Life is complex. The way a living system works can be described in a series of increasingly refined models, each fleshing out details of the previous model. Typically, description at one level raises questions about what is happening at the finer level. These questions induce hypotheses which drive experimental work which produces ever more detailed knowledge.
A paper about memory, just published, is an example of one incremental step in this process. In short, this research works out some of the fine detail at the molecular level for the process of forming visual memories.
In mammals, memories seem to be formed among the neurons in the hippocampus, and then transfered elsewhere in the brain. Memories that are still in the hippocampus are called "short term" and memories that are reformed elsewhere are called "long term." In various experiments, it has been shown that damage to the hippocampus can cause the inability to form new memories, but memories that are already established elsewhere are still available.
Putting it another way, memory is a pattern of changes in the way in which neurons interact at the synapses. Think of a memory as an impressionistic painting. Such a painting looks like what it looks like (at sufficient distance) because of the distribution of dabs of paint that the artist has placed on the canvas. Neural tissue that is not holding a memory is like the blank canvass; All the neurons have the default level of sensitivity. Just as the painter ads dots of paint to the canvas to make something that looks like a water lily, the neurons become sensitive at a different level than the default. The sum effect of the points of paint is a water lily, and the sum effect of the neural sensitivities is the memory.
This does happen at the synapse, so we need to understand what a synapse is and how it works.
A synapse is a gap between two neurons ... the pre-synaptic neuron and the post-synaptic neuron. The two neurons communicate in the following matter:
1) The pre-synaptic neuron reaches an electrical state that causes that particular neuron to "fire."
2) The pre-synaptic neuron "fires" by releasing special molecules ... usually peptides .... called neurotransmitters into the synapse.
3) The neurotransmitters attach to special molecules on the post-synaptic neuron. These molecules make up what are called "receptors" and the attachment of neurotransmitters to them cause the molecules to change their 'behavior."
4) The changes in behavior of the molecules at the receptor site cause changes in the post-synaptic neuron, which in turn results in that neuron firing, or being inhibited from firing.
Memory is thought to consist mainly of changes in how sensitive the receptor sites are. A neuron can become less sensitive to the firing of the pre-synaptic neuron if the receptor site molecules are altered in one way, and they can become more sensitive to the firing of the pre-synaptic neuron if the receptor site molecules are altered in a different way. The alteration it probably often caused by the firing itself. These alterations are changes in synaptic plasticity.
Now, back to a slightly more technical description of the process:
Two of the most important manifestations of synaptic plasticity are long-term potentiation (LTP) and long-term depression (LTD). By long-term, it is meant that the change sticks for a while (as opposed to most changes at the synapse which last milliseconds or seconds). What is being potentiated or depressed? The liklihood of the neuron firing, or the effects of it's firing.
That this happens is understood because it is possible to measure whether or not LTP or LTD has happened in a neuron, so the application of memory-causing effects can be followed up with measurement of LTP and LTD. Since one or the other occurs after a memory should be present, the link is likely to be important.
However, as pointed out in this paper, there is "surprisingly little direct evidence that the mechanisms [thought to be] responsible for either LTP or LTD are responsible for learning and memory." The present study looks specifically at LTD and the possible link to memory, using visual recognition memory as the focal system. Previous research had shown that in a particular region of the hippocampus evidence of LTD occurred in association with memory processing. The question remained, though, of how this process happens. What is the sub-cellular, molecule-level cause of LTD? In other words, what exactly is happening to the post-synaptic receptor sites?
One of the receptors involved in post-synaptic "listening" between two neurons is the AMPA receptor. It is thought that one of the ways that sensitivity can be reduced (that would be LTD) is that the AMPA receptors become internalized into the neuron, where they are no longer available for contact with molecules in the synapse. This happens when part of the AMPA receptor (the GluR2 subunit, if you must know) interacts with a certain protein (AP2).
So, you have AMPA receptors that help receive a neural signal from the up stream neuron. But if the GluR2 thingies built into the AMPA receptors interacts with AP2, the AMPA receptor gets "internalized" (sucked up into, sort of) the neuron. This causes the sensitivity of the neuron to be reduced. A blob of paint is thus added to the canvas.
If that is all there was, then brains would not be very smart. Every time a neuron fired, the post-synaptic neuron would undergo a change in sensitivity, decreasing that sensitivity, and eventually it would stop receiving signals. Naturally, there are other systems that increase sensitivity of the post synaptic neuron as well. And, most importantly, for each system that increases or decreases sensitivity (and there are probably many) there are a) many ways to do it and b) antagonistic systems in place. It is the complexity of the potential pathways to LTD and LTP that make memory both hard to understand and, presumably, a very powerful information processing, storage, and retrieval system.
In the present study, the researchers discovered that a certain molecule interferes with the interaction between AP2 and the GluR2 thingie. This molecule is a peptide ... a kind of shortened protein ... called pep-Delta-A849-Q853. It prevents AP2 and the GluR2 thingie from interacting, and thus, prevents the AMPA receptor from hiding out. They worked this out by applying this peptide in a number of different contexts, and they discovered that they could use this peptide to stop a mouse from forming visual memories in the hippocampus.
There are a couple of things to note as especially important in this study. First, this study confirms the relationship between neural plasticity (change in what happens at the synapse) and memory. Second, this study confirms that long-term depression (LTD) is important in visual memory. (LTP has been focused on somewhat more as an agent in memory.) Third, this study identifies one of the key steps in how this process works. This would be like discovering that the process of putting paint on the canvas to make the illy involved, say, a paintbrush, or any one of the other tools an artist uses, thus filling out the description of how it all works.
What this study does not address, but that the authors point out (in a press release) as being important is how these memories get from the hippocampus to somewhere else in the brain where we apparently keep them, and how they are stored there.
GRIFFITHS, S., SCOTT, H., GLOVER, C., BIENEMANN, A., GHORBEL, M., UNEY, J., BROWN, M., WARBURTON, E., BASHIR, Z. (2008). Expression of Long-Term Depression Underlies Visual Recognition Memory. Neuron, 58(2), 186-194. DOI: 10.1016/j.neuron.2008.02.022
A welcome flashback to my neuroscience days! Thanks for the summary/explanation. Ben
Nice post! I have a minor quibble with step 2 of your description of synaptic transmission. Most neurotransmitters are small molecules, not proteins. Some neurotransmitters are peptides, or short chains of amino acids. Insulin is the only protein that I know of that is currently classed as a neurotransmitter.
Yes, that is a valid quibble, and it somewhat dates me. The way I learned it, peptides are short proteins. But these days, proteins are "poly peptides" and thus are different.
So let's call it "usually peptides"
A few minor comments:
1) While certainly the hippocampus appears critical in forming memories, i'm not aware of any evidence that memories are ever actually stored IN the hippocampus (indeed, the specific function of the hippocampus is difficult to pin down). There are a few mentions of this in the post, and it strikes me a overly localize-y. The hippocampus can also function as a gateway in the formation of memories, and the behavioral outcome to structural damage would be similar.
2) There is, perhaps, a difference between "visual recognition memory" and "visual memory," to the extent that the memories in question are literally modal or amodal. The phrase visual memory invokes a clearly modal representation of information in the environment, while visual recognition memory is somewhat more vague (though potentially more accurate). The authors seem to stick to visual recognition memory, and I think they're really shooting for the 'recognition' part of things, anyway.
Again, nit-picky, just had to get it off my chest ;)
Brian: While it is a bit of an over simplifcation it is perfectly reasonable to assert that memories "live" in the hippocampus initially, and are then 'moved' to other areas where they become long term memories. So yes, they are stored there, temporarily.
I'm sure your second point is totally valid. I think the main thing they were doing, though, was using a system they had some control over to see what the molecules were doing.
... and where is the ref to the Cell article??
The Cell journal is different from Cell press that publishes Neuron, by almost a factor ten in impact factor.