Despite decades of research, much about how our memory works remains a mystery. How does an evening at a concert become a cherished recollection? How is that memory encoded and where is it stored? Scientists believe that the hippocampus orchestrates the process of turning experiences into permanent records, but that the memories themselves gradually get packed away into the cortex, becoming independent of their hippocampal conductor. How the hippocampus knows where to stow a memory is still a puzzle. In the February 18 Science, researchers led by Bruno Bontempi at the Université de Bordeaux in Talence, France, put forth an intriguing idea, suggesting that at the time of an experience, a set of cortical connections gets “tagged,” marking the synapses as future recipients of the data. In one indication of this, Bontempi and colleagues show that rats need to have cortical activity at the time of an experience in order to form an enduring memory, even though the memory will not be stored in the cortex until a couple of weeks later. The authors found that the tagging depends on activation of glutamate receptors. It also depends on epigenetic modifications in cortical neurons, that is, in how tightly DNA is packed. These findings hint at the mechanisms at work, but the exact nature of the tag remains unknown. Bontempi and colleagues also show that drugs that modify chromatin packing can strengthen a memory trace. This adds to a growing body of research that suggests fiddling with epigenetic mechanisms could be beneficial for treating memory disorders such as Alzheimer’s disease.
In an accompanying editorial, David Sweatt at the University of Alabama at Birmingham noted that the authors “move the field in a new direction with two conceptual advances,” i.e., the tagging of cortical synapses and the involvement of epigenetic mechanisms in memory storage.
Howard Eichenbaum at Boston University, Massachusetts, who was not involved in the work, points out that most previous studies in the field have supported the idea that the hippocampus participates in memory acquisition and directs the early stages of memory consolidation, but the cortex is not active until the late stages (see e.g., ARF related news story; Wittenberg and Tsien, 2002). “This study is different in that it suggests the cortex is important early on,” Eichenbaum said. “It is a new twist on the affair, so we have a lot to investigate and to understand.”
To sniff out mechanisms behind memory formation, first author Edith Lesburguères used a natural behavior of rats. In the wild, the animals learn which foods are tasty and safe by smelling them on the breath of other rats, a form of social learning. The process requires hippocampal activity and produces lifelong memories, making this paradigm a good model for human associative learning (see Galef and Whiskin, 2003). Lesburguères and colleagues fed cumin-flavored food to a “demonstrator” rat, then allowed it to interact with experimental “observer” rats. After just one exposure, observers showed a preference for cumin-flavored food over tastier flavors control rats typically prefer. The strength of the cumin preference provided an estimate of the memory’s robustness. In the first two weeks, the memory occupied the hippocampus, but by 30 days after rats first sniffed the new odor, the learned preference was firmly installed in the cortex, particularly in the orbitofrontal cortex. The researchers found that learning was accompanied by the growth of new dendritic spines in this brain region.
To test the route new memories take from hippocampus to cortex, the authors inactivated different brain regions at different times. When they targeted the animals’ hippocampi with AMPA receptor antagonists or the sodium channel blocker tetrodotoxin, the rats lost memories that were less than two weeks old, but not established ones. This is in keeping with memories being initially acquired in the hippocampus, then slowly transferred to the cortex. Inactivation of the orbitofrontal cortex 30 days after the memory was acquired also temporarily wiped out the cumin preference. In contrast, when the researchers silenced the orbitofrontal cortex one day after cumin exposure, rats chowed down cumin-flavored food, consistent with the memory still residing in the hippocampus.
The scientists tinkered with the process further. They specifically inactivated the orbitofrontal cortex with an AMPA antagonist right before exposing the rats to cumin. These rats had no trouble remembering they liked cumin one week later, when the hippocampus was controlling the memory. After 30 days, however, when the memory should have been uploaded to the cortex, the animals turned up their noses at the flavor. No new dendritic spines had developed in the orbitofrontal cortex. The authors saw a similar effect when they injected the AMPA antagonist through a guide cannula into the orbitofrontal cortex for 12 straight days after the rats got their first whiff of cumin. The researchers allowed cortical activity to resume for the next 18 days, a period when memories are normally transferred to the cortex. When tested at 30 days, however, the rats showed no preference for cumin, suggesting the cortex needs to be active in the process of memory consolidation from the beginning in order to form a permanent memory.
Since orbitofrontal cortex activity at the time of cumin exposure was essential for forming a long-term memory, the authors speculated that cortical-hippocampal connections were initially marked in some way that later allowed the cortical neurons to receive the memory trace from the hippocampus. The theory of synaptic tagging dates back over a decade. It describes molecular events occurring over minutes or hours that mark activated synapses for long-lasting changes (see Frey and Morris, 1998; Barco et al., 2008). The identities of the hypothetical tags are unknown, although some candidates have been proposed, including the somal protein Vesl-1S (see e.g., Okada et al., 2009). In contrast, the new work proposes a tagging mechanism that must persist over weeks and at a systems level, although in theory it could involve the same molecules as the short-term process described by Frey and Morris.
Lesburguères and colleagues further characterized the mechanism, showing that cortical tagging requires NMDA receptor activation in addition to AMPA activity. NMDA receptors are key molecules in synaptic plasticity. The researchers also showed that tagging is specific for a particular memory. They allowed observer rats to sniff cocoa flavor normally, then seven days later temporarily inactivated the orbitofrontal cortex before presenting cumin. When tested at 30 days, the rats had learned to like cocoa, but not cumin. In other words, blocking the tagging of the cumin memory did not interfere with consolidation of the cocoa memory.
Intriguingly, the scientists saw increased acetylation of histone H3 in the orbitofrontal cortex within the first hour after the animals smelled a new flavor. It was a transient phenomenon, being reversed over the next four hours. Since histone acetylation opens up packed chromatin and allows more genes to be transcribed, the finding hints that histone acetylation might act as an initial cortical tag that sets in motion longer-lasting changes. When Lesburguères and colleagues blocked acetylation by interfering with known signaling pathways (such as extracellular signal-regulated kinases and mitogen- and stress-activated protein kinases), the cortical memory did not form and no new spines grew, demonstrating an essential role for histone modification. Conversely, when rats were treated with histone deacetylase inhibitors during the first 12 days of memory consolidation in order to maintain histone acetylation, the cumin taste preference was strengthened.
The indication that memories can be bolstered in this way is encouraging data for scientists interested in treating memory disorders. A number of studies have already fingered histone modification as a way to reinforce learning (see ARF related news story on Fischer et al., 2007; ARF related news story on Guan et al., 2009; and ARF related news story on Peleg et al., 2010). Methylation, another form of chromatin modification, also modifies memory formation and might make a therapeutic target as well (see ARF related news story on Gupta et al., 2010).
If histone acetylation is part and parcel of the tagging process, then it raises a conundrum. As Sweatt notes in his commentary, epigenetic modifications, such as histone acetylation, generally lead to cellwide transcriptional changes, while cortical tags are assumed to be synapse-specific. How tagging and acetylation interact remains to be determined.
Another unanswered question, Eichenbaum told ARF, is whether the synapses that get tagged are the same ones that were activated when the rats first smelled the novel food. If so, it would reveal a great deal about how learning works and how memories are stored, in effect, suggesting that memories re-engage the same cortical network that had the original thought. Answering this question would be technically challenging, Eichenbaum said, but could conceivably be done using molecular labels that get switched on in activated synapses. Eichenbaum said the challenge would be to find labels that would last over several weeks, allowing a comparison of initially activated synapses with those that store the mature memory.
In future studies, Eichenbaum suggested, it would be fascinating to see if this same mechanism occurs in other learning paradigms, for example, in fear conditioning. For their part, Bontempi’s group proposes performing genomewide analyses of gene transcription while promoting histone acetylation at various time points during memory consolidation. This should help pin down some of the genes stimulated by acetylation, and might collar other molecules involved in the tagging process.—Madolyn Bowman Rogers
- The Essence of Time—Memory Studies Tackle Fourth Dimension
- Memories—Forgotten, But Not Gone?
- It’s an HDAC2 Wrap— Memory-suppressing DNA Modifier Identified
- Histone Acetylation: Epigenetic Achilles’ Heel of Memory in Aging
- Histone Methyls Solidify Memories
- Wittenberg GM, Tsien JZ. An emerging molecular and cellular framework for memory processing by the hippocampus. Trends Neurosci. 2002 Oct;25(10):501-5. PubMed.
- Galef BG, Whiskin EE. Socially transmitted food preferences can be used to study long-term memory in rats. Learn Behav. 2003 May;31(2):160-4. PubMed.
- Frey U, Morris RG. Synaptic tagging: implications for late maintenance of hippocampal long-term potentiation. Trends Neurosci. 1998 May;21(5):181-8. PubMed.
- Barco A, Lopez de Armentia M, Alarcon JM. Synapse-specific stabilization of plasticity processes: the synaptic tagging and capture hypothesis revisited 10 years later. Neurosci Biobehav Rev. 2008;32(4):831-51. PubMed.
- Okada D, Ozawa F, Inokuchi K. .
- Fischer A, Sananbenesi F, Wang X, Dobbin M, Tsai LH. Recovery of learning and memory is associated with chromatin remodelling. Nature. 2007 May 10;447(7141):178-82. PubMed.
- Guan JS, Haggarty SJ, Giacometti E, Dannenberg JH, Joseph N, Gao J, Nieland TJ, Zhou Y, Wang X, Mazitschek R, Bradner JE, DePinho RA, Jaenisch R, Tsai LH. HDAC2 negatively regulates memory formation and synaptic plasticity. Nature. 2009 May 7;459(7243):55-60. PubMed.
- Peleg S, Sananbenesi F, Zovoilis A, Burkhardt S, Bahari-Javan S, Agis-Balboa RC, Cota P, Wittnam JL, Gogol-Doering A, Opitz L, Salinas-Riester G, Dettenhofer M, Kang H, Farinelli L, Chen W, Fischer A. Altered histone acetylation is associated with age-dependent memory impairment in mice. Science. 2010 May 7;328(5979):753-6. PubMed.
- Gupta S, Kim SY, Artis S, Molfese DL, Schumacher A, Sweatt JD, Paylor RE, Lubin FD. Histone methylation regulates memory formation. J Neurosci. 2010 Mar 10;30(10):3589-99. PubMed.
- Lesburguères E, Gobbo OL, Alaux-Cantin S, Hambucken A, Trifilieff P, Bontempi B. Early tagging of cortical networks is required for the formation of enduring associative memory. Science. 2011 Feb 18;331(6019):924-8. PubMed.
- Sweatt JD. Neuroscience. Creating stable memories. Science. 2011 Feb 18;331(6019):869-70. PubMed.