Neuroscientist show how stress reshapes fear memories via the brain’s endocannabinoid system

Stress can shape how we form and recall memories, often sharpening our recollection of emotional or threatening events. However, new research published in Cell has shown that acute stress can disrupt this specificity, causing generalized memories that blur the boundaries between safe and dangerous situations. Neuroscientists discovered that stress alters the way memories are encoded in the brain, involving a larger-than-usual group of neurons, and identified a mechanism tied to the brain’s endocannabinoid system that could potentially be targeted for therapeutic interventions.

Our ability to form precise memories allows us to navigate the world safely and efficiently. For instance, animals that narrowly escape a predator in a specific location benefit from remembering the threat and avoiding that location in the future. However, if the memory becomes overly generalized—leading the animal to avoid all areas that remotely resemble the original location—it could hinder necessary behaviors like foraging. Misinterpreting safe environments as threatening is a hallmark of disorders like post-traumatic stress disorder (PTSD) and generalized anxiety disorder.

“My lab has a long-standing interest in how the brain encodes, stores and retrieves memories. We, along with several other groups around the world, have shown that in mice (at least, but we suspect much more broadly across species), memories are stored in the brain in a small population of brain cells (or neurons) known as an engram ensemble,” said study author Sheena Josselyn, a senior scientist at the Hospital for Sick Children in Toronto and a professor in psychology and physiology at the University of Toronto.

“These neurons in an engram ensemble are active at the time of the event and reactivated when the memory for that event is retrieved. Each memory has its own specific corresponding engram ensemble, allowing us to remember many different things. Previously, we noted that the size (that is the number of neurons) in an engram ensemble is relatively constant, despite memories having differing strengths and content. It is well-known that stress impacts memory, especially threat memories, but precisely how was unknown.”

The researchers conducted their study by using mice to model the effects of acute stress on memory formation and generalization. They designed experiments to test whether stress disrupts memory specificity and explored the underlying neuronal and molecular mechanisms involved in this process.

To begin, they exposed the mice to a memory training task involving auditory cues. In this task, one sound was paired with a mild foot shock, making it a threatening stimulus, while another sound had no consequence, serving as a safe stimulus. This setup allowed the researchers to measure how well the mice could distinguish between the two sounds. After the training, the mice were tested in a new environment to see if they would respond differently to the threatening and safe stimuli.

Some mice were subjected to acute stress before the memory training. This stress was induced either through restraint—a situation where the mice were confined to a small tube—or by injecting them with corticosterone, a hormone released during stress. Control mice were not exposed to these stressors. This allowed the researchers to compare memory formation and retrieval between stressed and unstressed groups.

The researchers observed a striking difference in behavior. Unstressed mice displayed specific memory recall, reacting defensively (freezing) only to the sound associated with the shock. In contrast, stressed mice showed generalized defensive reactions to both the threatening and safe sounds. This finding indicated that acute stress impaired the ability to form specific memories, leading to overgeneralization.

To delve deeper into the mechanisms, the researchers examined the neurons responsible for encoding memories. Using advanced imaging techniques, the researchers visualized the engram ensembles in the amygdala, a brain region critical for processing fear and emotions.

They found that the engram ensembles formed in stressed mice were larger than those in unstressed mice. Normally, memory formation is a highly selective process, involving a relatively small number of neurons. This sparse encoding allows for precise recall. However, stress disrupted this balance, leading to the inclusion of more neurons in the memory trace. This increase in the size of the engram ensemble corresponded to the generalized memory responses observed in stressed mice.

“We found that stress right before a threatening training event not only changed the quality of the memory for the threatening event, but also changed the size of the engram ensemble supporting this memory,” Josselyn told PsyPost. “Typically, a threat memory is remembered very specifically. We show fearful responses to the exact stimulus and little to other stimuli. However, when mice were stressed right before the threatening training event, the mice showed fearful responses to many, many stimuli. We call this threat generalization and it is a feature of many human disorders, including PTSD. We also found that the size of the engram supporting this memory was much larger than normal.”

To understand why stress caused these larger engram ensembles, the researchers turned their attention to inhibitory neurons in the amygdala, particularly those known as parvalbumin-positive (PV+) neurons. These neurons play a crucial role in maintaining the sparsity of engram ensembles by selectively excluding less active neurons from being part of the memory trace. The researchers discovered that stress impaired the activity of these inhibitory neurons, allowing more excitatory neurons to be recruited into the memory trace.

This disruption was traced to the brain’s endocannabinoid system, which is involved in regulating stress and emotional responses. Stress caused an increase in endocannabinoids, particularly anandamide, in the amygdala. These endocannabinoids acted on cannabinoid receptors located on PV+ neurons, reducing their ability to inhibit the recruitment of neurons into the engram ensemble. This mechanism explained why stress led to larger, less specific memory traces.

The researchers further tested the role of endocannabinoids by manipulating their levels pharmacologically. When stressed mice were treated with drugs that blocked the production or effects of endocannabinoids, their memory specificity was restored. Similarly, administering drugs that inhibited corticosterone, the stress hormone, also prevented the formation of generalized memories. These interventions confirmed the causal role of endocannabinoids and corticosterone in driving stress-induced memory generalization.

In another set of experiments, the researchers explored the implications of these findings for potential therapeutic applications. They used genetically engineered mice to selectively reduce cannabinoid receptor activity on PV+ neurons in the amygdala. This manipulation restored memory specificity in stressed mice, further validating the role of the endocannabinoid system in mediating stress effects on memory.

The researchers also discovered that the changes induced by stress were specific to threatening memories. The stressed mice’s responses to safe stimuli were generalized, but their overall learning and memory capabilities were not impaired. This specificity suggests that stress selectively affects the mechanisms underlying threat memory formation.

“The key takeaway from this study is that stress can change how our brains encode threat memories,” Josselyn explained. “Adding stress to the mix engages many more systems that can change the quality and specificity of any memory formed. These results may help inform the development of future treatment strategies for disorders such as PTSD.”

Although the findings provide evidence of stress-induced memory generalization and the neural mechanisms involved, the study has some limitations. It focused exclusively on aversive or threatening memories, leaving unanswered questions about whether stress similarly affects non-aversive or rewarding memories. Furthermore, translating these results to humans will require additional research. Mice serve as a useful model because their brains share many structural and functional similarities with human brains, particularly in regions like the amygdala that are involved in memory and emotional processing. Using animal models like mice allows scientists to control variables and explore mechanisms at a level of detail that is not feasible in human studies. However, there are inherent challenges in applying findings from one species to another.

“Our research used mice,” Josselyn said. “We tested a threat memory in mice as a proxy for threat memory generalization in humans. Of course, there can be caveats when going across this species divide. That being said, the only way to develop new effective treatments for a myriad of brain disorders is to study how the brain works. We believe that these types of studies are important in that grand endeavour.”

“Our goal is to understand how memories are encoded, stored and used in the brain. Memory is so important to our everyday lives. Our memories are who we are and help us navigate through the world. Disorders of memory exact a huge toll on both those afflicted, their families, communities and our society overall. Our goal is to help understand the neurobiological basis of memory to help fix memory when this process goes awry.”

The study, “Stress disrupts engram ensembles in lateral amygdala to generalize threat memory in mice,” was authored by Sylvie L. Lesuis, Sungmo Park, Annelies Hoorn, Paul W. Frankland, Matthew N. Hill, and Sheena A. Josselyn.