Essays on Science and SocietyEPPENDORF WINNER

Switching Memories ON and OFF

See allHide authors and affiliations

Science  07 Nov 2008:
Vol. 322, Issue 5903, pp. 874-875
DOI: 10.1126/science.1167228

Ireneo Funes, the fictional main character in Jorge Luis Borges' short story “Funes el Memorioso,” could remember in vivid detail every day of his life after he was thrown from a wild horse at a ranch in Fray Bentos, Uruguay. He had acquired a prodigious ability to store new information without any practice. Unlike Funes [and real “autistic savants” (1)], who could store information with a glance, most people learn new things only after many attempts.

Psychologists have identified two corresponding processes: short-term memory, which lasts from seconds to minutes; and long-term memory, which lasts for days, months, or even a lifetime (2, 3). It is now well accepted that making long-lasting memories is dependent on the ability of brain cells (neurons) to synthesize new proteins (2-7). Indeed, animals treated with a drug that blocks the production of new proteins cannot form long-term memories, yet their short-term memory is preserved (8). But how are memories stored? It is hypothesized that information is stored in the brain as changes in strength of the connections (synapses) between neurons (9, 10). Such changes in synaptic strength are observed when neuronal activity is recorded in the brain with microelectrodes: Relatively weak or infrequent stimulation elicits a short-lasting effect [early long-term potentiation (E-LTP)], whereas stronger or repeated stimulation elicits a sustained effect [late long-term potentiation (L-LTP)], lasting many hours instead of minutes. Similar to long-term memories, long-lasting changes in synaptic strength (L-LTP) are prevented by blocking protein synthesis (2, 5-7, 11, 12).

If making new proteins is the rate-limiting step required to store new long-lasting memories, how is this process turned on? If one were able to identify the triggering mechanism and switch it on, then stimulation normally eliciting short-lasting changes should evoke long-lasting ones. Could an increase in the ability to make new proteins explain extraordinarily long-lasting memories?

Our work on the translation initiation factor eIF2α suggests that this could be the case (13): eIF2α regulates two fundamental processes that are essential for the generation of long-lasting memories. Chemical modification (phosphorylation) enhances the activity of eIF2α and suppresses general protein synthesis (14, 15). Paradoxically, it also selectively stimulates the synthesis of a specific protein, ATF4 (CREB2) (16), a well-known memory repressor that blocks the new expression of genes needed for memories (2, 17, 18). In agreement with our idea, the activity of eIF2α is reduced in neurons exposed to L-LTP-inducing repeated stimulation, both increasing new protein synthesis and decreasing levels of the repressor (13, 19).

Eppendorf and Science are pleased to present the prize-winning essay by Mauro Costa-Mattioli, the 2008 winner of the Eppendorf and Science Prize for Neurobiology.

We predicted that if we reduced eIF2α activity (phosphorylation), animals would be able to generate long-term memories from stimuli that normally generate only a short-term change. Thus, we studied mice in which the activity of eIF2α is reduced genetically (eIF2α+/S51A mice) (13). Orienting oneself in a particular space is a function of the hippocampus, a brain region that is crucial especially for spatial learning and memory. The spatial memory of the eIF2α+/S51A mice was assessed in the Morris water maze (20). In this task, mice swimming in a pool of opaque water search for a submerged platform. To this end, they use visual cues that are placed on the wall of the testing room to remember the location of the hidden platform. Remarkably, we found that after initial training in the pool, eIF2α+/S51A mice found the platform faster, indicating that their spatial memory was enhanced (see the figure, panel A). Moreover, the eIF2α+/S51A mice exhibited enhanced learning and memory in other behavioral tests, such as the ability to learn that an auditory or visual stimulus predicts a foot shock and that a specific taste precedes nausea (13).

Control of eIF2α activity is critical for making long-term memories.

(A) eIF2α+/S51A mice exhibit enhanced long-term spatial memory in the Morris water maze. (B) A weak synaptic stimulation elicits a long-lasting response in hippocampal slices from eIF2α+/S51A mice. (C) Injection of Sal003 into the hippocampus immediately after training blocks long-term spatial memory. Dark syringes refer to either vehicle or Sal003 infusions across groups. (D) The induction of L-LTP is blocked by Sal003. Latency is the time required for the mice to find the submerged platform. fEPSP (field excitatory postsynaptic potential) measures changes in synaptic strength. [Figure reprinted from (13) with permission from Elsevier]

We next asked whether these mice also had a greater ability to generate long-lasting changes in synaptic strength. Indeed, weak synaptic stimulation in hippocampal slices from eIF2α+/S51A mice, which usually induces short-lasting changes in synaptic strength, generated L-LTP (see the figure, panel B). Accordingly, in another study we found that mice lacking the enzyme that activates eIF2α, GCN2, exhibit a similar phenotype (19). These data strongly support the idea that a reduction in the activity of eIF2α enhances LTP and mnemonic processes.

Would an increase in the activity of eIF2α have an opposite effect, that is, prevent long-lasting synaptic changes and memory storage? We tested this idea with Sal003, a drug that stimulates the activity of eIF2α by blocking eIF2α phosphatases (13). The Sal003-induced increase in eIF2α activity not only resulted in the inhibition of general protein synthesis but also increased the expression of ATF4 protein. Remarkably, animals injected with Sal003 in the hippocampus were unable to find the platform in the Morris water maze (see the figure, panel C) and they lost the ability to freeze in response to fear (13), clear indications that their memory was impaired. In addition, Sal003 blocked the long-lasting changes elicited by repeated synaptic activation (see the figure, panel D). The impaired ability to induce a sustained LTP was due to an increase in ATF4 protein levels, since L-LTP evoked in slices from mice lacking ATF4 was resistant to Sal003.

The ability to enhance memory formation by increasing new protein synthesis appears to be a widely conserved mechanism from sea slugs to mammals. Decreasing the activity of ATF4 in mice or ApCREB2 (the ortholog of ATF4) in the sea slug Aplysia lowers the threshold for long-lasting changes and memory (17, 18).

By combining cellular, molecular, neurophysiological, and behavioral methods, our experiments reveal a crucial step in memory processing: The activity of eIF2α “decides” whether a long-term memory is made from an experience. It therefore remains essential to identify the genes regulated by eIF2α and determine if de novo mutations in these genes underlie eidetic and other enhanced forms of memory.

As “Ireneo Funes died in 1889” (21), we shall never know whether eIF2α activity was exceptionally low in his brain; however, it remains a possibility.

References and Notes

Navigate This Article