William B. Levy - US grants

Brief, high-frequency conditioning stimulation of the entorhinal afferents to the dentate gyrus produces long-term potentiation (LTP) of this monosynaptic response. It is now clear that a diversity of modifications exists, all of which may occur simultaneously, although to varying extents and at different, although adjacent, locations. The four long-term, associative modifications are excitatory synaptic potentiation, excitatory synaptic depression, and potentiation and depression of what may be disynaptically activated, inhibitory synapses mediating feedforward inhibition (or its functional equivalent). Our electrophysiological research seeks to determine the independence of these four phenomena, particularly the temporally associated conditions of pre- and postsynaptic activity/inactivity that lead to each modification. Anatomical studies seek the cellular bases of these diverse modifications. It appears that larger synapses correlate with excitatory synaptic potentiation. Fewer synapses may be associated with excitatory synaptic depression. A primary aim of this project is to dissociate the correlates of excitatory synaptic potentiation from synaptic depression and mere afferent activity during conditioning. The proposed research would characterize further the physiological and anatomical changes. We week: 1) to continue characterization and quantification of the ultrastructural correlates of the LTP-conditioning paradigm; 2) to define and distinguish the variety of modifications accompanying LTP and LTP-like conditioning paradigms; 3) to characterize the modifiability and to determine inferentially the role of granule cell dendritic spines and dendritic branching patterns as determinants of the spatiotemporal summation of inputs upon the granule cell; 4) to determine if the absolute number of synapses is altered with conditioning, and 5) to relate our work on modification as a function of use (associative activity/inactivity) to contemporary trends and problems in neuroscience. The methodologies employed include quantitative light and electron microscopy, stereology, electron microscopic autoradiography, light microscopic-Golgi methods, extracellular neurophysiology, and computer modeling of anatomical data. Understanding how synaptic connectivity can be altered in the adult brain and the cellular bases of these alterations should contribute to the development of rational treatments for brain injury due to disease, trauma, stroke, and aging.

It has always been expected that understanding normal brain function, particularly the cellular bases of learning and memory, would lead to clinically relevant results, but no one dared guess how quickly this relevance would arrive. A hypothesis held by several researchers is that brain pathologies resulting from ischemia and anoxia develop because of excessive neural activity. It is not just activity, however, but activity that activates NMDA receptors. The activation of NMDA receptors requires associative activity of the very type we are studying. That is, the NMDA receptors are not there to kill cells but are there to mediate associative learning. The proposed research continues our NIH- funded research and uses anatomy, physiology, and biophysical modeling to extend our knowledge of long-term associative synaptic modification. The extension goes in two directions: one extension is to another synaptic system; the other extension is to relate long-term associative potentiation to induced brain pathologies. The other synaptic system is formed by entorhinal cortical (EC) afferents on the spiny pyramids of hippocampal CAl. Associative potentiation at these synapses is of interest in its own right but has an additional importance. The CAl pyramids represent a sensible transition which should allow a bridging of knowledge gained from the dentate granule cells to the pyramids of cerebral cortex. That is, the distal spine synapses of CAl are quite similar to the EC-DG synapses we have studied and to the layer-l spine synapses of cerebral cortex. In CAl we will study two types of changes: a homogeneous associative potentiation of the bilateral EC projections to the CAl molecular layer, and a heterogenous interaction in which the EC-CAl synapses generate the permissive event for CA3-CAl synaptic modification. The research will evaluate both the ultrastructural and physiological characteristics of these changes. Biophysical models will examine the implications of the morphological alterations for cellular spatiotemporal integration and the rules of LTP. By correlating the ultrastructural characteristics of LTP with the ultrastructural characteristics of epilepsy and of ischemic/traumatic brain injury, we will examine the relationship between associative modification and the relevant class of brain methodologies. The question is: to what extent is LTP a correlate of these pathologies? Key observations will be any ultrastructural correlates shared by LTP and these pathologies. LTP-like changes may be a transitional state before ischemic cell death while the epileptic state may require LTP.

The loss of memory or even the loss of the ability to store new declarative memories is devastating, as we know from the life of the patient H.M. Unfortunately, such types of memory loss, although perhaps not as severe, also occur from head injury, with aging, and sometimes with menopause. Because an adequate ability to learn and to remember is fundamental to normal cognitive behavior and because many cognitive behaviors hinge upon stored declarative memories, it is important to understand the role various brain structures play in forming such memories. it is our long-term goal to understand how the hippocampus initially forms declarative memories and then interacts with the cerebral cortex to store long-term memories there. Such knowledge should facilitate repairing the effects of aging or preventing the effects of menopause on memory processes. Thus, our goal here is to provide a quantitative understanding of hippocampal function by simulating biological plausible hippocampal-like networks with inputs that are relevant to cognitive/behavioral theories of hippocampal function. The specific aims of this proposal are: 1) to create a minimal, biologically plausible model of the hippocampus functioning as a cognitive map; 2) to test, and minimally modify if necessary, this model using other paradigmatic behaviors that require the hippocampus; and 3) to begin development of the simplest possible biological model that can sensibly predict the patterns of hippocampal cell firing for behavioral situations relevant to hippocampal function. Using simplified models of the hippocampus, we will demonstrate that the archetypal hippocampal anatomy and associated physiologies can reproduce the functions ascribed to the hippocampus, including context formation, spatial mapping (i.e., cognitive mapping), and flexible memory representations (Eichenbaum et al., '92). We will discover which functional properties of the hippocampus arise from the sparse recurrent connectivity of CA3 and which properties rely on other aspects of hippocampal anatomy and physiology. Technically, we will use the methods of computer simulation to model hippocampal cell firing and cognitive behavior. By studying reduced models which are gradually made more complex, we will achieve a fundamental understanding of the critical anatomy and the critical physiology for the functions which we study. Because our laboratory is involved in basic neurophysiological and anatomical studies of the hippocampal formation, we are strongly motivated to produce models with the greatest possible biological plausibility and relevance.

This research project develops new artificial neural network concepts for environment related prediction and spatial control that are inspired by the hippocampal function. Hippocampus is an enfolding of cerebral cortex into the lateral tissue of a cerebral hemisphere. The research combines a theory of brain development and a theory of hippocampal function with recent developments in statistical decision theory based upon recurrence and robustification of artificial neural networks. The major focus of this project is on a study of hippocampally inspired recurrent neural structures, adaptively developed connectivity, robust neural operations and mapping, and information-theoretic approaches for accessing and eventually reducing the computational demands on the networks. The combination of biological and engineering research produces a self re-enforcing, positive feedback, and the robust theory leads to a deeper understanding of the function of hippocampus, and new concepts for better implementations of spatial control based on robust decision theory.

9424286 Desmond This collaborative project between a young theorist and two experienced experimentalist to test a new theory of what happens in the brain during learning.. The theory says that learning in the mammalian cortex is guided by a balance between two distinct mechanisms. One is the continual activity-independent changes in the connections between neurons (called synapses) that tend to randomize the place on a neuron where an incoming afferent neuron establishes its connection. The second is an activity-dependent process in which the connections of afferent neurons that are active at the same time are stabilized if they happen by chance to be close together. This means that groups of frequently co- activated inputs to the brain come together to form neighboring synapses and, therefore, have a more powerful effect on their target neurons. This theory will be tested experimentally by looking anatomically at the spatial arrangement of afferent synapses on target neurons when the afferent neurons are artificially induced to have correlated activity. The theory will also be tested by simulating this type of biological experiment on a computer. The outcome of these studies promises to have a significant impact on the way we think about how the brain does the computations that allow us to learn complicated patterns.

DESCRIPTION (Verbatim from the Applicant's Abstract): The hippocampus is an important cognitive brain region by virtue of its critical role in learning and memory. In the adult hippocampus of the female rat, the CA3-CA1 synapses undergo phasic cycles of syanptogenesis and synapse shedding with each 4 or 5 day estrous cycle. Although this cycle requires the ovarian steroid estrogen, the excitatory pyramidal neurons in CA1 and CA3 lack the conventional, genomic estrogen receptor (ER alpha). This remarkable observation leads to many, unanswered questions about the control of ovarian steroid-dependent cycle of adult synaptogenesis in the hippocampus. It also begs us to ask whether this same cycle also occurs in cerebral cortex. Using the molecular biological techniques of PCR, Northern and Southern blots, and differential display, we will uncover a set of molecules that are selectively synthesized in an estradiol-dependent manner in hippocampal CA1 and CA3. The time course of selective synthesis will enable us to define a temporal cascade of expressed sequences from early time points, those presumably controlling transcription factors, through later ones, possibly including sequences involved in synapse assembly. By pharmacologically blocking estradiol-dependent synapse formation with an NMDA receptor antagonist, and independently, by hormonally undoing synapses already formed, we will identify a particularly interesting subset of these expressed transcripts. This selective subset will include estradiol-dependent transcripts specific to the synaptogenetic cascade. Interpreting the roles played by the estradiol-dependent transcripts will be facilitated by electrophysiological correlations, anatomical localization, and by structural similarities to known molecules. Using naturally cycling females and ovariectomized animals receiving estradiol replacement, we will electrophysiologically assess NMDA receptor function in CA1 and correlate this with altered gene expression. In situ hybridization, supplemented as necessary with microdissection, will permit the cellular localization of these novel gene products to pyramidal and/or nonpyramidal neurons and/or glia. In situ hybridization will subsequently be used 1) to search for similar expression patterns in cerebral cortex and hypothalamus, and 2) to compare this form of adult synaptogenesis with that in the developmentally immature, male rat. Finally, we begin a series of experiments, including cloning and mapping the newly discovered genes to mouse and human chromosomes. These studies will eventually culminate in testing the function of these novel gene products in the genetically altered mouse.

DESCRIPTION:(provided by applicant) Everybody knows that the purpose of the brain is to process information. But what does that mean? How should neuroscientists quantify such information processing? Surely the brain cannot be understood as a digital computer. And although information theory has something to say about how signals should or should not be passed between neurons, by itself there is much we find in the brain that Shannon's basic information theory does not seem to explain. For example how will we be able to compare the computations performed by one type of neuron with those of another type of neuron? Why are certain anatomies and physiologies preferred for one type of computation versus another type of computation? The proposed research seeks appropriate measures to quantify microscopic neuronal function that will make sensible quantitative aspects of neurons and their physiology. If we can successfully quantify and measure computation in a way that explains and predicts a somewhat diverse set of quantitative observations, then these measures will qualify as an appropriate language for quantifying information processing performed by the nervous system. The proposed approach will merge information theory with biologically inescapable issues; the principle issue being the cost of computation and communication. To establish the appropriate measures, the research will answer questions such as: Why are resting potentials around -70 mV? Why not smaller; why not larger? Why aren't energetically wasteful resting conductances smaller? Why not have brains half the size that compute twice as fast? Why do neurons fire in the frequency ranges observed? Why do synaptic failures occur in some systems and not in others and what is the explanation for the observed quantal failure rates? In answering these questions the research will advance some measures as conduits of our understanding while disqualifying other measures. Such qualification, or disqualification, arises from successful, or unsuccessful, quantitative matching of different sets of biological data. The essential organizing and interrelating principle is: identify those aspects of biology that quantitatively limit information processing in the brain. That is, the brain is a costly organ: food and water must be consumed to keep it working properly and, in terms of its absolute size, the brain is a burden for us to carry around. In performing such research, we will use mathematical analysis, computer-based calculations, and biophysical simulations. All of this work will draw on the most basic data about axons, dendrites, and synapses in the published literature. The proposed research promises to tie together diverse sets of anatomical and physiological observations, some of which are over fifty years old, well observed, often used, but never fully explained. The proposed research is necessarily theoretical theory being what is needed to produce a quantitative language for describing and understanding information processing. Because higher brain functions are built out of simpler bits of' computation such a solid foundation will benefit how neuroscientists study and understand higher order brain functions.

The wireless revolution in communication and computing has imbued electrical engineers with a steadily deepening appreciation for the importance of energy-efficient design. On the other hand, energy efficiency has for eons been a paramount consideration of living systems.

Statistical thermodynamics says that kTln2 is a lower bound for how many joules of energy must be expended in order either to encode/transmit/decode or to write/read/erase one bit of information in an environment at T degrees Kelvin, k being Boltzmann's constant. Biological encoding and decoding that performs close to the thermodynamic limit uses continuous time differential pulse position modulation (CTDPPM), arguably the most energy-efficient way to encode and transmit information. The co-PIs are applying their interdisciplinary expertise in neuroscience and information theory to address research topics such as: (i) Performance trade-offs in their bits/joule maximizing exact solution of a 1915 model by Schrodinger now known to be applicable to thresholding neurons, (ii) Extension of the Schrodinger-style optimization to short term dynamics of bits/joule maximizing CTDPPM systems; the present solution provides statistically optimum behavior only in the long term, (iii) Steps toward extension to networks containing multi-compartment elements, emphasizing hierarchical architectures with feedback governing short-term distribution of energy resources, and (iv) Intriguing connections with the burgeoning research discipline of network coding.