In a mouse model, scientists were able to identify select neural circuitry involved in a novel experience and turn on that same set of neurons during another experience, thereby artificially evoking the memory of things past. The result — a hybrid memory of the two experiences — is proving critically important in understanding the neurobiology of memory.
The animal studies may be similar to the neural footprints of human imagination; drawing on a memory of a past event during a new experience. If a mouse imagination can be compared to a human experience, “new memories are not formed de novo,” the investigators suggest. “Rather, coding of new learned information depends on a pre-existing circuit activity,” Mark R. Mayford, PhD, an associate professor in the department of cell biology at the Scripps Research Institute in La Jolla, CA, and his colleagues wrote in a March 23 paper that appears in Science.
“We were looking for a way to tap into the neurons that were changing with any given behavior,” explained Dr. Mayford. “Our goal was to have a technique that allows us to see the cells that are active in coding memory and labeling them genetically. We were able to manipulate an ensemble of neurons. This was not something we expected would work.”
The California scientists think that this may be a model for the way people build memories and incorporate them into existing ones and how these memories can change over time with new experiences.
Before they began asking questions about memory, they wanted to know how the brain represents the world. When you see a photograph of a landscape, for example, how do you recognize it as a place with which you may be familiar? They decided to design a way to put an analogous thought in an animal's head and record from it to see how the brain organizes such a thought.
They created a transgenic mouse that expresses a receptor in neurons. It is a G coupled protein receptor called hM3Dq that only gets expressed in cells in which the transcription factor c-Fos is activated. In other words, the receptor gene is under the control of the c-Fos promoter and when there is no activated c-Fos, no promoter is turned on and no receptor activation takes place.
When the mice are exposed to a particular situation, neurons get activated in specific memory circuits, c-Fos is activated, and the new receptor is expressed. They can then use an antibiotic to turn off the transgene and prevent further expression. Or, they can use an agonist called CNO (a clozapine derivative) that stimulates firing of the cells expressing the receptor.
In a sense, explained Neurology Today Associate Editor Kenneth L. Tyler, MD, Reuler-Lewin Family Professor and chair of the department of neurology at the University of Colorado-Denver School of Medicine, the scientists were able to recreate the original pattern of neuronal firing by using a drug to stimulate the neurons originally activated.
In the actual experiment, they put an animal in a new environment — they called Box A — and the experience would activate c-Fos. They coupled the expression of c-Fos with a specific receptor called DREADD, which serves as a marker of the cells that were activated. In other words, it tagged the memory or at least the neural circuitry involved in that memory.
A day later, the animal was put in another box — Box B — and delivered several bursts of electric shock that generally causes a conditioned fear response. But before the shock was administered, they used CNO that binds to the DREADD receptor and makes the cells fire. Put another way, they induced the memory of Box A, and in theory it looked as if the animal imagined what it was like in Box A while in Box B. The conditioned fear response was different than it would have been had the animals just been placed in Box B and exposed to the fear conditioning experiment.
“The model is a bit complex but they were trying to show that the animal really did behave (fear conditioning) like it was in Box A even in the new setting of Box B,” said Dr. Tyler.
They proved their finding in several different experiments. They used fluorescent markers to identify the new neural pattern of Box B and they were able to show that the additional neural pattern of Box A was also expressed at the same time. In the above experiment, they put the animal in Box B after the memories of Box A were triggered with CNO. In Box B, the animals wandered around for a bit before mild shock was delivered to their paws. This classic fear conditioning experiment was tied to the memories of both Box A and Box B. The next day, the scientists tested memory performance in the absence and presence of CNO.
The CNO-induced memory was stored and retrieved as a milder hybrid fear of both boxes.
“When you fire the Box A neurons at the same time the animals are being shocked in Box B, a hybrid synthetic memory is formed with the direct manipulation of a pattern of neurons,” explained Dr. Mayford. “The surprise was that the animals learned to be afraid of being in Box B and with the memory of Box A,” said Dr. Mayford.
The animals showed elements of both the CNO-induced artificial memory and the natural sensory cues from their experience in Box B. The artificially activated memory retained the spatial character of the neural ensemble, he added.
“Retrieval of a memory representation likely involves the reactivation of some neurons that were active during the initial learning,” the investigators wrote in the paper. “Our results imply a strong spatial component in encoding in this form of learning and support the idea that the internal dynamics of the brain at the time of learning contribute to memory encoding.”
Dr. Mayford said that the investigators are now exploring what neurons are responsible for recalling a memory.
EXPERTS WEIGH IN
Richard G. M. Morris, PhD, a professor of neuroscience at the Centre for Cognitive and Neural Systems at the University of Edinburgh, said that this study represents “a major step forward.” In a perspective piece in the same issue of Science, Dr. Morris — who developed the water maze that is standard research fare in rodents — and Tomonori Takeuchi, PhD, said “such a study could provide insights into the subtle interaction in cognition that occurs between representations of the physical world and our internal thoughts.”
“This is the closest to an experience that a mouse might have when imagining that it is somewhere other than where it is,” they said.
In a telephone interview, Dr. Morris said, the investigators paved the way for scientists to tap into internal memories to understand very basic and complex mechanisms that drive memory. He added that this technique may provide a way to explore the impact of memories on the laying down of new experiences in humans.
The advantage of such a technique, he said, is that scientists can now genetically activate specific spatial patterns of neurons. “We can ask fundamental questions about neural representations in recalling memories that we just could not do before now.”
(Dr. Morris's early work in developing the water maze led to the understanding of many aspects of memory and spatial orientation. His group also discovered NMDA's role in learning and memory and set the stage for a generation of studies with genetically engineered mouse models to explore memory.)
“There is so much we don't know and this study shows that we can engineer a memory and later try to express it,” Dr. Morris said. “Can a mouse imagine? Well, not the William Blake kind of imagination, but we now know that animals can be in one place and imagine they are partly there and partly someplace else.”
“In my mind, this takes us an additional step closer to understanding the mind of a mouse. So far, it doesn't tell us about human memory. If we imagine activation of a memory as a subset of cells firing, they re-created the firing of that subset in a different place. It is very clever.”
Still, scientists have a long way to go. Ultimately, they need a tool that allows them to record from thousands of cells at the same time. The Morris lab is now taking on the question of how the brain responds to a traumatic memory — a 9/11 experience, for instance — and why memories of trivial things that happen around that event take hold and last forever but may change over time as well.
Are animal models predictive of human behavior? No one is really certain that they are and that “is why we need to develop paradigms that could have more relevance to humans,” he said. “Animal work is not in a world of its own but is somehow related to how we think.”
Dr. Tyler said the approach is “incredibly clever.” The good news, he added, “is that you get the same neurons — the spatial context — but the bad news is that you can't recreate any temporal context or sequential pattern of activation. The drug turns them on all together at same time.”
Still, “this alone is a pretty cool advance. Other methods of stimulating neurons, like electrical stimulation, would get all the neurons in a particular area. This singles out only those activated in the original context.”