Monday, April 01, 2002

The Search for the Memory Switch

Memories Are Made of This: How Memory Works in Humans and Animals

By: Rusiko Bourtchouladze, Ph.D.

The fox in the most exciting brain-science hunt of the last two decades has a dreary name: CREB, the acronym for the even drearier-sounding “cAMP-response element binding protein.” But when it binds to DNA, CREB switches on dozens of genes, which, in turn, create new protein that is literally the stuff of our long-term memories.

No CREB, no protein, no long-term memories—and no mind or self as we know them, even in a fruit fly or mouse. Discovery of the CREB cycle helped to gain neuroscience the first Nobel Prize of the 21st century. It also launched several start-up companies to figure out how to switch CREB on and off: the fabled pill to power-up long-term memory and make whiz kids of us all.

Rusiko Bourtchouladze, born and educated in Georgia, in the old Soviet Union, knows the story of CREB from the inside. In 1992, she joined the memory research team at the Cold Spring Harbor Laboratory, New York; and from 1994 to 2001 worked at Eric Kandel’s Center for Neurobiology and Behavior at Columbia University. She is now Director of Model Systems for Helicon Therapeutics, Inc., a New York company seeking applications of memory research. Her new book, Memories Are Made of This, takes readers through the unfolding story of how advances in many fields have shaped our understanding of the way memory works. The climax—in the book and in memory research itself—is the molecular genetic approach to discovering how we literally synthesize a memory out of new protein in our brains. The following excerpt begins as scientists close in on the answer to that question. It is the story of a startling scientific success that raises, inevitably, anther question: Where do we go from here?


Excerpted from Memories Are Made of This: How Memory Works in Humans and Animals by Rusiko Bourtchouladze.  ©2002 by Rusiko Bourtchouladze. Used by arrangement with Columbia University Press.  All rights reserved. Reprinted with permission.


The stakes in the memory gene venture rose considerably during the 1960s and 1970s when several scientists made an intriguing observation. New proteins, synthesized within neurons during learning, appeared to be essential raw material for long-term memory, just as bricks are used to construct a house. But the significance of this goes beyond the trivial notion that neurons, like other cells, require protein synthesis for survival. If you give animals a drug that blocks this ability to make new proteins, they become incapable of forming long-lasting memories. They can learn, and they can even keep in mind the learnt task for an hour or so. But check their memory five or six hours later, or the next day, and you will be surprised by their stupidity—they have no clue about the task whatsoever.

What happened to memory? Where did it go? The answer is nowhere: the drug simply prevented the conversion of short-term memory into long-term memory. This mechanistic distinction between short-term memory and long-term memory—formation of long-term memory needs new proteins, while formation of short-term memory does not—is apparently universal and holds up for all creatures so far studied in laboratories, including Aplysia, fish, fruit flies, bees, chicks, mice, rats and rabbits. The implications are profound: since genes must be switched to make new proteins, they are, therefore, involved in long-term memory. What are these genes and what triggers them?

Back in Cold Spring Harbor, when we started hunting for genes involved in laying down long-term memories in mice, we had a few tentative hints where to search. Some of the earliest molecular events involved in learning in invertebrates are mediated by a molecular signaling system called the cyclic AMP (cAMP) pathway. For example, learning in Aplysia is disrupted if this signaling system is perturbed. Similarly, learning disabilities in fruit-fly mutants called dunce, rutabaga, amnesiac, Ria and Gsa stem from mutations in genes affecting some earlier molecular steps in the cAMP pathway. Two different animals, same pathway. Quite a hint here.

But how does learning trigger cAMP signaling? If I avoid some of the tongue-twisting names of the molecules of the cAMP pathway, a simplified chain would be as depicted above. When a cell receives a signal about some learning event, the intracellular concentration of cAMP increases. Cyclic AMP binds to a protein called cAMP-dependent protein kinase A (PKA). This is made up from four smaller proteins (subunits) bound together into one protein complex. As the concentration of cAMP increases, two of the PKA subunits bind to cAMP. As a result, PKA changes its shape and frees two active (catalytic) components of the kinase. These liberated proteins move to the nucleus where they activate a molecule known as a cAMP-response element binding protein, CREB. Activated CREB then binds to its designated sites on DNA, called cAMP-response elements (CRE-elements), and switches on dozens of other genes. 


When a cell receives a signal about some learning event, the concentration of cAMP increases within it. As it does, a molecule known as cAMP response element binding protein (or CREB) is activated, which can bind to DNA and switch on dozens of genes. These genes may synthesize the proteins that are the building blocks of new memories. In an experiment with mice where CREB was prevented from binding to DNA, no long-term strengthening of synaptic connections took place.

If the model is correct, and the chain works properly, the end result is synthesis of new proteins—the necessary “bricks” that, inserted here and there, strengthen the connectivity between synapses to form memories. If so, and CREB dictates whether or not a cell will make new proteins in response to cAMP signaling, then perhaps CREB is the key to long-term memory?

In 1990, Pramo Dash and Eric Kandel injected a “cocktail” of molecules containing artificially prepared CREs into a neural cell preparation isolated from Aplysia and studied the synaptic response. Whilst the cocktail did not affect short-term changes in neurotransmitter release (short-term facilitation), it completely blocked long-term strengthening of synaptic connections (also called long-term facilitation). The Columbia University researchers reasoned that the injected CRE competitors prevented internal CREBs from binding to DNA because they occupied all of the CRE sites. In doing so, they trapped all the CREB molecules, preventing them from doing their job—that is, activating the protein synthesis machinery necessary for long-term facilitation. These findings provided an important hint: CREB seemed to be the gateway to memory’s genes. But whether CREB had anything to do with memory in live and active animals awaited work on the fly and mouse.


Whenever man comes up with a better mousetrap, nature immediately comes up with a better mouse.

–James Carswell

There are several sorts of truth that I could tell about the decade of my work on genes and memory. One would be the version reported at conferences or published in scientific papers. It consists of smooth, logical stories, which always involve a set of well-designed, unambiguous experiments, build on earlier data and theoretical models, point to future directions, and acknowledge the work of colleagues. The second “truth” would be the version found in the reports in newspapers and popular magazines, with sensational titles such as: “Scientists Find Gene For Memory” or “Of Mice, Humans, and Genetic Mysteries.” The third version would be the story behind the conferences or scientific papers that reflects the true chronology of the experiments done—who did what, why, when, and how.

I will leave aside the second “truth.” This leaves me with two versions of the same story —the logical and the chronological. Fortunately, they fit well with one another; no matter where I start, I will end up on the same pathway. My only concern is to make the story as readable as possible and so I will mix the two. But what I shall try to avoid is telling you some of the “scientific intrigues” that are an inevitable part of doing science. These “intrigues” are mainly about who deserves credit for this or that idea, model or experiment: whose findings are “clear-cut” and whose are “fuzzy,” whose papers should be published in Cell or Nature and whose should be trashed; who was just lucky to succeed and who has worked hard for his or her luck. Scientists, as Jim Watson commented in the New York Times, “are like Michael Douglas’s characters—a little evil and very competitive.” (Watson did not, however, say anything about how female scientists fit this comparison.)

It was at the traditional lobster banquet after the annual Cold Spring Harbor Laboratory conference in November 1993 that Tim Tully asked me, “So, Rusiko, what are you planning to do with the CREB knockouts?”

“I want to give them training which will enable me to dissociate learning from memory, so I can see if CREB is exclusively involved in formation of long-term memories or has something to do with learning and short-term memory as well,” I answered.

A few months before this conversation, Alcino learned that cancer researcher Gunter Schutz and his fellow post-docs Edith Hummler, Timothy Cole, and Judy Blendy at the University of Heidelberg had genetically engineered a mouse that lacked the CREB gene. A collaboration between Schutz’s lab and our lab was arranged and the CREB mice were shipped to Cold Spring Harbor. There was much excitement about the  “CREB arrival,” both in our lab and in Tully’s lab. Jerry Yin, a molecular biologist in Tully’s lab, had some evidence that mutation of the CREB gene affected memory in fruit flies in a rather specific way: flies lacking the CREB gene could not form long-term memory but their short-term memory was perfectly normal.

Although promising and exciting, these findings needed further confirmation. As “fly people” say at Cold Spring Harbor: “Flies are flies, and mice are people.” The genome of a mouse is virtually identical to the genome of a human; we may not look much alike, but our genes do. Everyone was eagerly awaiting the outcome of the experiments on the CREB mice.

As Tim and Jerry continued their “Ebbinghaus type” experiment on flies, teaching them with massed or spaced training sessions how to remember a smell associated with a shock, I began running the CREB mice in the mazes. First, I trained them in a Pavlovian conditioning model, which I described in previous chapters. The results were clear-cut. When I tested mice either thirty or sixty minutes after a training session, both normal and CREB mutant mice showed an identical memory for fear: CREB mutants remembered perfectly well the context in which they had an aversive experience and they also remembered that the tone was a predictor of danger. In striking contrast, when CREB mutants were tested two hours or twenty-four hours after training, they showed no hints of fear. Clearly, in mice with a mutant CREB gene, short-term memory remains normal, but long-term memory does not form. Similarly, CREB mutants were much worse than wild-type mice at recalling the location of a platform in a water maze, suggesting that CREB may be a key regulator of spatial memory as well.

Given the importance of the hippocampus in generating long-term memory and the correlation of long-term potentiation in this region of the brain with information storage, the next step was to look at LTP in the CREB mice. So Bruno Franguelli made an in vitro comparison of the duration of LTP in hippocampal slices of the CREB-deficient mice with the wild-type mice. Short-term synaptic plasticity (described earlier in the chapter) was identical in the mutant and wild-type animals; but LTP was less in mutants than in normal mice, and it disappeared fast. So the CREB-deficient animals not only had severe deficiency in long-term memory but also had an abnormal long-term synaptic communication.

These findings took about 250 days of intensive work. The conclusion that the CREB gene was important for long-term memory but not for short-term memory or for learning was too serious and too demanding—could it be that the very same gene is critical for long-term memory, but not needed for short term memory? Is it really so? Although all experiments were done “blind”—that is, the experimenter did now know which mouse was a mutant and which was a wild-type—they needed to be repeated and reproduced. Diana Cioffi, and undergraduate student, and I divided a new shipment of mice into two groups. She trained one group of mice and I trained the other. Independently we got the same results—CREB mutants had good short-term memory but they could not form long-term memory.

I can’t say who was happier on learning about our results—we, the “mouse people,” or our colleagues, the “fly people.” Tim seemed so excited that the forgetfulness of our CREB mutant mice closely resembled what they saw in fruit flies with a deficient CREB gene that he invited us to celebrate our “CREB findings.” A few months later, the findings “Of Mice and Flies” were published side by side as three full-length articles in Cell, and were even highlighted on the cover of the distinguished journal. They were extremely well-received by both the scientific and the popular press.

What was so important about these experiments? First, they represent the first genetic demonstration in behaving animals that short-term memory and long-term memory are separate memory systems—the argument I advocated earlier in this book. Second, the similarity in results in fruit flies and mice, together with studies in Aplysia, indicated that CREB must be an evolutionarily conserved molecule involved in the switch of short-term memory into long-term memory. As Michael Greenberg, a leading neuroscientist at Harvard Medical School, wrote:

Studies in systems ranging from two mollusk neurons in culture to complex behaviors in mammals have revealed a molecular mechanism by which memories are generated...To me, the most striking finding is the involvement of CREB in the process of information storage in a living animal.

One must, however, keep in mind that CREB is present in many other cells of the body besides neurons. It is in fact believed to play a part in hormone metabolism, drug addiction and body clocks.

If it was largely Eric Kandel’s work that prompted my work on CREB and memory, it was my work on CREB mutants that brought me to Kandel’s lab. Kandel’s lifetime devotion to the study of the role of cyclic AMP signaling in memory is a monumental achievement, as Larry Squire says, and I could not agree more. So I reasoned that if I wanted to pursue my interests and study further how cAMP signaling is involved in memory processes, I should join Kandel’s group. I did, and I continued my scientific trip along the fascinating cAMP pathway.

When protein kinase A moves to the nucleus it activates CREB. What would happen to memory if PKA were inactivated? Would memory suffer? Ted Abel, Peter Nguyen, Mark Barad, Eric Kandel and I created mutant mice expressing a gene that inhibited the activity of PKA. Like the CREB mutants, these animals showed selective defects in synaptic communications. Early LTP, which lasts only an hour or two, was normal. In contrast, the late phase of LTP, which normally lasts for eight to ten hours and like tong-term memory requires protein synthesis, waned within one or two hours. But what about the memory capabilities of the mice?

Studies revealed several important findings. First, as we predicted, disruption of the PKA pathway leads to severe defects in long-term memory whilst short-term memory is good. Second, the deficit is only evident when task-solving requires the hippocampus, but not the amygdala, or a brain area called the gustatory cortex. This latter structure, as advocated by Yadin Dudai of the Weizmann Institute in Jerusalem, is crucial for forming memories about poisoned food (taste aversion). The discrepancy in memorizing different tasks—clearly favoring multiple memory system concepts—was not surprising because in these mutants the PKA pathway was most disrupted in the hippocampus, not in the amygdala or the gustatory cortex. So for short-term memory to be transformed into long-term memory, the pathway leading to the CREB gene must work properly.

Meanwhile, the research on CREB exploded. Reports about the crucial role of CREB in long-term memory and long-lasting neuronal communication began to sprout like mushrooms after rain. One after another, elegant experiments in flies, Aplysia and rats revealed that boosting CREB could produce memory akin to photographic memory in humans. New evidence supporting our findings about the role of CREB in spatial memory came from California. There, James McGaugh’s team drizzled a special chemical that blocks CREB into the hippocampus of the rat’s brain. Like our CREB mice, McGaugh’s rats with inactivated CREB lost the ability to form spatial memory. Further south, Ivan Izquierdo at the Institute for Biology of Porto Alegre in Brazil and Jorge Medina at the University of Buenos Aires in Argentina observed increased amounts of CREB protein in the hippocampus as rats memorized a fearful event in a step-down scheme. Activity of CREB appeared to be crucial for long-lasting memory of a variety of tasks in which rodents have to remember which food is safe to eat and which is not. Finally, there is tempting evidence that CREB malfunction may account for some human mental disorders as well.

As Mark Bear, a neuroscientist at Brown University, has commented, “the CREB system represents one of the most exciting developments in neurobiology over the past several years, because it seems to operate similarly in a number of different organisms.” Not surprisingly, research into the CREB switch has become a matter of not only academic but also commercial urgency as well. In 1997, two “memory companies” (according to a news release), Helicon Therapeutics and Memory Pharmaceuticals, were formed to search for memory-improving drugs based on the CREB switch.

Yet behind the euphoria that greets each new discovery lurks a nagging sense of dissatisfaction. Exactly how do genes affect neuronal plasticity? In which synapses, and in how many of them, are particular memories stored? Where exactly are the long-term memory cells located, those engrams for which Karl Lashley searched for forty years? How does CREB decide which memories should be stored and which should be ignored? How many genes does CREB target? How many pathways lead to CREB? Is CREB the only molecular switch for memory? And if not, how many more are there?

The truth is that we don’t have all the answers yet. But even in the past three years, from the moment I decided to write this book until the moment I am about to finish, we have accumulated colossal findings. They will be filtered and polished; some will be proved and others will be refuted by new evidence. It is a long way to go, but as Eric Kandel likes to put it, “we do have a nice beginning.” Memory is a mystery whose secrets have been slowly unfolding for years.


If the human race wants to go to hell in a basket, technology can help it get there by jet.

—Charles M. Allen

I have seen similar headings many times in scientific or popular literature and, I must admit, I don’t like them, even though the question is legitimate. Unless one writes about clear future strategy, what is usually summed up beneath these “goodbye notes” is what has been already said “between the lines” throughout the text. Sometimes, the notes are so long that one even begins to lose track of where one was to begin with. And I am of the opinion that if you give yourself more time to think, you eventually understand the main message of any story. I will, therefore, be brief.

The molecular genetics of memory will clearly be a major area of research. Yet the conclusions of such research should be tempered with caution. Basic genetic research has taught biologists that genes do not work in a vacuum. Rather, there is a complex interaction among many genes. If one further considers the role of the environment in gene expression, the factors involved in such aspects of life as behavior, personality, emotion, reasoning, thought, learning, and memory become immeasurable. “The ‘nature versus nurture’ debate in biology must be dismissed as oversimplified by contemporary genetic research,” says Tim Tully. Put simply, genes do not determine behavior; rather they influence it in concert with personality and the social environment.

This principle is particularly true when we talk about memory. Memory is what defines who we are and who others are in our own minds. Memory shapes our intellectual and moral personality, the way the think, smile, say hello and behave in day-to-day life. Indeed, it would be impossible to live as one person, with an individual history, or to possess our being in a continuous fashion, without the memory threads that constantly link our present to our past and prospective future. Science may never be able to fathom the complexity of memories, even though we may understand the basic blocks from which memory is made. Memory is what makes it possible for us to live but not to exist. Genes cannot determine memories, but they do affect our capacity to remember and to forget.

So what do I and other scientists ultimately hope to achieve by hunting for “memory” genes? I cannot, or course, speak for all scientists, but I believe at the heart of what we are trying to do is the hope that we will be able to gain a better understanding of the molecular mechanisms of which memories are made. This, in turn, will allow the development of drug treatments for patients suffering from memory lapses, including the terrible losses caused by diseases such as Alzheimer’s, dementia, or age-related mental decline. Perhaps the lost memories will never be recovered. But drugs that improve the long-term retention and storage of new information—perhaps as trivial as remembering a new acquaintance’s name or telephone number—will emerge, it seems to me, in the foreseeable future. Similarly, effective treatments could be developed to ease the memories of traumatic events, those horrible “flashbacks” that persist for years or sometimes for life in the victims of war, violent crime, bombs or sexual abuse.

My optimism is based on my memories of what we as humans have achieved in science, medicine, and technology over the past few decades. What is more, research on the biology of memory is still very young. These are not my romantic “goodbye notes,” rather they are my true scientific beliefs. But before they will be fulfilled, there will be many disappointments. These disappointments will arise from both the predictable and the unpredictable effects of memory compounds: there will be “no effects,” “side effects,” and even “lethal effects.” But perhaps the most serious worry about this technology will emerge from the abuse of memory drugs. Probably they will be taken not only by people who need them but also by healthy people who have a perfectly good memory. For example, they might be used and abused by schoolchildren simply to get better grades or to cover a school program that normally takes four years in one year; or by young adults so that they can boost their mental performance in order to land a better job. This is turn might create at least three problems. Can everyone afford such “smart” drugs? And if not, would it be fair to those who cannot afford memory drugs and have to work for years to achieve the same performance?

The second problem would be that even though the immediate outcome of such drugs might be a sharpening of memory, in the long run they might have an opposite effect—a dulling of memory. In short, what helps your grandmother might not help you. That should not surprise us. There is no reason to assume that, for most of us at most times, our molecular machinery that produces CREBs, enzymes, and neurotransmitters is not working more or less optimally. Normally, the brain is well buffered against the effect of arbitrary increases or decreases in circulating chemicals, and increasing their activity is no guarantee of increased mental performance. And even if it were, an overpowering memory, as brilliantly described by Alexander Luria, might create a social and intellectual misfit like Mr. Shereshevsky. People with exceptional, photographic memories often have enormous difficulty making even simple decisions, because at the same time they can think of fifty different options to choose from. Often, there is a great adaptive value in forgetting certain things. It clears the mind and allows us to concentrate on important values and think in perspective, rather than rummaging indefinitely in meaningless details. More does not necessarily mean better. Jorge Luis Borges describes this powerfully when he has the hero of Funes, el Memorioso say:

I have more memories in myself alone than all men have had since the world was a dreams are like your memory, sir, is like a garbage disposal.

As with everything else, memory drugs will have pros and cons. The intelligent thing to do with memory drugs will be to use them intelligently. By no means am I reducing our memories to single genes, or for that matter to any other molecules. However, if we are going to intervene pharmacologically in order to cure or prevent memory problems, we must know the workings of the proteins encoded by memory-related genes and we must know how to change the way they work

About Cerebrum

Bill Glovin, editor
Carolyn Asbury, Ph.D., consultant

Scientific Advisory Board
Joseph T. Coyle, M.D., Harvard Medical School
Kay Redfield Jamison, Ph.D., The Johns Hopkins University School of Medicine
Pierre J. Magistretti, M.D., Ph.D., University of Lausanne Medical School and Hospital
Helen Mayberg, M.D., Icahn School of Medicine at Mount Sinai 
Bruce S. McEwen, Ph.D., The Rockefeller University
Donald Price, M.D., The Johns Hopkins University School of Medicine
Charles Zorumski, M.D., Washington University School of Medicine

Do you have a comment or question about something you've read in CerebrumContact Cerebrum Now.