Rescuing Aging Memory: Stem Cells and Other Rising Stars

He not busy being born is busy dying
—Bob Dylan

I am not in my perfect mind. Methinks I should know you, and know this man; yet I am doubtful; for I am mainly ignorant what place this is, and all the skill I have remembers not these garments; nor I know not where I did lodge last night
—King Lear

In his melancholy refrain, Bob Dylan, knowingly or not, captures a fundamental fact about aging. When our bodies have matured, and our physical and mental functions have peaked (at different ages for different functions), we begin a gradual, almost imperceptible, decline. We are not dying, except by poetic license; we are aging. Our cognitive functions, including memory, are not exempt.

Scientists have long suspected that the memory decline of normal aging does not necessarily end in the devastating cognitive crack up of King Lear, who we now recognize as probably suffering from Alzheimer’s disease or other dementia. Physicians find themselves reassuring many patients who experience the first signs of memory impairment that they are not seeing the first symptoms of Alzheimer’s. One key difference between Alzheimer’s and age-related memory loss—a literally life-and-death difference—is that in normal memory loss we do not see the widespread, progressive death of nerve cells in the brain that we see in Alzheimer’s.¹

In this difference lies hope that the brain changes accompanying normal memory loss may be prevented or even treated—and the associated memory losses slowed, prevented, or even reversed.

In Alzheimer’s, by contrast, we face the problem of brain cells that have died; and until very recently the overwhelming consensus of scientists was that brain cells and cells of the central nervous system could not regenerate as do other cells. The death of brain and central nervous system cells, we were convinced, signaled inevitable, irreversible damage. The news from today’s brain research, however, is that this long-held consensus may need to be modified. In a host of experiments, scientists have found evidence that at least certain parts of the brain do regenerate cells. What is more, we are gaining insight into how brain and central nervous system cells may be coaxed to grow where and when we need them.

Today scientists are identifying those less-than-lethal changes in brain cells and brain circuits that seem to accompany normal gradual memory decline with age. The new discoveries that I will explain have revolutionized our theories of why our memories may decline with age. Understanding these discoveries will prepare the reader to appreciate the new hope for slowing, preventing, and reversing normal age-related memory loss.

While the science is moving very fast, successful clinical interventions for memory loss, in normal aging or Alzheimer’s, remain elusive. It may be some years before we can be certain if we hold the key to restoring memory, but the horizon is bright. I must add that, for a scientist who has spent almost 20 years preoccupied with the seemingly overwhelming degeneration of brain circuits associated with Alzheimer’s disease, the present optimism about restoration and regeneration of brain circuits impaired with age is a welcome shift in attitude. Science cannot yet treat memory loss in humans, but there is a growing confidence among neuroscientists that we should not view such loss as an inevitable accompaniment to aging. The approaches that I describe here, including the use of “stem cells,” transplanting genetically engineered cells, and an understanding of the effects of hormones on memory, all show solid experimental results in animals that bode well for humans.

Memory, Cognition, Dementia

I will begin by explaining in a bit more detail the concepts of memory, cognition, and dementia. A more precise grasp of these will enable us to understand the nature of memory loss, in both normal aging and Alzheimer’s, as well as the gulf of difference between them.

For our purposes, “memory” can be divided into two types.² “Declarative memory” refers to memories that can be described verbally. It is the ability to consciously recall the events of one’s life, the people and things that one has been exposed to, and general knowledge of the world in which we live. “Non-declarative memory,” (including procedural memory) is memory that is unconscious, and the memory involved in learning skills and procedures such as how to ride a bicycle; it also includes our behavioral dispositions, biases, and other unconscious determinants of how we respond as a result of prior experience.

“Cognition” describes a broader set of functions, including memory. Along with memory, cognition involves paying attention, thinking abstractly, and following a train of reasoning, as well as functions such as imagination, insight, and aesthetic appreciation.

“Dementia” is the state that exists following the loss of normal cognitive abilities. While Alzheimer’s disease profoundly affects declarative memory, it also affects cognition in general. When we speak of dementia, such as that accompanying Alzheimer’s, we are referring to a devastating, broad loss of cognitive abilities, going well beyond a problem with declarative memory. Perhaps the best way to understand dementia is through the definition offered by my colleague Dr. Peter Rapp: “Dementing disorders like Alzheimer’s slowly rob us of everything that defines who we are: our memory built up over a lifetime of experience, our capacity for reason and imagination, and ultimately our ability to think and reflect.” In contrast, normal age-related memory impairment is primarily a deficit in declarative memory, with other cognitive abilities largely unaffected. Clearly dementia represents a far more devastating condition than the deficits in declarative memory that we often see in normal aging, such as difficulty remembering recent events, and difficulty learning new spatial tasks or remembering words.

Just as cognition represents a much broader set of functions than declarative memory alone, so it is served by a much broader set of brain circuits [see illustration above]. The areas of the brain most closely involved with memory are the prefrontal cortex and the medial temporal lobe, and within the medial temporal lobe, the hippocampus. Interestingly, declarative memories are profoundly affected by damage to the hippocampus, but non-declarative and procedural memories are not. In this article I refer primarily to declarative memory, since it is this memory process that is so devastated in Alzheimer’s, as well as being affected in age-related memory impairment. While declarative memory also involves the prefrontal cortex, the hippocampus plays a particularly critical role in learning new things and laying down new memories.

Once new memories are formed they often reside primarily in the neocortex, particularly the areas that integrate diverse sensory or motor information, the association areas. This requirement for involvement of the neocortex’s association areas for the full breadth of cognition to occur is critically important, since it is the essential difference between isolated deficits in normal aging and the dementia of Alzheimer’s. Thus, it should come as no surprise that Alzheimer’s involves extensive neuron loss and circuit disruption in the neocortex, which is not seen in normal aging. The more modest deficits in declarative memory that are referred to as age-related memory impairment are likely to result primarily from impairments of hippocampal circuits, with neocortical circuits less affected. We will see the enormous potential significance of these differences.

The Devastation of Alzheimer’s Disease

In 1906, Alois Alzheimer described the “neuritic plaques” and “neurofibrillary tangles” that are strikingly obvious when we look at the brains of people who died of Alzheimer’s disease. These plaques and tangles seem to be at the heart of the damage caused by Alzheimer’s. Neuritic plaques may be created when a protein called “B amyloid” is deposited in brain tissue; they figure heavily in attempts to explain Alzheimer’s through understanding the genetics, biochemistry, and molecular pathology of amyloid. Neurofibrillary tangles, on the other hand, are the wreckage of neurons that have degenerated in the brain’s hippocampus and neocortex. Tangles seem to be closely linked with the pathology of proteins called “tau” and “neurofilament,” and to result from biochemical and cellular pathologies that are largely distinct from those that lead to formation of plaques.³

Whatever their origin and nature, plaques and tangles are important because they represent the disruption of brain circuits and the death of neurons, the essential events in a neurodegenerative disorder. They lead to the functional decline and ultimate dementia of the Alzheimer’s patient. Each tangle represents a dead neuron. Plaques, however, do not have a one-to-one link with degenerating neurons. In fact, plaques can occur in the absence of significant neuron death and do not necessarily imply significant functional decline on their own. This suggests that the role of plaques in neurodegeneration is either less direct, more complex, or both less direct and more complex than the role of tangles.

When we analyze the number of brain cells in certain layers and regions of the cerebral cortex of people who died of Alzheimer’s, and when we analyze the number of synapses (connections between brain cells) in those same layers and regions, we see an extensive loss of certain neurons and their associated synapses. But in which circuits does this loss occur? We see clearly that certain nerves running to the cortex from sites underneath the cortex are affected. The major cause of dementia in Alzheimer’s appears to be the loss of the circuits that connect related cortical areas—particularly the frontal, temporal, and parietal lobes, which are critical to cognitive processes.

But there is another circuit that is especially susceptible to plaques and tangles: the “perforant path”. In fact, this circuit is exquisitely vulnerable to both the neurodegenerative events that mark Alzheimer’s and the non-degenerative changes that mark normal aging. The perforant path provides the major connection between highly processed information in the neocortex and the hippocampal circuits that lay down new memories, especially declarative memories.

The Perforant Path: Getting at the Difference Between Alzheimer’s and Normal Aging

The circuit referred to as the perforant path is invariably devastated by extensive formation of neurofibrillary tangles in Alzheimer’s disease, even at its very earliest stages. We think that in Alzheimer’s this is the single most vulnerable circuit, and is likely the first circuit to degenerate. Nevertheless, if neurofibrillary tangles were to devastate the cells of the perforant circuit, and only those cells, without any disconnection of the circuits in the neocortex that make possible other elements of cognition, the Alzheimer’s patient would not display the disease’s ultimate total dementia. Certainly the patient would have a significant defect in declarative memory, particularly with respect to new memories, but this would still be far less debilitating than what we see in Alzheimer’s.

What happens to the perforant path in elderly people who do not have Alzheimer’s, who are neurologically normal? The vast majority of people older than 55 seem to have at least a few neurons in the entorhinal cortex with a biochemical profile that suggests that they are in transition to becoming tangles. But in an Alzheimer’s patient if we either count tangles or count remaining viable neurons, we clearly see that tangles have come to dominate the entorhinal cortex. Normal, healthy neurons are essentially gone. By contrast, in the elderly individual who does not have Alzheimer’s, there are so few tangles that investigators have concluded that there is no significant overall neuron loss.

Consider what this means. While the perforant path or circuit clearly is destroyed in Alzheimer’s, it does not appear to be significantly affected in normal aging (even though we find some hints of pathology). For a long time, scientists had assumed that age-related memory impairment would also be caused by neuron loss, although perhaps less loss than in Alzheimer’s. Not so.

The discovery that there is no significant loss of neurons in normal aging prompted us to look for some other source of functional decline.^4,5 This brings us to what has been discovered about more subtle changes in the neurochemistry of brain circuits and in cell structure that might affect how well brain cells communicate but would not lead to death of those cells. These changes will show us what we must deal with if we are to slow, halt, or even reverse normal age-related memory loss.

Characterizing Brain Cells and Circuits

Traditionally, scientists have categorized a given type of brain or nerve cell—or a given type of circuit in the brain—by looking at its physiological and anatomic characteristics. What information did a circuit transmit? Where did the circuit begin and end? Now, however, we can add to those characterizations of cells and circuits extensive biochemical and molecular information. Our characterization thus becomes more comprehensive; we can see more clearly which qualities of the circuit may relate to its role in a particular brain function.

The biochemical attributes of a given circuit or type of cell largely result from the effect of particular genes, the activiation of those genes to synthesize protein, how the protein is broken down and distributed, and how the cells in the circuit are activated to start sending a message. The sum of these attributes is the “neurochemical phenotype” of a class of circuits or cells. Thus, the neurochemical phenotype of a neuron includes its molecules that are related to signaling between cells, the neuron’s structure, its metabolic processes, and, in fact, any other function that is especially well developed in that class of neuron and is essential to its function in the brain.

Take an example. A neuron in the cerebral cortex that uses GABA (the chemical messenger that specializes in inhibiting neurons) has a neurochemical phenotype that differs in many fundamental ways from a neuron that uses glutamate (the chemical messenger that specializes in stimulating neurons). Therefore, if we want to really understand a brain circuit, we must look at its particular neurochemical profile, as well as its anatomic connections. Both will affect its functioning and its role in behavior.

Less-than-Lethal Changes

Now we can continue with our story. We are looking for ways to link age-related changes in the neurochemical phenotype, as well as other characteristics of key circuits in the cortex and hippocampus, to the declines in memory functioning that we see as people age. As we saw, scientists have pretty well disproved the idea that neurons inevitably die as we age, or at least they do not die at a level sufficient to cause cognitive decline as we age. In contrast to the dramatic cell death responsible for the devastating cognitive deficits of Alzheimer’s, we are seeing that the decline of the normal aging brain takes place without major loss of brain cells. Thus, it becomes increasingly important to investigate more subtle changes in the form and structure of these cells and their phenotype, changes that would impact function, but not be lethal. Let us look at three approaches to this investigation.

First, take gene expression and neurochemical phenotype. In understanding how the brain is organized, it is crucial to realize that fully half of our 100,000 genes are expressed (have their effect) solely or mostly in our brains. Thus, gene expression is important not only to different functions, but also, as we will see, to our understanding of how genes may be manipulated. For example, genes probably control how receptors for the chemical messenger glutamate change with age, particularly in the hippocampus, where glutamate receptors (especially one called the NMDA receptor) affect the flow of information through circuits—a function highly suggestive of a role for NMDA in age-related changes in memory.

The Critical NMDA Receptor

Studying NMDA receptors, hippocampal circuits, and aging is our second approach. Experiments with rats have shown that many aspects of chemical signaling in the hippocampus are unaffected by aging; some aspects may actually become stronger, but other aspects are weakened.⁶ It does not appear, however, that age simply decreases the number of synapses (gaps across which cells signal) in the hippocampus. Nor does age seem to involve a general deterioration of synapses. But one aspect of signaling across synapses that does appear to be weakened is a crucial process called “longterm potentiation,” known to scientists as LTP.

LTP is the process by which a given brain circuit, when it has been subjected to a certain frequency of electrochemical stimulation (50-200 stimulations per second), becomes more easily activated in the future.

LTP is often viewed as a cellular mechanism for learning because it is as if the circuit has “learned” to respond more readily once it has been activated in this manner. Although some circuits can display LTP without involving NMDA receptors, most of the time NMDA receptors must be activated for a circuit to display LTP. Scientists think that this is linked with the NMDA receptor’s well-known role in learning and memory.

Since we now see that structural degeneration of hippocampal circuits is not necessary for hippocampal function to be impaired, and that NMDA receptors have an important role in signaling within the hippocampus, perhaps age-related changes in the levels of NMDA receptors lead to age-related changes in memory. Unfortunately, studies have failed to resolve that question, some suggesting that there is a decrease in NMDA receptors with aging, others suggesting no decrease or, in humans without Alzheimer’s, perhaps even an increase. The important thing to bear in mind is that molecular shifts related to aging can occur in structurally intact circuits, and that these shifts can dramatically change the appearance and the function of the circuit—all in the absence of degeneration.

Given the importance of the perforant path’s connection between the entorhinal cortex and the hippocampus, as well as the importance of the NMDA receptor in age-related changes in memory, my colleagues and I decided to examine how glutamate receptors such as the NMDA receptor were distributed in this critical circuit. Our analysis demonstrated that aged monkeys show a decrease in NMDA receptors, compared with young adult monkeys, only in the specific portion of the hippocampus that receives input from the entorhinal cortex.⁷

This points, once again, to the perforant path as a key element in age-related changes. We also found that the perforant path was structurally intact in these aged animals and only NMDA receptors showed a decrease. We think that in aged monkeys the concentration of this key receptor can change in ways that affect specific circuits without gross changes in cells or fewer synapses. The changes in the concentration of NMDA receptors, compromising the transmission that those receptors make possible, could explain age-related shifts in LTP, and so changes in memory, without requiring any purely structural damage. Such age-related changes without structural degeneration have also been discovered for other circuits linked to cognition. The payoff, as I will describe later, is that gene therapy has been used to reverse these changes in aged monkeys.

Questions of Cause and Effect

Finally, we must also look at the issue of cause and effect. Although we are beginning to discover certain brain changes that occur in normal aging, how can we be sure that these are the cause of age-related memory loss? It is always difficult to show that molecular or structural change in the brain causes behavioral change, although our chances improve when we study brain circuits in animals whose behavior has been carefully analyzed such that the animals can be categorized with respect to their performance on various tasks. This approach has been useful for genetically modified mice as well as aged rats.

An important negative finding (discovering that something is not so) emerged from studies of these aged rats. When the number of hippocampal neurons was determined in young, age-impaired, and age-unimpaired mice, it turned out that there was no difference among these three groups. In other words, neuron loss could not explain age-related memory impairment.

This was an important impetus to look for more subtle shifts in brain circuit characteristics that might explain memory impairment. Again using animals with carefully characterized behavior, we looked at the structural integrity and synaptic function of key hippocampal circuits in the same three groups of animals: young, age-impaired, and age-unimpaired. We were able to correlate both molecular and behavioral changes with a specific hippocampal circuit. We found that circuit-specific alterations in molecules that control glutamate release in the hippocampus may contribute to the effects of aging on learning and memory, without actual degeneration of the neurons. Significantly, this again pointed to the entorhinal-hippocampal link as the culprit.

Estrogen and Aging

Reaching the end of our reproductive years may be the most fundamental event that initiates and affects human aging. Thus it is not surprising that scientists have begun to discover many connections among menopause, estrogen replacement therapy, and behavior—including cognitive function. For American women, the median age of menopause, the end of the menstrual cycle, is 51 years old. Ironically, this was also the life expectancy for American women a century ago before modern health care began to extend our life span. With today’s average life span of American women at 75-80 years, women now spend some 35 percent of their lives beyond their reproductive years. Understanding how estrogen levels decrease after menopause, and with what consequences, has become critical for women’s health.

The issues are complex with respect to the aging brain, because the interactions between our hormones (such as estrogen) and brain during reproductive “old age” are to some degree circular. Thus, the neuroendocrine system plays a critical role in regulating ovarian hormones such as estrogen, but age-related declines in estrogen also feed back to the brain. Here is where neurons in the hypothalamus directly regulate reproductive behavior and affect other brain regions that are associated with attention and memory. In fact, as described below, the effects on brain circuits of manipulating estrogen levels in younger animals are striking. As yet, however, we do not fully understand the links between estrogen levels and post-menopausal memory impairment in aging women, nor are the benefits of estrogen replacement therapy and the potential protective effects of estrogen against Alzheimer’s disease completely clear.

Looking at normal aging, postmenopausal women using estrogen replacement therapy performed better on tests of verbal recall and other measures of cognitive performance than non-users. Reinforcing this is the discovery that younger women whose ovarian estradiol (a form of estrogen) is suppressed with drugs show a decline in verbal memory, but that this decline can be ameliorated by estrogen replacement therapy. There is also evidence of a link between estrogen and neurodegenerative diseases associated with aging. Estrogen replacement following menopause reduces the risk of Alzheimer’s disease and delays the age of onset of its dementing stage. This means that estrogen may play a critical role in protection against the neurodegeneration typical of Alzheimer’s, as well as in maintaining the health of existing circuits, though it is unlikely to help once the degenerative cascade is underway.

We should keep in mind that, while the data on estrogen/brain interactions from animal studies is already compelling, the clinical data are not entirely consistent. This is not surprising, given variations in the formulation of hormone supplementation, dose regimen, duration of treatment, compliance with therapy, and the levels of circulating estrogen, as well as the difficulty in differentiating between the two cognitive functions, attention and memory.

Estrogen’s ability to affect many brain regions outside the hypothalamus suggests that the decline in estrogen levels during reproductive “old age” has far reaching consequences for the brain. Receptors for estrogen are widely distributed throughout the brain.⁸ For example, the estrogen receptor-alpha is found in the hippocampus, the cholinergic basal forebrain, and the neurons that project from the ventral tegmental area to the neocortex. Interestingly, all three of these systems are highly vulnerable to aging, but this vulnerability has never been adequately distinguished from their sensitivity to estrogen. Obviously, therefore, through multiple mechanisms estrogen can influence virtually the entire brain. It is not surprising that age-related or experimentally induced decreases in estrogen can affect cognitive processes.

How does estrogen affect vulnerable cells in the hippocampus and elsewhere in the brain? We are just beginning to answer that question. Increases in estrogen levels during natural ovarian cycles in young rats causes an increase in the density of the dendrites and synapses in the hippocampus that enable brain cells to communicate. Since NMDA receptors in the hippocampus are both estrogen- and age-dependent, there may be an interaction of estrogen with aging during menopause, resulting in an even greater decrease in NMDA receptors. Because these NMDA receptors in the hippocampus are critical in learning and memory, their decline during menopause could cause age-related changes in cognitive function. What are the effects of estrogen depletion and estrogen replacement therapy on the brain and cognition in both Alzheimer’s and normal age-related memory loss? We need to find out.

Can We Rescue Memory Circuits Affected by Aging?

We have seen that both normal aging and neurodegenerative disorders disrupt specific vulnerable circuits in the brain. That seems to suggest that the most successful interventions will be those that are sufficiently selective in the circuits they affect. Is there any evidence that damaged or degenerating circuits can be restored—and if so, how successful and selective is the restoration?

Although scientists have sought ways to restore the brain for decades, several recent developments have fostered a far more optimistic view about the aging brain and neurodegenerative disorders than was prevalent just a few years ago. Some of these developments and strategies are transplanting tissue, transplanting engineered cells, gene therapy, the use of stem cells, and the potential for exploiting the natural creation of nerve cells that occurs in the adult brain.

TRANSPLANTATION STRATEGIES. Transplanting embryonic brain tissue into a damaged adult brain was one of the first strategies that scientists tried in order to repair brain circuits. In a complicated operation, the surgeon removes the desired neurons from the embryo’s brain, then positions them in the adult’s brain to replace the damaged circuit. Transplantation of dopamine neurons has been successful in animals; the transplanted neurons survived, formed the appropriate connections, and the animal recovered its functioning. This procedure has even been tried as a treatment for human neurodegenerative diseases such as Parkinson’s.⁹ Initially, the surgery in humans had limited success, but in recent years the results have improved. We see long-term evidence of the survival of the transplant, a degree of functional recovery, and even evidence of release of dopamine from the transplanted cells. Transplanting fetal tissue in animal models, however—and more so in human patients—raises ethical and political questions.

A newer approach has been to inject genetically modified cells that secrete dopamine directly into the region of the brain that requires dopamine replacement.¹⁰ Genetically modified viruses have also been used in gene therapy to “deliver” a growth factor into the hippocampus. There the growth factor is successfully synthesized and secreted, promoting the improved health of the damaged circuit. The success of this approach, however, depends on the nature of the circuit. For example, the success with dopamine was possible because it was used to treat a unique circuit, one relatively uncomplicated in its organization. Similar results could not be achieved in the neocortical circuits that degenerate in Alzheimer’s, since these circuits have very complex and precise synaptic arrangements.

Recent primate studies have used gene therapy to reverse naturally occurring age-related weakness in two vulnerable circuits. One study targeted the type of neurons known to be vulnerable in Alzheimer’s.¹¹ The other study used a virus as a vehicle to deliver a growth factor intended to reverse the natural, age-related decline in dopamine function in a particular area of the brain. The gene therapy succeeded in both cases. Most exciting for our discussion is that in both studies the circuits had not degenerated; they were simply in poor health, having lost their high level of functioning. These experiments, therefore, may be particularly relevant to the circumstances that occur in normal aging.

THE PROMISE OF STEM CELLS. Transplantation techniques attempt to deal with the brain’s inability to repair itself. The brain cannot replace cells, in this case neurons, the way most tissue can—by duplicating exact copies of existing cells. Even if the adult brain could create new neurons (and it can do so in some regions, as discussed below), those new neurons would not be very helpful in most circumstances unless they were functionally equivalent to the cells that died. And they would have to be equivalent in terms of very specialized attributes and of participation in complex circuits that were established during early brain development.

Ideally, we would like to be able to replace dead or damaged neurons with cells that have not yet committed to becoming a particular cell type. Then we could “coax” those cells into the appropriate form and connections. Although recent, remarkable discoveries about what we call “stem cells” are still in their infancy, at least in terms of their application to brain aging and neurodegeneration, they may soon be very important for these purposes.¹²

Stem cells are cells that continue to proliferate without differentiating themselves into any particular organ or cell type. They retain what is referred to as a “pluripotent” (multi-potential) capacity. Because these cells can be nudged down a particular path of differentiation, scientists have become excited about their potential for restoring damage in multiple organ systems, including the brain. We know that undifferentiated embryonic stem cells can be kept alive for long periods, and then they differentiate themselves to become various “target” tissues. Relying on human embryonic stem cells for therapy presents many technical, political, and ethical problems, however, similar to the issues involved in using the tissues of human embryos for transplants.

Adult Stem Cell Breakthrough

We may not always have to rely on embryonic stem cells. Now it appears that adult stem cells from one particular organ retain the capacity to survive, differentiate, and presumably function in a different organ. For example, brain stem cells may retain their capacity to become blood cells, and bone marrow stem cells surviving in the brain may differentiate into both neurons and the glial cells of the brain’s white matter. Eventually, it may even be possible to replace damaged neural circuits with such an approach, done in a way analogous to the brain tissue transplants described earlier.

Recently, in an animal model, neural embryonic stem cells were shown to represent a promising therapeutic approach to recovery of function following traumatic spinal cord injury.¹³ The transplant-derived cells from the mouse survived and differentiated into brain cells in the rat, and the rat even recovered some functioning. The potential of such approaches, using neural stem cells for therapies directed at age-related pathology and traumatic injury, is profound. While this research is still in its infancy, the progress to date is astounding.

One difficult challenge with stem cell approaches is finding how, in any therapy, to achieve a high degree of circuit specificity. The good news is that the use of neural stem cells to achieve some recovery of function after spinal cord injury suggests that, at least sometimes, this circuit specificity may emerge. The catch is that, in the example above, results were achieved in a young animal with a traumatic injury; it is not clear, at this point, if similar success could be achieved in replacing brain circuits that are affected in aging. For example, if we tried to use stem cell therapy in Alzheimer’s to replace the neurons in the perforant path, how would we guide the neural stem cells into becoming the very highly differentiated neurons that reside in layer II of entorhinal cortex, and how would we promote formation of the appropriate complex circuits? Still more difficult, how would we replace the neurons in the cortical circuits connecting the frontal and temporal regions that are so damaged in Alzheimer’s, while leaving the intact circuits unaffected? While in some cases the brain may solve this problem itself, by continuing to generate neurons that can replace certain circuits throughout life, this may be a limitation in the approach.

NEW NEURONS IN THE ADULT HIPPOCAMPUS. Recent research has forced us to reconsider the long-accepted dogma that when neurons die they cannot be replaced by the generating of new neurons. Is neurogenesis a solution to memory loss? While scientists have known for three decades that some neurogenesis occurs in a part of the adult rat’s hippocampus, called the dentate gyrus, they viewed this as merely a curious remnant of earlier development. Now we know that this neurogenesis also occurs in nonhuman primates and even in humans—and may be occurring at an impressive rate.¹⁴ How the new neurons function in the dentate gyrus is not clear; but it does seem that they may form appropriate connections within the rest of the hippocampus. Furthermore, the prevailing notion that the dentate gyrus is the only region where such neurogenesis occurs has now been challenged by research suggesting that neurogenesis also occurs in the neocortex of nonhuman primates.¹⁵

Can neurogenesis in the adult brain be harnessed and exploited to counteract neurodegeneration? Certainly not in any comprehensive sense, since many of the most vulnerable neurons are not those that display neurogenesis; they are irreplaceable once they degenerate. Also, we must bear in mind that new neurons, far from improving function, may wreak havoc in highly ordered circuits. After all, complex circuits develop and are sculpted over time, in response to a complex array of genetic predispositions, developmental events, and environmental influences. Adding new neurons to an established system of complex connections might be problematic if those new neurons cannot replicate the fully developed patterns of connection. My colleague, Pasko Rakic, a scientist at Yale University, puts this issue in context when he says, “New neurons didn’t go to school!” In thinking about the implications of neurogenesis in the dentate gyrus, therefore, we must bear in mind that this may be a unique situation in terms of the ability to “educate” new neurons.

Several recent reports suggesting the potential importance of neurogenesis to aging all relate to the dentate gyrus section of the hippocampus. First, we have learned that neurogenesis in the dentate gyrus decreases in aging, and that decline may hurt our functioning in certain tasks.¹⁶ At the same time, however, we are aware of positive influences on neurogenesis that we may be able to use to prevent this age-related decrease. The simple process of being trained to perform a task actually enhances neurogenesis in the dentate gyrus of rats. So does living in an enriched environment. So does increased social interaction. Most surprisingly, perhaps, simple physical exercise, such as running, also increases the growth of these cells.¹⁷

Hormones also affect neurogenesis in the dentate gyrus. For example, estrogen seems to stimulate this neurogenesis in the adult female rat. More recently, scientists showed that the level of neurogenesis typical of a young animal could be restored in an aged rat by decreasing the old rat’s high levels of corticosteroids—the stress hormones generally seen in aged animals.¹⁸ These discoveries about neurogenesis are immensely intriguing in their implications for understanding aging and the neurobiological causes of age-related functional decline. This is an area for intense investigation, one of paramount importance in aging research.

Are there ways to promote adult neurogenesis in areas other than the dentate gyrus? If so, can we control this process so that it restores function, such as memory—or will it throw a monkey wrench into functioning? Even to be able to pose these questions as we think about enhancing the function of the aged brain reflects the present state of soaring optimism.