Scientists selectively erase memories from mice.
Memory is who we are, the essence of being human, the thread of individuality. Losing memory means losing the narrative of self, as happens in Alzheimer's. Memories can be evoked by associations of all kinds, especially by physical cues such as the famous madeleine. Where are memories stored? How are they stored? Are they material? Theists would refer to a "soul", and leave it at that, but scientists have pushed a bit further.
The going scientific model of memory is that it is a pattern of neural activity stored by enhancing the connections (synapses) between the most active neurons so that future nearby activations trigger re-activation of the same pattern. In other words, memories are stored in a distributed way in the mass of the brain, like holograms, and are reactivated when some approximation of the original activity takes place, such as when tasting a long-forgotten delicacy. Anatomically, there is also a complex short-term/long-term issue, where recent memories are stored in the hippocampus, before being transferred to the neocortex for long-term storage. Another system in the amygdala is both short- and long-term, rapidly storing emotionally laden memories such as fear responses, which are then virtually impossible to erase.
In molecular terms, memory storage involves strengthening the synapses between more-active neurons, essentially encoding past activity patterns by strengthening them and making them easier to activate. The best-known gene involved in this process encodes a transcription activator called CREB, for cyclic AMP response element binding protein. CREB binds to DNA and activates a variety of genes in response to the messenger molecule cAMP, which is generated inside activated neurons (in response to dopamine or serotonin stimulation, for instance), though it is used for many other signals in other cells. CREB appears to activate a bunch of genes that build up synapses near the sites of activation, making permanent what is marked temporarily by other activity-dependent molecules.
The current report in Science shows that if those few cells showing activated CREB during a training event are killed, then the memory is lost forever. The first trick is to use mice that are programmed with a diphtheria toxin receptor gene in a re-arrangable cassette. This cassette can be turned on by a DNA recombinase which is in turn activated by a CREB-inducible promoter sequence. Now they have mice whose memory/CREB-activated neurons specifically express a receptor for diphtheria and can be killed by applying diphtheria toxin, which the researchers do by injecting it directly into their amygdalas (which does not sound so easy, incidentally).
One control for the experiment is to treat other mice with another promoter-DNA recombinase setup not involving the CREB activator which in the end kills approximately the same number of cells in the amygdala (presumably randomly) after injection of diphtheria toxin. The experimenters then trained the mice to fear a sound, tested them for having learned it, (at which point select memory-involved cells in their brains would be expressing both CREB and the experimental diphtheria receptor), then injected the toxin, and assayed for both how many cells were killed, and how the mice now responded.
Where 50% of the mice had learned the response to start with, after the injection only 20% continued to respond- a significant erasure of memory. No difference was seen in the control mice with other neurons killed, or those not treated with toxin at all. The erasure persisted for 15 days, thus appearing to be permanent. The mice could also go on to learn and remember from later training experiences, showing that their memory systems were generally intact and relatively few cells had been deleted. Likewise, the experience did not erase earlier memories formed before the treatment with the CREB-DNA recombinase system that allowed actively learning cells to be deleted.
Fig 3. Bottom- Schematic of amygdala neurons after diphtheria toxin (DT) or saline (PBS). Blue, DNA-labeled neuronal nuclei; pink, neurons activated by memory; and white, ablated neurons. CREB-cre stands for CREB-activated DNA recombinase cre (introduced by vector injection), which recombines and activates the DT receptor gene.
What is the bottom line? That our memories (and thus conscious person-hood) are physically encoded in our brains. Hopefully this is another reason to take better care of them. It may be possible eventually to treat trauma victims in ways that impair their memories, or otherwise use this information in a beneficial way. I could certainly use better memory, or the ability to forget some things, like a certain film!
Notes on animal cruelty.. This type of research can be quite cruel to animals, and just because mice are small does not mean that they do not feel pain much as we do. Indeed we would not use them as models for this type of work if they didn't. So I'll note my hopes that the experimenters used anaesthesia properly at all stages of intervention (as claimed), minimized the aversive stimulus program, and euthanized humanely. In my experience in the lab, euthanization protocols leave a great deal to be desired. And would it be so difficult to train mice on a positive stimulus rather than always relying on the amygdala-fear system? Let's have a little compassion!
