OIST Institutional Repository
Not a member yet
2744 research outputs found
Sort by
Supersymmetry and Nonequilibrium Quantum Dynamics
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyIn this thesis I present two studies that use ideas and concepts from supersymmetric quantum mechanics to understand and control the nonequilibrium dynamics of a quantum many-body system. The two protocols I study involve the quenching of a spin-polarized Fermi gas and the adiabatic control of single particle states during the expansion of an infinite square well over a finite time interval. In the first study, I explore the survival probability and the work probability distribution for quenches within a hierarchy of potentials created using supersymmetric factorization methods. I show that in this setting one can take advantage of the degeneracy between supersymmetric potentials in order to find simplified expressions for these quantities. I also show that many-body revivals in these systems exist and are robust even at finite temperatures.
For the second study I explore a shortcut to adiabaticity (STA) based on counterdiabatic driving for the single particle states of the supersymmetric partner potentials of the infinite square. By calculating the fidelity, quantum speed limit time and the cost of driving a system, I compare the efficiency of the shortcuts between the ground state wavefunctions of three supersymmetric potentials and three wavefunctions that are isospectral to one another. The use of a supersymmetric setting allows me to distinguish between the dynamical effects stemming from the energy spectrum and from the distance between the states in Hilbert space. I also show that in the isospectral case one can develop an intertwining relationship between the counterdiabatic driving terms using the operators of their single particle Hamiltonians
Investigating the Role of Neurexins in the Early Evolution of the Nervous System
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyThe appearance of the nervous system was an evolutionary event that allowed animals to control their physiology and behavior. In the nervous system, chemical signaling occurs either via volume transmission mainly by neuropeptides, or synaptic (wired) transmission by ‘classical’ chemical neurotransmitters. It is hypothesized that during the early evolution of animals, the diffusion-driven peptidergic signaling was the major mode of cell-cell communication, even for neuron-less animals. Later on, a faster and more targeted synaptic signaling was needed, as animal bodies and behaviors increased in complexity. But it remains largely unclear how the synaptic machinery evolved. One crucial event was the establishment of cell-cell contacts to specify and stabilize the communication between functionally distinct heterologous cell types like sensory neurons and muscles. In this study, I investigated Neurexins (Nrxns), a family of core presynaptic cell adhesion molecules with critical roles in bilaterian chemical synapse, using the non-bilaterian model Nematostella vectensis, a member of Cnidaria which is the closest outgroup to Bilateria. A series of gene expression and structure analyses indicated a non-neural origin of Nrxns. Functional analysis of the epithelial (classical) Nrxn in N. vectensis revealed its major role in cell adhesion, in particular in the maintenance of junction between ectodermal and endodermal epithelia. Neural Nrxns, named delta-Nrxns, are distinctly expressed in neuronal cell clusters that exhibit both peptidergic and classical neurotransmitter signaling abilities. Knockdown of NvNrxnδ1 and NvNrxnδ2 resulted in abnormal behaviors of N. vectensis polyps, involving muscle contraction. Interestingly, the knockdown of the neuropeptideprecursor gene co-expressed with delta-Nrxns did not show the same behavioral abnormalities, reflecting an independent role of delta-Nrxns from peptidergic signaling. Pharmacology experiments suggested that the delta-Nrxns are required for chemical neurotransmission (i.e., synaptic signaling). This study provides molecular, functional, and cellular insights into the ancestral non-neural function of Nrxns and may explain how and why this cell adhesion molecule family was employed in the synaptic machinery of the ancestral nervous system
A Computational Model of Granule Cell Migration and Purkinje Cell Primary Dendrite Selection during Cerebellar Development
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyThis project aims to investigate the interrelationship between primary dendrite selection of Purkinje cells and migration of their pre-synaptic partner granule cells during cerebellar development. During development of the cerebellar cortex, each Purkinje cell grows more than three dendritic trees, among which a primary tree is selected to develop further, whereas others completely retract. Experimental studies suggested that this selection process is coordinated by physical and synaptic interactions with granule cells. However, technical limitations hinder a continuous observation on multiple populations of the cells. To reveal the mechanism underlying this selection process, we constructed a computational model of dendritic development and granule cell migration, using a new computational framework, NeuroDevSim. Comparisons of the resulting morphologies from the model demonstrated the roles of the selection stage in regulating the growth of the selected primary trees. This thesis presents the first computational model that simultaneously simulates growing Purkinje cells and the dynamics of granule cell migration, revealing the role of physical and synaptic interactions upon dendritic selection. Development of the model is expected to provide new insights in the development of neonatal Purkinje cells and help to track down how cerebellar cortex develops into a normal or abnormal structure
The Role of Serotonin Neurons in Mouse Reward-based Behaviors
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophySerotonin (5-HT) is an important neuromodulator in reward-driven learning and decision making. The dorsal raphe nucleus (DRN) sends diffuse 5-HT projections throughout the brain. The involvement of DRN 5-HT neurons in reward-based behaviors has been examined using various types of behavioral tasks; however, how DRN 5-HT affects computational processes of decision making remains unclear. Reinforcement learning (RL) is a theoretical framework to describe the decision making process. Previous studies based on the RL framework have proposed hypotheses on the role of 5-HT in decision making, such as temporal discounting and model-based value computation. The overall aim of this thesis is to examine these hypotheses by analyzing behaviors under optogenetic manipulation, thereby clarifying the role of DRN 5-HT neurons in reward-based behaviors. The first hypothesis is that 5-HT modulates the relative importance of future rewards. Previous behavioral studies showed that 5-HT activation enhances patience to wait for future rewards and vice versa. However, how 5-HT regulates persistence to act for future rewards remains unknown. In the first part of my thesis research, I trained mice to perform a free-operant lever-pressing task, in which motor action rather than stationary waiting was required to obtain delayed rewards. In testing the effects of optogenetic activation and inhibition of 5-HT neurons on sustained motor actions, I found that optogenetic activation or inhibition of 5-HT neurons did not affect persistence in motor actions but an effect of the activation on slowing down response vigor, suggesting a different role of 5-HT neurons in motor actions for future rewards compared to stationary waiting. The second hypothesis examined is that 5- HT affects model-based decision making. In model-based decision making, agents use their own internal models of action-outcome relationships to plan forward and to select actions. Previous computational studies proposed facilitation of model-based decision making by 5-HT neurons, but behavioral evidence of how 5-HT regulates the process is still limited. A two-step decision making task is an established behavioral task to understand model-based decision making. In the second half of my thesis project, I trained mice to perform the two-step decision making task and found that optogenetic inhibition of 5-HT neurons affected choice behaviors and reduced time to make decisions possibly reflecting the disruption of model-based decision making. By fitting behavioral data to a model-free/model-based hybrid model, I found that photoinhibition of 5-HT neurons decreased the weight of model-based decision making. These results revealed the role of 5-HT neurons reward-based behaviors and model-based computations
Developing Integrin-targeted Peptide Assemblies to Direct Cancer Cell migration
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyAdvances in mechanistic understanding of integrin-mediated adhesion highlight the importance of precise control of ligand presentation in directing cell migration. The development of top-down nanofabrication techniques, such as polymer blending and nanolithography, achieved control over the spatial presentation of integrin ligand at the sub-micron resolution, which promoted the mechanistic study of integrin-mediated adhesions and inspired biomaterial innovations. We sought to enhance the spatial resolution beyond sub-micron resolution to understand the subsequent cellular response at the molecular level. To address this challenge, we developed a bottom-up nanofabrication strategy reaching by far the highest spatial resolution of ligand presentation (48 ligands/100 nm2), which is beyond the submicron limit of the top-down technique. Via simple molecular engineering, we transformed a natural ECM-derived ligand into an assembling ligand. Co-assembly of the assembling ligand with non-functional motifs at various proportions forms biocompatible nanofilaments presenting different ligand densities. Peptide assemblies possessing various ligand densities to exert biphasic effects on cell migration, with fast migration occurring at low density because of promotion of nascent adhesion and lamellipodia formation, and inhibition occurring at high density due to the prevention of integrin and actin filament disassembly at the cell rear. Meanwhile, we illustrated the cellular response to extracellular super high-density ligands. When the cells are exposed to super high-density ligands, the stress-fiber-associated focal adhesions (FAs) slide inward, while the actin cytoskeleton together with the integrin receptors and adaptor proteins were stabilized at the cell rear, restricting the cell retraction and protrusion. By expressing vin 258, a mutant that possesses vinculin D1 domain exhibiting high affinity to talin and paxillin but lack of actin-binding domain, the cells successfully maintained the FA on the periphery but failed to preserve the actomyosin network and could not resume protrusion nor trailing edge retraction. Extra Rho activation preserves FAs on cell edge and is associated with actomyosin bundles and eased the full disassembly of FAs facilitating trailing edge retraction but failed to resume the cell protrusion. By contrast, the constantly activation of Tiam1/Rac1 signaling effectively rescued the cell migration restricted by the excessive binding interactions between integrins and the super high-density ligands. Together, this strategy may provide new insights in material design for manipulating and further understanding in ligand-density-dependent-modulation for manipulating and further understanding in ligand-density-dependent-modulatio
Molecular Dissection of Ancestral Glia
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyNervous systems of bilaterian animals generally consist of two cell types: neurons and glial cells. Glia participate in almost every process taking place in the nervous system of bilaterians, indicating a crucial role of glial cells in both neurophysiological functions and for the nervous system evolution. Therefore, tracing back the first glia and elucidating its ancestral function is important for understanding the evolutionary sophistications of the nervous system. Histological examinations have not so far revealed any morphological sign of glial cells in Cnidaria, the closest outgroup to Bilateria. This led to the hypothesis that glial cells appeared after the common bilaterian ancestor had branched off from Cnidaria. However, this view has not been well examined at the genetic level. In this work I sought to investigate gliogenic program conservation in non-bilaterian animals, i.e. basal metazoans (Cnidaria, Placozoa, Ctenophora, and Porifera). First, I performed a phylum-wide and genome-wide survey on representative species of bilaterians and basal metazoans to clarify the conserved genetic repertoire required for bilaterian glial development. I found that the glial cells missing (Gcm) is one of the evolutionary conserved glial transcription factors (TF). Second, in Nematostella vectensis, a cnidarian model with highly conserved genetic repertoire required for bilaterian glial development, a homolog of Gcm is expressed in specific neuronal cells. I analyzed the function of Gcm by knocking it down and performing RNA-seq and RT-qPCR analyses. I found that the Gcm knockdown in Nematostella embryos resulted in expression alterations of cell adhesion proteins, GABA and glutamate transporters, ion channels, metabolic and protein modifying enzymes, as well as zinc finger and Ets-related TFs. In addition, Gcm seems to control Notch-Delta signaling, which is one of the crucial neuro-gliogenic pathways in bilaterians. Immunostaining of a Gcm target protein visualized a novel class of cells with flat cell body and no clear neurite process, which were previously classified as neurons as they express neuronal markers (neuropeptides). The major finding of my thesis is that Nematostella Gcm-expressing cells demonstrate characteristics of both neurons and glia, suggesting a dual nature of ancestral cells. This may indicate that the ancestral gliogenic program was intertwined with the neurogenic program and separated later in the animal evolution
Spin–1 Magnets and Their Excitations
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyNature sometimes arranges itself in extremely curious ways, sowing the seed of very intriguing physics. Magnetic systems offer a rich variety of interesting features. They are traditionally studied in either their classical (S ! 1), or their extreme quantum limit (S = 12). However, magnetic degrees of freedom in spin systems span within a whole spectrum range and do not necessarily reduce to the specific case found at the extremities. Spin–1 magnets provide a good example of what happens to ground state and excitations properties for such instance. Indeed, a spin–1 is special, in the sense that, besides displaying dipolar degrees of freedom, a spin–1 can also exhibit on-site quadrupolar degrees of freedom, while retaining its quantum characteristics. Therefore, spin–1 systems are often used as examples to refer to spin-nematic order in magnetic insulators, Fe-based superconductors, or cold atoms. Unlike for spin–1 2, which in the classical limit can be represented by an O(3) vector, for spin–1, an O(3) vector does not completely describe all of what a spin–1 can do, namely intrinsically exhibiting quadrupoles. In this Thesis, I develop a united framework that enables us to treat dipolar and quadrupolar degrees of freedom of a spin–1 moment on an equal footing. My method is based on the extension of the usual su(3) algebra describing a quantum spin–1 into the u(3) algebra. Within the u(3) formalism, I derive equations of motion (EoM) for the objects living in the u(3) algebra. The u(3) approach enables the appropriate formulation for both classical and quantum derivations. Moreover, the EoM take a simple form suitable for numerical implementation. I illustrate this method by applying it to the well-known Bilinear-Biquadratic model on the triangular lattice for the ferroquadupolar state. This study is supported through classical low-temperature expansion in order to probe the thermodynamical properties, as well as quantum multi-bosons theory that allows to access dynamics. These results are validated by comparison with numerical simulations classical Monte Carlo (MC) and Molecular Dynamics (MD) respectively, both expressed in terms of u(3) objects. I show that at sufficiently low temperature numerical simulations can be corrected for the classical statistics, and the fully quantum zero-temperature analytical results are retrieved. Additionally, I confirm that our method is also applicable to anisotropic models, which is of experimental relevance. Finally, some new ideas, including the description of topological defects in spin–1 magnets and the generalization of the commonly used Self-Consistent Gaussian Approximation to the degrees of freedom of a spin–1 expressed within our u(3) framework are explored
Banp Regulates DNA Damage Response and Chromosome Segregation to Promote Cell-cycle Progression and Cell Survival in Zebrafish Retina
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyBtg3 associated nuclear protein (Banp) was initially identified as a nuclear matrix associated protein and is a tumor suppressor. Recently it was reported that Banp binds to the CGCG element containing motif enriched near the transcription initiation site of CpG island promoters, namely Banp motif, promotes the transcription in a DNA methylation dependent manner, and controls metabolic genes in pluripotent stem and differentiated neuronal cells. However, cellular roles of Banp in embryonic development remains to be elucidated. Here we report a novel role of Banp in cell-cycle progression and cell survival of zebrafish retinal progenitor cells (RPCs). In zebrafish banprw337 mutants, retinal progenitor cells showed mitotic cell accumulation and subsequent apoptosis. DNA replication stress and tp53-dependent DNA damage response were activated in banprw337 mutants. Inhibition of Tp53 significantly rescued apoptosis but not mitotic defect and DNA double strand break accumulation, suggesting that Banp is required for maintaining integrity of DNA during segregation and replication. Furthermore, live imaging of mitosis in banp morphant retinas revealed that chromosome segregation was not smoothly processed from prometaphase to anaphase, leading to prolonged M-phase. Bulk RNA-seq analysis show that mRNA expression of two chromosomal segregation regulators, cenpt and ncapg, were decreased in banprw337 mutants. Furthermore, ATAC-seq analysis showed that chromatin near their transcription start site was closed in banprw337 mutants and indeed Banp motif was found in this chromatin-closed region, suggesting that Banp directly regulates cenpt and ncapg transcription via Banp motif to promote chromosome segregation during mitosis. Our findings reveal that Banp is required for cell-cycle progression and cell survival by regulating replicative DNA damage response and mitotic chromosome segregation
Self-Organization of Action Hierarchy and Inferring Latent States in Deep Reinforcement Learning with Stochastic Recurrent Neural Networks
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyThe thesis aims to advance cognitive decision-making and motor control using reinforcement learning (RL) with stochastic recurrent neural networks (RNNs). RL is a computational framework to train an agent, such as a robot, to select the actions that maximize immediate or future rewards. Recently, RL has undergone rapid development by introducing artificial neural networks as function approximators. RL using neural networks, also known as deep RL, have shown super-human performance on a wide range of virtual and real-world tasks, such as games, robotic control, and manipulating nuclear fusion devices. There would not be such a success without the efforts of numerous researchers who developed and improved the deep RL algorithms. In particular, most of the works focus on designing or revising the RL objective functions by mathematical analysis and heuristic ideas. While the well-formulated loss functions are critical to the RL performance, relatively fewer efforts have been paid to developing and improving the architecture of the neural network models used in deep RL. The thesis discusses the benefits of using novel network architectures for deep RL. In particular, the thesis includes two of the authors’ original studies about developing novel stochastic RNN architectures for RL in partially observable environments. The first work proposes a novel, multiple-level, stochastic RNN model for solving tasks that require hierarchical control. It is shown that an action hierarchy, characterized by consistent representation for abstracted sub-goals in the higher level, self-develops during the learning in several challenging continuous robotic control tasks. The emerged action hierarchy is also observed to enable faster relearning when the sub-goals are recomposed. The second work introduces a variational RNN model for predicting state transitions in continuous robotic control tasks in which the environmental state is partially observable. By predicting subsequent observations, the models learn to represent the underlying states of the environment that are indispensable but not observable. A corresponding algorithm is proposed to facilitate efficient learning in partially observable environments. The proposed studies suggest that the performance of RL agents can be improved by adequate usage of stochastic RNNs structures, which provides novel insights for designing better model architectures for future deep RL studies
Estimating Protein-Protein Interactions with High Aspect Ratio Plasmonic Nanopillars
Full text linkOkinawa Institute of Science and Technology Graduate UniversityDoctor of PhilosophyProtein-protein interactions (PPIs) are crucial for biochemical processes within or among cells, which makes characterizing these interactions essential for understanding the fundamentals of every living organism. PPIs occur between two or more protein molecules that come into physical contact with each other, and are caused by combinations of electrostatic forces, van der Waals interactions, hydrogen bonding, and hydrophobic effects determined by the geometry of these molecules. Due to various interactions taking place even within a single protein molecule, quantifying magnitudes of PPIs in terms of net attraction forces or affinity between different biomolecules remains a challenge. In addition, protein characterization techniques aimed at measuring binding affinities are often time-consuming, require large volumes of analytes, or lack statistical power. To address these issues, I developed a high-throughput technique for quantitative estimation of magnitudes of PPIs inside a microchannel in real-time. I performed these measurements by utilizing the flexible nature of polymer-based nanopillars that were high in aspect ratio (≥ 10), coated with plasmonic metal films, and were positioned inside a microfluidic platform. In the recent years, plasmonic systems have gained significant interest since plasmonic nanostructures have been proven to be extremely sensitive to the refractive index changes in the surroundings, and thus allow highly accurate measurements of concentrations of biomolecules. In this thesis, I built a novel biosensing platform by reshaping the surface of the polystyrene thin film into an array of polystyrene nanopillars using anodized aluminum oxide membranes as templates, followed by deposition of thin films of Ag/Al mixture on the nanopillars’ surface. Next, I compared four most common surface chemistry methods for biomolecule immobilization on non-spherical plasmonic nanostructures, and identified that 11-mercaptoundecanoic acid linkers led to the most reliable and reproducible biosensing results when Ag/Al-coated polystyrene nanopillars were used. Finally, I used this platform to quantify changes in the magnitudes of the transverse resonance modes detected from the sensor surface and utilized that to quantitatively estimate magnitudes of PPIs within streptavidin-biotin and modified streptavidin-biotin biological systems, as well as proposed an analytical method for identifying optimized flow rate conditions for straight microchannels with rectangular cross section. Overall, this opto-microfluidic platform containing high aspect ratio plasmonic nanopillars can be applied to various biomolecular systems, laying foundations to high-throughput realtime detection and quantitative estimation of PPIs based on the detected optical signals with an additional potential of providing an option of multiplex and multiparametric screening