Dietary restriction enhances neurotrophin expression and neurogenesis in the hippocampus of adult mice


  • Jaewon Lee,

    1. Laboratory of Neurosciences, National Institute on Aging Gerontology Research Center, Baltimore, Maryland, USA
    2. Department of Anatomy and Neurobiology, University of Kentucky, Lexington, Kentucky, USA
    Search for more papers by this author
  • Kim B. Seroogy,

    1. Department of Anatomy and Neurobiology, University of Kentucky, Lexington, Kentucky, USA
    Search for more papers by this author
  • Mark P. Mattson

    1. Laboratory of Neurosciences, National Institute on Aging Gerontology Research Center, Baltimore, Maryland, USA
    2. Department of Anatomy and Neurobiology, University of Kentucky, Lexington, Kentucky, USA
    3. Department of Neuroscience, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
    Search for more papers by this author

Address correspondence and reprint requests to Mark P. Mattson, National Institute on Aging, GRC 4F01, 5600 Nathan Shock Drive, Baltimore, MD 21224, USA. E-mail:


The adult brain contains small populations of neural precursor cells (NPC) that can give rise to new neurons and glia, and may play important roles in learning and memory, and recovery from injury. Growth factors can influence the proliferation, differentiation and survival of NPC, and may mediate responses of NPC to injury and environmental stimuli such as enriched environments and physical activity. We now report that neurotrophin expression and neurogenesis can be modified by a change in diet. When adult mice are maintained on a dietary restriction (DR) feeding regimen, numbers of newly generated cells in the dentate gyrus of the hippocampus are increased, apparently as the result of increased cell survival. The new cells exhibit phenotypes of neurons and astrocytes. Levels of expression of brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT-3) are increased by DR, while levels of expression of high-affinity receptors for these neurotrophins (trkB and trkC) are unchanged. In addition, DR increases the ratio of full-length trkB to truncated trkB in the hippocampus. The ability of a change in diet to stimulate neurotrophin expression and enhance neurogenesis has important implications for dietary modification of neuroplasticity and responses of the brain to injury and disease.


AL, ad libitum


brain-derived neurotrophic factor




dietary restriction


glial fibrillary acidic protein


microtubule-associated protein-2


neural precursor cells




Tris-buffered saline.

The brain of adult mammals, including humans, contains populations of cells that can divide and differentiate into neurons and glia (Gage 2000). These neural precursor cells (NPC) are present in the subventricular zone and in the dentate gyrus of the hippocampus, Neurogenesis may allow the brain to respond to environmental demands such as increased intellectual challenge and brain injury. Indeed, studies of rodents have shown that the proliferation of NPC is reduced in association with age-related cognitive decline (Kuhn et al. 1996), and that suppression of NPC proliferation can impair learning and memory (Shors et al. 2001). In addition, ischemic and excitotoxic brain injuries (Parent et al. 1997; Liu et al. 1998), exposure to enriched environments (Kempermann et al. 1997; Nilsson et al. 1999; Young et al. 1999) and physical activity (van Praag et al. 1999) can increase the production and/or survival of new neural cells in␣the dentate gyrus of the hippocampus. The signaling mechanisms that mediate the effects of environmental stimuli on NPC proliferation, differentiation and survival are not yet established, but appear to involve neurotrophic factors (Cameron et al. 1998). Growth factors that have been shown to affect NPC include basic fibroblast growth factor, epidermal growth factor, and members of the neurotrophin family including brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT-3). Data suggest that BDNF and NT-3 can affect the proliferation, differentiation and/or survival of NPC from different brain regions including the subventricular zone and hippocampus (Vicario-Abejon et al. 1995; Lachyankar et al. 1997; Shetty and Turner 1998; Zigova et al. 1998; Benoit et al. 2001).

The impact of diet on brain function and susceptibility to␣neuropsychiatric and neurodegenerative disorders is increasingly appreciated (Young 1993). Dietary restriction (DR) can increase the lifespan of rodents and may ward off many different age-related diseases (Sohal and Weindruch 1996) including neurodegenerative disorders (Mattson 2000). Increasing numbers of reports have documented ‘antiaging’ effects of DR on the brain. Epidemiological data suggest that individuals with a low calorie intake are at reduced risk for Parkinson's (Logroscino et al. 1996) and Alzheimer's (Mayeux et al. 1999) diseases. In addition, rodents maintained on DR perform better on learning and memory tasks than do rats fed ad libitum (Idrobo et al. 1987; Ingram et al. 1987; Stewart et al. 1989). DR increases the resistance of neurons to degeneration and improves behavioral outcome in experimental animal models of Alzheimer's disease (Bruce-Keller et al. 1999; Zhu et al. 1999), Parkinson's disease (Duan and Mattson 1999), Huntington's disease (Bruce-Keller et al. 1999) and stroke (Yu and Mattson 1999). It was recently reported that levels of BDNF are increased in the hippocampus and cerebral cortex of rats maintained on a dietary restriction feeding regimen (Lee et al. 2000; Duan et al. 2001). Here we show that DR can enhance neurogenesis in the hippocampus of adult mice, and that this effect of DR on NPC is associated with increased production of BDNF and NT-3. Our findings suggest a contribution of enhanced neurogenesis to the beneficial effects of DR on hippocampal plasticity and resistance to neurodegenerative disorders.

Materials and methods

Mice, diets and BrdU administration

Fifty-six adult (8 weeks old) male C57BL/6 mice obtained from the National Cancer Institute were maintained under temperature- and light-controlled conditions (20–23°C, 12-h light/12-h dark cycle). Mice were divided into two groups (28 mice/group), an ad libitum (AL) group which had continual access to food, and a DR group which was provided food on alternate days. Previous studies have shown that rats and mice maintained on such an alternate day feeding schedule will consume less calories over time and live longer than animals fed AL (Goodrick et al. 1983). For evaluations of neurogenesis, 12 mice in each group were given a daily intraperitoneal injection of bromodeoxyuridine (BrdU; 50 mg/kg body weight) for 12 days. Half of the mice in each diet group were killed 1 day after the last BrdU injection and half were killed 1 month after the last BrdU injection. The remaining 16 mice in each diet group were processed for analyses of neurotrophin and neurotrophin receptor expression as described below. All procedures complied with National Institutes of Health guidelines and were approved by the Institutional Animal Care and Use Committee.

Quantification of newly produced neural cells

Mice were perfused transcardially with 4% paraformaldehyde and their brains were removed, postfixed at 4°C overnight, and transferred to a 30% sucrose solution. The cryoprotected brains were sectioned serially at 40 µm in the coronal plane using a freezing microtome. Every section which contained the hippocampal formation was saved. The protocol for immunostaining of brain sections with BrdU antibody was similar to that described previously (Nilsson et al. 1999). Briefly, free floating sections were treated with 0.6% H2O2 in Tris-buffered saline (TBS; pH 7.5) to block endogenous peroxidases, and DNA was denatured by exposing sections sequentially to heat, acid and base. The sections were incubated in TBS/0.1% Triton X-100/3% goat serum (TBS-TS) for 30 min, and incubated with primary anti-BrdU antibody (rat monoclonal, 1 : 400; Accurate Chemicals, Westbury, NY, USA) in TBS-TS overnight at 4°C. Sections were further processed using a biotinylated secondary goat anti-rat IgG antibody (Vector Laboratories, Burlingame, CA, USA, 1 : 200), avidin–peroxidase complex, and diaminobenzidine. Stained sections were mounted onto slides and counter-stained with cresyl violet to measure granule cell layer volume.

The total number of BrdU-positive cells in the dentate gyrus of each mouse was estimated using the optical fractionator technique (West 1993) assisted by a computer-based system, StereologerTM (SPA, Alexandria, VA, USA) using methods similar to those described previously (Long et al. 1998). Estimates of region volume were assessed using the Cavalieri point counting method (Gundersen and Jensen 1987). Cells in every sixth section throughout the entire rostro-caudal extent of the hippocampus were counted: the reference space consisted of the granular cell layer of the dentate gyrus. For each section, the reference space was delineated by outlining at low power (10 × objective; on-screen magnification = 271 ×); identification of BrdU-positive cells was accomplished at high power (63 × objective; on-screen magnification = 1714 ×). The dimension of the sampling frames were 92.2 µm in length by 92.2 µm in width and 12 µm in depth. The guard height for each section was 1 µm. The optical fractionator technique estimates the number of cells by multiplying the sum of cells counted by the reciprocal of the fraction of the region sampled. Volume densities were calculated by dividing the number of BrdU positive cells counted by the total volume sampled of the reference space. The volume of the sampled reference space was the number of disectors multiplied by the volume of one disector. All cell counts were performed by the same investigator (JL) blind to the group identification of each section.


Immunostaining for confocal analysis was performed on 40 µm coronal brain sections as follows. Sections were incubated for 1 h in a solution containing 2.5% normal horse serum, 2.5% normal goat serum, and 0.1% Triton X-100 in TBS. Primary antibodies were then added and the cultures were incubated overnight at 4°C. The primary antibodies used were a rat monoclonal antibody against BrdU (Accurate Chemicals, 1 : 200 dilution), rabbit polyclonal antibody against GFAP (Sigma, St Louis, MO, USA, 1 : 500 dilution) and a mouse monoclonal antibody against the neuron-specific nuclear antigen NeuN (Chemicon, Temecula, CA, USA, 1 : 500 dilution) and mature neuron-specific cytoskeletal antigen MAP2ab (Chemicon, 1 : 500 dilution). Cultures were then washed with TBS and incubated for 1 h in the presence of␣anti-rat IgG labeled with AlexaFluor-488, anti-rabbit IgG labeled with AlexaFluor-633 and anti-mouse IgG labeled with AlexaFluor-543 (Molecular Probes, Eugene, OR, USA; 1: 2000 dilution). Confocal images were acquired using a Zeiss 510 CSLM microscope.

In situ hybridization

Adjacent, coronal sections through the hippocampus were processed for the in situ hybridization detection of BDNF, NT-3, trkB, and trkC mRNAs by using 35S-labeled cRNA probes as described previously (Seroogy et al. 1994; Seroogy and Herman 1997; Numan and Seroogy 1999). Briefly, the slide-mounted sections were brought to room temperature, placed in 4% paraformaldehyde for 10 min, and washed sequentially in 0.1 m phosphate buffer (PB), 0.1␣m PB/0.2% glycine, and 0.25% acetic anhydride in 0.1 m triethanolamine. The sections were then dehydrated with increasing concentrations of ethanol, delipidated in chloroform, and air-dried. Sections were hybridized at 60°C overnight in a hybridization solution consisting of 50% formamide, 10% dextran sulfate, 1 × Denhardt's solution, 0.15 mg/mL yeast tRNA, 0.33 mg/mL denatured salmon sperm DNA, 40 mm dithiothreitol, 1 mm EDTA, 20 mm Tris-HCl and the 35S-labeled cRNA probe at a concentration of 1.0 × 106 cpm/50 µL/slide. Both sense and antisense cRNA probes for each neurotrophin and trk receptor were prepared by in vitro transcription using linearized DNA constructs in the presence of RNA polymerase (T3, T7 or SP6) and [35S]UTP (New England Nuclear; Boston, MA, USA). BDNF and NT-3 cDNA constructs (generous gifts from C. Gall and J. Lauterborn, University of California at Irvine) resulted in antisense transcripts that were 540 and 550 bases long, respectively. The cDNA constructs for trkB and trkC (kindly supplied by D. McKinnon, State University of New York at Stony Brook) resulted in antisense RNA transcripts that were 196 and 300 bases long, respectively. The trkB cRNA probe detects only the kinase-specific, full-length catalytic form of the receptor mRNA (Klein et al. 1990; Middlemas et al. 1991; Sternini et al. 1996), whereas the trkC cRNA probe recognizes mRNA transcripts for both the catalytic and non-catalytic isoforms of the receptor (Valenzuela et al. 1993; Dixon and McKinnon 1994; Albers et al. 1996). For posthybridization treatment, sections were washed several times in 4 x saline sodium citrate buffer (SSC; 1 × SSC = 0.15 m sodium chloride, 0.015 m sodium citrate, pH 7.0) containing 10 mm sodium thiosulfate, at␣37°C. The sections were then incubated in ribonuclease A (0.05 mg/mL) for 30 min at 45°C, followed by several washes in decreasing concentrations of SSC (2 ×, 0.5 × and 0.1 ×) at 37°C. The sections were then briefly rinsed in dH20, dipped in 95% ethanol, and air-dried. To generate film autoradiograms the sections were exposed to β-Max Hyperfilm (Amersham; Arlington Heights, IL, USA) for 11 days (BDNF and NT-3) or 7 days (trkB and trkC). In control procedures, prehybridization treatment of tissue with ribonuclease A (0.05 mg/mL; 45°C for 30 min), processing tissue with 35S-labeled sense strand transcripts for each probe, and processing tissue with no probe at all (positive chemography control), resulted in no specific hybridization signal. Film autoradiograms were analyzed with NIH Image public domain software (Image 1.62) to compare the densities of hybridization (mean corrected gray level) of each probe in various hippocampal subfields (dentate gyrus, CA1 and CA3) and in parietal cortex in each treatment paradigm. We did not attempt to quantify hybridization levels in specific subpopulations of cortical neurons; the analysis was made on the entire thickness of the cortex to provide a measure of overall levels of mRNA in the cortex. At least six measurements were taken for each probe from each animal. Statistical analysis included Student's unpaired t-test, and analysis of variance (anova) followed by Fisher's protected least significant differences procedure where appropriate. The NIH image software was also used to acquire images of representative sections from film autoradiograms.

Immunoblot analyses

Hippocampal and cerebral cortical tissues were homogenized in a sample buffer (62 mm Tris, 2 mm EDTA, 2 mm EGTA, 2% SDS, 10% glycerol and a protease inhibitor cocktail; pH 6.0). Solubilized proteins were separated by electrophoresis on a 7.5% SDS–acrylamide gel, and transferred to a nitrocellulose sheet. Following incubation of the membrane in blocking solution (5% non-fat milk in TTBS), the membrane was incubated overnight at 4°C in TTBS containing a primary antibody. The primary antibody was a rabbit polyclonal antibody that recognizes both full-length and truncated forms of trkB (Santa Cruz Biotechnology, Santa Cruz, CA, USA; 1 : 500 dilution). The membrane was then incubated for 1 h in TTBS containing HRP-conjugated secondary antibody (1 : 3000; Jackson Immunological Research Laborites Inc., West Grove, PA, USA) and immunoreactive proteins were visualized using a chemiluminescence-based detection kit according to the manufacture's protocol (ECL kit; Amersham Corp., Arlington Heights, IL, USA). Bands were quantified by densitometric scanning.


Mice were maintained on either an ad libitum diet (AL) or a dietary restriction (DR) feeding regimen in which they were fed every other day; their body weights after 3 months on the diets were: AL, 32.9 ± 0.46 g; DR, 27.2 ± 0.35 g (n = 24; p␣< 0.0001; paired t-test). Mice were then administered bromodeoxyuridine (BrdU) and killed either 1 day or 4 weeks later. BrdU-immunoreactive cells in the dentate gyrus of the hippocampus were quantified using unbiased stereological methods. At the 1 day time point the numbers of BrdU positive cells in the dentate gyrus were not significantly different in groups AL and DR (Figs 1a,b; Table 1). At the 4-week time point there were significantly more BrdU-positive cells in the dentate gyrus of mice in group DR compared to rats in the control group (Figs 1c,d; Table 1). The volume of the dentate gyrus was not significantly different in mice that had been maintained on AL and DR diets (data not shown).

Figure 1.

Dietary restriction increases neurogenesis in the dentate gyrus of adult mice. Photomicrographs of BrdU-positive cells in the dentate gyrus 1 day (a and b) and 4 weeks (c and d) after BrdU administration in mice that had been maintained fed on either ad libitum (a and c) or dietary restriction (b and d) for 3 months prior to␣BrdU administration. Quantification of labeled cells is shown in Table 1.

Table 1.  Proliferation, survival and survival rate of cells in the dentate gyrus of mice fed ad libitum in comparison with mice maintained on dietary restriction
 Ad libitumDietary restriction
  1. All mice received BrdU (50 mg per kg) for 12 days. Cell proliferation was assessed on 1 day after last injection. Survival of BrdU-labeled cells in the dentate gyrus were determined 4 weeks after last injection (n = 6 per group). All data presented as means ± standard error. *Significantly different from ad libitum group (p < 0.02).

Proliferation, 1 day3579 ± 222.33188 ± 116.8
Survival, 4 weeks992 ± 113.51404 ± 73.7*
Survival (%), 4 weeks28 ± 3.244 ± 2.3*

In order to determine the phenotypes of the newly generated cells, we performed either triple or double label confocal immunohistochemical analysis of hippocampi using antibodies against the astrocyte protein (GFAP) and the mature neuron-specific protein (NeuN or MAP2ab), in combination with the BrdU antibody. At one day after BrdU administration the vast majority of BrdU-positive cells were confined to the subgranular zone of the dentate gyrus and were not immunoreactive with either the GFAP or NeuN antibodies (Figs 2a and b). At 4 weeks after BrdU administration, BrdU-positive cells were scattered throughout the dentate gyrus. Essentially all BrdU-positive cells that were located in the granule cell layer were also NeuN and MAP2ab positive (Figs 2c, d, e, g and h, arrow). BrdU-positive cells located in the molecular layers of the dentate gyrus were mostly GFAP-positive (Fig. 2f, arrowhead), although some cells in the molecular layers were MAP2ab-positive (Fig. 2h).

Figure 2.

Confocal images documenting the phenotypes of newly generated cells in the dentate gyrus of mice that had been maintained on a dietary restriction feeding regimen. Sections were triple-labeled with either antibodies against BrdU (green, newborn cells), GFAP (white, astrocyte antigen) and NeuN (red, neuronal antigen) (a–d), or BrdU (green), GFAP (white) and MAP2ab (red) (g and h). (f) shows the molecular layer of the dentate gyrus in a section double labeled with antibodies against BrdU (green) and GFAP (red) (the arrowhead points to a double-labeled cell). Most of the BrdU-positive cells in the dentate gyrus did not exhibit a neuronal or astrocyte antigen 1 day after BrdU administration (a and b). Four weeks after BrdU administration, the majority of cells labeled with BrdU were also immunoreactive with the NeuN antibody (arrows in c and e). The neuronal phenotype of newborn cells on 4 weeks after BrdU administration was confirmed by immunofluorescence using antibodies to another neuronal marker MAP2, which labeled cell bodies and dendrites of granule neurons (arrows in g and h).

Previous studies have shown that environmental stimuli that increase neurogenesis in the dentate gyrus of the hippocampus also stimulate expression of the neurotrophins BDNF and NT-3 (Ballarin et al. 1991; Lee et al. 1997; Young et al. 1999). We therefore performed in situ hybridization analyses of brain sections from AL and DR mice to determine whether DR affects the expression of BDNF, NT-3 and/or their high-affinity receptors trkB and trkC. Examination of the pseudocolor densitometric images of autoradiograms revealed that the overall cellular pattern of expression of BDNF, NT-3, trkB and trkC was unchanged in mice that had been maintained on DR (Fig. 3). BDNF and trkB were expressed in pyramidal neurons in all regions of the hippocampus and in dentate granule cells. The BDNF mRNA hybridization signal appeared to be increased in CA1 and CA3 pyramidal neurons. NT-3 expression was confined to dentate granule cells and a small population of pyramidal neurons in region CA2, while trkC was expressed in all pyramidal neurons in all regions of the hippocampus and in dentate granule cells. The pattern of hybridization with each of the probes also appeared similar in the cerebral cortex of AL and DR mice (Fig. 3). Quantitative comparisons of levels of mRNAs encoding BDNF and NT-3 revealed significant effects of DR. Levels of BDNF mRNA were significantly increased by approximately 20% in CA1 and CA3 pyramidal neurons in hippocampi of DR mice compared to AL mice (Fig. 4). Levels of BDNF mRNA in dentate granule cells and cerebral cortical cells were unaffected by DR. Levels of NT-3 mRNA were significantly increased by approximately 30% in dentate granule cells of DR mice compared to AL mice (Fig. 4). There were no significant differences in trkB or trkC mRNA levels in hippocampal pyramidal cells, dentate granule cells or cortical cells in AL and DR mice.

Figure 3.

Pseudocolor densitometric images of film autoradiograms showing hybridization levels of BDNF, NT-3, trkB and trkC mRNAs in hippocampi from a control mouse fed ad-libitum (AL) and a mouse that had been maintained for 3 months on dietary restriction (DR). Arrows indicate regions of increased mRNA expression in the DR mice.

Figure 4.

Dietary restriction increases BDNF mRNA levels in CA1-3 pyramidal cells and NT-3 levels in dentate gyrus granule cells. Densitometric analyses of hybridization levels of BDNF, NT-3, trkB and trkC mRNAs were performed on autoradiograms of hippocampal brain sections from mice that had been maintained for 3 months on AL or DR diets. Values are the mean and SE of determinations made in six mice per group. *p < 0.02, **p < 0.01 compared to corresponding value for mice fed AL [anova with Fisher's protected least significant difference procedure (PLSD)].

The BDNF receptor trkB exists in cells in full-length and truncated forms; the full-length form is a functional receptor tyrosine kinase, while the truncated form may serve to negatively regulate trkB by sequestering BDNF. In order for the DR-induced increase of BDNF expression to enhance BDNF signaling in target neurons, it is essential that levels of functional trkB are maintained. We therefore determined relative levels of full-length and truncated trkB in hippocampal and cortical tissue from AL and DR mice (Fig. 5a). The ratio of full-length trkB to truncated trkB was significantly increased by approximately 25% in the hippocampus of DR mice compared to AL mice (Fig. 5b). Levels of full-length and truncated trkB in the cerebral cortex were not different in AL and DR mice.

Figure 5.

The ratio of full-length to truncated trkB is increased in the hippocampus of mice maintained on DR. (a) Proteins in homogenates of hippocampus and cerebral cortex from AL and DR mice were subjected to immunoblot analysis using antibodies that selectively recognize either full-length (145 kDa) or truncated (95 kDa) trkB. Each lane was loaded with a sample (50 µg) from a different mouse. (b)␣Densitometric analyses of full-length and truncated trkB levels were performed on immunoblots. Values are the mean and SE of determinations made in six mice per group. **p < 0.01 compared to corresponding AL value (anova with PLSD test).


The present findings establish an effect of diet on the expression of neurotrophins and on neurogenesis in the hippocampus of adult mice. DR had no significant effect on the proliferation of NPC; instead, DR enhanced neurogenesis by increasing the survival of newly generated cells. Some of the newly generated cells were localized to the dentate granule cell layer and expressed NeuN suggesting that they had differentiated into granule neurons, in agreement with previous reports that the majority of newly generated cells in the dentate gyrus migrate into the granule cell layer and display neuron-like properties (Parent et al. 1997; Young et al. 1999). A previous study showed that mice maintained in an enriched environment exhibit increased neurogenesis in the dentate gyrus, and a significant increase in the total number of dentate granule neurons, compared with littermates housed in standard cages (Kempermann et al. 1997). Another study showed that rats raised in an enriched environment exhibit increased neurogenesis (Nilsson et al. 1999). Similar to the effect of DR, the enriched environment did not increase NPC proliferation but did increase survival of the NPC progeny. This suggests that environmental enrichment and DR share a common mechanism of action in increasing the number of newly generated dentate cells. The progeny of many NPC may undergo apoptosis, as indicated by a decrease in the number of BrdU-positive cells with increasing time postlabeling. Consistent with this interpretation, environmental enrichment reduces spontaneous death of newly generated neural cells in the hippocampus (Young et al. 1999). We did not observe a significant effect of DR on the volume of the dentate gyrus, despite a significant increase in the survival of BrdU-labeled cells. This result is similar to that observed in animals maintained for several months in an enriched environment (van Praag et al. 1999). Perhaps a more prolonged period of DR would result in a measurable increase in hippocampal volume. On the other hand, it is possible that DR also affects the rate of loss of granule neurons or their size, or DR might affect gliogenesis or activation of glial cells (activated astrocytes increase in size). Indeed, it has been reported that DR can reduce activation of astrocytes (Major et al. 1997).

Our findings suggest a role for neurotrophins in mediating the positive effects of DR on neurogenesis. The expression of BDNF was increased in CA1 and CA3 pyramidal neurons, and the expression of NT-3 was increased in granule neurons of the dentate gyrus, in mice maintained on DR. The ratio of full-length to truncated trkB was increased in hippocampus which, in the presence of increased BDNF, would be expected to result in increased signaling via trkB. BDNF can promote the survival and differentiation of hippocampal NPC in culture (Lowenstein and Arsenault 1996; Shetty and Turner 1998) and of newly generated embryonic hippocampal and cortical neurons (Cheng and Mattson 1994; Mattson et al. 1995; Cheng et al. 1997; Hetman et al. 1999). Further evidence that BDNF mediates the effects of DR on neurogenesis comes from studies showing that stimuli that increase neurogenesis in the dentate gyrus also increase BDNF expression including seizure activity (Parent et al. 1997; Lowenstein and Arsenault 1996; Lee et al. 1997), ischemia (Lindvall et al. 1992; Liu et al. 1998) and an enriched environment (Cameron et al. 1998; Young et al. 1999). A study of primary hippocampal progenitor cells in culture showed that NT-3 and BDNF promote neuronal differentiation as indicated by increased expression of glutamate receptors (Sah et al. 1997). NT-3 promotes the maturation of the immature cells in the embryonic striatum into neurons that produce one or more neurotransmitters and (Vicario-Abejon et al. 1995), and also promotes differentiation of adult hippocampal NPC (Takahashi et al. 1999). Studies of mice lacking NT-3 revealed a requirement for this neurotrophin for the survival of certain populations of NPC and their neuronal and glial progeny (El Shamy et al. 1998; Kahn et al. 1999).

It has been reported that the proliferation of NPC is rapidly increased in response to kainic acid-induced seizures and ischemic stroke in the adult brain; these insults also increase the production of several different neurotrophic factors and␣cytokines including BDNF (Mattson and Lindvall 1997). Under basal conditions, NT-3 is expressed in the hippocampus predominately in dentate granule neurons and CA2 pyramidal neurons, and is down-regulated in dentate neurons following seizures (Rocamora et al. 1992; Lowenstein and Arsenault 1996). In contrast, we found that NT-3 expression was increased in dentate neurons in response to DR. Because BDNF and NT-3 can promote differentiation and survival of granule neurons, the DR-induced increases in BDNF and NT-3 levels may enhance the production of new granule neurons.

The mechanism whereby DR up-regulates BDNF and NT-3 production and enhances neurogenesis is not known. One possibility is that DR induces a mild metabolic stress response in neurons, a possibility supported by data showing that levels of the stress protein chaperones HSP-70 and GRP-78 are increased in neurons in the brains of rats and␣mice maintained on DR (Duan and Mattson 1999; Yu␣and Mattson 1999), and that more severe metabolic (ischemic) stress can induce neurogenesis (Liu et al. 1998). On the other␣hand, it is reasonable to consider that the effects of DR on neurotrophin expression and neurogenesis in the hippocampus are secondary to an effect of DR on behavior. For example, if the DR mice are more active such that they receive more exercise and more environmental exposure, this might account for our findings. Indeed, previous studies have suggested that motor activity is increased in rodents maintained on DR (Duffy et al. 1997), and enriched environments can induce increased expression of BDNF and NT-3 in the hippocampus (Torasdotter et al. 1996). However, increased physical activity is unlikely to account for the effect of DR documented in the present study because exercise increases both the proliferation and survival of NPC (van Praag et al. 1999), whereas DR increases survival only. Identification of the specific mechanism of action of DR on neurotrophic signaling pathways will require considerable further work.

Neurogenesis in the adult brain may be important for maintenance of certain neuronal populations in the face of continual cell loss, and may play critical roles in learning and memory (Shors et al. 2001) and the brain's response to injury (Parent et al. 1997; Liu et al. 1998). In this regard it is of considerable interest that DR can increase lifespan, can enhance learning and memory (Sohal and Weindruch 1996; Ingram et al. 1987), and can improve outcome following brain injury (Duan and Mattson 1999; Yu and Mattson 1999). Enhanced neurotrophin signaling may be the mechanism underlying these different beneficial effects of DR on the brain. Indeed, data from in vivo studies and analyses of synaptic plasticity in hippocampal slices have shown that BDNF plays an important role in learning and memory (Levine et al. 1995; Minichiello et al. 1999). The ability of a change in diet to affect neurotrophin expression and neurogenesis therefore has important implications for brain function in humans. For example, it may be possible to␣establish dietary regimens that enhance learning and memory, increase resistance of neurons to neurodegenerative conditions, and improve outcome following brain injury. In this view, brain healthspan might be increased through dietary manipulations.


We thank K. Lundgren for technical assistance, and M. Rao and D.␣Ingram for valuable discussions. Supported by the NIA and a␣grant to KBS from the NINDS (NS39128).