By Tim Stoddard
Joe Tsien still vividly recalls the gut-wrenching moment years ago when his elevator plummeted five stories inside the Tower of Terror, an amusement ride at Universal Studios in Orlando, Fla. “You only need to experience that kind of thing once and you’ll remember it forever,” he says.
Now Tsien is beginning to understand how that frightening moment was etched in his memory. With a team of MED neuroscientists, Tsien, a MED professor of pharmacology and director of the Center for Systems Neurobiology, is shedding light on how the brain forms memories of extraordinary physical experiences such as roller coaster rides, earthquakes, and car accidents. In a major breakthrough for memory research, the team has come up with a code-key for understanding how groups of neurons in mice encode memories.
Researchers have long known that the region of the brain called the hippocampus plays an important role in recording so-called episodic memories of highly stimulating events. They know that sensory information about smell, vision, touch, and sound are funneled from other regions of the brain to the hippocampus, exciting groups of neurons there in unique ways that represent the experience and the subsequent memory. But because of size restrictions, scientists could not monitor these patterns closely enough to visualize the patterns that were being formed: the hippocampus in mice is tiny and no one had devised a way of inserting enough bulky electrodes into adjacent nerve cells to see how different groups of neurons behaved.
In a technical feat, Tsien’s group developed an apparatus with very fine electrodes that can simultaneously record the activity of as many as 260 individual neurons in the mouse hippocampus. Previously, researchers had been able to monitor only about 29 neurons in mice. “Tsien has basically one-upped the current state of the art,” says Howard Eichenbaum, a UNI professor, a CAS psychology professor, and director of the Center for Memory and the Brain. “I think this could be a significant advance in the field.”
Researchers have been able to “listen” to individual neurons for over 50 years, Eichenbaum says, using electrodes that monitor the electrical impulses each cell generates in response to external stimuli. But memories aren’t formed by individual neurons. “Cells in the hippocampus are kind of like a scratchpad that you can write lots of memories onto at different times,” he says. “So it has more to do with their organization perhaps than it does with individual cells sitting there waiting for a particular stimulus to appear.” Researchers need to be able to record whole populations of cells, because “recording from a single cell in many ways is like trying to find out who’s popular by asking one voter at a time. You have to take a wide sample.”
Inspired by his memorable elevator ride, Tsien, who is also an ENG professor of biomedical engineering, set out to do this in his laboratory. He designed a series of elegant experiments that would startle mice in three different ways. At one station, mice received a sudden puff of air on the backs of their necks, which he says mimics the rush of air they might feel when an owl or another predator swoops down from the sky. Mice were also placed in boxes that were suddenly dropped about a foot, like elevators in a short free-fall. In a third scenario, mice felt a brief earthquake simulated by vibrating their cages for a few seconds on a vortex mixer.
During each episode, the team recorded distinct patterns of neural activity in response to each event. They also found that these patterns were composed of many different subgroups of neurons, which Tsien has dubbed neural cliques because of the way they act together in similar ways. “Some cliques are very specific only to the earthquake sensation, but some are also involved in both earthquakes and elevator drops, but not air puffs,” Tsien says. “The activation pattern of these coding units is the physical basis for memory.”
Tsien’s group could see the mouse brains remembering their startling experiences. During each air puff, earthquake, and elevator drop, the electrodes recorded a certain sequence of activity in the hippocampus. But then moments after each episode, when the mice were resting, the researchers also observed flashbacks, or faint electrical patterns similar to those seen during the actual stimulus. “We were quite surprised by the sensitivity of our methods,” Tsien says. “It seems the brains replay the experience, which is consistent with our own experiences. When you just come off a roller coaster ride, say, you can’t help talking about it — somehow, your brain is spontaneously getting you back into that state. We don’t know why or how the mouse brains did it, but we saw it. That’s where this method is so powerful: it not only allows us to decode the actual event, but also to visualize the post-event processing of learned information.”
Until now, researchers have mainly studied memory using behavioral tests, which are often imprecise and indirect. “By understanding the encoding units in the brain and the activation patterns of neural cliques,” Tsien says, “we can read the mouse’s mind reliably and directly without looking at its behaviors.”
Before coming to BU last year, Tsien had spent seven years researching memory in mice at Princeton University, where he made headlines in 1999 by creating a genetically engineered mouse capable of learning tasks in half the normal time. He dubbed the mouse strain Doogie after the precocious television doctor Doogie Howser. What makes the Doogie mice so smart is an extra copy of a gene that forges connections between two stimuli, such as a bell ringing and food. “It’s like a coincidence detector,” Tsien says. “We showed that if you knock out this receptor in the hippocampus, for example, the mice are impaired in memory. We know a lot about this at the molecular level, but the task of memory is being achieved at the network level.”
Tsien’s new technique of monitoring larger populations of neurons makes it possible to study these patterns in greater detail. He’s now curious to see if the same approach can be used to study other types of cognition in other regions of the brain and perhaps in other species of animals. He’s also hoping to repeat the experiments with Doogie mice to better understand how they learn so quickly. “It opens the door to many more experiments,” he says.