Sidebar: Society for Neuroscience
NEW MESSENGERS: Caltech’s Erin Schuman and colleagues discovered that one form of nitric oxide is important to long-term potentiation.
The relationship between short-term and long-term memory, as well as many other basic questions about learning and various forms of memories, have mystified psychologists and neuroscientists for decades. But now, new biological tools are helping to confirm long-held hunches about the molecular mechanisms governing memory and learning. Studies involving a menagerie of mutant mice and fruit flies-as well as the marine snail Aplysia and viral vectors-are shedding light on the biochemical mechanisms at work in these basic cognitive processes.
Neuroscience researchers point to a number of factors that account for this outpouring of information. "Our ability to use the expertise of molecular biology to ask very specific questions about neuronal function explains the explosion of new understanding we have," remarks Erin Schuman, an assistant professor of biology at the California Institute of Technology. "The budding marriage of molecular biology and traditional animal behavior studies has contributed greatly to that."
FORERUNNER: Long-running studies in the lab of Eric Kandel at Columbia using the marine snail Aplysia, right, spawned a host of investigations on CREB, a ubiquitous neuromolecular switch.
Researchers also hope that new insights will lead to a better understanding of why human memory fails, which may have direct implications for Alzheimer's disease as well as a host of memory and learning disorders in children.
For many years, scientists have been trying to unravel the molecules involved in the circuitry of long-term potentiation (LTP), a process that enhances the strength and efficiency of neural transmission. LTP is important in pathways of the hippocampus involved in storing memories about people, places, and things. Neuroscientists surmise that both the sending and the receiving neurons play a role in LTP, but "there has to be some kind of signal that travels back across the synapse," says Schuman.
This signal is called a retrograde messenger. The ubiquitous gas nitric oxide (NO)-important in a laundry list of physiological processes-has long been considered the likely candidate for this messenger.
"A couple of years ago, Eric Kandel and his colleagues tested a knockout mouse without the gene for the neuronal form of nitric oxide," relates Schuman. "But they still detected LTP. That was a problem for the whole hypothesis because everyone had assumed it was that form that was the retrograde messenger." Later Kandel and neuroscientist Solomon Snyder from Johns Hopkins University discovered another molecular player in LTP, the enzyme endothelial nitric oxide synthase (eNOS).
"That's where we come in," says Schuman. A group of Caltech neuroscientists led by Schuman recently demonstrated that endothelial NO is crucial for LTP (D.B. Kantor et al., Science, 274:1744-8, 1996). Using recombinant adenovirus vectors made in the lab of coauthor Norman Davidson, a Caltech professor, emeritus, of biology, the investigators inserted the modified virus into sections of rat hippocampus. They used viral vectors that encoded for a truncated form of eNOS or an eNOS that was fused to a transmembrane protein. Both of these vectors showed that membrane-targeted eNOS is required for LTP.
"Combining information from transgenic mice studies and the recombinant viral vector approach has been very useful," concludes Schuman. Although she and her colleagues have a much better understanding of how NO mediates neurotransmission, "we're only starting to probe what this all means for how memory works."
MUTANT MICE: Transgenic and other forms of aberrant mice help Richard Thompson’s lab at USC sort out how memories are stored during associative learning.
Of the three types of mutants Thompson works with, he says that fast-learning mice within a group of PKC-g (protein kinase Cg) knockout mice developed by Susumu Tonegawa's laboratory at the Massachusetts Institute of Technology (MIT) were the most intriguing. When Thompson's group was training the PKC-g mice to blink their eyes when they heard a tone, some mice began to respond much faster than their cohorts, most on the second day of training (C. Chen et al., Cell, 83:1233-42, 1995).
| Society for Neuroscience|
11 Dupont Circle, N.W., Suite 500, Washington, D.C. 20006
World Wide Web: http://www.sfn.org
President: Bruce S. McEwen
Executive director: Nancy Beang
Journal: Journal of Neuroscience
Carnegie Mellon University Center
Neurosciences on the Internet
WWW Virtual Library
"The other two mouse mutants show marked impairments in learning," notes Thompson. One is the GFAP (glial fibrillary acidic protein) knockout mouse developed by Shigeyoshi Itohara's laboratory at the University of Kyoto. These animals grow up without any of the protein GFAP, which is enlisted during repair to brain tissue.
"One of the theories on the anatomical basis of learning is that as you learn, neurons grow and form new synapses," Thompson explains. "GFAP has long been suspected to be one of the key players in that process."
When the USC group trained the GFAP knockouts, they found that these animals were markedly impaired in associative, Pavlovian learning (K. Shibuki et al., Neuron, 16:587-99, 1996). GFAP is normally found in glial cells, which wrap around the synapses that other neurons make with Purkinje cells.
"This was some of the first evidence that glia, as well as neurons, can play a key role in learning " says Thompson. These findings also support the notion that there are growth processes that occur with learning, because normally GFAP is associated with the growth of brain tissue.
The third type of mutant mouse that Thompson and his team are working with is called the stargazer mouse. It has been selectively bred for its unusual characteristics as opposed to the genetically engineered knockout varieties. "Stargazer tends to hesitate and look up at the sky much of the time," explains Thompson. "The reason for this is that it is actually having petite mal epileptic seizures."
Jeffrey Noebels from Baylor College of Medicine and Xiaoxi Qiao from USC discovered that stargazer has no brain-derived neurotrophic factor (BDNF), another growth factor found in the cerebellum. "This is really important, because this is a major factor in growth and development in the neurosystem," says Thompson.
In recent experiments, Lu Chen, a doctoral student in Thompson's lab, showed that these irreversibly hard-wired mice essentially can't learn at all after Pavlovian conditioning. "They're more impaired than any of those we've looked at so far," notes Thompson.
In a story that has been unfolding for the last 25 years, researchers have now discovered that two different types of memory share a highly conserved molecular switch. One form is important for remembering details about people, places, and things, while the other is crucial to motor and perceptual skills. Different forms of a molecule that regulates genes-cyclic AMP-response element-binding protein (CREB)-have been found to act in concert to switch both categories of memory from a short-term conformation to long-term permanency.
Kandel and his colleagues first found that the cyclic AMP system was important for learning and short-term memory in the early 1970s while studying Pavlovian conditioning in the gill-withdrawal reflex of Aplysia. They next found that Aplysia had long-term memory as well as short-term memory. What's more, the long-term memory had two characteristic features.
"The first is that, just like in higher forms, practice makes perfect," says Kandel. "Repetition is required for long-term memory." Second, Kandel explains, long-term memory differed from short-term memory in that it requires new protein synthesis and the expression of genes. From here, he and his colleagues asked what second-messenger pathways were important for the long-term process and found that it also required cyclic AMP.
Searching for the mechanism by which cyclic AMP turns on protein synthesis led Kandel to examine the role of CREB. Pramod Dash, a neuroscientist working in Kandel's lab, found that blocking the action of CREB selectively blocked long-term, but not short-term, processes. "That was the first demonstration that CREB was critically important for the long-term process," says Kandel.
Later the lab found a second form of CREB, dubbed CREB-2, which acts as a repressor of CREB. "When you remove CREB-2, it is easier to put information in long-term memory," explains Kandel. "If you just test your own experience you realize that the ease with which you put information in long-term memory varies a great deal." One mechanism to account for this, he surmises, is the degree to which this repressor is active.
In short, he notes, memory involves not only the turning on of positive regulators, but also the turning off of the CREB repressor, which restrains transcription. To sum up the CREB-mediated switch from short- to long-term memory, he explains that during short-term memory, pre-existing connections are strengthened, while long-term memory requires gene transcription and protein synthesis to grow new synaptic connections. "If we remember anything about this conversation tomorrow, it's because your head has undergone a real change, all without the use of drugs."
Although his lab has now identified other components of the switch, he says, "this is just the tip of the iceberg. Many other downstream genes still need to be delineated."
Thomas Carew, the John M. Musser Professor of Psychology and Biology at Yale University, explores the transition states between short-term and long-term memory. "We're interested not only in the content of memory, but especially its duration," he offers. "One of the questions that's been around for a long time is: Do you always have to experience short-term memory to lay down long-term memories?"
Carew's work with Aplysia suggests that in some cases the brain can bypass short-term processes and get right to the business of cementing long-term memories (J. Mauelshagen et al., Journal of Neuroscience, 16:7099-108, 1996; N.J. Emptage, T.J. Carew, Science, 262:253-5, 1993).
Although initial behavioral work associated with the CREB molecule started with Kandel's studies in Aplysia, more recent work with mice and fruit flies has begun to point to the universality of CREB's function as a molecular switch. Studies conducted by geneticists Seymour Benzer from Caltech, Chip Quinn from MIT, Lonnie Levin from Cornell University, Ron Davis from Baylor, and Marge Livingston at Harvard Medical School on Drosophila mutants showed that these learning-impaired fruit flies had defects in processes involving cyclic AMP.
|Photo: Marlena Emmons|
ON THE FLY: Jerry Yin and colleagues at Cold Spring Harbor study how gene activators and blockers affect long-term memory in mutant strains of Drosophila.
PARALLEL EVENTS: Using genetic techniques, UC-Berkeley’s Corey Goodman has found that synapse growth in fruit flies involves concurrent molecular signaling.
The simultaneous events at the Drosophila neuromuscular synapse, explains Goodman, are (1) the local growth of synaptic terminals caused by a down-regulation of Fasciclin-II, a cell-adhesion molecule; and (2) a signal up to the nucleus-via CREB-to turn on gene expression to make substances needed to strengthen the synapses (for a review, see K.C. Martin, E.R. Kandel, Neuron, 17:567-70, 1996).
Some novel knockout mice are showing that CREB plays a primary role in mammalian memory and learning. For the last several years, for example, Kandel's group at Columbia and Tonegawa's at MIT have been independently using knockout mice to study the relationship of LTP to memory storage. "Our labs showed that if you knocked out various kinases, you could interfere with LTP, and that interfered with spatial memory in the mice," says Kandel.
In other mouse studies, Cold Spring Harbor neuroscientist Alcino Silva published data describing the behavior and electrophysiology of a mutant mouse that lacks two forms of mammalian CREB (R. Bourtchuladze et al., Cell, 79:59-69, 1994). "We found in two different behavioral tasks that these mice have problems with learning and memory." The animals had normal short-term memory, but a selective defect for long-term memory. The electrophysiological studies also showed that the late-phase of LTP was compromised.
LASTING MEMORIES: "We’re interested not only in the content of memory, but especially its duration," notes Thomas Carew of Yale.
These and other studies, according to Kandel, "suggest the interesting possibility that in both vertebrates and invertebrates the switch from short-term to long-term memory may involve activation of CREB by cyclic AMP."
One of the latest chapters in the molecular basis of memory and learning unfolded in a series of five papers published in Cell and Science late last year (J.Z. Tsien et al., Cell, 87:1317-26, 1996; J.Z. Tsien et al., Cell, 87:1327-38, 1996; T.J. McHugh et al., Cell, 87:1339-49, 1996; A. Rotenberg et al., Cell, 87:1351-61, 1996; and M. Mayford et al., Science, 274:1678-83, 1996). In independent studies using newly developed hippocampus-specific-gene knockout mice and regulated expression of genes, Kandel and Mark Mayford at Columbia, along with Robert Muller at the State University of New York and Tonegawa and Matthew Wilson at MIT, are now looking at the role of LTP and spatial memory function.
Using different approaches to interfere with LTP, both groups found that the spatial memory of the mice-via the place cells of the hippocampus-was compromised. Researchers have long suspected that LTP is the process that maintains the circuitry for internal spatial maps. "[These findings] make it seem more probable that an LTP-like mechanism is important for this type of memory storage," states Kandel.
"I'm really excited about the current state of the field," remarks Carew. "And it's the CREB story begun in Kandel's lab that's pulling it all together. It seems that in fruit flies and mammals, and in Aplysia, this one molecule comes up time and again as an important molecular switch."