by Grigori Guitchounts
figures by Brian Chow
For decades, we have known that specialized neurons in the hippocampus of rodents called place cells reflect the animals’ location in space. Meanwhile, studies have also implicated the hippocampus in supporting memory formation. Could there be a link between the two seemingly detached functions? Yes! says the latest neurobiology research, showing that place cells not only encode an animal’s current location but also their memory for that location.
The 2014 Nobel Prize in Physiology or Medicine was awarded to a trio of neuroscientists for “their discoveries of cells that constitute a positioning system in the brain” (1). John O’Keefe, May-Britt Moser and Edvard Moser have contributed to neuroscience tremendously by observing the activity of special neurons in the hippocampus (Figure 1) and medial entorhinal cortex of rats called place cells and grid cells, respectively, that reflect an animal’s location in its environment. However, whether these neurons actually contribute to one’s sense of space has been difficult to test directly.
Figure 1. Comparison of human and mouse hippocampi, where place cells reside (not drawn to scale).
John O’Keefe first reported recordings of the activity of place cells in the hippocampus in a short paper in 1971 (2). Using the latest in transistor technology that allowed one to record neural activity in freely moving animals, O’Keefe implanted electrodes in rat hippocampus and observed the neuron activity therein as the rats explored a small rectangular arena. While most recorded neurons seemed to be activated by an assortment of behaviors (e.g. walking, sniffing) and sensory stimulation (e.g. whistles or clicks), a few displayed a remarkable firing pattern: they were activated when the rat was in one particular location within the arena (Figure 2). These neurons came to be known as place cells. Each place cell has a preferred location in the arena – its place field — where that cell would be most active.
Figure 2. As a mouse explores a rectangular arena, individual place cells in the hippocampus are active in certain spots within the arena, namely their place fields. The mouse’s path is indicated by the dotted line, and the location of each recorded place cell’s activity is denoted by colored circles
In the years since O’Keefe’s initial study, place cells have been shown to accurately encode an animal’s position in space. In one study, recordings from a population of place cells were used to train a computer program to predict a rat’s position. When fed with new recordings, the trained program was then able to predict the rat’s location within 10 cm (in a 10-meter-long track) (3), indicating that the activity of those neurons could encode information about the rat’s location. During sleep and quiet wakefulness, the hippocampus undergoes bursts of high activity called sharp wave ripples (SWP-R) that seem to re-play highly specific sequences of activity that were observed while the animal was awake and interacting with its environment (4). These replay events have been calculated to be extremely unlikely to occur by chance, and some have speculated that they represent the neurophysiological signature of dreaming.
Place cells became the basis for the idea that the hippocampus – at least in rodents – is used to gauge one’s position in an environment, akin to a cellphone’s GPS system. However, a large body of work in humans implicates the hippocampus as the seat of learning and memory for specific events (5). The case of Henry Molaison (known in dozens of studies as patient H.M.), whose hippocampi were surgically removed in 1953 as part of an epilepsy treatment plan, was the first piece of evidence for the idea that the hippocampus is crucial to forming new memories: H.M. remembered everything that’s happened to him prior to his surgery, but could not form any new memory, being forever stuck in the present moment.
The competing ideas – is the hippocampus used for navigation or memory? – are to this day not reconciled. While it is likely that the hippocampus is used for both, it has remained unclear whether the activity of individual place cells actually codes for one’s memory for position in a given environment or simply reflects the summation of a multitude of sensory inputs that are characteristic of that spot. In other words, is a place cell active in a location because the animal’s memory is activated by being there, or does it just respond to the animal being there now? Dissociating these two possibilities has been a challenge because one typically can’t separate the activity of a place cell from the animal being in that place cell’s place field, or preferred location. You would be correct to say, “sure, but if a place cell really does reflect memory of a place, shouldn’t it be active whenever the mouse remembers being in that place?”
So how do we peer into the mind of a mouse, to ask this question experimentally? In a recent paper published in the April issue of Nature Neuroscience, Gaetan de Lavilléon and colleagues in Paris leveraged the fact that place cells are active during sleep in order to explicitly test the idea that place cells represent the sense and memory of location rather than just sensation associated with that location (6). They reasoned that if a given place cell truly represents the animal’s memory for location, then pairing that cell’s activity with reward during sleep, when the animal isn’t actually in that cell’s place field, should still lead the animal to associate the location with reward when it awoke.
To do this, they implanted electrodes into the hippocampus of mice to record from place cells, and into the Medial Forebrain Bundle (MFB), electrical stimulation of which is known to produce intense feelings of pleasure. In one cohort of mice, the researchers allowed mice to explore a circular arena, during which they measured where each mouse spent its time, mapping each place cell to its place cell field in the arena. As expected, these mice explored the novel space homogeneously, not preferring any one spot over another Then, the researchers began the stimulation protocol: they picked one particular place cell and electrically stimulated the mouse’s MFB every time that cell was active. In a subsequent session in the same arena, instead of exploring the space uniformly, the mice tended to spend most of their time in the stimulated neuron’s place field. This isn’t surprising, since in this paradigm the mouse was in the arena during stimulation, and the place cell fired mostly in its place field. In this experiment, the place cell’s activity was still tied to the animal’s location.
Next, the researchers leveraged a separation of place cell activity from the animal’s location. They allowed a new cohort of mice to explore the same arena. As before, in this initial phase the mice ran around the entire arena. Having recorded a place cell and its place field, the researchers then put the mice into a separate sleep chamber, where the mice were allowed to sleep for one hour. As the animals slept, the recorded place cell’s every spike was immediately followed by electrical stimulation of the MFB. Next came the crucial test: when placed back into the arena after their nap, would the mice explore without preference as they did before (Figure 3)?
Figure 3. In the experiment reported in Nature Neuroscience, researchers first allowed mice to explore an arena, while mapping one place cell’s preferred location (top). Next, the mice were placed in a separate chamber, where they were allowed to sleep for one hour; as the mice slept, every time the recorded place cell was active, the researchers delivered rewarding electrical stimulation to the mouse’s medial forebrain bundle (MFB) (middle). When the mouse awoke, the researchers placed it back into the original arena, and observed its behavior (bottom). Amazingly, mice ran directly toward the area represented by the rewarded place cell, indicating that they had a memory for good feelings in that spot.
Amazingly, as the mice were put back into the arena, they headed straight for the stimulated neuron’s place field. Not only did they take less time to get to the stimulation-associated spot, but they also travelled more directly to it and spent about five times longer there after they underwent sleep stimulation. Interestingly, the effect was not long-lasting, virtually disappearing after the third trial. The authors speculate that because the mice were not getting any rewarding stimulation upon actual visitation to the place field, they learned after a few trials that they would not be likely to feel good in that spot after all. The fact that the mice ran directly to the spot that was represented by the stimulated place cell indicates that they had an accessible memory of having positive feelings in that location. If place cell activity only reflected the present moment of being in a given place, then sleep reward of place cell activity would not motivate the mice to run directly to the stimulated cell’s place field; instead, they would have explored the space uniformly, as before.
While several recent studies have established that it is possible to reactivate or create false memories (7, 8), this paper is the first to create false memories in sleeping animals, suggesting that inception, as Christopher Nolan would have us think, is in fact possible. Accordingly, the paper has received quite a lot of news coverage, focusing mainly on its creation of a false memory (9, 10). It is just as important, though, to note that these results validate the longstanding idea that place cells do in fact encode memory for physical locations (if they hadn’t, pairing place cell activity with MFB stimulation would not have lead the mice to prefer one spot over others). These results are also great evidence supporting the idea that replay activity during sleep has some functional role, and isn’t a byproduct of noisy activity in the brain.
Still, important questions remain. The memory aspect of this experiment is simplistic and in some ways, resembles classical conditioning. One might be tempted to go farther and ask whether memories for specific events, which must synthesize what happened where and when, could be implanted during sleep and used to influence behavior during wakefulness. To this end, scientists could electrically reward not the activity of an individual place cell, but sequences of activity across a population of cells, which are typically observed during more specific behaviors, such as running through a particular arm of a maze. If those sleep replay sequences represent memory for an experience, then upon waking the animals would be expected to preferentially engage in the behavior that produced the rewarded sequence of activity originally. It would appear that the simple yet powerful experiments by de Lavilléon and colleagues not only represent a neat addition to our understanding of memory and navigation, but may also pave a path for fascinating (though potentially sinister) techniques of neurobiological manipulation in the future.
Grigori Guitchounts is a second year PhD student in the neuroscience program at Harvard.
1. The Nobel Prize in Physiology or Medicine. <http://www.nobelprize.org/nobel_prizes/medicine/laureates/2014/>
2. J. O’Keefe, J. Dostrovsky, The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat. Brain Research 34, 171–175 (1971).
3. T. J. Davidson, F. Kloosterman, M. A. Wilson, Hippocampal Replay of Extended Experience. Neuron 63, 497–507 (2009).
4. D. Ji, M. A. Wilson, Coordinated memory replay in the visual cortex and hippocampus during sleep. Nature Neuroscience 10, 100–107 (2007).
5. J. C. Augustinack et al., H.M.’s contributions to neuroscience: a review and autopsy studies. Hippocampus 24, 1267–1286 (2014).
6. G. de Lavilléon, M. M. Lacroix, L. Rondi-Reig, K. Benchenane, Explicit memory creation during sleep demonstrates a causal role of place cells in navigation. Nature Neuroscience (2015), doi:10.1038/nn.3970.
7. S. Nabavi et al., Engineering a memory with LTD and LTP. (2014), doi:10.1038/nature13294.
8. S. Ramirez et al., Creating a false memory in the hippocampus. Science 341, 387–391 (2013).
9. M. Costandi. False memories implanted into the brains of sleeping mice. The Guardian. (2015). <http://www.theguardian.com/science/neurophilosophy/2015/mar/09/false-memories-implanted-into-the-brains-of-sleeping-mice>
10. J. Alford. Scientists Implant False Memories Into Sleeping Mice. IFL Science. (2015). <http://www.iflscience.com/brain/scientists-implant-false-memories-sleeping-mice>