Erasing memories

Erasing memories has long been a popular plot device for Hollywood scriptwriters. In the 2004 film Eternal Sunshine of the Spotless Mind, for example, Jim Carrey and Kate Winslet play a separated couple who undergo a radical treatment in order to abolish every trace of the relationship from their brains.

The ability to erase memories is no longer confined to the realms of science fiction. In the current issue of Neuron, researchers from the Medical College of Georgia, in collaboration with others from the Shanghai Institute of Brain Functional Genomics, report that they have rapidly erased new and old memories from the brains of mice in a controlled manner.

Using a combination of protein engineering and organic chemical synthesis, the researchers manipulated levels of a key enzyme called alpha calcium/ calmodulin-dependent protein kinase II (αCaMKII) at specified times in the mouse brain, and found that transiently increasing the level of the enzyme just before recall of a memory led to its selective erasure. 

CaMKII is found only in the brain, and it constitutes more than 1% of the proteins in that organ. It is highly enriched in the postsynaptic density (PSD), a dense network of more than 1,000 proteins, organized into 12 massive and precisely assembled multi-protein signalling pathways, located immediately underneath the membrane of neurons which use glutamate as a neurotransmitter. CaMKII is a key component of the PSD. It is attached to the NMDA receptor, and is activated by calcium currents entering through it. Upon activation, CaMKII dissociates from the receptor, then translocates within the PSD to regulate the activity of dozens of proteins in multiple signalling pathways. It does so by modifying their chemical structure: it catalyses the addition of a phosphate group to specific amino residues, a reaction which is called phosphorylation. CaMKII binds ATP, which contains 3 phosphate groups, and transfers one of them to each site on its target proteins.

In the new study, Joe Tsien and his colleagues created a strain of transgenic mice with twice the normal amount of αCaMKII in the forebrain. The enzyme had been engineered to contain a hidden cavity within the ATP-binding pocket. The cavity is accessible only to a specially designed inhibitor, a small organic molecule which is too large to fit into the ATP-binding site of the other, unmodified forms of CaMKII present in the brain. In these mice, the enzyme could be completely but reversibly suppressed by a specially designed inhibitor molecule which fits the pocket perfectly but does not bind to other forms of the enzyme. Because of its pharmacological properties, this inhibitor binds specifically to αCaMKII within 10 minutes, and then unbinds after 40 minutes, so that the enzyme is reactivated.

Experimental mice with increased levels of this modified enzyme were trained to perform 3 different memory tests and their performance was compared to control mice. In the object recognition test, the animals were placed in a large box and allowed to explore it. Two objects were then placed into the box and the time the mice spent exploring each was recorded. When the mice were removed from the box and returned one hour later, one of the objects had been replaced with another. The time spent exploring each object was recorded again, with a preference for the object encountered previously being taken as evidence for recognition of it.

In the fear conditioning task, the mice were placed in a compartmentalised chamber and taught to associate a sound with an electric shock delivered in one of the compartments. This involves pairing the sound with the shock and the compartment on several occasions. Later on, the mice should exhibit fear behaviour when presented with those things associated with the shock: if placed in the compartment in which they previously received the shock, or if they hear the sound associated with it, they should freeze if they have learnt the association.

The experimental animals - those expressing elevated levels of αCaMKII in the forebrain - were found to be severely impaired in all these tests. The impairments could have been due to the disruption any of the three stages of memory: acquisition, retention or retrieval. The researchers therefore devised an experiment to distinguish between each of these processes. Another group of experimental and control mice were taught the same memory tasks and the former was treated with the αCaMKII inhibitor 15 minutes before the recall tests.

This time, the experimental animals performed normally on all three tasks. They had learned and maintained their memories of the tasks in the presence of increased levels of αCaMKII, but the activity of the enzyme was temporarily suppressed during retrieval. This suggests that the impairments that occured as a result of increased αCaMKII levels in the first set of tests were down to deficits in recall of the memories and not acquisition or storage.

To investigate this possibility further, the researchers conducted the same set of experiments yet again, but this time the transgenics were treated with the αCaMKII inhibitor 15 minutes before and again 10 minutes after memory training, such that learning and retention, but not recall, took place while αCaMKII activity was supressed. Again, the transgenic animals were severely impaired in all three tasks compared to the controls, confirming that increased αCaMKII levels at the time of recall to impaired retrieval of newly formed memories. 

Another group of mice was then trained to perform the same memory tasks and the recall tests were carried out one month later. Those transgenic mice treated with the inhibitor 15 minutes before recall performed normally on the tests: just like the controls, they froze when placed in the electrically rigged compartment of the chamber, or when presented with the sound associated with the shock. However, when the inhibitor was administered 2 days before learning and then continuously for 28 days, so that it was withdrawn 2 days before the recall tests, the transgenic mice did not exhibit fear behaviour. All of these data demonstrate that overexpression of αCaMKII impairs retrieval, but not acquisition or retention, of both new and old memories.

Next, the researchers set out to establish whether the impairments are due to erasure of the memories or to an inability of the mice to gain access to them. The one month fear memory test was used again, but the mice were given two recall trials. In the transgenics, αCaMKII activity was supressed during the second trial. The rationale for this was that, if the impairments are due to recall deficits, the mice should successfully retrieve the memories in the second trial. If, on the other hand, they are due to erasure of the memories, then recall should fail in both trials. The transgenic mice treated with the αCaMKII inhibitor during the second trial still performed very badly in both trials, suggesting that the memories are in fact erased at the time of recall.

One final set of experiments showed that this recall-induced erasure is highly specific to the memory being actively retrieved. These experiments involved tests of sequential memory retrieval. Mice were subjected to the two fear conditioning tests. One month later, they were placed back into the compartment in which they had received an electric shock. As expected, the control animals quickly froze, but the transgenics did not. Then, all the mice were returned to their home cages, and some of the transgenics were injected with the αCaMKII inhibitor. When the animals were put back into the experimental set up and presented with the sound associated with the electric shock, those transgenics treated with the inhibitor froze when they heard it, just like the controls. When the period between learning and recall was extended to two weeks, the performance of the transgenics treated with the inhibitor was still indistinguishable from the controls. 

Further examination revealed that the specific memory erasure occurs very rapidly during recall. In all of the recall tests, the performance of the untreated transgenics with increased αCaMKII levels was high during the first minute of testing, but then decayed over the next 2 or 3 minutes. This suggests that memory recall in the presence of high αCaMKII levels rapidly triggered biochemical and physiological processes which ultimately led to the erasure of the memory being retrieved.

These findings show that both short-term and long-term memories, and memories of different types (object recognition and fear conditioning), depend on common molecular mechanisms, and that αCaMKII is a key player in all of these mechanisms. Last year, another group of researchers reported that memories could be erased by inhibiting another enzyme called protein kinase M zeta. This inhibitor only worked if applied for at least 2 hours after learning, and it had a general effect on all the memories encoded at the same time.

In this case, erasure of memories was inducible, rapid, and highly specific. This would of course be very useful in treating conditions such as post-traumatic stress disorder - what immediately comes to mind is a short-acting αCaMKII inhibitor that can be taken by a patient just before free recall of some traumatic event. Such treatments would pose ethical problems, and in any case are unlikely to be developed for use in humans.

Related:

• Memory lessons from Homer Simpson
• Synapse proteomics & brain evolution
• AMPA receptors & synaptic plasticity
• The neurogenetics of traumatic memories
  

Cao, X. et al (2008). Inducible and selective erasure of memories in the mouse brain via chemical-genetic manipulation. Neuron 60: 353-366 DOI: 10.1016/j.neuron.2008.08.027