Mathematical modeling of biological phenomena has experienced increasing interest since new high-throughput technologies give access to growing amounts of molecular data. These modeling approaches are especially able to test hypotheses which are not yet experimentally accessible or guide an experimental setup. One particular attempt investigates the evolutionary dynamics responsible for today's composition of organisms. Computer simulations either propose an evolutionary mechanism and thus reproduce a recent finding or rebuild an evolutionary process in order to learn about its mechanism. The quest for evolutionary fingerprints in metabolic and gene-coexpression networks is the central topic of this cumulative thesis based on four published articles. An understanding of the actual origin of life will probably remain an insoluble problem. However, one can argue that after a first simple metabolism has evolved, the further evolution of metabolism occurred in parallel with the evolution of the sequences of the catalyzing enzymes. Indications of such a coevolution can be found when correlating the change in sequence between two enzymes with their distance on the metabolic network which is obtained from the KEGG database. We observe that there exists a small but significant correlation primarily on nearest neighbors. This indicates that enzymes catalyzing subsequent reactions tend to be descended from the same precursor. Since this correlation is relatively small one can at least assume that, if new enzymes are no "genetic children" of the previous enzymes, they certainly be descended from any of the already existing ones. Following this hypothesis, we introduce a model of enzyme-pathway coevolution. By iteratively adding enzymes, this model explores the metabolic network in a manner similar to diffusion. With implementation of an Gillespie-like algorithm we are able to introduce a tunable parameter that controls the weight of sequence similarity when choosing a new enzyme. Furthermore, this method also defines a time difference between successive evolutionary innovations in terms of a new enzyme. Overall, these simulations generate putative time-courses of the evolutionary walk on the metabolic network. By a time-series analysis, we find that the acquisition of new enzymes appears in bursts which are pronounced when the influence of the sequence similarity is higher. This behavior strongly resembles punctuated equilibrium which denotes the observation that new species tend to appear in bursts as well rather than in a gradual manner. Thus, our model helps to establish a better understanding of punctuated equilibrium giving a potential description at molecular level. From the time-courses we also extract a tentative order of new enzymes, metabolites, and even organisms. The consistence of this order with previous findings provides evidence for the validity of our approach. While the sequence of a gene is actually subject to mutations, its expression profile might also indirectly change through the evolutionary events in the cellular interplay. Gene coexpression data is simply accessible by microarray experiments and commonly illustrated using coexpression networks where genes are nodes and get linked once they show a significant coexpression. Since the large number of genes makes an illustration of the entire coexpression network difficult, clustering helps to show the network on a metalevel. Various clustering techniques already exist. However, we introduce a novel one which maintains control of the cluster sizes and thus assures proper visual inspection. An application of the method on Arabidopsis thaliana reveals that genes causing a severe phenotype often show a functional uniqueness in their network vicinity. This leads to 20 genes of so far unknown phenotype which are however suggested to be essential for plant growth. Of these, six indeed provoke such a severe phenotype, shown by mutant analysis. By an inspection of the degree distribution of the A.thaliana coexpression network, we identified two characteristics. The distribution deviates from the frequently observed power-law by a sharp truncation which follows after an over-representation of highly connected nodes. For a better understanding, we developed an evolutionary model which mimics the growth of a coexpression network by gene duplication which underlies a strong selection criterion, and slight mutational changes in the expression profile. Despite the simplicity of our assumption, we can reproduce the observed properties in A.thaliana as well as in E.coli and S.cerevisiae. The over-representation of high-degree nodes could be identified with mutually well connected genes of similar functional families: zinc fingers (PF00096), flagella, and ribosomes respectively. In conclusion, these four manuscripts demonstrate the usefulness of mathematical models and statistical tools as a source of new biological insight. While the clustering approach of gene coexpression data leads to the phenotypic characterization of so far unknown genes and thus supports genome annotation, our model approaches offer explanations for observed properties of the coexpression network and furthermore substantiate punctuated equilibrium as an evolutionary process by a deeper understanding of an underlying molecular mechanism. / Die biologische Zelle ist ein sehr kompliziertes Gebilde. Bei ihrer Betrachtung gilt es, das Zusammenspiel von Tausenden bis Millionen von Genen, Regulatoren, Proteinen oder Molekülen zu beschreiben und zu verstehen. Durch enorme Verbesserungen experimenteller Messgeräte gelingt es mittlerweile allerdings in geringer Zeit enorme Datenmengen zu messen, seien dies z.B. die Entschlüsselung eines Genoms oder die Konzentrationen der Moleküle in einer Zelle. Die Systembiologie nimmt sich dem Problem an, aus diesem Datenmeer ein quantitatives Verständnis für die Gesamtheit der Wechselwirkungen in der Zelle zu entwickeln. Dabei stellt die mathematische Modellierung und computergestützte Analyse ein eminent wichtiges Werkzeug dar, lassen sich doch am Computer in kurzer Zeit eine Vielzahl von Fällen testen und daraus Hypothesen generieren, die experimentell verifiziert werden können. Diese Doktorarbeit beschäftigt sich damit, wie durch mathematische Modellierung Rückschlüsse auf die Evolution und deren Mechanismen geschlossen werden können. Dabei besteht die Arbeit aus zwei Teilen. Zum Einen wurde ein Modell entwickelt, dass die Evolution des Stoffwechsels nachbaut. Der zweite Teil beschäftigt sich mit der Analyse von Genexpressionsdaten, d.h. der Stärke mit der ein bestimmtes Gen in ein Protein umgewandelt, "exprimiert", wird. Der Stoffwechsel bezeichnet die Gesamtheit der chemischen Vorgänge in einem Organismus; zum Einen werden Nahrungsstoffe für den Organismus verwertbar zerlegt, zum Anderen aber auch neue Stoffe aufgebaut. Da für nahezu jede chemische Reaktion ein katalysierendes Enzym benötigt wird, ist davon auszugehen, dass sich der Stoffwechsel parallel zu den Enzymen entwickelt hat. Auf dieser Annahme basiert das entwickelte Modell zur Enzyme-Stoffwechsel-Koevolution. Von einer Anfangsmenge von Enzymen und Molekülen ausgehend, die etwa in einer primitiven Atmosphäre vorgekommen sind, werden sukzessive Enzyme und die nun katalysierbaren Reaktionen hinzugefügt, wodurch die Stoffwechselkapazität anwächst. Die Auswahl eines neuen Enzyms geschieht dabei in Abhängigkeit von der Ähnlichkeit mit bereits vorhandenen und ist so an den evolutionären Vorgang der Mutation angelehnt: je ähnlicher ein neues Enzym zu den vorhandenen ist, desto schneller kann es hinzugefügt werden. Dieser Vorgang wird wiederholt, bis der Stoffwechsel die heutige Form angenommen hat. Interessant ist vor allem der zeitliche Verlauf dieser Evolution, der mittels einer Zeitreihenanalyse untersucht wird. Dabei zeigt sich, dass neue Enzyme gebündelt in Gruppen kurzer Zeitfolge auftreten, gefolgt von Intervallen relativer Stille. Dasselbe Phänomen kennt man von der Evolution neuer Arten, die ebenfalls gebündelt auftreten, und wird Punktualismus genannt. Diese Arbeit liefert somit ein besseres Verständnis dieses Phänomens durch eine Beschreibung auf molekularer Ebene. Im zweiten Projekt werden Genexpressionsdaten von Pflanzen analysiert. Einerseits geschieht dies mit einem eigens entwickelten Cluster-Algorithmus. Hier läßt sich beobachten, dass Gene mit einer ähnlichen Funktion oft auch ein ähnliches Expressionsmuster aufweisen. Das Clustering liefert einige Genkandidaten, deren Funktion bisher unbekannt war, von denen aber nun vermutet werden konnte, dass sie enorm wichtig für das Wachstum der Pflanze sind. Durch Experimente von Pflanzen mit und ohne diese Gene zeigte sich, dass sechs neuen Genen dieses essentielle Erscheinungsbild zugeordnet werden kann. Weiterhin wurden Netzwerke der Genexpressionsdaten einer Pflanze, eines Pilzes und eines Bakteriums untersucht. In diesen Netzwerken werden zwei Gene verbunden, falls sie ein sehr ähnliches Expressionsprofil aufweisen. Nun zeigten diese Netzwerke sehr ähnliche und charakteristische Eigenschaften auf. Im Rahmen dieser Arbeit wurde daher ein weiteres evolutionäres Modell entwickelt, das die Expressionsprofile anhand von Duplikation, Mutation und Selektion beschreibt. Obwohl das Modell auf sehr simplen Eigenschaften beruht, spiegelt es die beobachteten Eigenschaften sehr gut wider, und es läßt sich der Schluss ziehen, dass diese als Resultat der Evolution betrachtet werden können. Die Ergebnisse dieser Arbeiten sind als Doktorarbeit in kumulativer Form bestehend aus vier veröffentlichten Artikeln vereinigt.

Deciphering the functioning of biological networks is one of the central tasks in systems biology. In particular, signal transduction networks are crucial for the understanding of the cellular response to external and internal perturbations. Importantly, in order to cope with the complexity of these networks, mathematical and computational modeling is required. We propose a computational modeling framework in order to achieve more robust discoveries in the context of logical signaling networks. More precisely, we focus on modeling the response of logical signaling networks by means of automated reasoning using Answer Set Programming (ASP). ASP provides a declarative language for modeling various knowledge representation and reasoning problems. Moreover, available ASP solvers provide several reasoning modes for assessing the multitude of answer sets. Therefore, leveraging its rich modeling language and its highly efficient solving capacities, we use ASP to address three challenging problems in the context of logical signaling networks: learning of (Boolean) logical networks, experimental design, and identification of intervention strategies. Overall, the contribution of this thesis is three-fold. Firstly, we introduce a mathematical framework for characterizing and reasoning on the response of logical signaling networks. Secondly, we contribute to a growing list of successful applications of ASP in systems biology. Thirdly, we present a software providing a complete pipeline for automated reasoning on the response of logical signaling networks. / Deciphering the functioning of biological networks is one of the central tasks in systems biology. In particular, signal transduction networks are crucial for the understanding of the cellular response to external and internal perturbations. Importantly, in order to cope with the complexity of these networks, mathematical and computational modeling is required. We propose a computational modeling framework in order to achieve more robust discoveries in the context of logical signaling networks. More precisely, we focus on modeling the response of logical signaling networks by means of automated reasoning using Answer Set Programming (ASP). ASP provides a declarative language for modeling various knowledge representation and reasoning problems. Moreover, available ASP solvers provide several reasoning modes for assessing the multitude of answer sets. Therefore, leveraging its rich modeling language and its highly efficient solving capacities, we use ASP to address three challenging problems in the context of logical signaling networks: learning of (Boolean) logical networks, experimental design, and identification of intervention strategies. Overall, the contribution of this thesis is three-fold. Firstly, we introduce a mathematical framework for characterizing and reasoning on the response of logical signaling networks. Secondly, we contribute to a growing list of successful applications of ASP in systems biology. Thirdly, we present a software providing a complete pipeline for automated reasoning on the response of logical signaling networks.

The cell is the central functional unit of life. Cell behaviors, such as cell division, movements, differentiation, cell death as well as cell shape and size changes, determine how tissues change shape and grow during regeneration and development. However, a generally applicable framework to measure and describe the behavior of the multitude of cells in a developing tissue is still lacking. Furthermore, the specific contribution of individual cell behaviors, and how exactly these cell behaviors collectively lead to the morphogenesis and growth of tissues are not clear for many developmental and regenerative processes. A promising strategy to fill these gaps is the continuing effort of making developmental biology a quantitative science. Recent advances in methods, especially in imaging, enable measurements of cell behaviors and tissue shapes in unprecedented detail and accuracy. Consequently, formalizing hypotheses in terms of mathematical models to obtain testable quantitative predictions is emerging as a powerful tool. Tests of the hypotheses involve the comparison of model predictions to experimentally observed data. The available data is often noisy and based on only few samples. Hence, this comparison of data and model predictions often requires very careful use of statistical inference methods. If one chooses this quantitative approach, the challenges are the choice of observables, i.e. what to measure, and the design of appropriate data-driven models to answer relevant questions. In this thesis, I applied this data-driven modeling approach to vertebrate morphogenesis, growth and regeneration. In particular, I study spinal cord and muscle regeneration in axolotl, muscle development in zebrafish, and neuron development and maintenance in the adult human brain. To do so, I analyzed images to quantify cell behaviors and tissue shapes. Especially for cell behaviors in post-embryonic tissues, measurements of some cell behavior parameters, such as the proliferation rate, could not be made directly. Hence, I developed mathematical models that are specifically designed to infer these parameters from indirect experimental data. To understand how cell behaviors shape tissues, I developed mechanistic models that causally connect the cell and tissue scales. Specifically, I first investigated the behaviors of neural stem cells that underlie the regenerative outgrowth of the spinal cord after tail amputation in the axolotl. To do so, I quantified all relevant cell behaviors. A detailed analysis of the proliferation pattern in space and time revealed that the cell cycle is accelerated between 3-4 days after amputation in a high-proliferation zone, initially spanning from 800 µm anterior to the amputation plane. The activation of quiescent stem cells and cell movements into the high-proliferation zone also contribute to spinal cord growth but I did not find contributions by cellular rearrangements or cell shape changes. I developed a mathematical model of spinal cord outgrowth involving all contributing cell behaviors which revealed that the acceleration of the cell cycle is the major driver of spinal cord outgrowth. To compare the behavior of neural stem cells with cell behaviors in the regenerating muscle tissue that surrounds the spinal cord, I also quantified proliferation of mesenchymal progenitor cells and found similar proliferation parameters. I showed that the zone of mesenchymal progenitors that gives rise to the regenerating muscle segments is at least 350 µm long, which is consistent with the length of the high-proliferation zone in the spinal cord. Second, I investigated shape changes in developing zebrafish muscle segments by quantifying time-lapse movies of developing zebrafish embryos. These data challenged or ruled out a number of previously proposed mechanisms. Motivated by reported cellular behaviors happening simultaneously in the anterior segments, I had previously proposed the existence of a simple tension-and-resistance mechanism that shapes the muscle segments. Here, I could verify the predictions of this mechanism for the final segment shape pattern. My results support the notion that a simple physical mechanism suffices to self-organize the observed spatiotemporal pattern in the muscle segments. Third, I corroborated and refined previous estimates of neuronal cell turnover rates in the adult human hippocampus. Previous work approached this question by combining quantitative data and mathematical modeling of the incorporation of the carbon isotope C-14. I reanalyzed published data using the published deterministic neuron turnover model but I extended the model by a better justified measurement error model. Most importantly, I found that human adult neurogenesis might occur at an even higher rate than currently believed. The tools I used throughout were (1) the careful quantification of the involved processes, mainly by image analysis, and (2) the derivation and application of mathematical models designed to integrate the data through (3) statistical inference. Mathematical models were used for different purposes such as estimating unknown parameters from indirect experiments, summarizing datasets with a few meaningful parameters, formalizing mechanistic hypotheses, as well as for model-guided experimental planning. I venture an outlook on how additional open questions regarding cell turnover measurements could be answered using my approach. Finally, I conclude that the mechanistic understanding of development and regeneration can be advanced by comparing quantitative data to the predictions of specifically designed mathematical models by means of statistical inference methods.

The cell is the central functional unit of life. Cell behaviors, such as cell division, movements, differentiation, cell death as well as cell shape and size changes, determine how tissues change shape and grow during regeneration and development. However, a generally applicable framework to measure and describe the behavior of the multitude of cells in a developing tissue is still lacking. Furthermore, the specific contribution of individual cell behaviors, and how exactly these cell behaviors collectively lead to the morphogenesis and growth of tissues are not clear for many developmental and regenerative processes. A promising strategy to fill these gaps is the continuing effort of making developmental biology a quantitative science. Recent advances in methods, especially in imaging, enable measurements of cell behaviors and tissue shapes in unprecedented detail and accuracy. Consequently, formalizing hypotheses in terms of mathematical models to obtain testable quantitative predictions is emerging as a powerful tool. Tests of the hypotheses involve the comparison of model predictions to experimentally observed data. The available data is often noisy and based on only few samples. Hence, this comparison of data and model predictions often requires very careful use of statistical inference methods. If one chooses this quantitative approach, the challenges are the choice of observables, i.e. what to measure, and the design of appropriate data-driven models to answer relevant questions. In this thesis, I applied this data-driven modeling approach to vertebrate morphogenesis, growth and regeneration. In particular, I study spinal cord and muscle regeneration in axolotl, muscle development in zebrafish, and neuron development and maintenance in the adult human brain. To do so, I analyzed images to quantify cell behaviors and tissue shapes. Especially for cell behaviors in post-embryonic tissues, measurements of some cell behavior parameters, such as the proliferation rate, could not be made directly. Hence, I developed mathematical models that are specifically designed to infer these parameters from indirect experimental data. To understand how cell behaviors shape tissues, I developed mechanistic models that causally connect the cell and tissue scales. Specifically, I first investigated the behaviors of neural stem cells that underlie the regenerative outgrowth of the spinal cord after tail amputation in the axolotl. To do so, I quantified all relevant cell behaviors. A detailed analysis of the proliferation pattern in space and time revealed that the cell cycle is accelerated between 3-4 days after amputation in a high-proliferation zone, initially spanning from 800 µm anterior to the amputation plane. The activation of quiescent stem cells and cell movements into the high-proliferation zone also contribute to spinal cord growth but I did not find contributions by cellular rearrangements or cell shape changes. I developed a mathematical model of spinal cord outgrowth involving all contributing cell behaviors which revealed that the acceleration of the cell cycle is the major driver of spinal cord outgrowth. To compare the behavior of neural stem cells with cell behaviors in the regenerating muscle tissue that surrounds the spinal cord, I also quantified proliferation of mesenchymal progenitor cells and found similar proliferation parameters. I showed that the zone of mesenchymal progenitors that gives rise to the regenerating muscle segments is at least 350 µm long, which is consistent with the length of the high-proliferation zone in the spinal cord. Second, I investigated shape changes in developing zebrafish muscle segments by quantifying time-lapse movies of developing zebrafish embryos. These data challenged or ruled out a number of previously proposed mechanisms. Motivated by reported cellular behaviors happening simultaneously in the anterior segments, I had previously proposed the existence of a simple tension-and-resistance mechanism that shapes the muscle segments. Here, I could verify the predictions of this mechanism for the final segment shape pattern. My results support the notion that a simple physical mechanism suffices to self-organize the observed spatiotemporal pattern in the muscle segments. Third, I corroborated and refined previous estimates of neuronal cell turnover rates in the adult human hippocampus. Previous work approached this question by combining quantitative data and mathematical modeling of the incorporation of the carbon isotope C-14. I reanalyzed published data using the published deterministic neuron turnover model but I extended the model by a better justified measurement error model. Most importantly, I found that human adult neurogenesis might occur at an even higher rate than currently believed. The tools I used throughout were (1) the careful quantification of the involved processes, mainly by image analysis, and (2) the derivation and application of mathematical models designed to integrate the data through (3) statistical inference. Mathematical models were used for different purposes such as estimating unknown parameters from indirect experiments, summarizing datasets with a few meaningful parameters, formalizing mechanistic hypotheses, as well as for model-guided experimental planning. I venture an outlook on how additional open questions regarding cell turnover measurements could be answered using my approach. Finally, I conclude that the mechanistic understanding of development and regeneration can be advanced by comparing quantitative data to the predictions of specifically designed mathematical models by means of statistical inference methods.

Ziel dieser Arbeit war es, den zentralen Kohlenstoffwechsel mit besonderem Fokus auf Regulation zu untersuchen, insbesondere durch die Auftrennung von zwei Regulationsebenen: metabolische Regulation, assoziiert mit direkten Wech- selwirkungen zwischen Metaboliten und Enzymen, sowie hierarchische Regulation, assoziiert mit Änderungen in Enzymmengenänderungen durch die Regulation von de novo Enzymproduktion. Unsere Untersuchungen basieren größtenteils auf drei Datensätzen aus glukoselimitierten Chemostatkulturen von S. cerevisiae. Im Kap. 2 wurden Extrazelluläre Bedingungen im Makroskopischen unter- sucht. Das wichtigsten Ergebnis dieser the- oretischen Analyse ist die Charakterisierung des Selektionsdruckes in einem Chemostatkultur. Im Kap. 4 wurde eine Analyse auf Systemebene des zentralen Kohlenstoffwech- sels durchgeführt. Unter Verwendung der Metaboliten- und der Flußdaten wurde ein kinetisches Modell konstruiert, welches wesentliche Teile des zentralen Kohlen- stoffwechsels umfaßt. Die meisten kinetischen Ausdrücke und Parameterwerte wurden aus einem bestehenden kinetischen Modells (Teusink-Modell) übernom- men. / In this work, we aimed to elucidate central carbon metabolism focusing on the aspect of regulation, especially by separating two regulatory levels: metabolic regulation, associated with direct interactions of metabolites and enzymes, and hierarchic regulation, associated with enzyme level change via regulation of de novo enzyme production. Our investigations were largely based on the analysis of three datasets from glucose limited continuous cultures of S. cerevisiae. Extracellular conditions on the macroscopic scale were investigated in Chapter 2. This was inspired by the perceived lack of clarity regarding an important aspect: concentration of glucose, the limiting nutrient and main carbon source in these cultures. The main outcome of this theoretical analysis was characterisation of the selection pressure in a chemostat culture, as selecting for cells which produce the growth rate, defined by the pre-set dilution rate, with lower external concentration of the limiting nutrient. Flux regulation on the scale of individual enzymes was investigated for selected reactions in Chapter 3. This analysis was based on the attempt to reproduce flux changes through these reactions, using enzyme kinetic expressions with inputs from the three aforementioned datasets. The notion of hierarchic and metabolic regulation was introduced and modified. System-level analysis of central carbon metabolism was undertaken in Chap- ter 4. Using the information on metabolite levels and flux, a kinetic model representing significant parts of central carbon metabolism was constructed. To get feasible flux distributions, constrained metabolic flux balance analysis was performed, using a stoichiometric network, constructed to be consistent with the model’s stoichiometry. Fitting the model resulted in two sets of parameters corresponding to steady states reproducing, the nominal data values of the anaerobic and the fully aerobic conditions.

Das Leben und Überleben einer Zelle wird auf verschiedenen Ebenen streng reguliert. Diese Ebenen sind eng miteinander verknüpft: (i) Signalwege leiten extrazelluläre Signale in den Zellkern, wo (ii) Genregulation sie zu Proteinen übersetzt, und (iii) Proteine kontrollieren metabolische Funktionen, die Nährstoffe zu Energie und zellulären Bausteinen konvertieren. Diese Systeme sind hochkomplex und werden oft nur einzeln betrachtet. Systembiologie ist ein interdisziplinäres Forschungsgebiet, das Methoden anbietet, um Informationen aus heutigen Hochdurchsatz-Experimenttechnologien zu extrahieren. Diese Methoden können effektiv sein, um die vorgenannten Systeme einzeln oder im Ganzen zu untersuchen. In dieser Doktorarbeit wende ich Methoden an, um Signalwege und Zellmetabolismus zu erforschen, und ich präsentiere neue Arbeitsabläufe für das Modellieren und Analysieren dieser Systeme. Beide Methoden sind auf großskalige Netzwerkrekonstruktionen fokussiert. Da die Erhältlichkeit von xperimentellen Daten eines der größten Probleme der Systembiologie darstellt, befassen sich die Methoden explizit mit dem Umgang mit Wissenslücken. Sie werden auf den Snf1 Signalweg und den Metabolismus von Hefezellen angewendet und vermitteln neue Erkenntnisse über diesen Modellorganismus. Des Weiteren präsentiert diese Arbeit eine eingehende Analyse vom metabolischen Reprogrammieren in Darmkrebszellen, welche bisher unbekannte Zusammenhänge von metabolischer Funktionalität und Onkogenen beinhaltet. Zum Abschluss stelle ich unseren Vorschlag für ein standardisiertes Datenaustauschformat vor, welches seinen Schwerpunkt auf Datentabellen der Systembiologie legt. Zusammenfassend behandelt diese Doktorarbeit die Signalwege und den Metabolismus von Zellen, inklusive neuer Modellierabläufe und biologischer Erkenntnisse. Diese Erkenntnisse werden in den Kontext unseres aktuellen Wissensstandes gesetzt und darauf aufbauend werden neue potentielle Ansatzpunkte für Experimente vorgeschlagen. / Cellular life is governed on different layers of regulation, which are tightly interconnected: (i) Signalling pathways transmit extracellular signals to the cells’ nucleus, where (ii) gene regulation translates these signals into proteins, and (iii) proteins control metabolic functions, which convert nutrients to energy and cell building blocks. Due to the complexity of each of these systems, they are often analysed individually or only partially. Systems Biology is an interdisciplinary field of research that offers techniques to harvest the information of todays high-throughput experiments. These techniques can be powerful approaches to investigate the aforementioned regulatory layers of a cell either individually or as a whole. In this thesis, I am employing means of Systems Biology to explore signalling pathways and metabolism, and I provide novel workflows for modelling and exploring these systems. Both workflows are focussed on accurate large-scale network reconstructions of the target system. Since one of the major problems in Systems Biology is the availability of experimental data, the workflows put emphasis on the handling of knowledge gaps. They are applied on the Snf1 pathway and metabolism in yeast and provide new findings about this model organism. Furthermore, this thesis presents an in-depth analysis of metabolic reprogramming in colorectal cancer cells, which yields previously unknown coherences of metabolic function and oncogenes. Finally, I am presenting a proposal for a standardised data format in Systems Biology, which is based on data tables. In summary, this thesis comprises works on signalling pathways and cell metabolism, which includes novel modelling workflows and new biological findings, analyses their impact on the scientific state of the art, and proposes directions for new experimental targets.

Mathematical modeling is an important tool in improving the understanding of complex biological processes. However, mathematical models are often faced with challenges that arise due to the limited knowledge of the underlying biological processes and the high number of parameters for which exact values are unknown. The method of generalized modeling is an alternative modeling approach that aims to address these challenges by extracting information about stability and bifurcations of classes of models while making only minimal assumptions on the specific functional forms of the model. This is achieved by a direct parameterization of the Jacobian in the steady state, introducing a set of generalized parameters which have a biological interpretation. In this thesis, the method of generalized modeling is extended and applied to different problems from cell biology. In the first part, we extend the method to include also the higher derivatives at the steady state. This allows an analysis of the normal form of bifurcations and thereby a more specific description of the nearby dynamics. In models of gene-regulatory networks, it is shown that the extended method can be applied to better characterize oscillatory systems and to detect bistable dynamics. In the second part, we investigate mathematical models of bone remodeling, a process that renews the human skeleton constantly. We investigate the connection between structural properties of mathematical models and the stability of steady states in different models. We find that the dynamical system operates from a stable steady state that is situated in the vicinity of bifurcations where stability can be lost, potentially leading to diseases of bone. In the third part of this thesis, models of the MAPK signal transduction pathway are analyzed. Since mathematical models for this system include a high number of parameters, statistical methods are employed to analyze stability and bifurcations. Thereby, the parameters with a strong influence on the stability of steady states are identified. By an analysis of the bifurcation structure of the MAPK cascade, it is found that a combination of multiple layers in a cascade-like way allows for additional types of dynamic behavior such as oscillations and chaos. In summary, this thesis shows that generalized modeling is a fruitful alternative modeling approach for various types of systems in cell biology. / Mathematische Modelle stellen ein wichtiges Hilfmittel zur Verbesserung des Verständnisses komplexer biologischer Prozesse dar. Sie stehen jedoch vor Schwierigkeiten, wenn wenig über die zugrundeliegende biologischen Vorgänge bekannt ist und es eine große Anzahl von Parametern gibt, deren exakten Werte unbekannt sind. Die Methode des Verallgemeinerten Modellierens ist ein alternativer Modellierungsansatz mit dem Ziel, diese Schwierigkeiten dadurch anzugehen, dass dynamische Informationen über Stabilität und Bifurkationen aus Klassen von Modellen extrahiert werden, wobei nur minimale Annahmen über die spezifischen funktionalen Formen getätigt werden. Dies wird erreicht durch eine direkte Parametrisierung der Jacobimatrix im Gleichgewichtszustand, bei der neue, verallgemeinerte Parameter eingeführt werden, die eine biologische Interpretation besitzen. In dieser Arbeit wird die Methode des Verallgemeinerten Modellierens erweitert und auf verschiedene zellbiologische Probleme angewandt. Im ersten Teil wird eine Erweiterung der Methode vorgestellt, bei der die Analyse höherer Ableitungen im Gleichgewichtszustand integriert wird. Dies erlaubt die Bestimmung der Normalform von Bifurkationen und hierdurch eine spezifischere Beschreibung der Dynamik in deren Umgebung. In Modellen für genregulatorische Netzwerke wird gezeigt, dass die so erweiterte Methode zu einer besseren Charakterisierung oszillierender Systeme sowie zur Erkennung von Bistabilität verwendet werden kann. Im zweiten Teil werden mathematische Modelle zur Knochenremodellierung untersucht, einem Prozess der das menschliche Skelett kontinuierlich erneuert. Wir untersuchen den Zusammenhang zwischen strukturellen Eigenschaften verschiedener Modelle und der Stabilität von Gleichgewichtszuständen. Wir finden, dass das dynamische System von einem stabilen Zustand operiert, in dessen Nähe Bifurkationen existieren, welche das System destabilisieren und so potentiell Knochenkranheiten verursachen können. Im dritten Teil werden Modelle für den MAPK Signaltransduktionsweg analysiert. Da mathematische Modelle für dieses System eine hohe Anzahl von Parametern beinhalten, werden statistische Methoden angewandt zur Analyse von Stabilität und Bifurkationen. Zunächst werden Parameter mit einem starken Einfluss auf die Stabilität von Gleichgewichtszuständen identifizert. Durch eine Analyse der Bifurkationsstruktur wird gezeigt, dass eine kaskadenartige Kombination mehrerer Ebenen zu zusätzliche Typen von Dynamik wie Oszillationen und Chaos führt. Zusammengefasst zeigt diese Arbeit, dass Verallgemeinertes Modellieren ein fruchtbarer alternativer Modellierungsansatz für verschiedene zellbiologische Probleme ist.

Mathematical modeling is an important tool in improving the understanding of complex biological processes. However, mathematical models are often faced with challenges that arise due to the limited knowledge of the underlying biological processes and the high number of parameters for which exact values are unknown. The method of generalized modeling is an alternative modeling approach that aims to address these challenges by extracting information about stability and bifurcations of classes of models while making only minimal assumptions on the specific functional forms of the model. This is achieved by a direct parameterization of the Jacobian in the steady state, introducing a set of generalized parameters which have a biological interpretation. In this thesis, the method of generalized modeling is extended and applied to different problems from cell biology. In the first part, we extend the method to include also the higher derivatives at the steady state. This allows an analysis of the normal form of bifurcations and thereby a more specific description of the nearby dynamics. In models of gene-regulatory networks, it is shown that the extended method can be applied to better characterize oscillatory systems and to detect bistable dynamics. In the second part, we investigate mathematical models of bone remodeling, a process that renews the human skeleton constantly. We investigate the connection between structural properties of mathematical models and the stability of steady states in different models. We find that the dynamical system operates from a stable steady state that is situated in the vicinity of bifurcations where stability can be lost, potentially leading to diseases of bone. In the third part of this thesis, models of the MAPK signal transduction pathway are analyzed. Since mathematical models for this system include a high number of parameters, statistical methods are employed to analyze stability and bifurcations. Thereby, the parameters with a strong influence on the stability of steady states are identified. By an analysis of the bifurcation structure of the MAPK cascade, it is found that a combination of multiple layers in a cascade-like way allows for additional types of dynamic behavior such as oscillations and chaos. In summary, this thesis shows that generalized modeling is a fruitful alternative modeling approach for various types of systems in cell biology. / Mathematische Modelle stellen ein wichtiges Hilfmittel zur Verbesserung des Verständnisses komplexer biologischer Prozesse dar. Sie stehen jedoch vor Schwierigkeiten, wenn wenig über die zugrundeliegende biologischen Vorgänge bekannt ist und es eine große Anzahl von Parametern gibt, deren exakten Werte unbekannt sind. Die Methode des Verallgemeinerten Modellierens ist ein alternativer Modellierungsansatz mit dem Ziel, diese Schwierigkeiten dadurch anzugehen, dass dynamische Informationen über Stabilität und Bifurkationen aus Klassen von Modellen extrahiert werden, wobei nur minimale Annahmen über die spezifischen funktionalen Formen getätigt werden. Dies wird erreicht durch eine direkte Parametrisierung der Jacobimatrix im Gleichgewichtszustand, bei der neue, verallgemeinerte Parameter eingeführt werden, die eine biologische Interpretation besitzen. In dieser Arbeit wird die Methode des Verallgemeinerten Modellierens erweitert und auf verschiedene zellbiologische Probleme angewandt. Im ersten Teil wird eine Erweiterung der Methode vorgestellt, bei der die Analyse höherer Ableitungen im Gleichgewichtszustand integriert wird. Dies erlaubt die Bestimmung der Normalform von Bifurkationen und hierdurch eine spezifischere Beschreibung der Dynamik in deren Umgebung. In Modellen für genregulatorische Netzwerke wird gezeigt, dass die so erweiterte Methode zu einer besseren Charakterisierung oszillierender Systeme sowie zur Erkennung von Bistabilität verwendet werden kann. Im zweiten Teil werden mathematische Modelle zur Knochenremodellierung untersucht, einem Prozess der das menschliche Skelett kontinuierlich erneuert. Wir untersuchen den Zusammenhang zwischen strukturellen Eigenschaften verschiedener Modelle und der Stabilität von Gleichgewichtszuständen. Wir finden, dass das dynamische System von einem stabilen Zustand operiert, in dessen Nähe Bifurkationen existieren, welche das System destabilisieren und so potentiell Knochenkranheiten verursachen können. Im dritten Teil werden Modelle für den MAPK Signaltransduktionsweg analysiert. Da mathematische Modelle für dieses System eine hohe Anzahl von Parametern beinhalten, werden statistische Methoden angewandt zur Analyse von Stabilität und Bifurkationen. Zunächst werden Parameter mit einem starken Einfluss auf die Stabilität von Gleichgewichtszuständen identifizert. Durch eine Analyse der Bifurkationsstruktur wird gezeigt, dass eine kaskadenartige Kombination mehrerer Ebenen zu zusätzliche Typen von Dynamik wie Oszillationen und Chaos führt. Zusammengefasst zeigt diese Arbeit, dass Verallgemeinertes Modellieren ein fruchtbarer alternativer Modellierungsansatz für verschiedene zellbiologische Probleme ist.

Embryonic development depends on the precise coordination of cell fate specification, patterning and morphogenesis. Although great strides have been made in the molecular understanding of each of these processes, how their interplay governs the formation of complex tissues remains poorly understood. New techniques for experimental manipulation and image quantification enable the study of development in unprecedented detail, resulting in new hypotheses on the interactions between known components. By expressing these hypotheses in terms of rules and equations, computational modeling and simulation allows one to test their consistency against experimental data. However, new computational methods are required to represent and integrate the network of interactions between gene regulation, signaling and biomechanics that extend over the molecular, cellular and tissue scales. In this thesis, I present a framework that facilitates computational modeling of multiscale multicellular systems and apply it to investigate pancreatic development and the formation of vascular networks. This framework is based on the integration of discrete cell-based models with continuous models for intracellular regulation and intercellular signaling. Specifically, gene regulatory networks are represented by differential equations to analyze cell fate regulation; interactions and distributions of signaling molecules are modeled by reaction-diffusion systems to study pattern formation; and cell-cell interactions are represented in cell-based models to investigate morphogenetic processes. A cell-centered approach is adopted that facilitates the integration of processes across the scales and simultaneously constrains model complexity. The computational methods that are required for this modeling framework have been implemented in the software platform Morpheus. This modeling and simulation environment enables the development, execution and analysis of multi-scale models of multicellular systems. These models are represented in a new domain-specific markup language that separates the biological model from the computational methods and facilitates model storage and exchange. Together with a user-friendly graphical interface, Morpheus enables computational modeling of complex developmental processes without programming and thereby widens its accessibility for biologists. To demonstrate the applicability of the framework to problems in developmental biology, two case studies are presented that address different aspects of the interplay between cell fate specification, patterning and morphogenesis. In the first, I focus on the interplay between cell fate stability and intercellular signaling. Specifically, two studies are presented that investigate how mechanisms of cell-cell communication affect cell fate regulation and spatial patterning in the pancreatic epithelium. Using bifurcation analysis and simulations of spatially coupled differential equations, it is shown that intercellular communication results in a multistability of gene expression states that can explain the scattered spatial distribution and low cell type ratio of nascent islet cells. Moreover, model analysis shows that disruption of intercellular communication induces a transition between gene expression states that can explain observations of in vitro transdifferentiation from adult acinar cells into new islet cells. These results emphasize the role of the multicellular context in cell fate regulation during development and may be used to optimize protocols for cellular reprogramming. The second case study focuses on the feedback between patterning and morphogenesis in the context of the formation of vascular networks. Integrating a cell-based model of endothelial chemotaxis with a reaction-diffusion model representing signaling molecules and extracellular matrix, it is shown that vascular network patterns with realistic morphometry can arise when signaling factors are retained by cell-modified matrix molecules. Through the validation of this model using in vitro assays, quantitative estimates are obtained for kinetic parameters that, when used in quantitative model simulations, confirm the formation of vascular networks under measured biophysical conditions. These results demonstrate the key role of the extracellular matrix in providing spatial guidance cues, a fact that may be exploited to enhance vascularization of engineered tissues. Together, the modeling framework, software platform and case studies presented in this thesis demonstrate how cell-centered computational modeling of multi-scale and multicellular systems provide powerful tools to help disentangle the complex interplay between cell fate specification, patterning and morphogenesis during embryonic development.

Die Regulation der Genexpression ist einer der wichtigsten zellulären Prozesse und steht in Zusammenhang mit der Entstehung diverser Krankheiten. Regulationsmechanismen können mit einer Vielzahl von Methoden experimentell untersucht werden, zugleich erfordert die Integration der Datensätze in umfassende Modelle stringente rechnergestützte Methoden. Ein Teil dieser Methoden modelliert die genomweite Genexpression als (lineares) Gleichungssystem über die Aktivität und Beziehungen von Transkriptionsfaktoren (TF), Genen und anderen Faktoren und optimiert die Parameter, sodass die gemessenen Expressionsintensitäten möglichst genau wiedergegeben werden. Trotz ihrer gemeinsamen Wurzeln in der mathematischen Optimierung unterscheiden sich die Methoden stark in der Art der integrierten Daten, im für ihre Anwendung notwendigen Hintergrundwissen, der Granularität des Regulationsmodells, des konkreten Paradigmas zur Lösung des Optimierungsproblems, und der zur Evaluation verwendeten Datensätze. In dieser Arbeit betrachten wir fünf solcher Methoden und stellen einen qualitativen und quantitativen Vergleich auf. Unsere Ergebnisse zeigen, dass die Überschneidungen der Ergebnisse sehr gering sind, was nicht auf die Stichprobengröße oder das regulatorische Netzwerk zurückgeführt werden kann. Ein Grund für die genannten Defizite könnten die vereinfachten Modelle zellulärer Prozesse sein, da diese vorhandene Rückkopplungsschleifen ignorieren. Wir schlagen eine neue Methode (Florae) mit Schwerpunkt auf die Berücksichtigung von Rückkopplungsschleifen vor und beurteilen deren Ergebnisse. Mit Floræ können wir die Identifizierung von Knockout- und Knockdown-TF in synthetischen Datensätzen verbessern. Unsere Ergebnisse und die vorgeschlagene Methode erweitern das Wissen über genregulatorische Aktivität können die Identifizierung von Ursachen und Mechanismen regulatorischer (Dys-)Funktionen und die Entwicklung von medizinischen Biomarkern und Therapien unterstützen. / Gene regulation is one of the most important cellular processes and closely interlinked pathogenesis. The elucidation of regulatory mechanisms can be approached by many experimental methods, yet integration of the resulting heterogeneous, large, and noisy data sets into comprehensive models requires rigorous computational methods. A prominent class of methods models genome-wide gene expression as sets of (linear) equations over the activity and relationships of transcription factors (TFs), genes and other factors and optimizes parameters to fit the measured expression intensities. Despite their common root in mathematical optimization, they vastly differ in the types of experimental data being integrated, the background knowledge necessary for their application, the granularity of their regulatory model, the concrete paradigm used for solving the optimization problem and the data sets used for evaluation. We review five recent methods of this class and compare them qualitatively and quantitatively in a unified framework. Our results show that the result overlaps are very low, though sometimes statistically significant. This poor overall performance cannot be attributed to the sample size or to the specific regulatory network provided as background knowledge. We suggest that a reason for this deficiency might be the simplistic model of cellular processes in the presented methods, where TF self-regulation and feedback loops were not represented. We propose a new method for estimating transcriptional activity, named Florae, with a particular focus on the consideration of feedback loops and evaluate its results. Using Floræ, we are able to improve the identification of knockout and knockdown TFs in synthetic data sets. Our results and the proposed method extend the knowledge about gene regulatory activity and are a step towards the identification of causes and mechanisms of regulatory (dys)functions, supporting the development of medical biomarkers and therapies.