![An illustration of the inner workings of a mouse brain during artificial hibernation. Workers prune the irrelevant dendritic spines from brain cells, leaving only the synapses necessary for long-term memory retention. [Luo-chu Yang]](https://www.genengnews.com/wp-content/uploads/2026/08/low-res-1.jpeg "DO NOT REUSE")
Memories can survive even after the brain temporarily loses more than half of its synaptic connections, according to the results of a study in mice that challenge the long-held view that long-term memories depend on stable individual synapses. The researchers, headed by a team at Okinawa Institute of Science and Technology (OIST), and including teams at the University of Tsukuba, Exploratory Research Center on Life and Living Systems (ExCELLS), and National Institutes of Physiological Sciences, used a mouse model of artificial hibernation to examine structural mechanisms underlying memory retention.
Their findings indicate that memory may be preserved not through individual synapses, but through resilient patterns of neural architecture, including specific clusters of connected synapses that remain protected during widespread hibernation-associated brain remodeling. These preserved structural motifs could act as a “core memory trace,” allowing the brain to rebuild functional networks after major disruptions.
Research lead Kazumasa Tanaka, PhD, head of OIST’s Memory Research Unit, said, “Previously, synaptic strengthening was thought to be key to memory recall, and that stronger synapses with larger dendritic spines were fundamental to long-term memory retention. Here, we show that not every synapse matters, and demonstrate instead the vital importance of engram architecture. The study indicates that small clusters of engram-engram synapses are preserved to enable accurate recall even after hibernation.”
Tanaka is senior and corresponding author of the team’s published paper in Science, titled “Artificial hibernation reveals synaptic engram architecture associated with memory retention.”
Understanding how memories are stored in the brain is one of the central challenges in neuroscience, the authors wrote. For decades, scientists have believed synaptic potentiation—the adaptive strengthening of our brain’s cellular connections—to be the key to memory retention. The newly reported research by Tanaka and colleagues now demonstrates the importance of higher-order synaptic architecture, suggesting that specific clustered patterns of connections between brain cells may be key to retaining long term memory.
“The structural underpinning of memory has been one of the most fundamental topics in neuroscience for decades,” the authors wrote. Much like computer storage, we humans require a physical memory trace to be stored in the brain. This physical trace, known as an engram, is encoded through a dedicated network of brain cells undergoing changes at their synapses, the junctions where they meet. When connecting brain cells repeatedly fire together, their synapses strengthen, increasing neurotransmitter release and triggering structural changes, such as larger dendritic spines, which expand the contact area between the two cells. Conversely, when a particular connection isn’t very active, the synaptic bonds between the cells weaken, and may eventually disappear altogether.
The synapses with larger, more stable dendritic spines have traditionally been seen as key for memory. While we might expect these connections to stay consistent over the course of a memory, recent studies have found that the structures and numbers of cells involved in a particular engram can change over time, without affecting recall. “… recent studies have revealed that synapses can be highly dynamic, with dendritic spines frequently appearing and disappearing, neuronal representations drifting over time, and memories remaining retrievable even after previously strengthened synaptic connections are disrupted,” the team continued. These observations raise a fundamental question, they noted. “How can memories remain stable despite extensive structural remodeling of neuronal networks?”
To investigate, the researchers turned to an artificial model of hibernation, as an experimental model for studying memory stability. In a hibernation state, decreased metabolism enables creatures to survive harsh, wintery conditions with little food, and can also causes brain activity to dramatically reduce.
In 2020, a team led by coauthor Takeshi Sakurai, PhD, at the International Institute for Integrative Sleep Medicine (WPI-IIIS), Tsukuba Institute for Advanced Research (TIAR), University of Tsukuba, induced artificial hibernation for the first time in mice. Their research uncovered the brain circuitry necessary for inducing hibernation, unlocking a new tool for neuroscience research. They later reached out to Tanaka to collaborate. “Our brains are incredibly complex. If hibernation can reduce and simplify brain activity and structure, it could make studying these convoluted systems a bit easier. That’s why I wanted to use artificial hibernation techniques to study memories,” he said.
By imaging mouse brains before, during and after artificial hibernation, the researchers found that more than half of synapses in the hippocampus region of the brain disappeared in hibernation, and neuronal firing rate, a measure of brain activity, reduced by roughly 70%. Synapse elimination didn’t seem to be influenced by dendritic spine size, with synapses involving both large and small spines equally likely to be removed.
“Artificial hibernation caused a profound reduction in neuronal activity and eliminated more than half of hippocampal synapses,” the investigators noted.
The surprise came in behavioral tests after artificial hibernation, which showed that the animals’ memory recall remained the same, or even improved in some instances. “Despite these large-scale structural changes, mice retained previously acquired memories and preserved neuronal representations of experience,” they stated.
![This diagram shows two characteristic patterns found within the study. On the left, we see a multi-synaptic bouton (MSB), where one presynaptic terminal joins to multiple different dendritic spines on different cells. The researchers found that MSBs were more likely to be conserved following hibernation, suggesting their essential role in memory retention. On the right, we see clustered engram patterns. Dendritic spines that are close together but that connect to various axons of neurons in a different section of the brain are active within the same engram. Again, the researchers found that clustered engram synapses were preferentially conserved after artificial hibernation, which suggests their importance in memory retention. [Luo-chu Yang]](https://www.genengnews.com/wp-content/uploads/2026/08/low-res-2-300x208.jpeg)
To investigate why memory retention may have stayed intact or improved, the researchers used a technique called CLEM—correlative light and electron microscopy. “CLEM combines fluorescence microscopy with high-resolution electron microscopy, explained co-author and technician Ai Takahashi. “By labelling samples with fluorescent tags, we can see different parts of the cell or organism in different colors. This can, for example, help us to pinpoint important proteins. Then using electron microscopy, we can zoom in on those same areas, to examine substructures in much higher detail.”
The team’s study is the first to use CLEM to observe engrams, a significant achievement given how small and sparse engram synapses are. “Successfully correlating the light microscopy and electron microscopy datasets to image engrams is a very technically challenging feat,” added Tanaka. “We hope our contributions to developing this method may provide new platforms for studying other important neuroscientific questions in future.”
The team fluorescently labelled the synapses thought to be involved in a particular memory trace and examined these before and after artificial hibernation. They found a significant decrease in synapses. However, certain clusters of synapses seemed to be spared. “Dendritic spines were eliminated regardless of their size, but many reappeared at the same dendritic locations after recovery,” they wrote. “Notably, synaptic connections between neurons encoding the memory were organized into spatially clustered groups.” Tanaka said, “This suggests that for long-term memory, only particular clusters of synapses matter—the rest may be dispensable. Interestingly, dendritic spine size, which has been shown to increase in initial memory encoding, doesn’t seem to play a factor in memory retention.”
Lin added, “We’ve observed this interesting correlation between clusters and memory retention, but not yet proven a causal link. As technology develops, it will be interesting to study these clusters in more detail, to answer these remaining questions.”
Looking forward, the researchers hope to continue their studies on memories, with the aim of understanding the mechanisms by which these core clusters are protected. “A deeper understanding of the mechanism by which hibernation protects the core memory trace and maintains the network integrity would reveal more comprehensive principles of memory,” the investigators concluded. “
Tanaka noted, “We’ve unlocked some insights into the architecture needed for memory retention. But there are many more questions to explore. How does the brain maintain this structure over time? How do different memories interplay? We have so much left to learn.”
The researchers also plan to study other aspects of artificial hibernation. “This work focused on mice, but the same neuronal circuitry for hibernation is well-conserved across many mammals, including humans,” Tanaka pointed out. “Therefore, through artificial hibernation studies, we may be able to discover new insights or applications that can translate to human health and neuroscience.”
