NeuroMolecular Medicine

, Volume 10, Issue 2, pp 59–66

Physical Activity and the Regulation of Neurogenesis in the Adult and Aging Brain


  • Klaus Fabel
    • Center for Regenerative Therapies Dresden (CRTD)DFG Forschungszentrum und Exzellenzcluster
    • Department of PsychiatryUniversity of Dresden
    • Center for Regenerative Therapies Dresden (CRTD)DFG Forschungszentrum und Exzellenzcluster
Original Paper

DOI: 10.1007/s12017-008-8031-4

Cite this article as:
Fabel, K. & Kempermann, G. Neuromol Med (2008) 10: 59. doi:10.1007/s12017-008-8031-4


The discovery that exercise regulates adult hippocampal neurogenesis, that is, the production of new neurons in the adult brain, was surprising news and changed quite fundamentally our view on how physical activity affects the brain. The everyday experience that not all athletes are necessarily smarter than more sedentary fellows and the scientific insight that adult hippocampal neurogenesis is actually a process that ranges on a very small scale raised important questions on the relevance of this finding. We propose that the exercise-related regulation of adult hippocampal neurogenesis is a qualitative rather than a quantitative event and that it is a particularly prominent and suggestive example of activity-dependent cellular plasticity. For rodents, the animals, in which most of this research has been done, cognition is almost inseparable from locomotion. Physical activity, especially exerted over longer periods of time, might indicate to the brain an increased chance of experience those situations rich in complexity and novelty that presumably benefit from more new neurons. We thus propose that it is not isolated physical activity that is “good for the brain”, but physical activity in the context of cognitive challenges. This would also explain why few new neurons could be beneficial for successful aging. We here review the current stage of the knowledge how this exercise-induced regulation of neurogenesis might work.


MouseLearning and memoryStem cellsDentate gyrusHippocampusGerontology

Introduction: Adult Hippocampal Neurogenesis

Adult neurogenesis produces only three types of neurons in the adult brain, two types of interneurons in the olfactory bulb and new excitatory granule cells in the hippocampus. Physical activity, mostly assessed in rodents in a paradigm of voluntary wheel running, positively affects adult hippocampal neurogenesis (van Praag et al. 1999a, b; Kronenberg et al. 2003; Snyder et al. 2005; Gould et al. 1999). More cognitive stimuli (such as exposure to an enriched environment or a learning task) increase neurogenesis in both the hippocampus and the olfactory bulb. Olfactory stimuli affected neurogenesis in the olfactory bulb and not the hippocampus (Rochefort et al. 2002). The literature on learning-induced neurogenesis in the adult hippocampus is not without contradictions (Dobrossy et al. 2003; Gould et al. 1999). However, the evidence of a functional connection between learning and neurogenesis has been increasing over the years.

The effects of physical exercise on hippocampal neurogenesis seem to be robust albeit relatively non-specific in that locomotion is not directly based on hippocampal function. As we will argue below, these effects can be best seen as a unspecific activation of “the neurogenic system”, thereby representing a potential that if used appropriately in the context of learning or environmental challenges leads to a stable and lasting improvement of hippocampal function.

On the other hand our increasing knowledge about the correlation of decreased adult hippocampal neurogenesis in the context of brain disease has led to the conclusion that exercise might be a putative way to counterbalance the negative effects of disease on hippocampal function by affecting adult neurogenesis (Kronenberg et al. 2006). We have previously argued that this might be of particular relevance in the context of aging and age-related cognitive impairment (Klempin and Kempermann 2007). Thus, interest in this research is largely driven by the idea to increase neurogenesis with therapeutic goals in mind.

The Regulation of Adult Hippocampal Neurogenesis

Regulation of adult hippocampal neurogenesis occurs on different conceptual levels. What ultimately constitutes “regulation” is the change in the number of new neurons that are generated. But this process might be looked at from the perspective of the individual cell that is produced or from the behavioral level and thus of the animal, in which adult hippocampal neurogenesis takes place. Between these, there are intermediate levels, the systems levels, and there are numerous sub-cellular levels, which are of mechanistic interest (Fig. 1). Researchers in neurobiology tend to focus on the regulation on the level of single cells and below, the cell’s individual genetic program, their responsiveness to growth factors and other extrinsic signals, the regulation of the cell cycle and cell survival machinery, and the transcriptional control of specific developmental programs. In addition, the developing new cells are found in a specific permissive environment, a so-called neurogenic niche, in which neurogenesis takes place and where cell–cell interactions of many types of cells exists and local secreted cytokines act on the local precursor cells (van Praag et al. 1999a; Palmer et al. 2000; Mercier et al. 2002). Behavioral changes affecting adult hippocampal neurogenesis must ultimately lead to changes in the transcriptional machinery governing neuronal development. Conversely, this modulated process of adult neurogenesis should result in behavioral benefits, somehow related to the cognitive representations associated with movements in the environment. Here, studies using single-unit recording already supported the idea that locomotion leads to the activation of well-defined neuronal units in the adult hippocampus compatible with the idea that such neuronal activity might drive the regulation of adult neurogenesis (O’Keefe 1976; Bland and Vanderwolf 1972).
Fig. 1

Multilevel analysis of adult hippocampal neurogenesis. See text for details

Regulation across conceptual levels leads to a complex network of interactions that are further complicated by the fact that obviously many processes involved in the regulation take place at the same time. This makes it necessary to explore each level of regulation with the aim to combine the results in an orchestrated way to finally give an answer to the question how neurogenesis is regulated by “activity”.

General Principles of Regulation: Precursor Cell Expansion and Enhanced Selective Survival

In one regard adult neurogenesis differs substantially from embryonic neurogenesis. The micro-environment for the newborn cells is the permissive niche, but beyond this exceptional neighborhood the brain lacks all embryonic neurogenic qualities. It is not programmed to foster neurogenesis. Also, the massive parallel expansion of precursor cell populations does not exist in the adult brain. Instead, neural precursors of all developmental stages can be found at any given time point and in close vicinity. Thus, the birth of a newborn neuron in the adult dentate gyrus is an individual and not a population event. Work in our laboratory has characterized the cell populations in the adult dentate gyrus (Kempermann et al. 2004). An undifferentiated stem-like precursor cell with radial glial properties gives rise to daughter cells which runs through three consecutive stages of transient amplifying precursor cells (type-2 and type-3 cells), differing by their proliferative potential and degree of neuronal differentiation. The precursor cells finally exit the cell cycle and reach a transient postmitotic stage (stage V) during which network connections are established, the selection for long-term survival occurs and finally stage VI where the terminal differentiation takes place (Fig. 2). The first phase of development until cell cycle exit is compressed into only a few days, and it is remarkable that quantitatively most of the net regulation actually takes place in this initial period and long before the new cells have become fully integrated.
Fig. 2

Stages of neural precursor differentiation in the adult dentate gyrus. A radial glia-like stem cell (Type-1 cell, Nestin+, GFAP+) is progressing (1) to a Nestin+GFAP-intermediate precursor cell (Type-2a cell) that undergoes amplification (2), acquires neural fate (Type-2b cell, Nestin+, Dcx+) and differentiates into a migratory progenitor cell (Type-3 cell, Nestin, Dcx+) finally maturating into a differentiated granular cell. The effects of running are exerted predominantly on the intermediateprogenitor cells (Type 2a/b). GFAP = glial fibrillary acid protein; Dcx = doublecortin

Adult hippocampal neurogenesis is dependent on the proliferation of precursor cells and an expansion of the pool of the rapid proliferating progenitor cells (type-2 and type-3) is the fastest way to increase this potential. In fact, these cells respond to the stimulus of acute physical activity (voluntary wheel running) with increased proliferation (Kronenberg et al. 2003). A similar response is seen in other non-specific activation paradigms, such as serotonergic (Encinas et al. 2006) or glutamatergic (Jessberger et al. 2005) activation. The radial glia-like type-1 cells show no (Kronenberg et al. 2003) or perhaps very little (Steiner et al. 2004) proliferative response to physical exercise. Because they have been shown to increase their rate of division in other contexts (Huttmann et al. 2003; Kunze et al. 2006), it is possible that “activity” quite generally affects their behavior but that their very low baseline level of proliferation (Filippov et al. 2003) makes it difficult to detect the changes.

Whereas cell proliferation increases the pool of cells that might develop into new neurons, the key decision in this process is made by the control of the selective survival of the newborn cells (Kempermann et al. 2003). This process is similar to the situation during embryonic brain development, where a large surplus of neurons is generated, of which only that fraction survives that is recruited into functional circuits. The remainder of the cells is eliminated by apoptosis (Biebl et al. 2000). Consequently, preventing apoptosis, increased adult hippocampal neurogenesis (Kuhn et al. 2005). This survival effect largely takes place on the level of the early postmitotic cells (Brandt et al. 2003; Plumpe et al. 2006). In a study based on about 30 different but related strains of mice, survival effects accounted for 85% of the variation in adult hippocampal neurogenesis, whereas “proliferation” explained only 19% (Kempermann et al. 2006).

As a final but more hypothetical mechanism, the fate choice of an undifferentiated precursor might be shifted toward the neural lineage, thereby increasing neurogenesis. To date, there is very little evidence that this mechanism plays a major role. The early precursor cells of the dentate gyrus have a glial phenotype. The transition from glial to neuronal phenotype takes place on the level of type-2 cells (Steiner et al. 2006). Presumably, sonic hedgehog is needed to maintain the precursor cells in their glial and proliferative state, but this has not been fully elucidated (Pozniak and Pleasure 2006).

In the context of electroconvulsive seizures, “activity” increased sonic hedgehog expression as well as expression of the downstream targets, suggesting that this would be a means by which the fraction of glia-like precursor cells is maintained during expansion (Banerjee et al. 2005). Another example is bone morphogenic proteins (BMPs) and Noggin in the adult subventricular zone. BMPs inhibited adult neurogenesis in the SVZ by directing newborn cells toward glial differentiation. In contrast, the antagonist in this system, Noggin, prevented glial differentiation and promoted neurogenesis (Lim et al. 2000). Such analysis has not yet been extended to the analysis of adult neurogenesis in the context of physical activity.

In addition, we have to make an important distinction between true changes of fate when a given cell would decide between maturing into a neuron rather than a mature glial cell, from this transition from glial to neuronal, which lies in the nature of adult neurogenesis with its glia-like precursor cells.

Adult Hippocampal Neurogenesis and Wheel Running

Physical exercise in rodents robustly increases neurogenesis in the adult hippocampus (van Praag et al. 1999b; Kronenberg et al. 2003), whereas neurogenesis in the adult olfactory bulb did not seem to be influenced by exercise (Brown et al. 2003).

Most investigations of exercise-induced neurogenesis in mice and rats have been performed using the running wheel paradigm. Mice given free access to a running wheel run as much as 3–8 km a night, which has been estimated to reflect their natural physical activity. They produce 2–3 times the number of newborn cells in the SGL of the dentate gyrus, an effect that can be observed as early as 24 h after the start of exercise but the most profound effect can be detected after 3 days of exercise (van Praag et al. 1999a; Kronenberg et al. 2003). Interestingly, forced treadmill training had a similar upregulating effect, but here fewer details are known as yet (Kim et al. 2004; Uda et al. 2006).

In acute settings voluntary wheel running primarily affects the proliferation of transient amplifying precursor cells (types-2 and -3) (Kronenberg et al. 2003, 2006). This acute effect, however, wears off, if the exercise is continued. At that stage an additional survival-promoting effect becomes visible. This effect manifests itself in a persistent increase in late progenitor cells (type-3) and early postmitotic neurons even at the time, when the acute pro-proliferative has already declined again (Kronenberg et al. 2006). On even longer time-scales an increased maintenance of intermediate precursor cells becomes prominent, counteracting the normal age-related decrease in precursor cell proliferation. Presumably, in the absence of additional, more specific stimuli, however, this increased potential is not translated into an increase in net neurogenesis (Kronenberg et al. 2006). We thus propose that both types of activity have to come together to elicit a lasting net effect on adult hippocampal neurogenesis: increased expansion of the precursor cell pool and increased recruitment.

Molecular Mechanism for the Activity-dependent Regulation of Adult Hippocampal Neurogenesis

The dual effects of exercise on adult hippocampal neurogenesis must be reflected in a corresponding complexity on the level of molecular mechanisms mediating this control.

Many factors have been shown to increase in both the circulation and the adult brain after physical exercise. Among these are growth factors released predominantly from muscle tissue during exercise such as vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), and insulin-like growth factor (IGF) (Carro et al. 2000; Gomez-Pinilla et al. 1997; Trejo et al. 2001). VEGF, for example, links exercise and adult hippocampal neurogenesis in that increased levels of VEGF can be measured in exercising individuals and experiments using local perfusion into the brain have shown to increase neurogenesis in the subventricular zone and the dentate gyrus (Schanzer et al. 2004). Intriguingly, the neurogenic region in the dentate gyrus is organized in close association to endothelial cells forming a supportive “vascular niche”(Palmer et al. 2000). The influence of paracrine secreted growth factors such as VEGF in this system seems to serve as a supporting factor for both endothelial cells and NPCs in the vascular niche (Fabel et al. 2003a, b). Peripheral blockade of VEGF by soluble VEGF receptor led to an inhibition of the effect of running on neurogenesis in the adult dentate gyrus (Fabel et al. 2003a). Similar findings have been obtained for IGF1 (Carro et al. 2000; Trejo et al. 2001).

In addition, correlative data and plausibility argue in favor of the idea that within the hippocampus higher levels of the neurotrophin brain-derived neurotrophic factor (BDNF) and serotonin play a major regulatory role in exercise-driven adult hippocampal neurogenesis. However, no causal relationship has yet been proven.

Here, BDNF fits particularly well into concepts of function-dependent regulation of adult hippocampal neurogenesis, because stress and glucocorticoids, which decrease adult hippocampal neurogenesis, also down-regulate BDNF, whereas physical activity up-regulates BDNF and neurogenesis in the hippocampus (Adlard et al. 2005; Johnson and Mitchell 2003).

The fact that exercise is often associated with higher levels of corticosteroids indicates that the link between BDNF and corticoids cannot be fixed. Here exercise appears to override the negative effects of high levels of corticosteroids on BDNF protein expression. Voluntary physical activity may maintain neurogenesis-promoting neurotrophin levels in the brain (Adlard and Cotman 2004; Zheng et al. 2006).

Exercise itself can even be considered a form of stress. Consequently, interaction effects can be observed if more than one stressor, including exercise, is present. The analysis of such interaction effects is complex and has not yet been convincingly achieved. However, it has been proposed that under certain conditions exercise might even increase the susceptibility to glucocorticoid-induced suppression of neurogenesis (Stranahan et al. 2006). The exact contribution of the individual regulatory partners thus has to be determined for each different situation.

With respect to serotonin it has been demonstrated that running increases serotonin levels in the hippocampus (Samorajski et al. 1987; Gomez-Merino et al. 2001; Soares et al. 2007). Another study showed that physical exercise potentiates the upregulation of BDNF seen after the application of antidepressants, which act through the serotonergic pathway (Ivy et al. 2003; Russo-Neustadt et al. 1999).

More Neurons for Learning? Can We Enhance Our Cognitive Function by Exercise?

It seems that quite generally the activation of brain function triggers adult hippocampal neurogenesis. This activation appears rather independent of the nature of the stimulus. A great variety of factors and situations have been shown to up-regulate precursor cell proliferation. We hypothesize that physical activity is the quintessential type of this stimulus. We assume that the activation of adult hippocampal neurogenesis (in the sense of expanding the pool of precursor cells that can get recruited potentially) would be a non-specific response to a new environment, a new task, or anything that might let us initiate a physical response.

More specific, in terms of higher cognitive skills and learning, a likely explanation seems to be that adult neurogenesis is beneficial in situations, in which we need to respond to something new posing a particular challenge to the learning capacities of the hippocampus. We have previously hypothesized that adult neurogenesis might contribute to hippocampal function by allowing to optimize the network in the dentate gyrus to those levels of complexity and novelty previously experienced by the individual (Kempermann 2002). In that sense, adult hippocampal neurogenesis is an investment for the future. By exercise it seems that we enhance the amount of undifferentiated neuronal precursor cells in the hippocampus, possibly expanding the basis for possible upcoming learning challenges.

At the core of this idea lies the assumption that brain activity itself, triggered by locomotion in a new and complex environment, ultimately exerts the effect on adult hippocampal neurogenesis. An alternate hypothesis is that physical exercise simply leads to an increased circulation and release of growth factors from muscle tissue, leading to the activation of neurogenesis. Both ideas are not as mutually exclusive as they might appear. In both cases, a peripheral stimulus is translated to the neurogenic niche and into a pro-proliferative stimulus on the precursor cells. In our model, however, the positive effect of running on adult hippocampal neurogenesis is less of an accidental by-product of peripheral events but an important, evolutionary conserved mechanism of brain–body interaction.


One of the most profound modifiers of neurogenesis in the adult brain is aging. The total number of newborn neurons decreases in a nonlinear hyperbolic way over the lifespan (Kronenberg et al. 2006; Altman and Das 1965; Kuhn et al. 1996; Seki and Arai 1995; Cameron and McKay 1999). Adult hippocampal neurogenesis has a peak in early adulthood (6–8 weeks in mice) quickly decreasing thereafter. It does not seem to go away, however, and even in very old animals a low level of adult hippocampal neurogenesis has been detected; this also seems to apply to humans (Eriksson et al. 1998). One idea has been that chronically elevated levels of corticosteroids have a main influence on the age-dependent decline of adult hippocampal neurogenesis. Convincingly, adrenalectomy restored adult hippocampal neurogenesis to a level similar to younger animals (Cameron and McKay 1999). As it seems, in older animals corticosteroids have stronger effects on the more immature neuronal precursors in the hippocampus since the profile of corticosteroid receptor expression shifts toward this direction (Garcia et al. 2004), possibly increasing the sensitivity of the immature progenitor cells to corticosteroids. We have recently reviewed the topic of adult hippocampal neurogenesis in aging in greater detail (Klempin and Kempermann 2007). Sustained voluntary exercise during a period of 3–9 months of age significantly prevented the age-dependent decline in cell proliferation; however, it did not maintain net neurogenesis on a level similar to younger animals (Kronenberg et al. 2006). Corticosteroids were not studied in that experiments, but overall, physical exercise is often associated with higher rather than lower levels of circulating corticoid hormones (Feduic et al. 2006). Again, only consideration of the receptor expression and the many other regulatory factors will allow elucidating the full mechanism. Recent data suggest that repeated physical exercise may, at least in some cases, diminish detrimental effects of stress on limbic-hypothalamic-pituitary-adrenocortical axis by influencing both the serum levels of corticoids and the expression of their receptors (Zheng et al. 2006; Park et al. 2005).

In addition, effects on the supporting vascular niche seem to play a role as well. Vascular support is decreasing, possibly because glucocorticoid levels alter the endothelial status (Heiss et al. 1996). In addition, glucocorticoids down-regulated VEGF and VEGF receptors in models of tumor angiogenesis and vascular cell interactions during angiogenesis (Heiss et al. 1996; Nauck et al. 1998). Independent of the underlying mechanism, however, we propose that physical activity might contribute to successful aging by increasing the potential for neurogenesis represented by the pool of proliferative precursor cells.

Depressed Neurons: Exercise as an Anti-depressant

The possible link between exercise and adult neurogenesis has also received much attention due to the fact that failing adult hippocampal neurogenesis might be involved in the pathogenesis of major depressive disorder, in the context of exercise see (Ernst et al. 2006; Zheng et al. 2006). Patients with longstanding depression have a tendency to have a smaller hippocampus than healthy controls (Videbech and Ravnkilde 2004). Although not massive this shrinkage is too large to be explained by a lacking contribution of new neurons to hippocampal volume alone, however. Still, the local coincidence is suggestive. In addition, multiple classes of antidepressants increase hippocampal cell proliferation and neurogenesis in a chronic time course (Encinas et al. 2006; Malberg et al. 2000; Lucassen et al. 2004). This increase is again thought to be the result of the effects of increased serotonergic signaling on BDNF (D’Sa and Duman 2002). Moreover, blockade of hippocampal neurogenesis in behavioral models of depression even blocked the actions of selective serotonin re-uptake inhibitor fluoxetine (Santarelli et al. 2003). Clinical studies have demonstrated that serum levels of BDNF in drug-naive patients with major depression are significantly decreased compared to normal controls (Huang et al. 2007).

Physical exercise also acts as an antidepressant (Lawlor and Hopker 2001). To investigate this link and the possible role of adult hippocampal neurogenesis within it, an animal model of depression, the flinders sensitive Line (FSL) of rats, was used. These (at face value) depressed rats indeed showed a reduced level of adult hippocampal neurogenesis. Given access to a running wheel, however, the levels of adult hippocampal neurogenesis normalized (Bjornebekk et al. 2006).


Although the regulation of adult hippocampal neurogenesis by physical activity should not be taken as a safe road to successful aging, we believe it is rather obvious that this intriguing example of activity-dependent plasticity harbors great potential. The reason is that physical activity is easy to obtain and it is only a sedentary society that has estranged us from a rather natural behavior. We argue that the link between locomotion and cognition goes back to our phylogenic roots. Adult neurogenesis might be an exception, not the rule, but the activity-dependent regulation of adult hippocampal neurogenesis might reflect an evolutionary well-conserved and important principle. As such, the impact on research on this topic might go beyond the actual functional contribution of newborn neurons in response to “activity”. This said, exercise might have counterbalancing effects against the loss of physical and cognitive function associated with aging because of the positive association with adult neurogenesis. Whether this link is causal and what, in fact, the exact functional contribution of adult neurogenesis to hippocampal function across the life span needs further investigation.

Copyright information

© Humana Press 2008