John Lisman - US grants

DESCRIPTION (Adapted from applicant's abstract): This proposal would continue a long-term investigation of phototransduction mechanisms in Limulus photoreceptors. CaM-kinase has been implicated in the light-dependent phosphorylation of arrestin, but its function is unknown. The possible role of CaM-kinase in regulating equivalent light and other transduction processes will be tested by intracellular injection of active kinase or kinase inhibitors. A second set of experiments will attempt to elucidate the late stages of transduction which remain unknown. These experiments are predicated on the hypothesis that the Ca2+ elevation produced by the phospholipase C (PLC)-IP3 pathway triggers a cyclase; the resulting elevation of cyclic nucleotide could activate the cyclic nucleotide-gated channels observed in earlier experiments on excised patches. If confirmed, this hypothesis would provide a complete pathway from rhodopsin to channel and a framework for understanding the degeneration produced by light in invertebrates. A final set of experiments builds upon the earlier observation that CaM may also be involved in an early step of transduction. Physiological and biochemical experiments are proposed to localize this step. Preliminary evidence suggests that it may be at PLC. These results, if confirmed, may reveal altogether new regulatory mechanisms for phospholipase and Ca2+, insight that may be important in understanding the role of this enzyme in vertebrate cones, where its role remains completely unclear.

Postsynaptic long-term potentiation (pLTP) in the frog sympathetic ganglion will be used as a model system for the study of mechanisms of information storage in the nervous system. Previous work by others has demonstrated that pLTP occurs in the cell body of ganglion cells, that pLTP lasts for at least several hours, and that external Ca is required for induction. The initial goals of this proposal are to further described the phenomenon: the full time course of pLTP ill be characterized; voltage-clamp methods will be use to determine whether the conductance that is potentiated is the nicotinic ACh conductance, as previously concluded by others on the basis of current-clamp recordings. In a second set of experiments, Ca buffers and Ca itself will be injected intracellularly to determine whether a rise in Ca is necessary and sufficient for induction of pLTP. The third set of experiments is aimed at identifying the biochemical process responsible for the long duration of pLTP. The PI has proposed the theory that the Ca-calmodulin dependent protein kinase II has the requisite biochemical properties to be switched "on" by a rise in Ca and to maintain this "on" state long after Ca returns to baseline. The possibility that such a storage process is responsible for maintenance of pLTP will be tested using both biochemical and physiological methods. A monoclonal antibody that inhibits the function of this kinase will be injected to see if pLTP can be blocked. The kinase itself will be injected to see if pLTP can be induced. Biochemical assays will be used to determine whether the activity of this enzyme is turned "on" during induction of pLTP and whether it remains "on" for the duration of pLTP. Parallel studies on other kinases of interest will also be conducted. These studies will provide the first systematic test of the most explicit model of neuronal information storage yet proposed. The information acquired may well be relevant to the general questions of memory storage.

Our general goal is to develop an integrated view of the biochemical processes that govern basal synaptic transmission and LTP in the hippocampal CA1 region. Aim 1 investigates the functional relationship of the protein kinases, CaMK, PKC and Src, all of which have been implicated in LTP. Internal perfusion of kinase inhibitors and alpha-CaMK knockout mice will be used to test the hypothesis that the primary role of PKC and Src in LTP induction is to modulate the NMDA conductance. Other experiments test the possibility that sustained PKC and CaMK activities are responsible for the maintenance of LTP. A critical issue, whether LTP can be reversed by subsequent addition of kinase inhibitor, will be further explored. The possibility that LTP can be easily reversed only in a brief time window after induction will be tested by laser-induced uncaging of kinase inhibitor just after LTP induction. Recent biochemical work suggests another possibility: that CaMK maintains LTP by forming a structural complex with NMDAR-2B. New inhibitors allow this to be tested. Aim 2 concerns our finding that LTP maintenance can be reversed by cyclic nucleotide analogs such as Rp-cAMPS. The possibility that this is due to inhibition of persistent PKA will be explored. However, it seems more likely that reversal is due to activation of cyclic nucleotide gated channels (CNGC). It is known that Ca2+ elevation can produce weakening, but the consequences of Ca2+ entry through CNGC has not been explored. We will determine whether local uncaging of cyclic nucleotides can raise Ca2+ and reverse LTP. Aim 3 builds on our observation that an actin-dependent process is required to maintain basal AMPA-mediated transmission. It appears that cytoskeleton-based vesicle transport systems are required to sustain the rapid turnover of AMPA channels. Consistent with this view, preliminary results indicate that a microtubule-motor system and a vesicle fusion process is required for maintaining AMPA transmission.

9723466 Lisman The long-term goal of this project is to understand the network and cellular basis of short-term memory, the type of memory used in remembering brief lists of items such as a phone number. Psychophysical tests on humans have shown that there is a limit of about 7 items that can be stored in short-term memory. The proposed research will extend previous theoretical work showing how a single brain network can keep 7 memories active at the same time using a multiplexing scheme. This scheme is closely tied to physiological observations on brain oscillations; multiplexing is achieved by a clocking mechanism, which is apparent in brain recordings as nested oscillations in which about 7 gamma (20-80Hz) oscillations occur within a single theta (4-10Hz) cycle. Memories become serially active in sequential gamma cycles. The psychophysical analysis of human memory provides a rich quantitative data set. The goal of the proposed work is to a develop a more detailed physiological model that correctly predicts this data. The first specific aim is to provide a specific model of how memories are scanned during a recognition memory test. Computer simulation methods will be used to model brain networks in a physiologically plausible way. The second aim is to understand a surprising constancy observed by Cavanagh: the time required to scan a memory and the number of memories that can be held both vary with the complexity of the memory; however the product of the two appears to be a universal constant. The third aim is to understand whether there are fundamental network constraints that limit how fast gamma oscillations can be; if they were faster, more memories could be held, but there may be reasons why gamma cannot be faster. Preliminary results related to the first two aims indicate that theta oscillation frequency may decrease as the number of stored memories increases. Furthermore, gamma frequency may decrease as the complexity of stored items increases. These are important findings because they provide a way of relating easily manipulatable aspects of short term memory to brain signals and because they are very testable. As the fourth goal of this proposal, collaborative efforts will be initiated to test these predictions.

Synapses on CA1 hippocampal neurons have become a leading model system for understanding CNS synapses and the activity-dependent functional changes that are thought to underlie memory. Structural studies of these synapses have revealed a remarkable diversity of synapse size and diversity in the structure and size of the dendritic spines on which the synapses occur. The function of this diversity is unknown. CNS synapses have other structural specializations, the function of which is also unknown. The general goal of this project is to study the function of single identified synapses and to relate their function to their structure. Critical to this goal is the newly developed method of optical quantal analysis. The method is based on the finding that active synapses can be identified by Ca2+ signals in the spine head using a high-speed confocal microscope. These signals, in conjunction with whole-cell recording, make it possible to characterize an identified synapse by determining the quantal-analysis parameters, p and q. 3D EM reconstructions will be made of the same synapses that are physiologically characterized. Ultrastructural features, such as the size and type of active zone, will be determined. By comparing structure to function, it will be possible to determine whether size is the primary determinant of synaptic strength or whether strong modulatory processes are also at work. Optical quantal analysis will also be used to study the effect of LTP at individual synapses. The data obtained may resolve the controversy regarding the presynaptic/postsynaptic locus of LTP expression. The ability to monitor the function of individual CNS synapses and the ability to relate their function to their structure should have wide ranging applications and wide ranging implications for understanding the synaptic malfunctions that underlie disorders of memory.

DESCRIPTION (provided by applicant): Over the 14 year tenure of this Program, Brandeis University has built a multidisciplinary program encompassing faculty in Biology, Biochemistry, Chemistry and Psychology. The Program educates students in the full range of Neuroscience topics, from the molecular biology of ion channels to the cognitive effects of aging. The uniqueness of this Brandeis Neuroscience Program has less to do with the breadth of educational options than with the fact that most students actually take advantage of that breadth. Each member of the faculty collaborates with multiple others, and most projects involve several levels of analysis. Students are part of an intellectually and spatially integrated neuroscience community. The resulting cohesion is reflected in every aspect of the program: coursework, rotations, thesis supervision, the general availability of advanced instrumentation, and the collegiality of interactions. Students enter Neuroscience from a remarkably wide variety of different backgrounds (Psychology, Biology, Biochemistry, Chemistry, Physics, Computer Science, Engineering, and Mathematics). The breadth of opportunity and interaction at Brandeis University allows each student to develop according to her/his individual needs, and results in a low attrition rate. While the Program is relatively small, over the last five years it has graduated 28 students (five of whom were members of underrepresented minority groups). A strong aspect of the Program is the integration of computational and experimental neuroscience. Emphasis is placed on making sure that students gain exposure to quantitative issues. Students graduate with excellent credentials, and go on to obtain excellent positions in academia and industry. The program of course work, rotations, multiple small-group colloquia, proposition examinations, and participation in teaching necessarily consumes the bulk of a student's time during the first two years. They cannot (and would not) have their laboratory research as their sole focus prior to their third year at Brandeis University. Therefore, they are not supported on research grants in these first critical years. This Training Program provides crucial funding to support students while they develop a broad set of intellectual skills. There are 23 mentors in the Program, and funds for ten trainees per year are requested. RELEVANCE: The goal of the Neuroscience Program is to understand brain function and to provide new strategies and therapies for the treatment of diseases of the brain. Training in neuroscience is particularly challenging because of the very broad range of information required. The Training Program is designed to give young scientists the intellectual and technical depth to make major contributions to the treatment of brain disease.

DESCRIPTION (provided by applicant): CaMKII is a leading candidate as a molecular memory. In a first aim, we will explore the hypothesis that CaMKII and phosphatase-1 (PP1) in the postsynaptic density (PSD) form a bistable switch. Monte Carlo simulations will be developed. The model will be used to address a fundamental theoretical question: how is the stability of stored information limited by stochastic fluctuations in the reactions of the small group of molecules at synapses? In a parallel set of biochemical experiments, we will directly test whether the CaMKII/PP1 system in isolated PSDs can act as a bistable switch. In Aim 2 we examine how synaptic strength can be bi-directionally modified by different patterns of synaptic activation. The moderate Ca2+ elevation during the induction of depotentiation reduces CaMKII phosphorylation through a phosphatase cascade that involves calcineurin, I1 and PP1. Since <1% of synapses are activated during induction active synapses are point sinks/sources, producing dendritic gradients of diffusible molecules such as I1. The buildup of such gradients could potentially explain kinetic aspects of induction and heterosynaptic effects. Because the role of gradients has not been previously considered, it will be useful to make a diffusion model to study how different factors affect the spatial/temporal gradient of possible importance in plasticity. In related physiological experiments, we will test the role of I1 in depotentiation. Aim 3 relates to recent work indicating that LTP produces a stable increase in synapse size. To understand the principles that might underlie such structural stability, we have formed a team with expertise in physical chemistry, structural biology, neuroscience and physics. Simulations will be used to explore the principles that could underlie structural stability. Together these aims address the deepest issues regard the mechanism by which memories are stored in the brain. The insights derived are likely to be of importance in understanding diseases of memory and suggest strategies for therapeutic intervention.

LTP is triggered by Ca^* entry through the NMDAR; subsequently Ca^* activates calmodulin (CaM), which then activates CaMKIl. Despite extensive studies demonstrating the pivotal role of CaMKIl in LTP and memory, the mechanisms of its activation in living cells is not known. The goal ofthe proposed work is to understand these mechanisms in quantitative detail. This requires methods to measure biochemical events in single spines near the limit of optical resolution and a sophisticated modeling framework for simulating these reactions. Because the experimental and computational methods were not previously available, this will be the first attempt to account for the measured activation of an enzyme in a living cell. Aim 1. Measurements will be made of critical quantitative properties of the system, including free CaMKIl, CaM, Ng and CaMKIl. This will be done used calibrated optical methods. Aim 2. To model CaM activation requires information about the spatial/temporal gradients of Ca^* in spines. 2-photon uncaging of glutamate will be used to activate NMDARs in a controlled way; the resulting Ca^* elevation in the bulk spine cytoplasm will be measured. Aim 3. Using fluorescence lifetime methodology (FLIM)), the time course of CaMKIl activation in single spines will be measured. Computer simulations will then be used to predict how the elevation of Ca^*, as determined in Aim 2, leads to CaMKIl activation. The parameters determined in Aim 1 are needed for this calculation. This predicted CaMKIl activation will be compared to the measured activation. Aim 4. Neurogranin (Ng) is an abundant postsynaptic protein that binds CaM and may be important in controlling the CaM that is available to activate CaMKIl. The effects of Ng knockout will be studied. RELEVANCE (See instructions): The proposed research is relevant to addiction, which involves persistent changes in synaptic strength. The role of CaMKIl in the persistence of synaptic strength has recently been demonstrated; notably biochemical attack of CaMKIl has reversed synaptic strength. There is therefore the possiblity that agents that attack CaMKIl can be used to reverse the synaptic changes that underlie addiction.

This is a proposal to host a workshop at Brandeis University on the role for learning of the neuromodulator dopamine--a key player in the brain's reward system. The format is to be a "discussion meeting" for an invited group of about 25 researchers, most of whom are working to understand how the neuromodulator dopamine affects synaptic plasticity in the brain, and thus plays a role in learning. Much of this work is new and unpublished. Only recently have investigators begun to realize that the role of dopamine in learning may be fundamental, so this is the best moment to bring those researchers together to discuss the effects of dopamine and the conditions under which it is released. This small gathering of active researchers should lead to more collaboration, and thus ensure that this new field will advance as rapidly and efficiently as possible.

DESCRIPTION (provided by applicant): Large amplitude, global EEG oscillations in the delta/theta frequency range are normally characteristic of slow-wave sleep. In schizophrenia, however, delta power is high during wakefulness in frontal and central regions. This has been termed a thalamocortical dysrhythmia. Many symptoms of schizophrenia can be induced by NMDAR antagonists; work in rats shows that a delta frequency dysrhythmia can be induced by injection of NMDAR antagonist into the thalamus. The first goal of our work is to understand how NMDAR antagonists generate this abnormality. Our preliminary results point to a cellular mechanism: cells of the nucleus reticularis (nRT) of the thalamus are hyperpolarized by NMDAR antagonist; this deinactivates T-type Ca2+ channels which then generate delta frequency bursting. Our second goal is to understand the molecular mechanism of this effect. Our preliminary results confirm in situ hybridization showing that thalamic cells contain a rare form of NMDAR subunit, NR2C. This subunit is weakly blocked by Mg2+ at resting potential and thus generates a significant inward current in response to ambient glutamate; block of this current leads to the hyperpolarization that produces delta- frequency bursting. Thus our work points to a molecular/cellular mechanism for dysrhythmia. The third goal of our work is to understand how dopamine interacts with these processes. Our preliminary results suggest that D2 action may be synergistic with NMDA hypofunction in producing dysrhythmia. We will study this process in vivo to determine whether the delta oscillations induced by NMDA hypofunction can be reduced by D2 antagonist. A critical aspect of the dysrhythmia hypothesis, as proposed by R. Llinas, is that subregions of the thalamus generate abnormal low frequency oscillations in associated subregions of cortex; this produces local deficits in information processing that underlie the symptoms of the disease. A final goal of our work is to test this hypothesis. We will use a CRE-recombinase method to produce postdevelopmental knockout of NMDARs in subregions of the nRT. We will test whether NMDAR knockout in anterior nRT can produce enhancement of delta power in the frontal/central regions affected in schizophrenia, without affecting occipital cortex. We will test whether behaviors mediated by these regions are selectively affected by these oscillations. Because the CRE- recombinase method produces chronic changes, it provides a model system for identifying the chronic processes that underlie schizophrenia and drug therapy. If successful, this mouse will model the EEG symptoms of schizophrenia and provide a system in which potential therapeutic targets can be identified, based on the known pharmacology of the thalamus, and then tested for their ability to reverse the EEG symptoms.

DESCRIPTION (provided by applicant): Intellectual Merit: Although considerable progress has been made in understanding how episodic memory is stored in the brain, fundamental questions remain. We will use a combination of experimental and computational approaches to study two major questions. Aim 1: Network mechanisms of sequence recall in the hippocampus: Our ability to recall a sequence of events is a core component of episodic memory. The hippocampus is necessary for episodic memory, and electrophysiological recordings from this structure have begun to reveal some of the processes involved. After a rat experiences a spatial path, it recalls sequences of places along that path during brief events called sharp waves. Recent experiments show that these waves are necessary for memory consolidation. Sharp waves are generated intrahippocampally, but the specific hippocampal subregions that produce them are not known. According to one hypothesis, CA3 alone generates the sharp wave. Another hypothesis (the ping-pong hypothesis) posits that sharp waves are generated by the combined action of the dentate gyrus (DG) and CA3.This hypothesis has its origins in the work of Sompolinsky and Kleinfeld, who argued that each cycle in sequence recall requires two substeps: a) a chaining substep in which cells representing one item stimulate the cells representing the next item in the sequence and b) an autoassociative substep that corrects minor errors produced in each chaining step, thereby avoiding concatenation of error. Lisman proposed how this could be mapped onto hippocampal circuitry: CA3 cells that represent the nth item in the sequence excite the n+1 item in the DG (by the known backprojections); then, the n+1 item in DG is sent to CA3 for error correction (pattern completion). This, in turn, initiates te next cycle. To distinguish between the ping-pong and CA3-alone hypotheses, J. Leutgeb will record simultaneously from dentate and CA3. She will determine whether both dentate and CA3 cells fire during sharp waves and whether they fire at a different phase of gamma oscillations, as would be expected from a ping-pong process. To test whether accurate sequence recall in CA3 requires the dentate, Leutgeb will use optogenetic methods. If the dentate is required, accurate sequence recall should be disrupted by artificially inducing activity in the dentate or by blocking activity in the dentate. The data analysis will be done in the Lisman laboratory. Aim 2: Neural coding in the hippocampus: Two key experiments have provided insight into how spatial and sensory information are encoded in the hippocampus. O'Keefe showed that different spatial positions in a sequence are represented in different phases of a theta cycle. The Moser lab demonstrated that sensory information is encoded by rate remapping: sensory information associated with a place is encoded by a change in the firing RATE of the place cells that encode that position. It is unclear whether these major ideas are compatible: the increased number of spikes during rate remapping might smear theta phase and thereby compromise phase coding. To determine whether this is the case, Lisman will analyze an existing data set and a new data set to be obtained by Leutgeb. Our working hypothesis is that rate remapping increases the number of spikes in a brief burst; since the spikes in a burst have nearly the same theta phase, phase coding would not be significantly degraded. Broader Impact: Understanding memory is a major goal of neuroscience because of the increasing incidence of memory problems in an aging population. The proposed experiment will identify a key component of the neural circuitry that underlies episodic memory. This may enable more targeted therapeutic strategies. A second contribution will be a MATLAB tutorial for the Summer Program in Neuroscience, Ethics and Survival (SPINES). This is an NIMH-sponsored course aimed at helping grad students and postdocs from under-represented minority and disadvantaged groups to develop knowledge and skills. Lisman and Leutgeb will lecture on major advances in episodic memory and describe how computation is important in this endeavor. This will be followed by a one-month tutorial on the computational platform MATLAB.

The mechanism of memory is one the major mysteries of biology. Recent work suggests that as a result of learning synapses grow and that the size of the synapse is what stores the components of memories. The aim of the proposed work is to visualize directly this growth process in brain tissue using a newly developed super-resolution microscope, and to understand why such structures have stable size once learning has occurred. Instability would result is loss of memory, so evolution is thought to have favored ways of maximizing stability. To gain insight into the mechanism of stability, physical and computational model systems will be used. If the principles that underlie stability in the face of variable size can be understood, the outcome of this work could open the door to a new era in nanotechnology in which these principles could be utilized, leading potentially to novel solutions to problems in self-assembly. Additional contributions of this project include the organization and instruction of a course in the scientific programming language, MATLAB, in an enrichment course for students from groups under-represented in science and technology, and the opportunity for US trainees to participate in an international collaboration.

This proposal focuses on supramolecular structures that do not have fixed size but can exist in multiple different sizes, all of which are stable. Thus, if a stimulus causes the transition from one stable state to another, the structure has information storage capability (memory). The investigators termed this type of structure variable-size stable structures (VSSS). Interest in VSSS arises from two seemingly unrelated fields: neuroscience and the physics of nanostructures. The molecular basis of memory is one the most fundamental unsolved problems in neuroscience. Evidence strongly suggests that synapses grow to encode memory. Thus, memory storage in the brain appears to be a structural problem, and efforts need to be made to understand the structural principles that make memory storage possible. The project integrates cutting-edge optical microscopy with theoretical modeling. Utilizing a newly-available super-resolution microscope, the investigators will make the first effort to observe synaptic growth during synaptic plasticity in real time. The goal of the theoretical efforts is to develop a physical theory of VSSS and evaluate different models, including ones that have emerged from the study of synapses. Questions to be addressed include: (i) The importance of cooperative interactions among multiple components to generating stable yet kinetically accessible and reconfigurable assemblages. (ii) Design principles that lead to self-terminating assembly, such as growth by finite-size modules. (iii) Mechanisms by which nonequilibrium energy consumption changes the limits of VSSS. An ultimate goal is a generalized theory for nonequilibrium self-assembly capable of describing VSSS.

This project is jointly funded by the Neural Systems Cluster in the Division of Integrative Organismal Systems and by the Physics of Living Systems Program in the Physics Division.

? DESCRIPTION (provided by applicant): Activity-dependent synaptic modifications (LTP/LTD) are the leading candidate for the mechanism of memory. LTP involves induction, maintenance, and expression processes. This proposal seeks to elucidate the maintenance processes that store memory information. The critical test of any hypothesis regarding memory maintenance is the erasure test in which an inhibitor is applied after LTP/memory is established. If this blocks LTP/memory and the effect persists after the inhibitor is removed, the inhibitor must have erased a maintenance process. We have conducted the erasure test using an inhibitor of CaMKII (CN-peptide). We found that application of this peptide after LTP induction produced erasure of saturated LTP. Moreover, after LTP was erased, LTP could be reinduced, indicating that plasticity mechanisms remained intact. The overall goal of this proposal is to conduct further critical tests of the role of CaMKII in maintenance of LTP and to extend this work by examining the role of CaMKII in the maintenance of behavioral memory. Memory maintenance is likely to depend on many different processes. The results of the proposed work could provide the first strong evidence identifying one of these processes. The goal of Aim 1 is to test a specific variant of the CaMKII hypothesis. Biochemical experiments show that activation of CaMKII causes it to bind to the NMDAR. We have shown that CN-peptide, a peptide that produces a reduction in the complex, reverses LTP. These results suggest that the CaMKII/NMDAR complex forms during LTP and is responsible for the maintenance of LTP. However, there have been no previous methods for monitoring the complex dynamics at synapses during and after LTP induction. Thus, crucial information regarding the complex formation and persistence is lacking. We have developed and validated a novel optical method based on FLIM-FRET. Our preliminary evidence demonstrates that LTP induction produces complex formation in spines and that the formation is synapse specific. We will determine the duration of the complex under conditions that either induce short-lasting LTP (early LTP) or produce both early and late LTP. We will also examine how the duration of the complex depends on factors that enhance (e.g., BDNF) or prevent (e.g., protein synthesis inhibitors) late LTP. These experiments will provide a strong test of whether the complex has the persistence required to be a molecular memory. The goal of Aim 2 is to conduct the erasure test at the behavioral level. We will use conditioned place aversion, a hippocampal-dependent form of memory. After learning is achieved, HSV virus will be used to deliver a dominant-negative form of CaMKII (K42M) to the CA1 region, producing transient (several-day) expression of this mutant kinase. Memory retention will then be tested a week later. Our preliminary results indicate that this form of behavioral memory can be erased by this procedure. In additional experiments, we will test the more specific hypothesis that behavioral memory is dependent on the complex of activated CaMKII with the NMDAR (or other PSD proteins to which CaMKII binds).

Project Description In schizophrenia (SZ), the power of low-frequency EEG oscillations (delta/theta; 1-7 Hz) is elevated in the awake state in subregions of the thalamocortical system. NMDAR antagonist induces similar low-frequency oscillations and also produces many of the symptoms of SZ, thus raising the possibility that the abnormal delta oscillations in SZ are causal in producing symptoms of the disease. Our previous work using the NMDA hypofunction model shows that T-type Ca channels and NR2C are critical for generation of abnormal delta oscillations. We further showed that optogenetically inducing delta oscillation in the nucleus reuniens of the thalamus is sufficient to interfere with working memory, a cognitive process that shows deficits in SZ. This supports the hypothesis that abnormal delta in SZ could be causal in generating disease symptoms. Independent support for this framework for understanding SZ has come from two genome-wide studies; these identified the same isoform of the T-type channel as a risk gene for SZ. Thus, there is strong rationale for further understanding of how abnormal delta can produce symptoms of SZ. Furthermore, elucidation of the cellular and molecular mechanisms may suggest new strategies for disease treatment. In Aim 1, we will further analyze how optogenetic stimulation of the reuniens at delta frequency interferes with working memory. The experiments are designed to determine whether the oscillations interfere with encoding or recall processes. The goal of Aim 2 is to test, in vivo, our understanding of delta generation and to determine whether drugs that reduce delta ameliorate symptoms in an animal model. Our specific hypothesis is that drugs that inhibit T- channel function directly, or reduce their function by depolarizing cells (thus producing inactivation of T channels), will reduce delta oscillations and ameliorate behavioral deficits. We have developed an in vivo assay in which we can evoke delta oscillations by injection of ketamine into the thalamus; we will use this model to evaluate drugs for their ability to reduce the power of these oscillations. We will further test drugs using the Df(16)A+/- mice that have been generated to model the human chromosomal deletion 22q11.2 that is the largest known risk factor for SZ. Consistent with the importance of delta oscillations, these mice have elevated delta power in the awake state. With this model, we can test for drugs that reduce delta power and determine whether these drugs ameliorate the working memory deficits in these animals. In Aim 3, we test a novel hypothesis about the negative symptoms of SZ, symptoms that have been particularly difficult to understand and treat. This explanation is built on a proposal by Graybiel/Surmeier according to which activity in the parafasicular/centro-medial (PF/CM) nucleus of the thalamus preferentially activates the indirect (NoGo) pathway of the basal ganglia. Activity in the indirect (NoGo) pathway is thought to inhibit behavior and could thus produce avolition. We will use optogenetic methods to test whether imposing delta oscillations in PF/CM produces avolition.

Experiments show that interference with CaMKI I after LTP can erase LTP, a strong indication of the importance of CaMKll in the LTP maintenance process. CaMKll holoenzymes contain 12 catalytic subunits. In the ON state, each subunit is phosphorylated and therefore active. When a site becomes dephosphorylated, it can be refreshed (rephosphorylated) by a neighboring active subunit. Aim 1. Is refresh dependent on activity? The CaMKll refresh process requires low levels of Ca, but whether achievement of this level is dependent on spontaneous neural activity is not known. Therefore, a fundamental question of interest is whether the maintenance of LTP or memory requires neural activity. In Aim 1A, we will test this in acute hippocampal slices. In Aim 1B, we will explore whether activity is necessary for maintaining memory at the behavioral level using Drosophila. Although it is known that CaMKll is important for Drosophila memory, experiments have not yet tested whether CaMKll is important in memory maintenance. Given the importance of this issue for interpreting the effects of activity on memory, we will conduct the critical erasure test for determining whether CaMKll mediates memory storage in Drosophila. Aim 2. Does CaMKll subunit exchange occur in vivo: a potential mechanism for molecular refresh? According to theoretical models, switch stability could long outlive the lifetime of any subunit if CaMKll underwent protein turnover by subunit exchange: a newly inserted unphosphorylated subunit could be phosphorylated by a neighboring phosphorylated subunit, thereby providing a molecular refresh. We will make the first attempts to test whether subunit exchange occurs in living cells (Drosophila and hippocampus) and characterize its activity-dependence. Aim 3. Computational modelling: what kinds of neural activity are required to refresh CaMKll phosphorylation? The level of resting Ca2+, and that during spontaneous action potentials or mEPSPs can be estimate, as well as the rate of these reactions. We will use a verified computational model of CaMKll to determine whether these brief Ca2+ events are sufficient to refresh the phosphorylated state of CaMKll and thus ensure the stability of stored information. The results will bear importantly on the fundamental question of whether refresh reactions are mediated by spontaneous activity, or alternatively, are dependent on a network process that replays memories. RELEVANCE (See instructions): Understanding the processes that store memory at synapses will have major implications for several health problems. In particular, addiction has been demonstrated to involve persistent changes in CaMKll at synapses in the basal ganglia networks that are critical for addictive behaviors. The proposed work may provide ways to turn off CaMKll and thus reduce addictive behaviors. CaMKll has also been strongly implicated in memory disorders and stroke.

In this project we explore question of how the information replayed during the SPW-R is represented, how and in which sub-networks this information may get encoded during learning, and the molecular and network mechanisms that lead to storage of information at hippocampal synapses. Project 3.1 seeks to develop a method for artificially inducing a memory using new optogenetic methods that allow small groups of neurons to be excited in awake behaving animals. A measure of success will be if the artificial memory is replayed during a SPW-R. The value of such a system will be that it will provide a controlled system for studying the biophysics of memory encoding. Project 3.2 seeks to erase a behaviorally encoded memory. Recent work suggests that memory is stored synaptically by the abundant synaptic protein, CaMKII. This work was based on the ability of dominant negative CaMKII to erase conditioned place avoidance. We will test whether dominant negative CaMKII in CA3 or DG can erase the ability of hippocampal system to replay a memory during the SPW-R. Project 3.3 explores the question of how spatial information is represented in memory. One view is that spatial memory is represented as a continuous path. We will test the alternative hypothesis that memory can be represented selectively by reward sites. Project 3.4 takes advantage of recent advances in long-term optical recording of identified cells. This will allow us to make the first observation of the long-term stability of the stored sequence content replayed during SPW-Rs, as well as longitudinally identifying the network elements associated with this long-term replay, thereby providing important insight into the mechanisms of consolidation.

Although neuroscience has provided a great deal of information about how neurons work, the fundamental question of how neurons function together in a network to produce cognition has been difficult to address. Our group has been at the forefront of developing methods that allow large scale monitoring of identified neurons, monitoring of voltage signals by optical means and elucidation of subcellular events in dendrites, all of which can now be done in awake behaving animals. We propose to use these methods to provide a deep understanding of how the neurons of the hippocampal region generate the sharp-wave ripple (SPW- R). This remarkable signal has been shown to depend on prior learning and to produce high-speed replay of memory sequences (e.g. a path along a track). The function of this signal is memory consolidation; disruption of SPW-Rs results in strong deficits in memory-guided behavior. Because much is known about the hippocampal cell types involved and their network connections, understanding the SPW-R is a tractable target for the first major effort to elucidate the cellular/network mechanism of a mammalian brain signal at an analytical level comparable to that achieved in the study of simple invertebrate systems. Project 1 is aimed at understanding the external and intra-hippocampal pathways that control the initiation of SPW-Rs. Project 2 deals with the events that occur during the SPW-R, including the timing of activity in identified cell types and understanding the fundamental network architecture by which memory sequences are produced. Project 3 deals with how the information that is replayed during the SPW-R is encoded. We will attempt to create an artificial memory and then determine whether the memory is replayed during a SPW-R; we will also interfere with molecular mechanisms of memory storage to determine whether we can erase the memories that are replayed during the SPW-R. Project 4 builds upon recent work indicating that differentially projecting CA1 pyramidal cells have distinct properties and will test the possibility that SPW- Rs in distinct output channels may carry different information and affect different behaviors. In Project 5 we will develop the first non-reduced computational model of the hippocampus, incorporating information about cell types and connections. This will be a major new resource for our group and the research community that will permit unprecedentedly close interplay between experiment and computation. To the extent that the model can account for the experimental observations, we can use it to understand underlying network principles and design interventional experiments to validate this understanding. To the extent that the model cannot explain results, it will help point us to aspects of network function that require further elucidation. Taken together, Projects 1-5 provide a tractable path to a major breakthrough in understanding how a cognitively important brain signal is generated.