|
magnified
scene by clicking image
Figure
1. Hippocampal place fields on two different linear track mazes. Colored points
represent spiking of individual neurons in area CA1 of the hippocampus. Different
cells (colors) fire at different locations on the tracks. (Mehta et al., Neuron,
25:707-715, 2000) |
|
|
Introduction
Research in the Wilson laboratory focuses on the study of mechanisms that underlie formation and maintenance of distributed memories in freely behaving animals addressing the basic question of what does a memory look like at the level of neuronal ensembles and how is it influenced by experience?
Combining behavioral electrophysiology with molecular genetics
Through collaboration with the laboratory of Dr. Susumu Tonegawa, we have been able to combine the measurement of ongoing neuronal activity with manipulation of molecular genetic targets to allow the study of how specific cellular mechanisms regulate neural function to produce learning and memory at the behavioral level. Recent experiments made use of mice engineered by Dr. Kazutoshi Nakazawa in the Tonegawa laboratory in which deletion of the NMDA receptor was restricted to the CA3 subregion of the hippocampus. The highly recurrent pattern of connectivity within CA3 has led to suggestions that this region of the hippocampus might serve as an auto-association network in which memories can be retrieved through the mechanism of pattern completion. In pattern completion, the network is able to retrieve complete memories given only a subset of the original cues that were present when it was formed.
This was tested behaviorally by training mice to find a hidden platform located in a pool of water using the spatial arrangement of four visual cues outside of the maze to navigate. To see how well the animals could retrieve spatial memories formed in the pool when given only partial cueing information, several of the visual cues outside the tank were removed. Normal mice easily remembered the location of the platform when tested in this way, but the mice with altered NMDA receptors in the CA3 area of the hippocampus were impaired. By placing electrodes into the hippocampus of these mice, Michael Quirk and Kazutoshi Nakazawa were able to observe the patterns of neural activity related to spatial memory formation and retrieval in these animals. Spatial experience creates unique patterns of activity in the hippocampus that are believed to correspond to the memory of that experience. Individual neurons called place cells fire in response to memory of specific locations in space.
When animals explored an environment in which all four visual cues were present, the patterns of place cell activity in the genetically altered mice appeared to be normal. But when cues were removed, the cells responded much more weakly than cells in normal animals. The weakened response was attributed to the inability of the CA3 area to successfully retrieve the complete memory of that environment when cued with partial information.
magnified
scene by clicking image
Figure
2. Place fields of 3 cells (rows) in the deep entorhinal cortex on two different
linear track mazes. Color indicates average firing rate as a function of location
on the maze. Individual cells fire in a similar manner for both the W and U track
suggesting that they generalize across environments. (Frank et al., Neuron, 27:169-178,
2000) |
|
|
Spatial-temporal sequences in hippocampal memory processing
While a model based upon the simple associative memory properties of hippocampal synaptic connections can explain the basic properties of spatial context memory formation and retrieval observed in these animals, it is known that the timing of pre and post synaptic activity further regulates synaptic plasticity in a temporally asymmetric fashion. Presynaptic activity that arrives prior to postsynaptic output within a narrow time window on the order of tens of milliseconds, leads to synaptic potentiation, while reversing this order leads to synaptic depression. Mayank Mehta proposed that this property of synaptic plasticity in the hippocampus would lead to spatially asymmetric changes in spatial receptive fields that would reflect the history of behavior in a given environment, allowing animals to form memory of temporally sequenced spatial experience. This prediction was confirmed when it was found that place fields in area CA1 rapidly take on a spatially asymmetric shape consistent with the history of behavior and the temporally asymmetric properties of NMDA-mediated synaptic plasticity in this area (figure 1).
This result suggests that the hippocampus and adjoining areas might be involved in the processing of temporally sequenced experience that in the spatial domain would appear as spatial trajectories. Precisely such responses were observed during spatial tasks involving multiple paths. Loren Frank recorded simultaneously from the cells in the superficial and deep layers of the entorhinal cortex that provide the input to the hippocampus and also receive the output of the hippocampus. Cells in each of these areas were found to show dependence upon spatial trajectories or paths. Neurons in the deep entorhinal cortex that process the output of the hippocampus also appeared to generalize across environments (figure 2), capturing regularities in spatial behavior that might be used to construct generalized models that are derived from specific unique experience, but are more applicable to novel circumstances-a property that might distinguish neocortical and hippocampal memory systems.
Examining the reactivation of memory during sleep
In addition to these studies of learning and memory in awake behaving animals we have been exploring the nature of sleep and its role in the processing of memory. Previous theories have suggested that sleep states may be involved in the process of memory consolidation, in which memories are transferred from short to longer-term stores and possibly reorganized into more efficient forms.
Our observations of the temporal sequence dependency of responses of hippocampal neurons and neurons of adjacent cerebral cortex suggested that memory of temporally ordered events might also be maintained within these structures.
In order to observe temporally sequenced event memory we examined the activity of hippocampal neurons during periods of sleep. The study of sleep provides an opportunity to identify mnemonic activity that results from behavior in a context in which sensory and behavioral input no longer contributes to that activity. Hence, it can be argued that this activity is a direct reflection of the residual influence of experience on neural substrates and therefore must necessarily be derived from underlying mechanisms of memory.
magnified
scene by clicking image
Figure
3. (A) Activity of 10 hippocampal place cells during sleep before (sleep 1) and
after (sleep 2) running on a behavioral task (run). Periods of REM sleep are shown
in gray. (B) The pattern of activity over several minutes of behavior shows close
correspondence to the pattern of activity over several minutes of REM. (Louie
et al., Neuron, 29:145-156, 2001) |
|
|
Albert Lee working in the lab discovered that temporalsequence memory of recent experience was reactivated incompressed form during periods of slow-wave sleep immediately following that experience. This mnemonic reactivation took place predominantly during brief periods of coordinated network discharge known as "ripples" lasting approximately 100 milliseconds. Earlier findings by Thanos Siapas showed that these hippocampal discharge events tended to occur in conjunction with a neocortical rhythmic event known as the sleep spindle suggesting that these events might serve as the initial stage of a process of memory consolidation i
nvolving communication between the hippocampus and neocortex.
We have also found evidence of long timescale, sequential event memory that is believed to be a critical function of hippocampal circuits by examining the patterns of place cell activity across ensembles of hippocampal neurons in the rat during rapid eye movement (REM) sleep. Kenway Louie discovered that memories of event sequenceslasting minutes were replayed on a similar timescale in over 40 percent of REM episodes 24 hours or more after the awake experience (figure 3). The correspondence was sufficiently robust to allow reconstruction of the spatial trajectories being replayed on a second by second basis over the course of an entire REM episode.
Conclusion
Overall, these results indicate a primary role of the hippocampus in the storage and retrieval of associative memories. The temporal component of memory established through specific behavior within that environment would further be encoded through temporally asymmetric modifications of synapses within this region allowing paths or trajectories to be incorporated into the hippocampal memory system. The generalization of these paths or sequential events in the neocortex as evidenced by the responses in deep entorhinal cortex would provide the means to construct models of behavior derived from experience but able to guide the animal under varied conditions. The retrieval and replay of these memories during sleep might provide a mechanism by which this mnemonic information in the hippocampus is gradually incorporated into neocortical circuits. |
|
|
|
|
|