and derived from them nonredundant representative chains, with interaction sites mapped from multiple complexes. Alternative machine learning techniques are used, including Support Vector Machines and Neural Networks, so as to evaluate the relative effects of the choice of a representation and a specific learning algorithm. The effects of induced fit and uncertainty of the negative (noninteracting) class assignment are also evaluated. Several representative methods from the literature are reimplemented to enable direct comparison of the results. Using rigorous validation protocols, the authors estimated that the new method yields the overall classification accuracy of about 74% and Matthews correlation coefficients of 0.42, as opposed to up to 70% classification accuracy and up to 0.3 Matthews correlation coefficient for methods that do not utilize RSA prediction-based fingerprints. The new method is available at http://sppider.cchmc.org.     

Statistical aspects of van't Hoff analysis: a simulation study

The thermodynamics of biological interactions is frequently studied by the van't Hoff analysis whereby data on variation of the binding constant K(D) with temperature are used to obtain estimates of standard enthalpy (Delta H degrees ), entropy (Delta S degrees ), and heat capacity (Delta C degrees P) of complex formation. A Monte Carlo simulation demonstrates that the absolute error of the above parameters is proportional to the relative error of KD and independent of the actual values of KD and of the way they vary with temperature. The error of Delta H degrees is approximately the same as that of T Delta S degrees (within 14% in the temperature range 5-45 degrees C). The error depends both on the number of temperature points within the experimental temperature range and on the size of the range, but it is more sensitive to the latter. Using the linear form of the van't Hoff equation to fit data with non-zero Delta C degrees P gives erroneous Delta H degrees and DeltaS degrees estimates at standard temperature except for the case when the T points are placed symmetrically with respect to the standard temperature. With the range of Delta C degrees P values usual for protein-protein interactions, the KD error must be very low to confidently infer that Delta C degrees P is non-zero or to claim that two interactions have different Delta C degrees P.     

Expression of NLRP3 Inflammasomes in Neurogenic Niche Contributes to the Effect of Spatial Learning in Physiological Conditions but Not in Alzheimer’s Type Neurodegeneration

A common feature of neurodegenerative disorders, in particular Alzheimer's disease (AD), is a chronic neuroinflammation associated with aberrant neuroplasticity. Development of neuroinflammation affects efficacy of stem and progenitor cells proliferation, differentiation, migration, and integration of newborn cells into neural circuitry. However, precise mechanisms of neurogenesis alterations in neuroinflammation are not clear yet. It is well established that expression of NLRP3 inflammasomes in glial cells marks neuroinflammatory events, but less is known about contribution of NLRP3 to deregulation of neurogenesis within neurogenic niches and whether neural stem cells (NSCs), neural progenitor cells (NPCs) or immature neuroblasts may express inflammasomes in (patho)physiological conditions. Thus, we studied alterations of neurogenesis in rats with the AD model (intra-hippocampal injection of Aβ1-42). We found that in Aβ-affected brain, number of CD133+ cells was elevated after spatial training in the Morris water maze. The number of PSA-NCAM+ neuroblasts diminished by Aβ injection was completely restored by subsequent spatial learning. Spatial training leads to elevated expression of NLRP3 inflammasomes in the SGZ (subgranular zones): CD133+ and PSA-NCAM+ cells started to express NLRP3 in sham-operated, but not AD rats. Taken together, our data suggest that expression of NLRP3 inflammasomes in CD133+ and PSA-NCAM+ cells may contribute to stimulation of adult neurogenesis in physiological conditions, whereas Alzheimer’s type neurodegeneration abolishes stimuli-induced overexpression of NLRP3 within the SGZ neurogenic niche.

     

A genetic component of extinction risk in mammals

Genetic factors may play an important role in species extinction but their actual effect remains poorly understood, particularly because of a strong and potentially masking effect expected from ecological traits. We investigated the role of genetics in mammal extinction taking both ecological and genetic factors into account. As a proxy for the role of genetics we used the ratio of the rates of nonsynonymous (amino acid changing) to synonymous (leaving the amino acid unchanged) nucleotide substitutions, K_a / K_s. Because most nonsynonymous substitutions are likely to be slightly deleterious and thus selected against, this ratio is a measure of the inefficiency of selection: if large (but less than 1), it implies a low efficiency of selection against nonsynonymous mutations. As a result, nonsynonymous mutations may accumulate and thus contribute to extinction. As a proxy for the role of ecology we used body mass W, with which most extinction‐related ecological traits strongly correlate. As a measure of extinction risk we used species’ affiliation with the five levels of extinction threat according to the IUCN Red List of Threatened Species. We calculated K_a / K_s for mitochondrial protein‐coding genes of 211 mammalian species, each of which was characterized by body mass and the level of threat. Using logistic regression analysis, we then constructed a set of logistic regression models of extinction risk on ln(K_a / K_s) and lnW. We found that K_a / K_s and body mass are responsible for a 38% and a 62% increase in extinction risk, respectively. Given that the standard error of these values is 13%, the contribution of genetic factors to extinction risk in mammals is estimated to be one‐quarter to one‐half of the total of ecological and genetic effects. We conclude that the effect of genetics on extinction is significant, though it is almost certainly smaller than the effect of ecological traits. Synthesis Mutation provides the material for evolution. However, most mutations that play a role in evolution are slightly deleterious and thus may contribute to extinction. We assess the role of mitochondrial DNA mutations in mammalian extinction risk and find it to be one‐quarter to one‐half of the total of mutation and body mass effects, where body mass represents an integral measure of extinction‐related ecological traits. Genetic factors may be all the more important, because ecological traits associated with large body mass would both promote and protect from extinction, while mutation accumulation caused by low effective population size seems to have no counterbalance.     

Genomic and experimental evidence for multiple metabolic functions in the RidA/YjgF/YER057c/UK114 (Rid) protein family

Background

It is now recognized that enzymatic or chemical side-reactions can convert normal metabolites to useless or toxic ones and that a suite of enzymes exists to mitigate such metabolite damage. Examples are the reactive imine/enamine intermediates produced by threonine dehydratase, which damage the pyridoxal 5''-phosphate cofactor of various enzymes causing inactivation. This damage is pre-empted by RidA proteins, which hydrolyze the imines before they do harm. RidA proteins belong to the YjgF/YER057c/UK114 family (here renamed the Rid family). Most other members of this diverse and ubiquitous family lack defined functions.

Results

Phylogenetic analysis divided the Rid family into a widely distributed, apparently archetypal RidA subfamily and seven other subfamilies (Rid1 to Rid7) that are largely confined to bacteria and often co-occur in the same organism with RidA and each other. The Rid1 to Rid3 subfamilies, but not the Rid4 to Rid7 subfamilies, have a conserved arginine residue that, in RidA proteins, is essential for imine-hydrolyzing activity. Analysis of the chromosomal context of bacterial RidA genes revealed clustering with genes for threonine dehydratase and other pyridoxal 5''-phosphate-dependent enzymes, which fits with the known RidA imine hydrolase activity. Clustering was also evident between Rid family genes and genes specifying FAD-dependent amine oxidases or enzymes of carbamoyl phosphate metabolism. Biochemical assays showed that Salmonella enterica RidA and Rid2, but not Rid7, can hydrolyze imines generated by amino acid oxidase. Genetic tests indicated that carbamoyl phosphate overproduction is toxic to S. enterica cells lacking RidA, and metabolomic profiling of Rid knockout strains showed ten-fold accumulation of the carbamoyl phosphate-related metabolite dihydroorotate.

Conclusions

Like the archetypal RidA subfamily, the Rid2, and probably the Rid1 and Rid3 subfamilies, have imine-hydrolyzing activity and can pre-empt damage from imines formed by amine oxidases as well as by pyridoxal 5''-phosphate enzymes. The RidA subfamily has an additional damage pre-emption role in carbamoyl phosphate metabolism that has yet to be biochemically defined. Finally, the Rid4 to Rid7 subfamilies appear not to hydrolyze imines and thus remain mysterious.

Electronic supplementary material

The online version of this article (doi:10.1186/s12864-015-1584-3) contains supplementary material, which is available to authorized users.     

Functional changes in the snail statocyst system elicited by microgravity

Background

The mollusk statocyst is a mechanosensing organ detecting the animal''s orientation with respect to gravity. This system has clear similarities to its vertebrate counterparts: a weight-lending mass, an epithelial layer containing small supporting cells and the large sensory hair cells, and an output eliciting compensatory body reflexes to perturbations.

Methodology/Principal Findings

In terrestrial gastropod snail we studied the impact of 16- (Foton M-2) and 12-day (Foton M-3) exposure to microgravity in unmanned orbital missions on: (i) the whole animal behavior (Helix lucorum L.), (ii) the statoreceptor responses to tilt in an isolated neural preparation (Helix lucorum L.), and (iii) the differential expression of the Helix pedal peptide (HPep) and the tetrapeptide FMRFamide genes in neural structures (Helix aspersa L.). Experiments were performed 13–42 hours after return to Earth. Latency of body re-orientation to sudden 90° head-down pitch was significantly reduced in postflight snails indicating an enhanced negative gravitaxis response. Statoreceptor responses to tilt in postflight snails were independent of motion direction, in contrast to a directional preference observed in control animals. Positive relation between tilt velocity and firing rate was observed in both control and postflight snails, but the response magnitude was significantly larger in postflight snails indicating an enhanced sensitivity to acceleration. A significant increase in mRNA expression of the gene encoding HPep, a peptide linked to ciliary beating, in statoreceptors was observed in postflight snails; no differential expression of the gene encoding FMRFamide, a possible neurotransmission modulator, was observed.

Conclusions/Significance

Upregulation of statocyst function in snails following microgravity exposure parallels that observed in vertebrates suggesting fundamental principles underlie gravi-sensing and the organism''s ability to adapt to gravity changes. This simple animal model offers the possibility to describe general subcellular mechanisms of nervous system''s response to conditions on Earth and in space.     

Linear regression models for solvent accessibility prediction in proteins.

The relative solvent accessibility (RSA) of an amino acid residue in a protein structure is a real number that represents the solvent exposed surface area of this residue in relative terms. The problem of predicting the RSA from the primary amino acid sequence can therefore be cast as a regression problem. Nevertheless, RSA prediction has so far typically been cast as a classification problem. Consequently, various machine learning techniques have been used within the classification framework to predict whether a given amino acid exceeds some (arbitrary) RSA threshold and would thus be predicted to be "exposed," as opposed to "buried." We have recently developed novel methods for RSA prediction using nonlinear regression techniques which provide accurate estimates of the real-valued RSA and outperform classification-based approaches with respect to commonly used two-class projections. However, while their performance seems to provide a significant improvement over previously published approaches, these Neural Network (NN) based methods are computationally expensive to train and involve several thousand parameters. In this work, we develop alternative regression models for RSA prediction which are computationally much less expensive, involve orders-of-magnitude fewer parameters, and are still competitive in terms of prediction quality. In particular, we investigate several regression models for RSA prediction using linear L1-support vector regression (SVR) approaches as well as standard linear least squares (LS) regression. Using rigorously derived validation sets of protein structures and extensive cross-validation analysis, we compare the performance of the SVR with that of LS regression and NN-based methods. In particular, we show that the flexibility of the SVR (as encoded by metaparameters such as the error insensitivity and the error penalization terms) can be very beneficial to optimize the prediction accuracy for buried residues. We conclude that the simple and computationally much more efficient linear SVR performs comparably to nonlinear models and thus can be used in order to facilitate further attempts to design more accurate RSA prediction methods, with applications to fold recognition and de novo protein structure prediction methods.