This thesis mainly concerns modeling, approximation and inference of gene regulatory dynamics on the basis of gene expression patterns. The dynamical behavior of gene expressions is represented by a system of ordinary dierential equations. We introduce a gene-interaction matrix with some nonlinear entries, in particular, quadratic polynomials of the expression levels to keep the system solvable. The model parameters are determined by using optimization. Then, we provide the time-discrete approximation of our time-continuous model. We analyze the approximating model under the aspect of stability. Finally, from the considered models we derive gene regulatory networks, discuss their qualitative features of the networks and provide a basis for analyzing networks with nonlinear connections.

abstract: Our ability to understand networks is important to many applications, from the analysis and modeling of biological networks to analyzing social networks. Unveiling network dynamics allows us to make predictions and decisions. Moreover, network dynamics models have inspired new ideas for computational methods involving multi-agent cooperation, offering effective solutions for optimization tasks. This dissertation presents new theoretical results on network inference and multi-agent optimization, split into two parts - The first part deals with modeling and identification of network dynamics. I study two types of network dynamics arising from social and gene networks. Based on the network dynamics, the proposed network identification method works like a `network RADAR', meaning that interaction strengths between agents are inferred by injecting `signal' into the network and observing the resultant reverberation. In social networks, this is accomplished by stubborn agents whose opinions do not change throughout a discussion. In gene networks, genes are suppressed to create desired perturbations. The steady-states under these perturbations are characterized. In contrast to the common assumption of full rank input, I take a laxer assumption where low-rank input is used, to better model the empirical network data. Importantly, a network is proven to be identifiable from low rank data of rank that grows proportional to the network's sparsity. The proposed method is applied to synthetic and empirical data, and is shown to offer superior performance compared to prior work. The second part is concerned with algorithms on networks. I develop three consensus-based algorithms for multi-agent optimization. The first method is a decentralized Frank-Wolfe (DeFW) algorithm. The main advantage of DeFW lies on its projection-free nature, where we can replace the costly projection step in traditional algorithms by a low-cost linear optimization step. I prove the convergence rates of DeFW for convex and non-convex problems. I also develop two consensus-based alternating optimization algorithms --- one for least square problems and one for non-convex problems. These algorithms exploit the problem structure for faster convergence and their efficacy is demonstrated by numerical simulations. I conclude this dissertation by describing future research directions. / Dissertation/Thesis / Doctoral Dissertation Electrical Engineering 2017

Chez les mammifères, l'empreinte génomique parentale est un mécanisme épigénétique restreignant l'expression d'une centaine de gènes à un seul allèle, déterminé selon son origine parentale. Les gènes affectés et les mécanismes sous-jacents à leur expression mono-allélique sont essentiellement déterminés par une marque épigénétique différentielle portés par les allèles maternel et paternel. D'un point de vue fonctionnel et au niveau physiologique, l'empreinte est actuellement comprise comme un mécanisme contrôlant la quantité de ressources attribuées par la mère à sa progéniture. Les gènes soumis à empreinte s'inscrivent dans un même réseau transcriptionnel (IGN), et plusieurs études indiquent qu'ils contrôleraient l'équilibre entre prolifération et quiescence de cellules souches adultes. A travers cette étude, nous montrons une induction coordonnée de la plupart des gènes soumis à empreinte lors de la sortie du cycle cellulaire, que celle-ci soit réversible (quiescence) ou non (différenciation). De plus, dans un modèle de pré-adipocytes 3T3-L1, la perturbation de la dynamique d'expression de plusieurs de ces gènes semble conforter l'hypothèse d'un contrôle des transitions entre différents états cellulaires (prolifération, quiescence et différenciation) par l'IGN. Outre l'identification d'une fonction cellulaire commune aux gènes soumis à empreinte, nos résultats ouvrent la voie d'une meilleure compréhension des mécanismes de régulation de la quiescence. De plus, nos conclusions permettent de suggérer un nouveau scénario pour la sélection de l'empreinte parentale au cours de l'évolution des mammifères. / Mammalian genomic imprinting is an epigenetic mechanism that restrains the expression of about a hundred genes to a single allele, in a parent-of-origin specific manner. The identity of imprinted genes and the molecular basis of their monoallelic expression mostly rely on a differential epigenetic marking of the parental alleles. Presently, imprinting is understood as a mechanism aimed at controlling the amount of maternal resources allocated to the offspring. Imprinted genes belong to the same transcriptional network (IGN) and, according to different reports, they seem to control the balance between proliferation and quiescence of adult stem cells. In this study, we show that most imprinted genes are induced upon cell cycle exit, whether reversible (quiescence) or not (differentiation). In addition, within the 3T3-L1 preadipocytes cell line, impairing the dynamics of expression of several imprinted genes impairs the transitions between different cellular states, namely proliferation, quiescence and differentiation. Our results highlight the existence of a common cellular function of imprinted genes, and provide a new frame to understand cellular quiescence, at a molecular level. Furthermore, they suggest a new plausible scenario for the implementation of genomic imprinting during mammalian evolution.

Les Lymphocytes T auxiliaires sont nécessaires pour la production de cytokines adaptées au type d'infection. Différentes sous-populations ont été décrites, parmi lesquelles les Th1, Th2, et Th17, pro-inflammatoires et les iTregs, anti-inflammatoires, exprimant Foxp3. La décision prise par une cellules T naïve de se différentier en l'une de ces populations est étudiée ici.Des découvertes récentes ont montré que les nutriments peuvent modifier la différentiation, mais elles ont négligé la glutamine en dépit de son importance comme source principale d'azote. Dans cette étude, un manque de glutamine induit une expression ectopique de Foxp3 en cours de différentiation en Th1 mais pas en Th2, tout en altérant la différentiation des Th1 et Th17. Cela suggère que, dans des environnements métaboliquement pauvres comme au sein de tumeurs solides, le manque de glutamine pourrait supporter une réponse anti-inflammatoire et donc néfaste.Dans l'optique de comprendre comment la détection de la glutamine influence le réseau de régulation de la différentiation des lymphocytes auxiliaires, une approche de modélisation mathématique a été suivie, consistant d'équation différentielles, et conçue pour capturer les propriétés de cette différentiation. Pour la phase d'apprentissage du modèle, les cinétiques d'expression des principaux facteurs de transcription et cytokines ont été mesurées in vitro en conditions normales, en présence de glutamine. Ces données ont décelé des retards majeurs en terme de transcription, traduction et sécrétion des cytokines, qui à leur tour façonnent l'ordre des évènements qui décident l'issue de la différentiation. Le modèle a reproduit avec succès la dynamique des différentiation 'canoniques', montrant que celles-ci peuvent être expliquées par un réseau de régulation relativement simple. Cependant, le modèle n'a reproduit qu'une partie des propriétés de plasticité des lymphocytes T, et a besoin d'être affiné. Ce n'est qu'alors qu'il pourra être utilisé pour comparer différentes hypothèses mécanistiques sur l'impact de la glutamine sur la différentiation. / T helper cells are required to produce cytokines adapted to the type of infection. Several subsets have been defined, including pro-inflammatory Th1, Th2, Th17; and anti-inflammatory, Foxp3+ iTreg cells. The fate-determining decision of a naive T cell to differentiate into a defined subset was investigated here.Recent findings showed that metabolic constituents impact T cell differentiation, but so far the influence of glutamine on T cell differentiation has been neglected although being the main source of nitrogen. In this study, deprivation of glutamine induced an abnormal expression of Foxp3 under Th1 but not under Th2 condition, while impairing Th1 and Th17 differentiation. Thus, in poor metabolic micro-environments like solid tumours, a lack of glutamine would initiate a detrimental anti-inflammatory response.A mathematical modelling approach using Ordinary Differential Equations was chosen to capture the properties of T cell differentiation, first in normal conditions with glutamine. In order to train the model, kinetics of the master transcription factors and cytokines expression were measured under different T cell differentiation polarizing conditions. The in vitro data revealed major delays in transcription, translation and secretion of cytokines, which shaped the order of fate decision events. The model could successfully reproduce the dynamics of differentiation, confirming that the 'canonical' differentiation in vitro can be explained by a simple regulatory network. However, it only partially reproduced the plastic behaviour of T cells. The mathematical model will be utilized to compare different mechanistic hypotheses linking glutamine sensing to differentiation.

<div><div><p>Important cellular processes such as migration, differentiation, and development often rely on precise timing. Yet, the molecular machinery that regulates timing is inherently noisy. How do cells achieve precise timing with noisy components? We investigate this question using a first-passage-time approach, for an event triggered by a molecule that crosses an abundance threshold. We investigate regulatory strategies that decrease the timing noise of molecular events. We look at several strategies which decrease the noise: i) Regulation performed by an accumulating activator, ii) Regulation dues to a degrading repressor, iii) Auto-regulation and the presence of feedback. We find that either activation or repression outperforms an unregulated strategy. The optimal regulation corresponds to a nonlinear increase in the amount of the target molecule over time, arises from a tradeoff between minimizing the timing noise of the regulator and that of the target molecule itself, and is robust to additional effects such as bursts and cell division. Our results are in quantitative agreement with the nonlinear increase and low noise of <i>mig-1</i> gene expression in migrating neuroblast cells during <i>Caenorhabditis elegans</i> development. These findings suggest that dynamic regulation may be a simple and powerful strategy for precise cellular timing.</p><p>Autoregulatory feedback increases noise. Yet, we find that in the presence of regulation by a second species, autoregulatory feedback decreases noise. To explain this finding, we develop a method to calculate the optimal regulation function that minimizes the timing noise. Our method reveals that the combination of feedback and regulation minimizes noise by maximizing the number of molecular events that must happen in sequence before a threshold is crossed. We compute the optimal timing precision for all two-node networks with regulation and feedback, derive a generic lower bound on timing noise, and compare our results with the neuroblast migration during <i>C. elegans</i> development, as well as two mutants. We finds that indeed our model is aligned with the experimental findings.</p></div></div><div><p>Furthermore, we apply our framework of temporal regulation to explain how the stopping point of the migrating cells in <i>C. elegans</i> depends on the body size. Considering temporal regulation, we find the termination point of the cell for various larval sizes. We discuss three possible mechanisms: i) No compensation; here the migration velocity is constant across the mutants of <i>C. elegans</i>, and this results in the migration distance to be constant but the relative position to be different across various sizes; ii) Total compensation; here the velocity is compensated with body size, hence resulting in the same relative position of cells across mutants; and iii) Partial compensation; here the velocity of migration is correlated with body size to some degree, resulting in a non-linear relationship between termination point and body size. We find that our partial compensation model is consistent with experimental observations of cell termination.</p><p>Finally, we look at the detection of traveling waves by single-celled organisms. Cells must use temporal and spatial information to sense the direction of traveling waves, as seen in cAMP detection by the <i>amoeba </i><i>Dictyostelium</i>. If a cell only uses spatial information to sense the direction of the wave then the cell will move forward when the wave hits the front of the cell, and move backward when the wave hits the back of the cell, resulting in neutral movement. Cells must use temporal information along with spatial information to effectively move towards the source. Here we develop a mechanism by which cells are able to integrate the spatial and temporal information through a system of inhibitors. We find the optimal time to release the inhibitors for maximizing the precision of directional sensing.</p></div>

Microbial fuel cell (MFC) technology is a renewable energy source that employs microorganisms as biocatalysts to degrade substrates into electrons and protons, and then transfer the electrons to the anode electrode. Electron transfer rates by microorganisms depend on many factors as well as on their diverse electron transfer mechanisms. The present study compared cytochromes, flavoproteins, electron transfer complexes, redoxins and other extracellular membrane proteins that have direct involvement in electron transfer mechanisms in Escherichia coli str. K-12 MG1655, Rhodopseudomonas pulastris DX-1 and Shewanella oneidensis MR-1. Escherichia coli str. The results showed that K-12 MG1655 had a more diverse range of extracellular proteins for electron transfer mechanisms compared to Rhodopseudomonas pulastris DX-1 and Shewanella oneidensis MR-1. Escherichia coli str. K-12 MG1655 expressed more flavoproteins, redoxin and electron transfer complex related proteins that had direct involvement in electron transfer mechanisms compared to two other bacterial species indicating that it may be able to transfer more electrons when employed in MFC technique. Escherichia coli str. K-12 MG1655 expressed 16 cytochromes, 9 flavoproteins, 6 redoxins, 6 electron transport complexes, 1 hypothetical and 1 oxidoreductase proteins. On the other hand, Rhodopseudomonas pulastris DX-1 and Shewanella oneidensis MR-1 expressed 26 and 35 cytochromes proteins. But these two bacterial species expressed less flavoproteins and redoxin related proteins and they didn’t express any electron transport complexes or hypothetical and oxidoreductase related proteins for electron transfer. STRING and SMART results suggested that the identified proteins transferred electrons either by connecting with other types of identified proteins in the constructed gene network or independently by taking part in oxidation-reduction reaction, metal ion reduction reaction or by their FMN binding activities.

The copious information generated from transcriptomes gives us an opportunity to learn biological processes as integrated systems; however, due to numerous sources of variation, high dimensions of data structure, various levels of data quality, and different formats of the inputs, dissecting and interpreting such data presents daunting challenges to scientists. The goal of this research is to provide improved and new statistical tools for analyzing transcriptomes data to identify gene expression patterns for classifying samples, to discover regulatory gene networks using natural genetic perturbations, to develop statistical methods for model fitting and comparison of biochemical networks, and eventually to advance our capability to understand the principles of biological processes at the system level. / Ph. D.

Gene expression is tightly regulated by complex transcriptional regulatory mechanisms to achieve specific expression patterns, which are essential to facilitate important biological processes such as embryonic development. Dysregulation of gene expression can lead to diseases such as cancers. A better understanding of the transcriptional regulation will therefore not only advance the understanding of fundamental biological processes, but also provide mechanistic insights into diseases. The earlier versions of high-throughput expression profiling techniques were limited to measuring average gene expression across large pools of cells. In contrast, recent technological improvements have made it possible to perform expression profiling in single cells. Single-cell expression profiling is able to capture heterogeneity among single cells, which is not possible in conventional bulk expression profiling. In my PhD, I focus on developing new algorithms, as well as benchmarking and utilising existing algorithms to study the transcriptomes of various biological systems using single-cell expression data. I have developed two different single-cell specific network inference algorithms, BTR and SPVAR, which are based on two different formalisms, Boolean and autoregression frameworks respectively. BTR was shown to be useful for improving existing Boolean models with single-cell expression data, while SPVAR was shown to be a conservative predictor of gene interactions using pseudotime-ordered single-cell expression data. In addition, I have obtained novel biological insights by analysing single-cell RNAseq data from the epiblast stem cells reprogramming and the leukaemia systems. Three different driver genes, namely Esrrb, Klf2 and GY118F, were shown to drive reprogramming of epiblast stem cells via different reprogramming routes. As for the leukaemia system, FLT3-ITD and IDH1-R132H mutations were shown to interact with each other and potentially predispose some cells for developing acute myeloid leukaemia.

京都大学 / 新制・課程博士 / 博士(薬学) / 甲第23844号 / 薬博第851号 / 新制||薬||242(附属図書館) / 京都大学大学院薬学研究科薬学専攻 / (主査)教授 山下 富義, 教授 石濱 泰, 教授 金子 周司 / 学位規則第4条第1項該当 / Doctor of Pharmaceutical Sciences / Kyoto University / DFAM

Chez tout organisme, l’évolution et l’acclimatation aux changements du milieu de vie sont orchestrés par de nombreux acteurs moléculaires. Parmi eux, les facteurs de transcription (FTs) jouent un rôle clé en régulant l’expression des gènes. Identifier les FTs impliqués dans la production de composés d’intérêt est donc une étape importante dans un contexte biotechnologique. Le laboratoire dispose d’une souche mutante de la microalgue haptophyte Tisochrysis lutea produisant deux fois plus de lipides de réserve que la souche sauvage en condition de privation azotée. Compte tenu du rôle clé des FTs dans l’établissement du phénotype, cette thèse vise à identifier les FTs impliqués dans la mise en place de ce phénotype mutant.Un pipeline bio-informatique d’identification et classification des FTs présents dans le génome de T. lutea a été élaboré. Le manque de donnée chez les haptophytes constituant un vide dans l’étude de l’histoire évolutive des microalgues, une étude comparative des FTs présents dans le génome d’algues de différentes lignées a été réalisée. Celle-ci révèle que l’étude des FTs aide à comprendre et illustrer l’histoire évolutive des microalgues par la mise en évidence de présences/absences de familles de FTs spécifiques de lignée.Afin de comprendre l’établissement du phénotype de la souche mutante de T. lutea, des données transcriptomiques ont permis la construction de réseaux de co-expression et de régulation des gènes chez les deux souches. Leur analyse croisée a identifié sept FTs candidats potentiellement liés au phénotype mutant. Une approche de p-RT-PCR a confirmé l’implication de deux FTs dans la remobilisation de l’'azote en condition de stress azoté. / In every organism, evolution and acclimation to environmental changes are orchestrated by numerous molecular players. Among them, transcription factors (TFs) play a crucial role by regulating gene expression. Therefore, identify TFs involved in the production of high value products is a significant step in a biotechnological context. The laboratory has at its disposal a mutant strain of the haptophyte microalga Tisochrysis lutea producing twice more storage lipids than the wild type strain when exposed to nitrogen deprivation. Given the key role of TFs in phenotype establishment, this PhD aim at identify the TFs involved in that of the mutant phenotype of T. lutea.A TFs identification and classification pipeline was elaborated and applied to T. lutea’s genome. Since the lack of data in haptophytes constitutes a limit in studies on microalgae evolutionary history, a comparative study of TFs identified in the genome of microalgae belonging to different lineages was carried out. This study reveals that TFs could be used to understand and illustrate microalgae evolutionary history through the highlight of lineage specific presence/absence of TF families.Aiming at understanding T. lutea’s mutant strain phenotype establishment, transcriptomic data were used to build gene co-expression networks and gene regulatory networks for both strains. Their comparative analysis identified seven TFs potentially liked to the mutant phenotype. A q-RT-PCR approach confirmed the involvement of two TFs in nitrogen recycling under nitrogen deprivation.