David B. Jaffe - US grants

IBN-9511309 Jaffe The hippocampal formation is a region of the brain that is important for certain aspects of learning and memory. The goal of this project is to examine how an important component of the hippocampus, the CA3 region, might process, store and recall specific patterns of information. A computational model of the CA3 neural network will be used to test hypotheses. Specific parameters regarding the firing properties of inhibitory neurons and certain aspects of synaptic transmission will be obtained using state-of-the-art electrophysiological recording methods from neurons in hippocampal slices. The computer network will contain 1,000 to 10,0000 neurons based on published data and the experimental data from inhibitory neurons. A key feature of the network will be that excitatory connections between pyramidal neurons will potentiate or depress based on biophysical rules for synaptic potentiation and depression. The CA3 network will be trained with different patterns of input to test the following: 1) can the network can discriminate between input patterns, 2) can the network complete partially presented patterns, and 2) is information distributed across the network. This work will be important for our understanding of how this region of the hippocampus is involved in associative information storage.

IBN: 9634401 PI: Hiromoto Specific functional areas of the brain can contain 100,000 to 1 million neurons. When trying to model functional aspects of these neural networks it is critically important that the model is of a sufficiently large scale (greater than 1,000 neurons) to examine emergent properties of the network. Large-scale biophysical neural networks on the order of 10,000 neurons require efficient computational resources that are beyond the normal scope of single high-speed computer workstations. Advances in interprocessor communications and algorithms for computing large-scale problems now permit us to apply these methods to neurophysiological questions. In this work, we intend to develop tools that can be used to investigate how networks of biophysical models of neurons behave based on known anatomical connectivity, cellular neurophysiology, and rules for synaptic plasticity. Our project has two specific aims. First, we will develop algorithms and supporting software that exploit the inherent properties of parallel computational architectures for simulating the synchronous activity of neuron models interconnected by synapses that undergo activity-dependent changes in strength. Such changes may accompany learning. The simulator developed here will be user-modifiable at many levels allowing for the study of different brain regions having varied anatomical and physiological characteristics. Second, graphic visualization tools will be developed to better study and analyze the large amounts of data produced by the large-scale neural simulations. These will include methods to examine the states of individual neurons as well as the correlated activity of different neurons within the network.

Debilitating memory impairment is the most consequential result of Alzheimer's disease. Pathological changes in selective regions of the hippocampal formation, an area of the CNS important for the acquisition and consolidation of certain types of information into long-term memory (8), are found in Alzheimer's patients (7). Long-term potentiation (LTP) of synaptic transmission is a leading candidate mechanism for memory storage in the CNS and can be expressed at all major excitatory synapses within the hippocampus (1). A general hypothesis is that the hippocampus stores information, through LTP, for later consolidation into long-term memories in other brain areas. We are therefore interested in how the hippocampus processes and stores information. The mossy fibers are a major excitatory afferent pathway into the CA3 region of the hippocampus. They are axons of dentate granule neurons and synapse onto CA3 pyramidal neurons. Synaptic transmission at the mossy fibers is through excitatory amino acids, but they also release a number of opioid peptides (6). Mossy fiber LT? (MF-LT?) has a number of unique properties compared to most other excitatory synapses of the hippocampus. Most notably, it is not dependent upon N-methyl-D-aspartate receptors (2). MF-LT? is also dependent upon opioid receptor activation (5). Furthermore, there appear to be two forms of MF-LT? (9). One form is triggered by solely a presynaptic mechanism (11). The other form is dependent upon postsynaptic depolarization and an increase in postsynaptic calcium (4, 10)and is therefore Hebbian; it requires conjunctive presynaptic and-postsynaptic activation (3). This is important because neural networks connected by Hebbian synapses have the potential for associative information storage. Recent work in our laboratory suggests that there are at least two potential sources of calcium available to trigger this form of MF-LT?, including voltage-gated calcium entry and the release of calcium from intracellular s tores. I We propose to study the biophysical and pharmacological mechanisms involved in triggering mossy fiber LTP. We will use the in vitro hippocampal slice preparation and a combination of extracellular, whole-cell, and fluorescence imaging methods for these studies. The results of these experiments will be important for our understanding of the mechanisms underlying associative information storage in the hippocampal formation.

Debilitating memory impairment is the most consequential result of Alzheimer's disease. Pathological changes in selective regions of the hippocampal formation, an area of the CNS important for the acquisition and consolidation of certain types of information into long-term memory (8), are found in Alzheimer's patients (7). Long-term potentiation (LTP) of synaptic transmission is a leading candidate mechanism for memory storage in the CNS and can be expressed at all major excitatory synapses within the hippocampus (1). A general hypothesis is that the hippocampus stores information, through LTP, for later consolidation into long-term memories in other brain areas. We are therefore interested in how the hippocampus processes and stores information. The mossy fibers are a major excitatory afferent pathway into the CA3 region of the hippocampus. They are axons of dentate granule neurons and synapse onto CA3 pyramidal neurons. Synaptic transmission at the mossy fibers is through excitatory amino acids, but they also release a number of opioid peptides (6). Mossy fiber LT? (MF-LT?) has a number of unique properties compared to most other excitatory synapses of the hippocampus. Most notably, it is not dependent upon N-methyl-D-aspartate receptors (2). MF-LT? is also dependent upon opioid receptor activation (5). Furthermore, there appear to be two forms of MF-LT? (9). One form is triggered by solely a presynaptic mechanism (11). The other form is dependent upon postsynaptic depolarization and an increase in postsynaptic calcium (4, 10)and is therefore Hebbian; it requires conjunctive presynaptic and-postsynaptic activation (3). This is important because neural networks connected by Hebbian synapses have the potential for associative information storage. Recent work in our laboratory suggests that there are at least two potential sources of calcium available to trigger this form of MF-LT?, including voltage-gated calcium entry and the release of calcium from intracellular s tores. I We propose to study the biophysical and pharmacological mechanisms involved in triggering mossy fiber LTP. We will use the in vitro hippocampal slice preparation and a combination of extracellular, whole-cell, and fluorescence imaging methods for these studies. The results of these experiments will be important for our understanding of the mechanisms underlying associative information storage in the hippocampal formation.

The long-term goal of my research is to understand how the properties of individual neurons participate within a network to promote associative learning and the normal cognitive processing of information. In both neocortex and hippocampus, 40-100 Hz (gamma) oscillations are observed during specific behaviors. A widely held hypothesis suggests that perception and the binding of features during learning or memory retrieval results from the temporal relationship between coactivated neurons. Gamma oscillations could provide a means for the long-range synchronization of neural activity. Both theoretical and experimental evidence point to inhibitory interneurons as a major source of rhythmogenesis in the hippocampal formation, a region of the brain important for certain aspects of learning and memory. A knowledge of the basic mechanisms underlying interneuron excitability is therefore critical for our understanding their role in hippocampal information processing. There is evidence that unitary synaptic inputs from pyramidal neurons have a high probability of triggering interneuron firing. Active dendritic conductances have been proposed as a mechanism for achieving the observed high firing probability. Specific Aim 1 will test the hypothesis that dendritic spike initiation is a common feature of hippocampal CA3 interneurons. We will first directly test the hypothesis that action potentials are initiated in the dendrites of hippocampal interneurons. Dual patch-clamp recordings from the dendrites and soma of hippocampal CA3 interneurons will be used to determine if spikes originate in the dendrites or in the soma. Next, we will examine whether dendritic spike initiation occurs across specific classes of hippocampal CA3 interneurons. Specific Aim 2 will test the hypothesis that the properties and distribution of voltage-gated conductances favors dendritic spike initiation. Given that differences in voltage-gated Na+ or K+ channels between the soma and dendrites may account for dendritic spike initiation, we will characterize the types of voltage-gated channels found in the dendrites and soma of hippocampal interneurons. Finally, Specific Aim 3 will test the hypothesis that the properties of excitatory synaptic transmission promote a high-firing probability in CA3 interneurons. We will test whether frequency facilitation or the spatial summation of spontaneous synaptic activity enhances the likelihood of interneuron firing in response to minimal excitatory stimulation. The results from this study will further our understanding of how inhibitory interneurons operate within cortical neural networks, including the hippocampal formation.

DESCRIPTION (provided by the applicant): I am a mathematician with a Ph.D. from the University of California at Berkeley. Through theory and computation, including large-scale software development, I solved longstanding open problems in the fields of algebraic geometry and error-correcting codes. I have grown increasingly interested in the computational aspects of biology and recognize that a number of important and challenging biological problems can be addressed using the mathematical tools I am familiar with. Stimulating and productive interactions with researchers at the Whitehead Institute, including work on whole genome shotgun assembly, have confirmed my enthusiasm for this area of research and revealed my need for deeper and more formal training to understand the biological context of these problems. I am committed to a course of retraining which will enrich my biological background and facilitate a career in genomics through coursework, on the job training, and research experience. The grant will enable me to continue and extend my work on fundamental open problems of the field, including cost-effective and accurate sequence acquisition and assembly, cross-species genomic alignment, and through the latter, effective location of genes and regulatory regions, thereby shedding light on vast riches of new genomic data. I will carry out my work at the Whitehead Institute, where an interdisciplinary spirit and stimulating intellectual environment meshes with a cutting-edge production facility. Dr. Eric Lander will direct my project. His background and research interests make him ideally suited to supervising me. He also earned his doctorate in pure mathematics and has worked successfully at the interface of pure mathematics and molecular biology as well as in genome research. I look forward to joining his team, which will design the sequencing and assembly of the mouse genome, do the sequencing itself, and from there make deductions about both the human and mouse genomes.

computer system design /evaluation