Computational Methods for Reconstruction and Analysis of Genome-Scale Metabolic Networks

Metabolism is the cellular subsystem responsible for generation of energy from nutrients and production of building blocks for larger macromolecules. Computational and statistical modeling of metabolism is vital to many disciplines including bioengineering, the study of diseases, drug target identification, and understanding the evolution of metabolism. In this thesis, we propose efficient computational methods for metabolic modeling. The techniques presented are targeted particularly at the analysis of large metabolic models encompassing the whole metabolism of one or several organisms. We concentrate on three major themes of metabolic modeling: metabolic pathway analysis, metabolic reconstruction and the study of evolution of metabolism. In the first part of this thesis, we study metabolic pathway analysis. We propose a novel modeling framework called gapless modeling to study biochemically viable metabolic networks and pathways. In addition, we investigate the utilization of atom-level information on metabolism to improve the quality of pathway analyses. We describe efficient algorithms for discovering both gapless and atom-level metabolic pathways, and conduct experiments with large-scale metabolic networks. The presented gapless approach offers a compromise in terms of complexity and feasibility between the previous graph-theoretic and stoichiometric approaches to metabolic modeling. Gapless pathway analysis shows that microbial metabolic networks are not as robust to random damage as suggested by previous studies. Furthermore the amino acid biosynthesis pathways of the fungal species Trichoderma reesei discovered from atom-level data are shown to closely correspond to those of Saccharomyces cerevisiae. In the second part, we propose computational methods for metabolic reconstruction in the gapless modeling framework. We study the task of reconstructing a metabolic network that does not suffer from connectivity problems. Such problems often limit the usability of reconstructed models, and typically require a significant amount of manual postprocessing. We formulate gapless metabolic reconstruction as an optimization problem and propose an efficient divide-and-conquer strategy to solve it with real-world instances. We also describe computational techniques for solving problems stemming from ambiguities in metabolite naming. These techniques have been implemented in a web-based sofware ReMatch intended for reconstruction of models for 13C metabolic flux analysis. In the third part, we extend our scope from single to multiple metabolic networks and propose an algorithm for inferring gapless metabolic networks of ancestral species from phylogenetic data. Experimenting with 16 fungal species, we show that the method is able to generate results that are easily interpretable and that provide hypotheses about the evolution of metabolism.

Quantifying diet-induced metabolic changes of the human gut microbiome

The human gut microbiome is known to be associated with various human disorders, but a major challenge is to go beyond association studies and elucidate causalities. Mathematical modeling of the human gut microbiome at a genome scale is a useful tool to decipher microbe-microbe, diet-microbe and microbe-host interactions. Here, we describe the CASINO (Community And Systems-level INteractive Optimization) toolbox, a comprehensive computational platform for analysis of microbial communities through metabolic modeling. We first validated the toolbox by simulating and testing the performance of single bacteria and whole communities in vitro. Focusing on metabolic interactions between the diet, gut microbiota, and host metabolism, we demonstrated the predictive power of the toolbox in a diet-intervention study of 45 obese and overweight individuals and validated our predictions by fecal and blood metabolomics data. Thus, modeling could quantitatively describe altered fecal and serum amino acid levels in response to diet intervention.

DENITRIFICATION IN SOILS: FROM GENES TO ENVIRONMENTAL OUTCOMES

Denitrification is an important process of global nitrogen cycle as it removes reactive nitrogen from the biosphere, and acts as the primary source of nitrous oxide (N2O). This thesis seeks to gain better understanding of the biogeochemistry of denitrification by investigating the process from four different aspects: genetic basis, enzymatic kinetics, environmental interactions, and environmental consequences. Laboratory and field experiments were combined with modeling efforts to unravel the complexity of denitrification process under microbiological and environmental controls. Dynamics of denitrification products observed in laboratory experiments revealed an important role of constitutive denitrification enzymes, whose presence were further confirmed with quantitative analysis of functional genes encoding nitrite reductase and nitrous oxide reductase. A metabolic model of denitrification developed with explicit denitrification enzyme kinetics and representation of constitutive enzymes successfully reproduced the dynamics of N2O and N2 accumulation observed in the incubation experiments, revealing important regulatory effect of denitrification enzyme kinetics on the accumulation of denitrification products. Field studies demonstrated complex interaction of belowground N2O production, consumption and transport, resulting in two pulse pattern in the surface flux. Coupled soil gas diffusion/denitrification model showed great potential in simulating the dynamics of N2O below ground, with explicit representation of the activity of constitutive denitrification enzymes. A complete survey of environmental variables showed distinct regulation regimes on the denitrification activity from constitutive enzymes and new synthesized enzymes. Uncertainties in N2O estimation with current biogeochemical models may be reduced as accurate simulation of the dynamics of N2O in soil and surface fluxes is possible with a coupled diffusion/denitrification model that includes explicit representation of denitrification enzyme kinetics. In conclusion, denitrification is a complex ecological function regulated at cellular level. To assess the environmental consequences of denitrification and develop useful tools to mitigate N2O emissions require a comprehensive understanding of the regulatory network of denitrification with respect to microbial physiology and environmental interactions.

Hierarchical modeling gave plausible estimates of associations between metabolic syndrome and components of antiretroviral therapy

OBJECTIVE: Hierarchical modeling has been proposed as a solution to the multiple exposure problem. We estimate associations between metabolic syndrome and different components of antiretroviral therapy using both conventional and hierarchical models. STUDY DESIGN AND SETTING: We use discrete time survival analysis to estimate the association between metabolic syndrome and cumulative exposure to 16 antiretrovirals from four drug classes. We fit a hierarchical model where the drug class provides a prior model of the association between metabolic syndrome and exposure to each antiretroviral. RESULTS: One thousand two hundred and eighteen patients were followed for a median of 27 months, with 242 cases of metabolic syndrome (20%) at a rate of 7.5 cases per 100 patient years. Metabolic syndrome was more likely to develop in patients exposed to stavudine, but was less likely to develop in those exposed to atazanavir. The estimate for exposure to atazanavir increased from hazard ratio of 0.06 per 6 months' use in the conventional model to 0.37 in the hierarchical model (or from 0.57 to 0.81 when using spline-based covariate adjustment). CONCLUSION: These results are consistent with trials that show the disadvantage of stavudine and advantage of atazanavir relative to other drugs in their respective classes. The hierarchical model gave more plausible results than the equivalent conventional model.

Modeling hybridoma cell metabolism using a generic genome-scale metabolic model of Mus musculus

The reconstructed cellular metabolic network of Mus musculus, based on annotated genomic data, pathway databases, and currently available biochemical and physiological information, is presented. Although incomplete, it represents the first attempt to collect and characterize the metabolic network of a mammalian cell on the basis of genomic data. The reaction network is generic in nature and attempts to capture the carbon, energy, and nitrogen metabolism of the cell. The metabolic reactions were compartmentalized between the cytosol and the mitochondria, including transport reactions between the compartments and the extracellular medium. The reaction list consists of 872 internal metabolites involved in a total of 1220 reactions, whereof 473 relate to known open reading frames. Initial in silico analysis of the reconstructed model is presented.

11β-Hydroxysteroid dehydrogenase inhibitors and selectivity modeling of 11β-HSDs

11β-hydroksisteroididehydrogenaasientsyymit (11β-HSD) 1 ja 2 säätelevät kortisonin ja kortisolin määrää kudoksissa. 11β-HSD1 -entsyymin ylimäärä erityisesti viskeraalisessa rasvakudoksessa aiheuttaa metaboliseen oireyhtymän klassisia oireita, mikä tarjoaa mahdollisuuden metabolisen oireyhtymän hoitoon 11β-HSD1 -entsyymin selektiivisellä estämisellä. 11β-HSD2 -entsyymin inhibitio aiheuttaa kortisonivälitteisen mineralokortikoidireseptorien aktivoitumisen, mikä puolestaan johtaa hypertensiivisiin haittavaikutuksiin. Haittavaikutuksista huolimatta 11β-HSD2 -entsyymin estäminen saattaa olla hyödyllistä tilanteissa, joissa halutaan nostaa kortisolin määrä elimistössä. Lukuisia selektiivisiä 11β-HSD1 inhibiittoreita on kehitetty, mutta 11β-HSD2-inhibiittoreita on raportoitu vähemmän. Ero näiden kahden isotsyymin aktiivisen kohdan välillä on myös tuntematon, mikä vaikeuttaa selektiivisten inhibiittoreiden kehittämistä kummallekin entsyymille. Tällä työllä oli kaksi tarkoitusta: (1) löytää ero 11β-HSD entsyymien välillä ja (2) kehittää farmakoforimalli, jota voitaisiin käyttää selektiivisten 11β-HSD2 -inhibiittoreiden virtuaaliseulontaan. Ongelmaa lähestyttiin tietokoneavusteisesti: homologimallinnuksella, pienmolekyylien telakoinnilla proteiiniin, ligandipohjaisella farmakoforimallinnuksella ja virtuaaliseulonnalla. Homologimallinnukseen käytettiin SwissModeler -ohjelmaa, ja luotu malli oli hyvin päällekäinaseteltavissa niin templaattinsa (17β-HSD1) kuin 11β-HSD1 -entsyymin kanssa. Eroa entsyymien välillä ei löytynyt tarkastelemalla päällekäinaseteltuja entsyymejä. Seitsemän yhdistettä, joista kuusi on 11β-HSD2 -selektiivisiä, telakoitiin molempiin entsyymeihin käyttäen ohjelmaa GOLD. 11β-HSD1 -entsyymiin yhdisteet kiinnittyivät kuten suurin osa 11β-HSD1 -selektiivisistä tai epäselektiivisistä inhibiittoreista, kun taas 11β-HSD2 -entsyymiin kaikki yhdisteet olivat telakoituneet käänteisesti. Tällainen sitoutumistapa mahdollistaa vetysidokset Ser310:een ja Asn171:een, aminohappoihin, jotka olivat nähtävissä vain 11β-HSD2 -entsyymissä. Farmakoforimallinnukseen käytettiin ohjelmaa LigandScout3.0, jolla ajettiin myös virtuaaliseulonnat. Luodut kaksi farmakoforimallia, jotka perustuivat aiemmin telakointiinkin käytettyihin kuuteen 11β-HSD2 -selektiiviseen yhdisteeseen, koostuivat kuudesta ominaisuudesta (vetysidosakseptori, vetysidosdonori ja hydrofobinen), ja kieltoalueista. 11β-HSD2 -selektiivisyyden kannalta tärkeimmät ominaisuudet ovat vetysidosakseptori, joka voi muodostaa sidoksen Ser310 kanssa ja vetysidosdonori sen vieressä. Tälle vetysidosdonorille ei löytynyt vuorovaikutusparia 11β-HSD2-mallista. Sopivasti proteiiniin orientoitunut vesimolekyyli voisi kuitenkin olla sopiva ratkaisu puuttuvalle vuorovaikutusparille. Koska molemmat farmakoforimallit löysivät 11β-HSD2 -selektiivisiä yhdisteitä ja jättivät epäselektiivisiä pois testiseulonnassa, käytettiin molempia malleja Innsbruckin yliopistossa säilytettävistä yhdisteistä (2700 kappaletta) koostetun tietokannan seulontaan. Molemmista seulonnoista löytyneistä hiteistä valittiin yhteensä kymmenen kappaletta, jotka lähetettiin biologisiin testeihin. Biologisien testien tulokset vahvistavat lopullisesti sen kuinka hyvin luodut mallit edustavat todellisuudessa 11β-HSD2 -selektiivisyyttä.

Magnitude differences in bioactive compounds, chemical functional groups, fatty acid profiles, nutrient degradation and digestion, molecular structure and metabolic characteristics of protein in newly developed yellow-seeded and black-seeded canola lines.

Recently, new lines of yellow-seeded (CS-Y) and black-seeded canola (CS-B) have been developed with chemical and structural alteration through modern breeding technology. However, no systematic study was found on the bioactive compounds, chemical functional groups, fatty acid profiles, inherent structure, nutrient degradation and absorption, or metabolic characteristics between the newly developed yellow- and black-seeded canola lines. This study aimed to systematically characterize chemical, structural, and nutritional features in these canola lines. The parameters accessed include bioactive compounds and antinutrition factors, chemical functional groups, detailed chemical and nutrient profiles, energy value, nutrient fractions, protein structure, degradation kinetics, intestinal digestion, true intestinal protein supply, and feed milk value. The results showed that the CS-Y line was lower (P ≤ 0.05) in neutral detergent fiber (122 vs 154 g/kg DM), acid detergent fiber (61 vs 99 g/kg DM), lignin (58 vs 77 g/kg DM), nonprotein nitrogen (56 vs 68 g/kg DM), and acid detergent insoluble protein (11 vs 35 g/kg DM) than the CS-B line. There was no difference in fatty acid profiles except C20:1 eicosenoic acid content (omega-9) which was in lower in the CS-Y line (P < 0.05) compared to the CS-B line. The glucosinolate compounds differed (P < 0.05) in terms of 4-pentenyl, phenylethyl, 3-CH3-indolyl, and 3-butenyl glucosinolates (2.9 vs 1.0 μmol/g) between the CS-Y and CS-B lines. For bioactive compounds, total polyphenols tended to be different (6.3 vs 7.2 g/kg DM), but there were no differences in erucic acid and condensed tannins with averages of 0.3 and 3.1 g/kg DM, respectively. When protein was portioned into five subfractions, significant differences were found in PA, PB1 (65 vs 79 g/kg CP), PB2, and PC fractions (10 vs 33 g/kg CP), indicating protein degradation and supply to small intestine differed between two new lines. In terms of protein structure spectral profile, there were no significant differences in functional groups of amides I and II, α helix, and β-sheet structure as well as their ratio between the two new lines, indicating no difference in protein structure makeup and conformation between the two lines. In terms of energy values, there were significant differences in total digestible nutrient (TDN; 149 vs 133 g/kg DM), metabolizable energy (ME; 58 vs 52 MJ/kg DM), and net energy for lactation (NEL; 42 vs 37 MJ/kg DM) between CS-Y and CS-B lines. For in situ rumen degradation kinetics, the two lines differed in soluble fraction (S; 284 vs 341 g/kg CP), potential degradation fraction (D; 672 vs 590 g/kg CP), and effective degraded organic matter (EDOM; 710 vs 684 g/kg OM), but no difference in degradation rate. CS-Y had higher digestibility of rumen bypass protein in the intestine than CS-B (566 vs 446 g/kg of RUP, P < 0.05). Modeling nutrient supply results showed that microbial protein synthesis (MCP; 148 vs 171 g/kg DM) and rumen protein degraded balance (DPB; 108 vs 127 g/kg DM) were lower in the CS-Y line, but there were no differences in total truly digested protein in small intestine (DVE) and feed milk value (FMV) between the two lines. In conclusion, the new yellow line had different nutritional, chemical, and structural features compared to the black line. CS-Y provided better nutrient utilization and availability.

Metagenomic systems biology: frameworks for modeling and characterizing the gut microbiome

Thesis (Ph.D.)--University of Washington, 2014

Top-down systems biology modeling of host metabotype-microbiome associations in obese rodents

Covariation in the structural composition of the gut microbiome and the spectroscopically derived metabolic phenotype (metabotype) of a rodent model for obesity were investigated using a range of multivariate statistical tools. Urine and plasma samples from three strains of 10-week-old male Zucker rats (obese (fa/fa, n = 8), lean (fal-, n = 8) and lean (-/-, n = 8)) were characterized via high-resolution H-1 NMR spectroscopy, and in parallel, the fecal microbial composition was investigated using fluorescence in situ hydridization (FISH) and denaturing gradient gel electrophoresis (DGGE) methods. All three Zucker strains had different relative abundances of the dominant members of their intestinal microbiota (FISH), with the novel observation of a Halomonas and a Sphingomonas species being present in the (fa/fa) obese strain on the basis of DGGE data. The two functionally and phenotypically normal Zucker strains (fal- and -/-) were readily distinguished from the (fa/fa) obese rats on the basis of their metabotypes with relatively lower urinary hippurate and creatinine, relatively higher levels of urinary isoleucine, leucine and acetate and higher plasma LDL and VLDL levels typifying the (fa/fa) obese strain. Collectively, these data suggest a conditional host genetic involvement in selection of the microbial species in each host strain, and that both lean and obese animals could have specific metabolic phenotypes that are linked to their individual microbiomes.

Colonization-induced host-gut microbial metabolic interaction

The gut microbiota enhances the host's metabolic capacity for processing nutrients and drugs and modulate the activities of multiple pathways in a variety of organ systems. We have probed the systemic metabolic adaptation to gut colonization for 20 days following exposure of axenic mice (n = 35) to a typical environmental microbial background using high-resolution (1)H nuclear magnetic resonance (NMR) spectroscopy to analyze urine, plasma, liver, kidney, and colon (5 time points) metabolic profiles. Acquisition of the gut microbiota was associated with rapid increase in body weight (4%) over the first 5 days of colonization with parallel changes in multiple pathways in all compartments analyzed. The colonization process stimulated glycogenesis in the liver prior to triggering increases in hepatic triglyceride synthesis. These changes were associated with modifications of hepatic Cyp8b1 expression and the subsequent alteration of bile acid metabolites, including taurocholate and tauromuricholate, which are essential regulators of lipid absorption. Expression and activity of major drug-metabolizing enzymes (Cyp3a11 and Cyp2c29) were also significantly stimulated. Remarkably, statistical modeling of the interactions between hepatic metabolic profiles and microbial composition analyzed by 16S rRNA gene pyrosequencing revealed strong associations of the Coriobacteriaceae family with both the hepatic triglyceride, glucose, and glycogen levels and the metabolism of xenobiotics. These data demonstrate the importance of microbial activity in metabolic phenotype development, indicating that microbiota manipulation is a useful tool for beneficially modulating xenobiotic metabolism and pharmacokinetics in personalized health care. IMPORTANCE: Gut bacteria have been associated with various essential biological functions in humans such as energy harvest and regulation of blood pressure. Furthermore, gut microbial colonization occurs after birth in parallel with other critical processes such as immune and cognitive development. Thus, it is essential to understand the bidirectional interaction between the host metabolism and its symbionts. Here, we describe the first evidence of an in vivo association between a family of bacteria and hepatic lipid metabolism. These results provide new insights into the fundamental mechanisms that regulate host-gut microbiota interactions and are thus of wide interest to microbiological, nutrition, metabolic, systems biology, and pharmaceutical research communities. This work will also contribute to developing novel strategies in the alteration of host-gut microbiota relationships which can in turn beneficially modulate the host metabolism.

Metabolic phenotype modulation by caloric restriction in a lifelong dog study

Modeling aging and age-related pathologies presents a substantial analytical challenge given the complexity of gene−environment influences and interactions operating on an individual. A top-down systems approach is used to model the effects of lifelong caloric restriction, which is known to extend life span in several animal models. The metabolic phenotypes of caloric-restricted (CR; n = 24) and pair-housed control-fed (CF; n = 24) Labrador Retriever dogs were investigated by use of orthogonal projection to latent structures discriminant analysis (OPLS-DA) to model both generic and age-specific responses to caloric restriction from the 1H NMR blood serum profiles of young and older dogs. Three aging metabolic phenotypes were resolved: (i) an aging metabolic phenotype independent of diet, characterized by high levels of glutamine, creatinine, methylamine, dimethylamine, trimethylamine N-oxide, and glycerophosphocholine and decreasing levels of glycine, aspartate, creatine and citrate indicative of metabolic changes associated largely with muscle mass; (ii) an aging metabolic phenotype specific to CR dogs that consisted of relatively lower levels of glucose, acetate, choline, and tyrosine and relatively higher serum levels of phosphocholine with increased age in the CR population; (iii) an aging metabolic phenotype specific to CF dogs including lower levels of liproprotein fatty acyl groups and allantoin and relatively higher levels of formate with increased age in the CF population. There was no diet metabotype that consistently differentiated the CF and CR dogs irrespective of age. Glucose consistently discriminated between feeding regimes in dogs (≥312 weeks), being relatively lower in the CR group. However, it was observed that creatine and amino acids (valine, leucine, isoleucine, lysine, and phenylalanine) were lower in the CR dogs (<312 weeks), suggestive of differences in energy source utilization. 1H NMR spectroscopic analysis of longitudinal serum profiles enabled an unbiased evaluation of the metabolic markers modulated by a lifetime of caloric restriction and showed differences in the metabolic phenotype of aging due to caloric restriction, which contributes to longevity studies in caloric-restricted animals. Furthermore, OPLS-DA provided a framework such that significant metabolites relating to life extension could be differentiated and integrated with aging processes.

Structural modeling of high-affinity thyroid receptor-ligand complexes

Understanding the molecular basis of the binding modes of natural and synthetic ligands to nuclear receptors is fundamental to our comprehension of the activation mechanism of this important class of hormone regulated transcription factors and to the development of new ligands. Thyroid hormone receptors (TR) are particularly important targets for pharmaceuticals development because TRs are associated with the regulation of metabolic rates, body weight, and circulating levels of cholesterol and triglycerides in humans. While several high-affinity ligands are known, structural information is only partially available. In this work we obtain structural models of several TR-ligand complexes with unknown structure by docking high affinity ligands to the receptors` ligand binding domain with subsequent relaxation by molecular dynamics simulations. The binding modes of these ligands are discussed providing novel insights into the development of TR ligands. The experimental binding free energies are reasonably well-reproduced from the proposed models using a simple linear interaction energy free-energy calculation scheme.

Modeling the age-specific per capita growth and reproduction of Rhizobius lophanthae (Col: Coccinellidae)

A per capita model for the growth, development and reproduction of the coccinellid predator Rhizobius lophanthae (Blaisd) feeding on the oleander scale (Aspidiotus nerii Bouche (Homoptera: Diaspididae)) was developed. A thermal threshold for development of 9.4 degrees C was found. Under conditions of unlimited food, the relationship of mass at time t+1 to that at t (in days at 25 degrees C) suggests an 8.7 percent growth rate per mg larvae per day at 25 degrees C. An adult female beetle produces approximately 20 eggs per day while consuming an average of 8.5 scales/day. This is approximately 2.16 eggs per scale consumed above the maintenance level of 1.88 scales per day. More precisely, this compensation point is 0.12 mg of prey/mg of predator/day at 25 degrees C and the egestion rate is 1 - beta = 0.63.

Molecular modeling databases: A new way in the search of protein targets for drug development

DBMODELING is a relational database of annotated comparative protein structure models and their metabolic, pathway characterization. It is focused on enzymes identified in the genomes of Mycobacterium tuberculosis and Xylella fastidiosa. The main goal of the present database is to provide structural models to be used in docking simulations and drug design. However, since the accuracy of structural models is highly dependent on sequence identity between template and target, it is necessary to make clear to the user that only models which show high structural quality should be used in such efforts. Molecular modeling of these genomes generated a database, in which all structural models were built using alignments presenting more than 30% of sequence identity, generating models with medium and high accuracy. All models in the database are publicly accessible at http://www.biocristalografia.df.ibilce.unesp.br/tools. DBMODELING user interface provides users friendly menus, so that all information can be printed in one stop from any web browser. Furthermore, DBMODELING also provides a docking interface, which allows the user to carry out geometric docking simulation, against the molecular models available in the database. There are three other important homology model databases: MODBASE, SWISSMODEL, and GTOP. The main applications of these databases are described in the present article. © 2007 Bentham Science Publishers Ltd.

Metabolic profiling by ultra-performance liquid chromatography-mass spectrometry and parallel factor analysis for the determination of disease biomarkers in Eucalyptus

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)