OIST Institutional Repository
Not a member yet
2744 research outputs found
Sort by
Phosphoenolpyruvate Regulates the JunB-Dependent Pathogenic Th17 Transcriptional Program
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyAerobic glycolysis, a metabolic pathway essential for effector T cell survival and proliferation, regulates the differentiation of autoimmune T helper (Th)17 cells, but the mechanism underlying this regulation is largely unknown. Here, we identify a glycolytic intermediate metabolite, phosphoenolpyruvate (PEP), as a negative regulator of Th17 differentiation. PEP supplementation or inhibition of downstream glycolytic enzymes in differentiating Th17 cells increases intracellular PEP levels and inhibits the expression of Th17 signature molecules, such as IL-17A. However, PEP supplementation does not significantly affect metabolic reprogramming, cell proliferation, and survival of differentiating Th17 cells. Mechanistically, PEP regulates the JunB-dependent pathogenic Th17 transcriptional program by inhibiting the DNA-binding activity of the JunB/BATF/IRF4 complex. Furthermore, daily administration of PEP to mice inhibits the generation of Th17 cells and ameliorates Th17-dependent autoimmune encephalomyelitis. These data demonstrate that PEP links aerobic glycolysis to the JunB-dependent pathogenic Th17 transcriptional program, suggesting the therapeutic potential of PEP for autoimmune diseases
Social Interaction Under the Free Energy Principle: A Cognitive Robotics Approach
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyMany things we do or think involve other people. We (strive to) understand, predict, and coordinate our actions with them in various social contexts. Many scientists investigate related cognitive mechanisms of social behavior. However, how individual and collective dynamics allow us to coordinate our actions in different social settings is not fully understood. Recently, the free energy principle has drawn attention with the expectation that it could provide a unified theory of the brain to model action, perception, and learning, among other cognitive skills. This thesis investigates how action, perception of actions, and learning of two agents translate into a mutual social context. I use a neurorobotics experimental setup to systematically study dyadic synchronized imitative interaction by extending the frameworks of predictive coding and active inference under the free energy principle. In the proposed model, the top-down processes encode prior beliefs, which allow robots to generate an action and predict the future action of the other robot. The two robots observe each other’s actions through a reciprocally coupled action-perception loop. The bottom-up inference process approximates the posterior belief by minimizing free energy whenever prediction and observation differ. In a wide range of experiments, I explored how regulating free energy complexity, i.e., the divergence of the prior and the approximate posterior belief can guide individual behavior and behavior coordination in the dyadic context. I show that three types of dyadic behavior coordination dynamically emerge due to the coregulated optimization processes in the robots’ reciprocally coupled action-perception loop, including (1) one robot is leading and the other robot is following (and vice versa), (2) robots are ignoring each other, and (3) robots are spontaneously taking turns in leading and following. Finally, I show that slowly oscillating the regulation of free energy complexity during an interaction leads to rather agreed-upon, intentional turntaking behavior. The contribution of this thesis is shedding light on the mechanisms under the free energy principle that lead to the dynamic emergence of different types of behavior coordination in imitative interaction. By discussing qualitative and quantitative differences between those behaviors, including the aforementioned two types of turn-taking, essential aspects of autonomous behavior coordination in synthetic and empirical studies are clarified
Unraveling the Nature of Excitons and their Interactions through Time-Resolved Photoemission and Optical Spectroscopy
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyThe exciton – a coulomb-bound electron-hole pair, was first conceptualized by Frenkel and Wannier in the 1930s. Since then, it has been integral to understanding the optoelectronic response in semiconductors, particularly in low-dimensional semiconductors. Therein, excitons have large binding energies due to quantum confinement and reduced dielectric screening, thus dominating the optical response of the material even at room temperature. Despite their importance, a critical fundamental property of excitons remains inaccessible – the momentum of the constituent electrons and holes! Such a measurement would immediately reveal valuable information, such as their direct or indirect nature, their wave function, their size, and the nature of their interactions. Resolving the momentum coordinates of excitons requires the development of a new instrumentation platform that probes the excitons in time, energy, space, and momentum. This thesis describes the need for such an instrumentation platform and its development, namely the development of time-resolved momentum microscopy. It then describes three studies on the nature of excitons and their interactions.
First, we study the interlayer excitons in a WSe2/MoS2 heterostructure. Using time-resolved Momentum Microscopy, we resolve the momentum coordinates of the constituent electrons and holes within the interlayer exciton, directly measuring its size and confinement within the moir unit cell. Next, we demonstrate Floquet effects in monolayer WS2, in the absence of optical fields, resulting from the time-periodic oscillations in the electron self-energy due to excitons. The strong amplitude of the time-periodic perturbation allows us to observe the hybridization of the original band structure with the exciton-dressed one. Finally, we use traditional µ-TAS to study the exciton-exciton annihilation process in bilayer black phosphorus. We show that it is possible to alter the dimensionality of the exciton-exciton annihilation process from one dimensional-like to two dimensional-like by tuning the exciton density and temperature. In conclusion, this thesis answers some fundamental questions about excitons and their interactions in two dimensional semiconductors and paves the way for uncovering novel non-equilibrium phenomena in two dimensional materials
Simulations of Multiphase Turbulent Flows
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyThe flow of an incompressible fluid can be described exactly and succinctly using the NavierStokes equation. However, the nonlinearity of this equation leads to flow structures with detail at many length scales, known as turbulence. The only exact theory in turbulence was made in 1941 by Kolmogorov. In this thesis, we probe Kolmogorov’s predictions in the case of multiphase systems by making direct numerical simulations of droplets, particles, and solid phases in turbulent flows.
Firstly, we show how the coalescence of droplets can reduce drag in a turbulent channel flow. Following on from this, in the remainder of the thesis we look more closely at the turbulent energy cascade using simulations of statistically homogeneous and isotropic flows. We show that the balance of turbulent forces and surface tension means large and small droplets form distinct shapes. Considering particles in turbulent flows, we see that isotropic and anisotropic particles couple to the cascade at different length scales, and in some cases can enhance the flow. Finally, we show that the plasticity of a fluid enhances its turbulent behaviour.
The results presented in this thesis find cases in which Kolmogorov’s predictions do not hold. In each case, we aim to explain why. These results can have wide-reaching implications for health, industry and the environment, including heart disease, micro-plastic dispersal, mudslides, and cloud formation
Geometry and Topology in Memory and Navigation
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyGeometry and topology offer rich mathematical worlds and perspectives with which to study and improve our understanding of cognitive function. Here I present the following examples: (1) a functional role for inhibitory diversity in associative memories with graph- ical relationships; (2) improved memory capacity in an associative memory model with setwise connectivity, with implications for glial and dendritic function; (3) safe and effi- cient group navigation among conspecifics using purely local geometric information; and (4) enhancing geometric and topological methods to probe the relations between neural activity and behaviour. In each work, tools and insights from geometry and topology are used in essential ways to gain improved insights or performance. This thesis contributes to our knowledge of the potential computational affordances of biological mechanisms (such as inhibition and setwise connectivity), while also demonstrating new geometric and topological methods and perspectives with which to deepen our understanding of cognitive tasks and their neural representations
Population Dynamics of Microorganisms in Spatially Structured Environments
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyMicrobial populations live and grow in spatially structured environments. These structures lead to spatial patterns in populations and alter the course of their natural evolution. Such phenomena are theoretically studied using spatially explicit models. However, these models are still poorly understood due to their analytical and numerical complexity. In this thesis, we study two systems of microorganisms living and proliferating in different spatially structured environments. The first system consists of populations of Escherichia coli growing in rectangular microchannels with two open ends. We study such populations with a lattice model in which cells shift each other along lanes as they reproduce. The model predicts rapid diversity loss along the lanes, with neutral mutations appearing in the middle of the channel being the most likely to fixate. These theoretical predictions are in agreement with our experimental observations. The second system is constituted by planktonic microorganisms that are transported by chaotic oceanic currents. To replicate their dynamics, we employ an individual-based coalescence model. The model predicts the effect of oceanic currents on the biodiversity of planktonic communities, as observed in metabarcoding data sampled from oceans and lakes around the world
Machine Learning Applications for the Study and Control of Quantum Systems
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyIn this thesis, I consider the three main paradigms of machine learning – supervised, unsupervised, and reinforcement learning – and explore how each can be employed as a tool to study or control quantum systems. To this end, I adopt classical machine learning methods, but also illustrate how present-day quantum devices and concepts from condensed matter physics can be harnessed to adapt the machine learning models to the physical system being studied. In the first project, I use supervised learning techniques from classical object detection to locate quantum vortices in rotating BoseEinstein condensates. The machine learning model achieves high accuracies even in the presence of noise, which makes it especially suitable for experimental settings. I then move on to the field of unsupervised learning and introduce a quantum anomaly detection framework based on parameterized quantum circuits to map out phase diagrams of quantum many-body systems. The proposed algorithm allows quantum systems to be directly analyzed on a quantum computer without any prior knowledge about its phases. Lastly, I consider two reinforcement learning applications for quantum control. In the first example, I use Q-learning to maximize the entanglement in discrete-time quantum walks. In the final study, I introduce a novel approach for controlling quantum many-body systems by leveraging matrix product states as a trainable machine learning ansatz for the reinforcement learning agent. This framework enables us to reach far larger system sizes than conventional neural network-based approaches
Population Dynamics of Microorganisms in Spatially Structured Environments
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyMicrobial populations live and grow in spatially structured environments. These structures lead to spatial patterns in populations and alter the course of their natural evolution. Such phenomena are theoretically studied using spatially explicit models. However, these models are still poorly understood due to their analytical and numerical complexity. In this thesis, we study two systems of microorganisms living and proliferating in different spatially structured environments. The first system consists of populations of Escherichia coli growing in rectangular microchannels with two open ends. We study such populations with a lattice model in which cells shift each other along lanes as they reproduce. The model predicts rapid diversity loss along the lanes, with neutral mutations appearing in the middle of the channel being the most likely to fixate. These theoretical predictions are in agreement with our experimental observations. The second system is constituted by planktonic microorganisms that are transported by chaotic oceanic currents. To replicate their dynamics, we employ an individual-based coalescence model. The model predicts the effect of oceanic currents on the biodiversity of planktonic communities, as observed in metabarcoding data sampled from oceans and lakes around the world
Using Optogenetics to Spatially Control Cortical Dynein Activity in Mitotic Human Cells
No full textSeveral light-inducible hetero-dimerization tools have been developed to spatiotemporally control subcellular localization and activity of target proteins or their downstream signaling. In contrast to other genetic technologies, such as CRISPR-mediated genome editing, these optogenetic tools can locally control protein localization on the second timescale. In addition, these tools can be used to understand the sufficiency of target proteins’ function and manipulate downstream events. In this chapter, I will present methods for locally activating cytoplasmic dynein at the mitotic cell cortex in human cells, with a focus on how to generate knock-in cell lines and set up a microscope system
Protein Sequence, Structure, and Dynamics Reveal Insights in the Divergence of Protein Functions
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyProteins participate in every important aspect of known living. The amino acid sequence of which a protein is composed contains information about the physicochemical properties, the three-dimensional structure, and its function. However, connecting protein sequence to function is still an open challenge, particularly for protein families with complex inter-relationships, i.e., heteromeric interactions. Such is the case of the Epidermal Growth Factor (EGF) receptor system, comprising four or more paralogs of the EGF receptor interacting with seven or more paralogs of the peptide ligand.
In this thesis, I use the evolutionary history of the EGF receptor system to show how phylogenetic patterns of evolution relate to functional divergence at the protein sequence level. By combining measures of residue conservation and residue co-evolution I developed a method to identify residues responsible of a specific protein function. Mutations on the residues highlighted by this method altered the auto-phosphorylation level of the EGF receptor and affected cellular growth. Next, I studied a fish-specific gene duplication of the EGF receptor and used it to describe and model a rare pattern of sequence evolution. I showed that this pattern could be related to functional divergence, thus providing a way to identify the occurrence of the event and the residues responsible of it. Ultimately, I analyzed whole protein families using protein similarity networks. My results showed how the networks made from structural similarity of predicted 3D-models give a better representation of the protein functions compared to sequence similarity networks, thus supporting a paradigm shift from sequence-based to predicted-structure-based bioinformatics software.
Overall, my thesis shows a deep interconnection between functional divergence and protein sequence evolution that can be exploited for prediction of function or identification of evolutionary events. The conceptual foundations of this study could be used in other fields where gene duplication and functional residues play an important part, as for example protein engineering and the study of copy number variation in cancer biology