Michael Pfaff

Integrated metabolic spatial-temporal model for the prediction of ammonia detoxification during liver damage and regeneration

The impairment of hepatic metabolism due to liver injury has high systemic relevance. However, it is difficult to calculate the impairment of metabolic capacity from a specific pattern of liver damage with conventional techniques. We established an integrated metabolic spatial‐temporal model (IM) using hepatic ammonia detoxification as a paradigm. First, a metabolic model (MM) based on mass balancing and mouse liver perfusion data was established to describe ammonia detoxification and its zonation. Next, the MM was combined with a spatial‐temporal model simulating liver tissue damage and regeneration after CCl4 intoxication. The resulting IM simulated and visualized whether, where, and to what extent liver damage compromised ammonia detoxification. It allowed us to enter the extent and spatial patterns of liver damage and then calculate the outflow concentrations of ammonia, glutamine, and urea in the hepatic vein. The model was validated through comparisons with (1) published data for isolated, perfused livers with and without CCl4 intoxication and (2) a set of in vivo experiments. Using the experimentally determined portal concentrations of ammonia, the model adequately predicted metabolite concentrations over time in the hepatic vein during toxin‐induced liver damage and regeneration in rodents. Further simulations, especially in combination with a simplified model of blood circulation with three ammonia‐detoxifying compartments, indicated a yet unidentified process of ammonia consumption during liver regeneration and revealed unexpected concomitant changes in amino acid metabolism in the liver and at extrahepatic sites. Conclusion: The IM of hepatic ammonia detoxification considerably improves our understanding of the metabolic impact of liver disease and highlights the importance of integrated modeling approaches on the way toward virtual organisms. (Hepatology 2014;;60:2039–2050)

Identification of rheumatoid arthritis and osteoarthritis patients by transcriptome-based rule set generation

IntroductionDiscrimination of rheumatoid arthritis (RA) patients from patients with other inflammatory or degenerative joint diseases or healthy individuals purely on the basis of genes differentially expressed in high-throughput data has proven very difficult. Thus, the present study sought to achieve such discrimination by employing a novel unbiased approach using rule-based classifiers.MethodsThree multi-center genome-wide transcriptomic data sets (Affymetrix HG-U133 A/B) from a total of 79 individuals, including 20 healthy controls (control group - CG), as well as 26 osteoarthritis (OA) and 33 RA patients, were used to infer rule-based classifiers to discriminate the disease groups. The rules were ranked with respect to Kiendl’s statistical relevance index, and the resulting rule set was optimized by pruning. The rule sets were inferred separately from data of one of three centers and applied to the two remaining centers for validation. All rules from the optimized rule sets of all centers were used to analyze their biological relevance applying the software Pathway Studio.ResultsThe optimized rule sets for the three centers contained a total of 29, 20, and 8 rules (including 10, 8, and 4 rules for ‘RA’), respectively. The mean sensitivity for the prediction of RA based on six center-to-center tests was 96% (range 90% to 100%), that for OA 86% (range 40% to 100%). The mean specificity for RA prediction was 94% (range 80% to 100%), that for OA 96% (range 83.3% to 100%). The average overall accuracy of the three different rule-based classifiers was 91% (range 80% to 100%). Unbiased analyses by Pathway Studio of the gene sets obtained by discrimination of RA from OA and CG with rule-based classifiers resulted in the identification of the pathogenetically and/or therapeutically relevant interferon-gamma and GM-CSF pathways.ConclusionFirst-time application of rule-based classifiers for the discrimination of RA resulted in high performance, with means for all assessment parameters close to or higher than 90%. In addition, this unbiased, new approach resulted in the identification not only of pathways known to be critical to RA, but also of novel molecules such as serine/threonine kinase 10.

The netgenerator algorithm: reconstruction of gene regulatory networks

Mathematical models of gene regulatory networks aim to capture the causal regulatory relationships by fitting the network models to monitored time courses of gene expression levels. In this paper, the NetGenerator algorithm is presented that generates mathematical models in form of linear or nonlinear differential equation systems. The problem of finding the most likely interactions between genes is solved by a structure identification method. This can also be effectively supported by the incorporation of available expert knowledge. Using favorable parameter identification methods from a system identification point of view allows to fit accurate and sparsely connected models. By the inclusion of higher order submodels, the algorithm enables the identification of gene-gene interactions with significantly time delayed gene regulation.

Gene Expression Data Mining for Functional Genomics using Fuzzy Technology

Methods for supervised and unsupervised clustering and machine learning were studied in order to automatically model relationships between gene expression data and gene functions of the microorganism Escherichia coli. From a pre-selected subset of 265 genes (belonging to 3 functional groups) the function has been predicted with an accuracy of 63–71 % by various data mining methods described in this paper. Whereas some of these methods, i.e. K-means clustering, Kohonen’s self-organizing maps (SOM), Eisen’s hierarchical clustering and Quinlan’s C4.5 decision tree induction algorithm have been applied to gene expression data analysis in the literature already, the fuzzy approach for gene expression data analysis is introduced by the authors. The fuzzy-C-means algorithm (FCM) and the Gustafson-Kessel algorithm for unsupervised clustering as well as the Adaptive Neuro-Fuzzy Inference System (ANFIS) were successfully applied to the functional classification of E. coli genes.

Knowledge acquisition and knowledge based control in bioprocess engineering

Knowledge acquisition is still a major bottle-neck with respect to efficient computer control design of knowledge based systems in bioprocess engineering. In this paper an approach towards the automatic generation of fuzzy rules is presented and applied to data of an industrial antibiotic fermentation. Fuzzy rules generated describe the relationship between the kinetics of the preculture and the antibiotic yield of the main culture. The terms used in these rules were derived by clustering employing the fuzzy-C-means method. In order to rate and select rules and finally to optimize parameters of membership functions of fuzzy variables different criteria are discussed in relation to the aim of the knowledge based control. Results are presented with respect to process monitoring. Genetic algorithms proved suitable for optimization procedures due to the existence of multiple local optima.

Optimality in the zonation of ammonia detoxification in rodent liver

Abstract The rodent liver eliminates toxic ammonia. In mammals, three enzymes (or enzyme systems) are involved in this process: glutaminase, glutamine synthetase and the urea cycle enzymes, represented by carbamoyl phosphate synthetase. The distribution of these enzymes for optimal ammonia detoxification was determined by numerical optimization. This in silico approach predicted that the enzymes have to be zonated in order to achieve maximal removal of toxic ammonia and minimal changes in glutamine concentration. Using 13 compartments, representing hepatocytes, the following predictions were generated: glutamine synthetase is active only within a narrow pericentral zone. Glutaminase and carbamoyl phosphate synthetase are located in the periportal zone in a non-homogeneous distribution. This correlates well with the paradoxical observation that in a first step glutamine-bound ammonia is released (by glutaminase) although one of the functions of the liver is detoxification by ammonia fixation. The in silico approach correctly predicted the in vivo enzyme distributions also for non-physiological conditions (e.g. starvation) and during regeneration after tetrachloromethane (CCl4) intoxication. Metabolite concentrations of glutamine, ammonia and urea in each compartment, representing individual hepatocytes, were predicted. Finally, a sensitivity analysis showed a striking robustness of the results. These bioinformatics predictions were validated experimentally by immunohistochemistry and are supported by the literature. In summary, optimization approaches like the one applied can provide valuable explanations and high-quality predictions for in vivo enzyme and metabolite distributions in tissues and can reveal unknown metabolic functions.

Network Analysis of the Kinetics of Amino Acid Metabolism in a Liver Cell Bioreactor

The correlation of the kinetics of 18 amino acids, ammonia and urea in 18 liver cell bioreactor runs was analyzed and described by network structures. Three kinds of networks were investigated: i) correlation networks, ii) Bayesian networks, and iii) dynamic networks that obtain their structure from systems of differential equations. Three groups of liver cell bioreactor runs with low, medium and high performance, respectively, were investigated. The aim of this study was to identify patterns and structures of the amino acid metabolism that can characterize different performance levels of the bioreactor.

Dynamic model of amino acid and carbohydrate metabolism in primary human liver cells

Human liver cell bioreactors are used in extracorporeal liver support therapy. To optimize bioreactor operation with respect to clinical application an understanding of the central metabolism is desired. A two-compartment model consisting of a system of 48 differential equations was fitted to time series data of the concentrations of 18 amino acids, ammonia, urea, glucose, galactose, sorbitol and lactate, measured in the medium outflow of seven liver cell bioreactor runs. Using the presented model, the authors predict an amino acid secretion from proteolytic activities during the first day after inoculation of the bioreactor with primary liver cells. Furthermore, gluconeogenetic activites from amino acids and/or protein were predicted.

Prediction of the performance of human liver cell bioreactors by donor organ data

Human liver cell bioreactors are used in extracorporeal liver support therapy. To optimize bioreactor operation with respect to clinical application an early prediction of the long-term bioreactor culture performance is of interest. Data from 70 liver cell bioreactor runs labeled by low (n=18), medium (n=34) and high (n=18) performance were analyzed by statistical and machine learning methods. 25 variables characterizing donor organ properties, organ preservation, cell isolation and cell inoculation prior to bioreactor operation were analyzed with respect to their importance to bioreactor performance prediction. Results obtained were compared and assessed with respect to their robustness. The inoculated volume of liver cells was found to be the most relevant variable allowing the prediction of low versus medium/high bioreactor performance with an accuracy of 84 %.