Susan L. Wearne - US grants

The broad objective of this research program is to determine the biophysical and cellular substrate of the velocity storage neural integrator (VSNI) in the vestibulo-oculomotor system. Existing neural models of vestibular integrators rely largely on recurrent positive feedback networks to implement the integration. Anatomic evidence for the requisite recurrent axon collateral feedback in identified integrator neurons in goldfish and in mammals, however, is lacking. Preliminary analyses suggest that VSNI neurons behave as fractional order integrators; however, neither fractional nor integrative dynamical behaviors is presently understood at the neural level. The proposed research will develop the mathematical, analytic and methodological techniques to be used in a subsequent larger study of the relative contributions of cellular and network properties to velocity storage neural integration. The specific aims are (1) to test the hypothesis that VSNI neurons in goldfish are fractional order integrators, and quantify their integrative properties; (2) to test the hypothesis that the observed range of branching geometries and heterogeneity of dendritic process types in Area II neurons is causally related to the diversity of observed response dynamics; (3) to abstract the contributions of dendritic branching topology, dendritic nonuniformity and intrinsic membrane currents to fractional and integrative response dynamics via biophysically realistic modeling of VSNI neurons. The expected results are: characterization of the range of integrative response dynamics in VSNI neurons; correlation between fractal structural properties and possibly fractional integrative response dynamics and an estimate of the relative roles of intrinsic structural and cellular factors in producing these dynamics. The unique features of this project are (1) use of the goldfish preparation in which velocity storage neurons are easily identified, finite in number for realistic modeling and structure-function experiments that include both single cell sharp and patch electrode recordings; (2) use of new mathematical methods for relating fractional integrative dynamics to fractal dendritic structures, including computation of 3-D fractal dimension and new analytic techniques for deriving fractional differential equations from their physical substrates. This research will provide the methodology and pilot results for future model-based studies of the neural basis of velocity storage and angular VOR spatial orientation. These results will impact on current system-level models of vestibulo-ocular function, and on general theoretical models of persistent neural activity and short-term memory in multiple areas of neuroscience.

We propose to acquire an SGI Onyx 3400 supercomputer, capable of state-of-the-art performance in simulation, computation and visualization, to upgrade a well-established Computational Core facility, and to provide the basic computational engine of the recently formed Advanced Imaging Program at Mount Sinai School of Medicine (MSSM). A recent initiative at MSSM to expand the imaging and computational sciences, together with the ongoing evolution of bandwidth and resolution in imaging technology is generating enormous multi-dimensional datasets (on order 20-4090 GB in this proposal) whose analysis, reconstruction and visualization demands computational speed and memory far exceeding available resources. No machine current exists at MSSM for visualization and simulation of large multi- dimensional datasets in excess of approximately 1 GB. The Advanced Imaging Program is a distributed program that incorporates all levels of microscopic analysis, MR microscopy and several imaging modalities in humans and animal models, and promotes the exchange of ideas and collaborative initiatives amongst researchers that is essential to ensure the continued advancement of scientific research at MSSM. In addition to the major user group, the supercomputing/visualization facility will be available to collaborators at neighboring institutions (MYU, SUNY Stony Brook), providing a centralized resource for high-performance computational and visualization, and fostering interdisciplinary and inter- institutional collaborations amongst the wider New York research community. Each of the research programs described by the major users of the proposed Onyx 3400 supercomputer is current constrained either by computational limitations or by unmet visualization requirements that are beyond the scope of any available computational resources. The projects that will benefit from the proposed equipment represents a broad range of basic scientific and clinically related studies, which heavily utilize well-developed imaging facilities at MSSM, but are lacking a centralized computational and visualization facility. The acquisition of the proposed supercomputer will have a major impact on research at MSSM by establishing a common facility by computationally intensive analysis and display of high-resolution imaged data. As a shared computational facility, it will promote development and exchange of novel quantitative imaging and image analysis tools between the major users, and foster multi-disciplinary interactions with investigators from diverse areas with common computational and visualization requirements.

[unreadable] DESCRIPTION (provided by applicant): Age-related cognitive impairment, which affects the vast majority of elderly people worldwide, is accompanied by subtle changes in the geometry of dendritic arborizations, and reduced densities and distributions of dendritic spines. Understanding exactly how these structural changes might produce the observed cognitive deficits requires accurate 3D representations of neuronal morphology, and biophysical modeling that can relate structural changes to altered neuronal firing patterns. Recent experimental work indicates that the combined effects of these global (dendritic topology) and local (spine geometry) variations critically affect neuronal dynamic behavior and the extent of synapse-specific plasticity. To date however, no tools capable of resolving and simulating neuronal morphology and dynamics on both global and local scales have been available. This application directly addresses both of these needs. Existing methods of acquiring neuronal morphology for quantitative analysis or compartment modeling are limited in resolution, or are prohibitively time-consuming. The central goal of this project is to develop an automated software system for digitization, 3D reconstruction and biophysically-based simulation of detailed neuronal morphology, that reliably captures detail on spatial scales spanning several orders of magnitude and that avoids the subjective errors that arise during manual tracing. This system will be used to obtain a mechanistic understanding of the role of structural changes in age-related decrements in working memory and cognition. Two specific aims will address this broad objective: 1) To develop an automated user interface for our prototype volume integration system and to extend the prototype to handle multiphoton image stacks that contain entire layer III pyramidal neurons; 2) To develop software tools for automatic parameter extraction from confocal and multiphoton imaged data, suitable for simulation with standard compartment modeling packages. By determining simple morphologic indices for altered efficacy of synaptic transmission, neural integration and action potential forward and backpropagation, the tools developed in this study will provide insight into fundamental mechanisms of memory induction and maintenance that underlie normal cognitive function. Such insight will lead to enhanced strategies for delaying and ameliorating the impairment of learning, memory and cognitive performance that accompanies normal aging and age-related pathologies. [unreadable] [unreadable]

Mount Sinai School of Medicine (MSSM) seeks funding for an Abilene/Internet2 to better handle the demands of image analysis and processing. A recent National Center for Research Resources (NCRR) grant allowed the institution to upgrade the computer network in the main research building. The wide area network now presents a bottleneck for collaboration with other institutions. The connection, itself, will also support collaborative projects in the following areas:

3-dimensional Morphometric Analysis of Mammalian Neurons: A collaboration between MSSM, investigators from the Courant Institute of Mathematical Sciences (NYU), New York University School of Medicine, SUNY at Stony Brook, and the University of New South Wales, Australia. Researchers at MSSM study the structure-function relationship of neurons in the brain, employing advanced 3D microscopic imaging techniques as well as 3D image processing, rendering, analysis and modeling. CNIC investigators are using confocal and multi-photon microscopes to image (in 3D) entire pyramidal neurons in the cortex at the highest resolution possible. This creates approximately 20-gigabyte data sets for a single neuron. After the data is collected, it is exported to computers, where image stitching, volume rendering, deconvolution and 3D-image analysis is done. This award will greatly facilitate the capabilities of off-site collaborators to access these large datasets as well as manipulate and analyze the data using unique software developed by CNIC investigators and hosted at MSSM.
Neural Mechanisms of Vestibular Function: MSSM researchers and collaborators at Washington University, the University of Utah and NASA-Ames are studying vestibular functions. Electrophysiological and microscopic studies are being performed to determine the basis for regional variations in afferent response dynamics across the vestibular sensory epithelium. 600 Mbytes to 1.2-gigabyte data sets are created using a multi-photo laser. The ability to manipulate the data in 3D via access to volume rendering and 3D image analysis software hosted on MSSM computers will provide invaluable information to off-site collaborators. Furthermore, due to the length of time required to compile the data for any one experiment, it will be greatly beneficial for collaborators to review the data as it is collected (in real time) on the multi-photon microscope. This will avoid any delay in feedback and permit the off-site collaborators to modify experimental conditions.

[unreadable] DESCRIPTION (provided by applicant): Neurons of the velocity storage neural integrator store a 'short-term memory' of head velocity that provides a central representation of head and body orientation in space, allowing animals to navigate with respect to an inertial reference frame. Much effort has been focused on modeling vestibular integrators at the systems level and by recurrent feedback networks, however the biophysical mechanisms that subserve neural integration remain unknown. Individual integrator neurons exhibit a corresponding variability in integrator storage capacity and dendritic branching structure, but to date no studies have attempted to relate these. This is primarily due to technical difficulties in obtaining the requisite structure/function to begin biophysical modeling in mammals, and lack of computational and analytic techniques to perform precise geometric modeling. A multidisciplinary approach combining mathematical, experimental and imaging expertise will address these inadequacies in the well-developed goldfish model, in which integrator neurons are easily identified, and finite in number for realistic modeling and structure-function experiments. Our central hypothesis, that spatially-extended single cell properties including dendritic topology and its interaction with active membrane and synaptic properties are essential elements of neural integration, will be tested by (1) characterizing the diversity in integrator neurons and its role in oculomotor behavior and plasticity; (2) parametrising 3-D dendritic branching structure with novel imaging, image analysis and geometric techniques; (3) verifying the relationships determined between function in Aim 1 and morphology in Aim 2 by compartment modeling of high-resolution morphologic data, and evaluation of circuit models containing reduced versions of biophysically realistic neuron models. Development of new mathematical techniques for relating neural dynamics to local and global properties of dendritic structures is a unique feature of this project. In the long term, we want to understand the contributions of cellular properties and interactions between realistic neurons to the persistent neural activity that underlies vestibular neural integrators specifically, and fundamental mechanisms of short-term or working memory in multiple areas of neuroscience. A mechanistic understanding of neural integration will yield basic information leading to rational strategies for prevention, treatment and reversal of balance and equilibrium dysfunction, and for understanding the structural determinants of disorders and loss of short-term memory. [unreadable] [unreadable]

[unreadable] DESCRIPTION (provided by applicant): Cognitive impairment in normal aging and neurodegenerative disease is accompanied by altered morphologies on multiple scales: from the fine-grained geometry of individual spines to the global topologies of multi-neuron and vasculature networks that are distorted by space-occupying histopathologic lesions. A mechanistic understanding of the role of these structural changes in producing the observed cognitive deficits requires accurate 3D representations of neuronal morphology, and realistic biophysical modeling that can directly relate structural changes on multiple scales to altered neuronal firing patterns. To date however, no tools capable of resolving, digitizing and analyzing neuronal morphology on both local and global scales, and in true 3D, have been available. The central goal of this project is development of an automated analysis system for digitization, 3D reconstruction and geometric analysis of detailed and accurate neuronal morphology, capable of handling morphologic details on scales spanning local spine geometry through complex tree topology to the gross spatial arrangement of multi-neuron networks. As a specific example we will analyze morphologic changes in a Tg2576 mouse model of Alzheimer's disease (AD), in which amyloid deposition, altered cortical microvasculature and neural abnormalities provide easily identifiable examples of pathologic lesions. Four Specific Aims will address this broad objective: (1) To develop a semi-automated system for 3D tree extraction and spine analysis from laser scanning microscopy (LSM) imaged data, with sub-voxel resolution for accurate neuronal morphometry at the finest scales; (2) to image and digitize in 3D individual neurons, multineuron and vasculature networks, and senile plaques from human and Tg2576 mouse models of AD; (3) to develop tools for global analysis of spatially complex cellular structures in 3D; (4) to distribute and maintain all software, and develop a database-driven web repository for distribution of digitized neurons and networks. By providing true 3D morphometry of complex neural structures on multiple scales, the tools developed in this study will enable future multiscale biophysical modeling studies capable of testing hypothesized mechanisms by which altered dendritic structure, spine geometry and network branching patterns in normal aging and neurodegenerative disease determine pathologies of working memory and cognitive function. Such studies will provide crucial insight into general mechanisms of memory induction and maintenance that underlie normal cognitive function, its dysfunction in diseased states, and potential mechanisms for its restoration. [unreadable] [unreadable]