Human hippocampal astrocytes: Computational dissection of their transcriptome, sexual differences and exosomes across ageing and mild-cognitive impairment.

The role of astrocytes in Alzheimer's disease is often disregarded. Hence, characterization of astrocytes along their early evolution toward Alzheimer would be greatly beneficial. However, due to their exquisite responsiveness, in vivo studies are difficult. So public microarray data of hippocampal homogenates from (healthy) young, (healthy) elder and elder with mild cognitive impairment (MCI) were subjected to re-analysis by a multi-step computational pipeline. Ontologies and pathway analyses were compared after determining the differential genes that, belonging to astrocytes, have splice forms. Likewise, the subset of molecules exportable to exosomes was also determined. The results showed that astrocyte's phenotypes changed significantly. While already 'activated' astrocytes were found in the younger group, major changes occurred during ageing (increased vascular remodelling and response to mechanical stimulus, diminished long-term potentiation and increased long-term depression). MCI's astrocytes showed some 'rejuvenated' features, but their sensitivity to shear stress was markedly lost. Importantly, most of the changes showed to be sex biassed. Men's astrocytes are enriched in a type 'endfeet-astrocytome', whereas women's astrocytes appear close to the 'scar-forming' type (prone to endothelial dysfunction, hypercholesterolemia, loss of glutamatergic synapses, Ca+2 dysregulation, hypoxia, oxidative stress and 'pro-coagulant' phenotype). In conclusion, the computational dissection of the networks based on the hippocampal gene isoforms provides a relevant proxy to in vivo astrocytes, also revealing the occurrence of sexual differences. Analyses of the astrocytic exosomes did not provide an acceptable approximation to the overall functioning of astrocytes in the hippocampus, probably due to the selective cellular mechanisms which charge the cargo molecules.

Mapping the transcriptomic changes of endothelial compartment in human hippocampus across aging and mild cognitive impairment.

Compromise of the vascular system has important consequences on cognitive abilities and neurodegeneration. The identification of the main molecular signatures present in the blood vessels of human hippocampus could provide the basis to understand and tackle these pathologies. As direct vascular experimentation in hippocampus is problematic, we achieved this information by computationally disaggregating publicly available whole microarrays data of human hippocampal homogenates. Three conditions were analyzed: 'Young Adults', 'Aged', and 'aged with Mild Cognitive Impairment' (MCI). The genes identified were contrasted against two independent data-sets. Here we show that the endothelial cells from the Younger Group appeared in an 'activated stage'. In turn, in the Aged Group, the endothelial cells showed a significant loss of response to shear stress, changes in cell adhesion molecules, increased inflammation, brain-insulin resistance, lipidic alterations, and changes in the extracellular matrix. Some specific changes in the MCI group were also detected. Noticeably, in this study the features arisen from the Aged Group (high tortuosity, increased bifurcations, and smooth muscle proliferation), pose the need for further experimental verification to discern between the occurrence of arteriogenesis and/or vascular remodeling by capillary arterialization. This article has an associated First Person interview with the first author of the paper.

Influence of Glucose Availability and CRP Acetylation on the Genome-Wide Transcriptional Response of Escherichia coli: Assessment by an Optimized Factorial Microarray Analysis.

Background: While in eukaryotes acetylation/deacetylation regulation exerts multiple pleiotropic effects, in Escherichia coli it seems to be more limited and less known. Hence, we aimed to progress in the characterization of this regulation by dealing with three convergent aspects: the effector enzymes involved, the master regulator CRP, and the dependence on glucose availability. Methods: The transcriptional response of E. coli BW25113 was analyzed across 14 relevant scenarios. These conditions arise when the wild type and four isogenic mutants (defective in deacetylase CobB, defective in N(ε)-lysine acetyl transferase PatZ, Q- and R-type mutants of protein CRP) are studied under three levels of glucose availability (glucose-limited chemostat and glucose-excess or glucose-exhausted in batch culture). The Q-type emulates a permanent stage of CRPacetylated, whereas the R-type emulates a permanent stage of CRPdeacetylated. The data were analyzed by an optimized factorial microarray method (Q-GDEMAR). Results: (a) By analyzing one mutant against the other, we were able to unravel the true genes that participate in the interaction between ΔcobB/ΔpatZ mutations and glucose availability; (b) Increasing stages of glucose limitation appear to be associated with the up-regulation of specific sets of target genes rather than with the loss of genes present when glucose is in excess; (c) Both CRPdeacetylated and CRPacetylated produce extensive changes in specific subsets of genes, but their number and identity depend on the glucose availability; (d) In other sub-sets of genes, the transcriptional effect of CRP seems to be independent of its acetylation or deacetylation; (e) Some specific ontology functions can be associated with each of the different sets of genes detected herein. Conclusions: CRP cannot be thought of only as an effector of catabolite repression, because it acts along all the glucose conditions tested (excess, limited, and exhausted), exerting both positive and negative effects through different sets of genes. Acetylation of CRP does not seem to be a binary form of regulation, as there is not a univocal relationship between its activation/inhibitory effect and its acetylation/deacetylation stage. All the combinatorial possibilities are observed. During the exponential growth phase, CRP also exerts a very significant transcriptional effect, mainly on flagellar assembly and chemotaxis (FDR = 7.2 × 10-44).

Sexual Dimorphism and Aging in the Human Hyppocampus: Identification, Validation, and Impact of Differentially Expressed Genes by Factorial Microarray and Network Analysis.

Motivation: In the brain of elderly-healthy individuals, the effects of sexual dimorphism and those due to normal aging appear overlapped. Discrimination of these two dimensions would powerfully contribute to a better understanding of the etiology of some neurodegenerative diseases, such as "sporadic" Alzheimer. Methods: Following a system biology approach, top-down and bottom-up strategies were combined. First, public transcriptome data corresponding to the transition from adulthood to the aging stage in normal, human hippocampus were analyzed through an optimized microarray post-processing (Q-GDEMAR method) together with a proper experimental design (full factorial analysis). Second, the identified genes were placed in context by building compatible networks. The subsequent ontology analyses carried out on these networks clarify the main functionalities involved. Results: Noticeably we could identify large sets of genes according to three groups: those that exclusively depend on the sex, those that exclusively depend on the age, and those that depend on the particular combinations of sex and age (interaction). The genes identified were validated against three independent sources (a proteomic study of aging, a senescence database, and a mitochondrial genetic database). We arrived to several new inferences about the biological functions compromised during aging in two ways: by taking into account the sex-independent effects of aging, and considering the interaction between age and sex where pertinent. In particular, we discuss the impact of our findings on the functions of mitochondria, autophagy, mitophagia, and microRNAs. Conclusions: The evidence obtained herein supports the occurrence of significant neurobiological differences in the hippocampus, not only between adult and elderly individuals, but between old-healthy women and old-healthy men. Hence, to obtain realistic results in further analysis of the transition from the normal aging to incipient Alzheimer, the features derived from the sexual dimorphism in hippocampus should be explicitly considered.

Q-GDEMAR: a general method for the identification of differentially expressed genes in microarrays with unbalanced groups.

Microarray analysis is a powerful tool to simultaneously determine the pattern of transcription of large amounts of genes. For data post-processing distinct computational methods are currently used that, however, lead to different results regarding the genes expressed differentially. Herein, a new methodology for microarray data analysis named Q-GDEMAR is presented. It combines the quantile characterization of the entire distribution together with the Gaussian deconvolution of the central region of the microarray data distribution. Three discriminant variable variants are proposed that allow us to summarize data and compare groups even when their size is strongly unbalanced. In addition, a simple procedure to compute the false discovery rate (FDR) is also presented. The performance of the method is compared with that observed when using LIMMA (Linear Models Microarray) software as reference. In 58 out of 68 cases, Q-GDEMAR showed a higher sensitivity than LIMMA to detect differentially expressed genes (p = 1 × 10(-10)). The proposed method does not produce biased information, detecting genes with high sensitivity equally well at both tails of the distribution (p = 0.7428). Moreover, all detected genes were associated with very low levels of FDR (median value = 0.67%, interquartile range = 0.87%). Q-GDEMAR can be used as a general method for microarray analysis, but is particularly indicated when the conditions to be compared are unbalanced. The superior performance of Q-GEDEMAR is the consequence of its higher discriminative power and, the fact that it yields a univocal correspondence between the p-values and the values of the discriminating variable. Q-GDEMAR was tested only using Affymetrix microarrays. However, given that it operates after the step of data standardization, it can be used with the same quality features on any of the available mono- or dual-channel microarray platforms.

Analysis of cell adhesion during early stages of colon cancer based on an extended multi-valued logic approach.

Cell adhesion in the normal colon is typically associated with differentiated cells, whereas in cancerous colon it is associated with advanced tumors. For advanced tumors growing evidence supports the existence of stem-like cells that have originated from transdifferentiation. Because stem cells can also be transformed in their own niche, at the base of the Lieberkühn's crypts, we conjectured that cell adhesion can also be critical in early tumorigenesis. To assess this hypothesis we built an annotated, multi-valued logic model addressing cell adhesion of normal and tumorigenic stem cells in the human colon. The model accounts for (i) events involving intercellular adhesion structures, (ii) interactions involving cytoskeleton-related structures, (iii) compartmental distribution of α/ß/γ/δ-catenins, and (iv) variations in critical cell adhesion regulators (e.g., ILK, FAK, IQGAP, SNAIL, Caveolin). We developed a method that can deal with graded multiple inhibitions, something which is not possible with conventional logical approaches. The model comprises 315 species (including 26 genes), interconnected by 269 reactions. Simulations of the model covered six scenarios, which considered two types of colonic cells (stem vs. differentiated cells), under three conditions (normal, stressed and tumor). Each condition results from the combination of 92 inputs. We compared our multi-valued logic approach with the conventional Boolean approach for one specific example and validated the predictions against published data. Our analysis suggests that stem cells in their niche synthesize high levels of cytoplasmatic E-cadherin and CdhEP(Ser684,686,692), even under normal-mitogenic stimulus or tumorigenic conditions. Under these conditions, E-cadherin would be incorporated into the plasmatic membrane, but only as a non-adhesive CdhE_ß-catenin_IQGAP complex. Under stress conditions, however, this complex could be displaced, yielding adhesive CdhE_ß-catenin((cis/trans)) complexes. In the three scenarios tested with stem cells, desmosomes or tight junctions were not assembled. Other model predictions include expected levels of the nuclear complex ß-catenin_TCF4 and the anti-apoptotic protein Survivin for both normal and tumorigenic colonic stem cells.

Model identification in presence of incomplete information by generalized principal component analysis: application to the common and differential responses of Escherichia coli to multiple pulse perturbations in continuous, high-biomass density culture.

In a previous report we described a multivariate approach to discriminate between the different response mechanisms operating in Escherichia coli when a steady, continuous culture of these bacteria was perturbed by a glycerol pulse (Guebel et al., 2009, Biotechnol Bioeng 102: 910-922). Herein, we present a procedure to extend this analysis when multiple, spaced pulse perturbations (glycerol, fumarate, acetate, crotonobetaine, hypersaline plus high-glycerol basal medium and crotonobetaine plus hypersaline basal medium) are being assessed. The proposed method allows us to identify not only the common responses among different perturbation conditions, but to recognize the specific response for a given stimulus even when the dynamics of the perturbation is unknown. Components common to all conditions are determined first by Generalized Principal Components Analysis (GPCA) upon a set of covariance matrices. A metrics is then built to quantify the similitude distance. This is based on the degree of variance extraction achieved for each variable along the GPCA deflation processes by the common factors. This permits a cluster analysis, which recognizes several compact sub-sets containing only the most closely related responsive groups. The GPCA is then run again but is restricted to the groups in each sub-set. Finally, after the data have been exhaustively deflated by the common sub-set factors, the resulting residual matrices are used to determine the specific response factors by classical principal component analysis (PCA). The proposed method was validated by comparing its predictions with those obtained when the dynamics of the perturbation was determined. In addition, it showed to have a better performance than the obtained with other multivariate alternatives (e.g., orthogonal contrasts based on direct GPCA, Tucker-3 model, PARAFAC, etc.).

Analysis of the Escherichia coli response to glycerol pulse in continuous, high-cell density culture using a multivariate approach.

Pulse experiments in continuous-culture are a valuable tool in microbial physiology research. However, inferences become difficult when the cell response is followed by monitoring many biochemical variables or when several types of perturbations are compared. Moreover, there is no objective criterion to delimit the time-window, so that the recorded responses will render valid inferences. Hence, we have investigated the capability of a multivariate approach to deal with complex data from a previously described series of pulse experiments. Data are concerned with 12 biochemical variables that were monitored when an anaerobic, steady-continuous culture of E. coli O74K74 was disturbed by six types of pulses (glycerol, fumarate, acetate, crotonobetaine, hypersaline plus high-glycerol basal medium and crotonobetaine plus hypersaline basal medium). Our analysis determined the instantaneous uptake rate for the pulsed metabolite (Dynamical Chemical-Balances), reduced the multivariate observations to one response curve (Principal Component Analysis) and determined the optimal time-window (Cluster Analysis). Finally, input-output data were filtered (Orthogonal Signal Correction) while both blocks were mathematically connected (Partial Least-squares Regression). This systematic approach allowed us to detect several relevant patterns not previously revealed: (i) Glycerol uptake rate did not follow a Michaelian kinetics but showed a biphasic dependence on glycerol concentration; noticeably, net uptake decreased 136-fold despite the high availability of glycerol in the milieu. (ii) The structure of the bacterial response changed during time the glycerol-disturbance lasted (2 h), hence analyses had to be limited to the early response (time from 0 to 5 min). (iii) By mathematically relating the input (glycerol uptake rate) with the output (12 biochemical responses) it was possible to identify which of the monitored variables were primary targets of the glycerol disturbance (namely: ATP, formate, acetyl-CoA synthase, isocitrate dehydrogenase, and isocitrate lyase), which were secondarily responsive (ethanol) and those that were independent (acetate, carnitine, lactate, and NADH/NAD ratio). Identification was achieved even though all the analyzed variables were affected by the pulse. (iv) Some variables exhibited uncorrelated dynamics despite their close functional relationship (ATP and NADH/NAD ratio, ethanol and lactate; carnitine and the crotonobetaine hydratase complex; acetate and the enzymes phosphotransacetylase, acetyl-CoA synthase and isocitrate lyase). The results are discussed in terms of E. coli transcriptional control, and it is concluded that glycerol pulse produces a stressing effect. The consequent activation of the polyamine-dependent mechanisms involved in such stressing effect provides a unified explanation for how glycerol uptake is down-regulated in the presence of high glycerol availability and how acetate can be produced without de novo biosynthesis.

A computer model of oxygen dynamics in human colon mucosa: implications in normal physiology and early tumor development.

Based on the geometry of the colon mucosa, we built a model to compute the oxygen supply, the oxygen diffusion across the interstitial matrix, and the oxygen consumption by cryptal and stromal cells. By using an iterative algorithm, we have been able to solve a set of discretized (time and space) oxygen balance equations and determine the three-dimensional distribution of pO(2) in the mucosa. Although significant longitudinal and radial pO(2) variations were found, cells appeared to operate at their maximum respiratory capacity, regardless of their location in the tissue. The estimated oxygen extraction fraction was 47%, while the capillary oxygen permeability was 1.57 x 10(-5) cm m s(-1). We concluded that cellular metabolism in normal colon mucosa is not limited by oxygen supply, thus prompting the idea that oxygenation does not determine the characteristic microenvironments occurring along the normal Lieberkhün crypts. In an extended model, simulation of an aberrant crypt focus (ACF)--the earliest stage in the adenomatous polyp-carcinoma sequence--showed instead that respiratory activity decreased when the capillary array symmetry is disrupted due to the ACF growth. A unified explanation about the alternative of a hypoxic-independent and/or a hypoxic-dependent early angiogenic response associated to the development of ACF is proposed.

Dynamics of sulfur amino acids in mammalian brain: assessment of the astrocytic-neuronal cysteine interaction by a mathematical hybrid model.

A mathematically hybrid model was used to analyze three mechanisms by which cysteine could be produced in the brain to be used as preferential substrate in the synthesis of neuronal glutathione. In that way, the fluxes of sulfur-compounds at the brain-blood barrier were integrated with their transport in astrocytes and neurons, and with their metabolism in astrocytes. We concluded that cysteine, in contrast with its precursor cystine, would not be taken up from the blood at the blood-brain barrier, but instead it must be lost continuously from astrocytes. Cysteine efflux is produced because the uptake of cystine in astrocytes is much greater than their cysteine demand to synthesize glutathione, hypotaurine and taurine. Once in the interstitial parenchyma, cysteine would be taken for the neurons, as backwardly by the endothelial cells. Remarkably, a close sulfur-macro balance can be maintained only if the surplus of the produced cysteine is transferred from the endothelial cells to the blood together with significant amounts of other sulfur-compounds, probably taurine and hypotaurine. In addition, the results obtained shown that alternative mechanisms of cysteine generation (i.e., nonenzymatic-thiol-disulfide exchange reaction, enzymatic cleavage of the glutathione effluxed from astrocytes) are not quantitatively significant under physiological conditions, in situ.

Canonical sensitivities: a useful tool to deal with large perturbations in metabolic network modeling.

The dynamic range of metabolic models can be extended to deal with large perturbations by introducing the related concepts of "generalized" kinetic order and "canonical" sensitivities. Generalized kinetic orders are built as a well-defined non linear combination of the canonical sensitivities coefficients, which in turn are obtained by a least-squares regression on central composite factorial design data. In a such way, the whole domain of the operating variables is mapped without need to determine locally neither the first nor the second order model derivatives. The method was validated through numerical simulations, its predictions being compared with those coming from a Michaelis-Menten formalism taken as reference. In parallel, two variants of the Power-law formalism (S-system, least-squares GMA) also were tested. The canonical sensitivities method produced the widest range to predict metabolite concentrations and metabolic fluxes at the steady states. In addition, the variation pattern for the logarithmic gains and for the characteristic eigenvalues have been accurately determined from a unique overall model, being both required to make realistic analysis in metabolic engineering. The achieved information also can be expressed in terms of those typical coefficients derived from the Metabolic Control Analysis (MCA). Even if current first order Power-law or MCA formalisms were used, the canonical sensitivities approach provides a significant advantage, since complete sets of homologous, accurate, locally valid metabolic coefficients can be simultaneously recovered from the array proposed, being representative of the whole range of the operating variables instead of a unique nominal condition as is usual.