To persist, memories surf molecular waves from thalamus to cortex

Memories are avid travelers: After they form in the hippocampus, they pass through intermediate regions such as the thalamus before finally stabilizing in the cortex. Yet not all memories survive this journey and become consolidated. The ones that persist do so thanks to a cascade of transcriptional changes in a thalamocortical circuit, a mouse study published today in Nature shows.

The work suggests that long-term memory is enabled by the sequential activation of molecular cascades in regions beyond the hippocampus, says study investigator Priya Rajasethupathy, associate professor of neurosciences and behavior at Rockefeller University.

The study also provides insight into what happens long after a memory is formed, says Hidehiko Inagaki, research group leader at the Max Planck Florida Institute for Neuroscience, who was not involved with the work. “People have studied a lot how expression changes overnight when the memory is first consolidated,” he says. “But here in this paper, they look at over many weeks—that’s a time scale that not so many people have studied mechanistically.”

Rajasethupathy and her team focused on the thalamus because previous work showed that connections stemming from this area mediate the maintenance but not the initial learning of a memory. “[The thalamus] is saying that something’s happening at the time of learning that’s assigning value, saying, ‘Hey, I want to be able to remember this in the future,’” she says.

As a mouse learns something new, neurons in the thalamus transition through multiple transcriptional states—each defined by a different expression profile and orchestrated by transcriptional factors that function as molecular timers for memory persistence, the work shows.

These waves of transcription show that memory formation and stabilization is not on or off but rather a sequential and dynamic process, Inagaki says. “Memory maintenance is not just a one-time thing, but it is a very active process which needs to keep happening,” he says. “It’s an active and continuous process.”

R

ajasethupathy and her colleagues created a virtual-reality task that enabled them to examine the differences between memories that are consolidated versus memories that are forgotten. During the task, thirsty mice navigated through a corridor until they reached areas in which they received a drink of water. The mice encountered one reward area frequently and another reward area less frequently.

After five days of habituation and one week of training, the mice showed increased licking whenever they entered either reward area, indicating that they had learned to anticipate a drink of water in those locations, the study shows.

In the 15 to 30 days after the end of the training, however, the mice retained only the frequent-reward memory. Consolidation did not occur when the team used optogenetics to disrupt a circuit from the thalamus to the anterior cingulate cortex. “With this paradigm, you’re able to isolate the maintenance phase of memory,” Rajasethupathy says.

During memory consolidation, neurons in the thalamus turned on genes associated with synaptic plasticity while neurons in the anterior cingulate cortex turned on chromatin remodeling genes. The neurons progressed through distinct phenotypes during the early, intermediate and late stages of learning, a computational analysis that maps a cell’s transitions found.

Disrupting some of the genes revealed that they indeed play a role in memory stabilization. For example, mice without CAMTA1 or TCF4 in the thalamus had impaired recall in the days to weeks after training, suggesting that these factors are involved in the refining stage. Meanwhile, mice without ASH1L, which has a role in histone methylation, in the anterior cingulate cortex had impairments in the weeks to months after training, suggesting that this gene is involved in epigenetically preserving the memory.

This shows that “you kind of have to go through these intermediate states” that allow for a stability-plasticity trade-off, Rajasethupathy says. “You want to be able to promote and demote as experience and time shapes your sense of how important a memory is.”

The work identifies potential molecules that control memory stabilization, Inagaki says, and the next steps are to examine the mechanisms by which they can alter neuronal properties and ultimately affect behavior and memory persistence.