The hippocampus is a critical brain structure for navigation, context-dependent learning, and episodic memory. It is composed of anatomically heterogeneous subregions, including CA1, CA3, dentate gyrus and subiculum. These regions differ in their anatomical inputs as well as in their internal circuitry. A major feature of the CA3 region is its recurrent collateral circuitry, by which the CA3 pyramidal cells make excitatory synaptic contacts on each other. In contrast, pyramidal cells in the CA1 region are not extensively interconnected. These anatomical differences have inspired numerous theoretical models of differential processing capacities of these two regions. In particular, David Marr (1971) conceptualized the hippocampus as an autoassociator network that performs pattern storage and retrieval. An autoassociator has three basic requirements: (a) massive internal recurrency among principal cells; (b) strong, sparse synapses from external afferents; (c) plasticity at the synapses between cells. Hippocampal CA3 satisfies all three requirements. First, each CA3 pyramidal cell may receive contact from 4% of other pyramidal cells. Second, CA3 cells receive a small number of inputs from mossy fibers of dentate granule cells, whose transmission fidelity at CA3 is very high. Third, plasticity of LTP/LTD has been demonstrated at CA3 synapses. Therefore, pattern would be presented in CA3 over the mossy fibers, where recurrent collateral synapses between coactive pyramidal cells would undergo plasticity to store the pattern. Later, if a partial input of that pattern is presented along the weaker entorhinal afferents, some CA3 pyramidal cells would become active. After several iterations of activity through the recurrent collaterals, more CA3 cells would be activated until the entire stored pattern is retrieved. This pattern completion concept occurs during memory retrieval. Nakazawa et al previously showed that mice that lack CA3 NRs are unable to recall spatial memory with only partial cue in Morris Water Maze hidden platform task, indicating that CA3 NRs are essential for spatial pattern completion (2002). However, the evidence for possible involvement of non-spatial pattern completion in the hippocampus is scarce. Further, there is no relevant task that can test both pattern completion and pattern separation, an ability to disambiguate contexts with different contextual (episodic) histories. In this project, we first designed an experiment that allows for the examination of non-spatial pattern completion and separation. Catherine Cravens demonstrated that by reducing startle effect and using distinguishable tone frequencies, C57BL/6N (B6) mice are able to perform both pattern completion and separation dependent upon tone frequencies. Mice were trained on a series of CS (tone)-US (foot shock) pairings. Twenty-four hours later the freezing of mice was observed in an auditory cue test where mice were presented, in alteration, with the training tones and tones whose frequency was different from that of the training tones. In the cue test, she first determined the maximum intensity of tone in which no startle responses were elicited. Under this condition, B6 mice are expected to use pattern completion for associative memory recall between the two different frequencies of tones if they show high levels of freezing on both tones. She then established the lower range limit of tone frequencies in which B6 mice are able to recall memory, indicative of pattern completion. This paradigm can be directly translated to examine pattern completion and separation in CA3 N-methyl-D-asparate (NMDA) receptor knockout mice to specifically examine the role of the hippocampus in pattern completion and separation (unpublished). We are in progress of testing CA3-NR1 knockout mice and their control littermates to examine deficiency in non-spatial pattern completion under this neww paradigm. Second research focus of this CA3 project is to seek the behavioral and physiological consequence of CA3 NMDA receptor knockout in nucleus accumbens (NAcc). It is well known that hippocampus sends a strong efferent into NAcc, probably providing the context information for the process of reinforcement learning in the NAcc. However, since NAcc also receives inputs from at least the cortex, amygdala, and the ventral tegmental area (VTA), how the hippocampal input modulate the NAcc activity and its physiological importance has not been clarified. Taking advantage of having CA3 NMDA receptor knockout mice, Dr. Juan Belforte and I have set up the recording environment with in vivo multi-electrodes from awake behaving mice, confirming that this system is useful for recording from the hippocampal CA1. Our recording from NAcc is in progress. We have also needed to establish relevant behavioral tasks, which could measure the function of NAcc and can be combined with in vivo physiology. Here Melissa Tanner has established the behavioral task called ?novel object to place? task, which has been reported to be defective following hippocampal lesion. She is now testing CA3 NMDA receptor knockout mice, whether any behavioral phenotypes are observed in the mutant animals. Another specific feature of the hippocampal CA3 is that this area is a generator of a peculiar brain wave pattern, so-called ?sharp wave?. The hippocampus displays fast (approximately 200 Hz) network oscillations called ?ripples?, which are superimposed on slower sharp waves. Such sharp wave-ripple complexes (SPW-R), which spread into CA1 and further extra-hippocampus, have been implicated in memory consolidation. Taking advantage of having KA-1 BAC promoter, which has shown to direct the transgene expression confined to CA3 pyramidal cells (Nakazawa et al., unpublished), we have started to develop CA3 pyramidal cell-restricted inducible transgenic mice. We use macrolide-inducible system, in which a trans-activator ET4 induces ETR-element dependent transgene transcription in the presence of erythromycin in vitro. To test this system in vivo, we have attempted to generate two mouse strains, one is CA3-restricted ET4 transgenic line and another is a transgenc line in which Kir2.1 fused with GFP (green fluorescent protein) is under the control of ETR-minimal promoter. Kir2.1 is an inwardly rectifying potassium channel, overexpression of which has shown to hyperpolarize the neurons, leading to the electrical shut down. By crossing these two lines, we expect to elicit membrane hyperpolarization only in CA3 pyramidal cells upon drug administration and hopefully suppress SPW-R generation. The key issue is whether the inducibility is sufficient to suppress SPW. Our screening of the two lines is underway.
| Kishimoto, Yasushi; Nakazawa, Kazu; Tonegawa, Susumu et al. (2006) Hippocampal CA3 NMDA receptors are crucial for adaptive timing of trace eyeblink conditioned response. J Neurosci 26:1562-70 |
| Nakazawa, Kazu; McHugh, Thomas J; Wilson, Matthew A et al. (2004) NMDA receptors, place cells and hippocampal spatial memory. Nat Rev Neurosci 5:361-72 |
