Address correspondence and reprint requests to Daniel M. Michaelson, Department of Neurobiochemistry, The George S. Wise Faculty of Life Sciences, Tel Aviv University, Tel Aviv 69978, Israel. E-mail: email@example.com
Neurodegeneration in Alzheimer's disease (AD) is associated with the activation of neurogenesis. The mechanisms underlying this crosstalk between neuronal death and birth and the extent to which it is affected by genetic risk factors of AD are not known. We employed transgenic mice expressing human apolipoprotein E4 (apoE4), the most prevalent genetic risk factor for AD, or expressing human apoE3 (an AD-benign allele), in order to examine the hypothesis that apoE4 tilts the balance between neurogenesis and neuronal cell death in favor of the latter. The results showed an isoform-specific increase in neurogenesis in the hippocampal dentate gyrus (DG) under standard conditions in apoE4-transgenic mice. Environmental stimulation, which increases neurogenesis in the DG of apoE3-transgenic and wild-type mice, had the opposite effect on the apoE4 mice, where it triggered apoptosis while decreasing hippocampal neurogenesis. These effects were specific to the DG and were not observed in the subventricular zone, where neurogenesis was unaffected by either the apoE genotype or the environmental conditions. These in vivo findings demonstrate a linkage between neuronal apoptosis and the impaired neuronal plasticity and cognition of apoE4-transgenic mice, and suggest that similar interactions between apoE4 and environmental factors might occur in AD.
The above findings prompted us to postulate that the cognitive and synaptic deficits of environmentally stimulated apoE4 mice result from impairment of the balance between neurogenesis and apoptosis. Here, we examined this hypothesis by housing young transgenic mice expressing human apoE4 or human apoE3 on a null mouse apoE background, as well as wild-type and apoE-deficient mice, in an enriched environment, and investigating the resulting effects on neurogenesis and neuronal death. We found that neurogenesis in the hippocampus of apoE4 mice is increased compared with apoE3 mice under standard housing conditions, and that environmental enrichment, which stimulates neurogenesis in apoE3 mice, triggers apoptosis in apoE4 mice.
Materials and methods
Human apoE3- and apoE4-transgenic mice were generated on an apoE-deficient C57BL/6J background by the use of human apoE3 and apoE4 genomic constructs, as previously described (Xu et al. 1996). The experiments were performed with the apoE3-453 and apoE4-81 lineages, which express similar levels of brain apoE (Levi et al. 2003). The apoE-transgenic mice were back bred with genetically homogeneous apoE-deficient mice (The Jackson Laboratory, Bar Harbor, ME, USA; catalogue no. N10JAX) for more than 10 generations and were heterozygous for the human apoE transgene and homozygous for mouse apoE deficiency. The apoE genotype of the mice was confirmed by PCR analysis as previously described (Levi et al. 2003). Wild-type C57BL/6J mice were used as controls.
Three-week-old male apoE3- and apoE4-transgenic mice and corresponding wild-type and apoE-deficient mice were housed for 24 weeks in standard cages or in environmentally enriched cages (n = 5 males per cage and per mouse group in each environment). The environmentally enriched cages were equipped with a running wheel for spontaneous exercise, a three-dimensional labyrinth, bedding, a ladder, a house, chains, and wooden blocks, as described (van Praag et al. 2000).
Immunohistochemistry and immunofluorescence staining
Mice were anesthetized with ketamine and xylazine and perfused transcardially with phosphate-buffered saline (PBS) and then with 4% paraformaldehyde in 0.1 m phosphate buffer, pH 7.4. Their brains were removed, fixed overnight in 4% paraformaldehyde in PBS, and placed in 30% sucrose for 48 h. Frozen coronal sections (25 μm) were cut on a sliding microtome and collected serially. The free-floating sections were immunostained as described (Matsumori et al. 2005), with the following primary antibodies: mouse anti-neuronal nuclei (anti-NeuN) monoclonal antibody (1 : 500; Chemicon, Temecula, CA, USA; Gil et al. 2005), mouse anti-nestin monoclonal antibody (1 : 200; Chemicon), goat anti-doublecortin (DCX) antiserum (DCX C-18, 1 : 200 and DCX N-19 which are directed, respectively, at the C and N terminals of DCX, 1 : 200; Santa Cruz Biotechnology, Santa Cruz, CA, USA; Kronenberg et al. 2003), goat anti-Ki-67 (1 : 200; Santa Cruz Biotechnology; Gaulke et al. 2005), and rabbit anti-cleaved caspase 3 antiserum (1 : 200, Cell Signaling Technology, Beverly, MA, USA; Knuesel et al. 2005). The bound primary Abs were visualized with the aid of the appropriate biotinylated second Abs and peroxidase-conjugated avidin (ABC Elite Kit; Vector Laboratories, Burlingame, CA, USA). The peroxidase-immunostained sections were viewed and photographed with a 60 × objective and a Nikon DS-5M camera (Nikon Instech, Tokyo, Japan). The intensities of immunohistochemical staining and the numbers of stained cells in the indicated brain areas were determined with the aid of the Image-Pro Plus system (version 5.1; Media Cybernetics, Silver Spring, MD, USA). Cells were quantified on three randomly selected and coded coronal sections of the DG for each mouse. DCX as well as ki-67 and nestin-stained cells, whose distribution in the DG was relatively sparse, were counted individually by the program, whereas the more densely packed cells labeled with NeuN and cleaved caspase 3 were quantified by placing a rectangular cursor (250 × 250 μm) over the granular layer of the DG and calculating the percentage of the area that was stained.
Double immunofluorescence staining
Co-localization of DCX and nestin was evaluated by double immunostaining with fluorescent chromogens, as described (Kronenberg et al. 2003). In brief, sections were first blocked by incubation with 0.1% Triton X-100 and 5% normal serum in PBS for 15 min at 23°C. Two primary antibodies (goat-anti DCX, 1 : 200, and mouse anti-nestin; 1 : 200) were then dissolved in the blocking solution and incubated with the sections for 24 h at 4°C. The bound primary monoclonal antibodies were visualized by incubation of the sections with Cy2-conjugated donkey anti-goat (for DCX) for 1 h at 23°C, and then (for nestin) with horse anti-mouse biotinylated secondary antibody coupled to streptavidin Cy3 (1 : 1000; Jackson Immuno-Research Laboratories, West Grove, PA, USA). The sections were then mounted on dry gelatin-coated slides, and fluorescence was visualized using a 60 × objective with a Nikon TE2000-S microscope (Nikon Instech) equipped with a CoolSnap HQ camera (Photometrices, Tucson, AZ, USA). Stained cells in the DG were counted with the aid of Image-Pro Plus. For co-location, staining was analyzed using a confocal scanning laser microscope (LSM 510; Zeiss, Oberkochen, Germany). Digital images were processed with Adobe Photoshop 7.0. Cells in DG-containing fields were quantified in three randomly selected and coded coronal sections for each mouse. Other than making general adjustments for color, contrast, and brightness, the images were not manipulated.
Terminal deoxynucleotidyl transferase – mediated dUTP nick-end labeling (TUNEL) was performed according to the manufacturer's instructions (apoptosis kit; MBL, Woburn, MA, USA). In brief, prefixed 25-μm coronal sections were rinsed in PBS, then quenched in 3% H2O2, rinsed in PBS, and incubated with terminal deoxynucleotidyl transferase (TdT) at 37°C for 1 h in a humid chamber. The slices were then washed and incubated with anti-digoxigenin peroxidase conjugate at 23°C for 30 min, and labeling was visualized by application of diaminobenzidine dihydrochloride (DAB; Young et al. 1999). Three randomly selected and coded coronal sections of the hippocampus were quantified for each mouse by placing a rectangular cursor (250 × 250 μm) over the granular cell layer of the DG and calculating the percentage of the area that was stained.
All numerical analyses were performed using SPSS version 12 (SAS Institute, Cary, NC, USA). The four groups in each of the two environments were compared by anova (utilizing a 4 × 2 design) and further analyzed, where appropriate, by Fisher's post-hoc test.
Neurogenesis was monitored immunohistochemically, using the microtubule-associated protein DCX, which is expressed transiently in newly formed neurons (Brown et al. 2003), as a marker. Figure 1(a) depicts representative DCX-stained hippocampal sections obtained from apoE3- and apoE4-transgenic mice that were exposed to standard or enriched environments. As shown, the DG hippocampal subfield contains DCX-positive cells, whose perikarya are in the subgranular zone and whose neurites project into the DG. Under standard housing conditions, more of these DCX-positive neurons were seen in the DG of the apoE4 mice than of the apoE3 mice (Fig. 1a). Environmental stimulation increased the numbers of DCX-positive neurons in the apoE3 mice, but had the opposite effect in the apoE4 mice (Fig. 1a). Similar results were obtained with an anti-DCX antiserum directed at the protein's N-terminal domain (pAb DCX N-19 not shown). Analysis of these results in the apoE3 and apoE4 mice and in apoE-deficient and wild-type mice [all values are expressed as percentages (mean ± SD) of those of the wild-type mice housed under standard conditions] revealed a significant effect of mouse group × treatment (p < 0.0001; Fig. 1b). Further analysis revealed that, under standard conditions, significantly more DCX-positive cells were present in the apoE4 mice than in the apoE3 or wild-type mice (250 ± 25% compared with 70 ± 20 and 100 ± 20%, respectively; p < 0.001). The numbers of DCX-positive neurons of the corresponding apoE-deficient mice (160 ± 35%) were also substantially lower than those of the apoE4 mice, but were higher than those of the apoE3 and the wild-type mice (Fig. 1b). As expected, the numbers of DCX-positive neurons of the apoE3 and the wild-type mice rose significantly after environmental stimulation, as did those of the apoE-deficient mice (230 ± 60, 210 ± 20, and 220 ± 31%, respectively; p < 0.001). In contrast, environmental stimulation of the apoE4 mice reduced the numbers of their DCX-positive neurons (170 ± 25%).
Similar results were obtained when DCX staining was analyzed as per cent of the DG area which was positively stained and double labeling experiments utilizing GFAP confirmed that DCX is associated with neurons and not with astroctyes (not shown).
As DCX is a transient marker of newly formed neurons, the observed results may be as a result of either the effects of apoE and environmental stimulation on the time that DCX is expressed or of the increased proliferation and production of new neurons. In order to differentiate between these possibilities, we measured the effects of apoE and environmental conditions on cellular proliferation in the DG utilizing Ki-67 as a marker (Jessberger et al. 2005; Matsumori et al. 2005). This revealed that, under regular conditions, the level of Ki-67-positive cells in the DG was higher in the apoE4 mice than in the other mice groups and that, following environmental stimulation, it decreased in the apoE4 mice but increased in the other mice groups (Fig. 2). Furthermore, double labeling experiments revealed that, under regular conditions, the level of the DCX and Ki-67 double-labeled cells in the DG was higher in the apoE4 mice than in the other groups (not shown). These findings suggest that, under regular conditions, apoE4 stimulates neurogenesis in the DG but that it reduces neurogenesis following environmental stimulation.
Importantly, and unlike in the DG, the number of DCX-positive neurons in the subventricular zone was not affected by either the apoE genotype or environmental stimulation (not shown), which suggests that the observed effects of environmental stimulation on neurogenesis in the DG and the biphasic effects of apoE4 on neurogenesis are specific to this brain area.
We next examined the effects of the apoE genotype and environmental conditions on the different stages of the neurogenesis cascade. Neurogenesis progresses via two major stages of progenitor cells mitosis, which can be distinguished by cells that are either nestin-positive or nestin + DCX-positive, and are followed by an early post-mitotic stage in which the cells take on neuronal morphology and are DCX-positive and nestin-negative (Kempermann et al. 2004b). Measurements of the effects of environmental stimulation on the numbers of hippocampal progenitor cells in apoE3 and apoE4 mice and in apoE-deficient and wild-type mice revealed a significant effect of group × treatment for the cells that were positive for both nestin and DCX and for cells that were DCX-positive and nestin-negative (p < 0.01, 2-way anova), whereas the numbers of cells that were nestin-positive and DCX-negative were not affected by either apoE or the environmental conditions. The results obtained with the apoE3 and apoE4 mice are depicted in Fig. 3. As shown, the numbers of DCX-positive cells that are either nestin-positive or nestin-negative were significantly higher in apoE4 mice under standard conditions than in the corresponding apoE3 mice (respectively, 280 ± 24 and 130 ± 28% in the apoE4 mice compared with 116 ± 31 and 86 ± 13% in the apoE3 mice). Environmental stimulation increased the numbers of DCX-positive, nestin-negative cells in the apoE3 mice, but decreased the numbers of these cells and of the DCX-positive, nestin-positive cells of the apoE4 mice (Fig. 3). Unlike the above, the numbers of nestin-positive, DCX-negative cells were not affected by either apoE genotype or environmental stimulation. The numbers of the different progenitor cells of the wild-type and apoE-deficient mice were similar to those of the apoE3 mice (not shown), except that the numbers of DCX-positive, nestin-negative cells in the apoE-deficient mice were higher under standard conditions (139 ± 15%) and the numbers of DCX- and nestin-positive cells in these mice decreased following environmental stimulation (50 ± 20%). Most of the DCX-positive cells did not contain nestin (Fig. 3). Accordingly, the effects of the apoE genotype and environmental stimulation on the numbers of the DCX-positive, nestin-negative cells (Fig. 3) were similar to their effects on the total numbers of DCX-positive cells (compare Figs 1 and 3).
These findings suggest that the stimulation of neurogenesis in apoE4 mice under standard conditions is mediated by activation of progenitor cells that are positive for both DCX and nestin, and of DCX-positive, nestin-negative progenitor cells. In contrast, environmental stimulation of neurogenesis in apoE3 (Fig. 3) and wild-type mice (not shown) is associated only with activation of cells that are both DCX-positive and nestin-negative.
We next examined the possibility that the specific decrease in the number of newly formed DCX-positive neurons in environmentally stimulated apoE4 mice is associated with activation of neuronal cell death and apoptosis. This was first investigated by use of the TUNEL assay, which measures cleaved DNA (Young et al. 1999). As shown in Fig. 4, the densities of TUNEL-positive cells in apoE3 and apoE4 mice, and in wild-type and apoE-deficient mice were low and similar under standard conditions, and were enhanced in the DG of the apoE4 mice following environmental stimulation (Fig. 4). This isoform-specific effect was statistically significant (p < 0.0001 for group × treatment; 2-way anova) and was associated with an increase of almost 3-fold in the density of TUNEL-positive cells in the DG of the environmentally stimulated apoE4 mice (Fig. 4). This effect was similar in the subgranular and granular layers of the DG. Furthermore, it was specific to the DG and no TUNEL-positive cells were detectable in other hippocampal subfields. The type of cell death in the DG of environmentally stimulated mice was examined by utilizing the apoptotic marker activated caspase 3 (Bingham et al. 2005). This demonstrated that the density of activated caspase 3 staining in the DG, like the TUNEL staining, is low under standard conditions in the different mouse groups and shows an isoform-specific increase in the apoE4 mice after environmental stimulation (p < 0.001 for group × treatment, 2-way anova; Fig. 5). In addition to the co-localization of TUNEL and activated caspase 3 in the DG, suggesting that neuronal death in this area in the apoE4 mice was apoptotic, activated caspase 3 was also associated with dendritic fields in the DG (Fig. 5).
Using the neuronal marker NeuN, we then assessed the extent to which the pro-apoptotic effects of apoE4 and environmental stimulation affect the overall number of hippocampal neurons. The results indicated that environmental stimulation mediates a decrease in the density of NeuN-positive neurons in the DG of the apoE4 mice and evokes the opposite effect in the corresponding apoE3 mice (Fig. 6). These effects were statistically significant (p < 0.001 for group × treatment, 2-way anova) and were associated with a decrease in the density of NeuN-positive neurons in the DG of the environmentally stimulated apoE4 mice to 68 ± 17% (p < 0.002) and corresponding increases in the densities of the apoE3, wild-type, and apoE-deficient mice to 135 ± 15, 133 ± 12, and 130 ± 16% (p < 0.02). Immunohistochemical assessment of the level and spatial distribution of apoE in the hippocampi of the apoE3 and apoE4 mice revealed, in agreement with previous immunoblot findings (Levi et al. 2003), that they were the same in the apoE3 and apoE4 mice housed under standard conditions and were similarly increased in these mice after environmental stimulation (data not shown). These findings suggested that the observed effects of apoE4 cannot be attributed to differences in the brain levels of apoE3 and apoE4.
Using the environmental enrichment paradigm and transgenic mice expressing human apoE, we examined the effects of apoE4 on the balance between neurogenesis and neuronal cell death. The results showed that in apoE4-transgenic mice housed under standard conditions neurogenesis in the hippocampal DG is increased, whereas in apoE4-transgenic mice housed in an enriched environment, which normally increases such neurogenesis, apoptosis is triggered and neurogenesis is reduced. These effects are specific to the DG and were not observed in the subventricular zone.
Hippocampal neurogenesis in the adult DG starts from a radial glia-like stem cell and progresses through successive stages of amplifying lineages of progenitor cells that are defined by specific markers (e.g. cells that are nestin-positive and DCX-negative, followed by cells that are DCX- and nestin-positive and then by cells that are DCX-positive and nestin-negative). This is followed by early post-mitotic morphological differentiation of the DCX-positive, nestin-negative cells to mature neurons detectable by neuronal markers such as NeuN (Kempermann et al. 2004b). Our analysis of the neurogenesis stage that is stimulated by environmental enrichment in wild-type and apoE3 mice revealed, in line with previous publications (Kempermann et al. 1997), that the first lineage to be stimulated under these conditions is that of the DCX-positive, nestin-negative cells (Fig. 3). In contrast, the neurogenesis lineage which is stimulated by apoE4 under standard conditions is further upstream in the progenitor lineage and is that of cells positive for both DCX and nestin (Fig. 3). This suggests that stimulation of neurogenesis by apoE4 under standard conditions and by apoE3 following environment stimulation are mediated by different mechanisms. Furthermore, the finding that under standard conditions the level of neurogenesis in the DG of apoE-deficient mice is higher than in the apoE3 and wild-type mice, but lower than that in the DG of the corresponding apoE4 mice (Fig. 1), suggests that neurogenesis under standard conditions is down-regulated by apoE3 and mouse apoE, and that the pro-neurogenesis effects of apoE4 under standard conditions are as a result of the loss of this inhibitory effect, as well as of the gain of a pro-neurogenesis effect. The extent to which these effects are because of apoE4-mediated changes in the kinetics of formation or in the survival of newly born neurons is not yet known.
The stimulation of neurogenesis in the DG of apoE3 and wild-type mice following environmental stimulation (Fig. 1) was associated with a corresponding increase in the total number of NeuN-positive neurons (Fig. 6). The stimulation of neurogenesis in the apoE4 mice under regular conditions was also associated with an increase in the total number of NeuN-positive neurons in the DG, which, however, was not statistically significant (Fig. 6). The functional role of neurogenesis in the adult hippocampus is still debated (Abrous et al. 2005), and further studies are required for assessing the functional effects of neurogenesis in the apoE4 mice under standard conditions.
The present finding that apoE4 activates apoptosis in the hippocampus of environmentally stimulated mice is in line with previous in-vitro cell cultures findings (Michikawa and Yanagisawa 1998; Hashimoto et al. 2000; Ji et al. 2002; DeKroon et al. 2003; Frey et al. 2006; Ji et al. 2006), and provides the first in-vivo linkage between neuronal cell death and the impaired neuronal plasticity and cognition of apoE4 mice. The extent to which other neurodegeneration-related forms of programmed cell death, such as autophagy (Nixon et al. 2005), are also affected by apoE4 is not yet known. The present finding that the extent of apoptosis in the DG of apoE-deficient mice after environmental stimulation is low and similar to that of the corresponding apoE3 and wild-type mice suggests that the pro-apoptotic phenotype of apoE4 mice is attributable to a gain of toxic function. This pro-apoptotic toxic function acquired by apoE4 might be triggered by the increase in hippocampal apoE induced by environmental enrichment. Environmental stimulation of apoE3 and wild-type mice improves their learning and memory (Levi et al. 2003) and is associated with concomitant increases in the density of hippocampal synapses (Levi et al. 2003) and in the numbers of total and newly formed DG neurons (Figs 1, 3 and 6). In contrast, environmental stimulation of apoE4 mice decreases the numbers of total and newly formed DG neurons (Figs 1 and 6) and has no effect on their hippocampal synaptic density and cognitive performance (Levi et al. 2003). These findings suggested that the cognitive improvements in apoE3 mice following environmental stimulation may be related to the corresponding increase in numbers of newly formed DG neurons and synapses in these mice. The ability of environmentally stimulated apoE4 mice to maintain their basal hippocampal synaptic level and cognitive performance, despite the fact that their DG neurons undergo apoptosis, might be attributable to a compensatory synaptic mechanism which, following neuronal cell death in the DG (Figs 4, 5 and 6), is able to maintain a constant level of synapses (Levi et al. 2003, 2005).
Over-expression of the anti-apoptotic protein bcl-2 and lack of the pro-apoptotic protein bax lead to a significant reduction in the number of apoptotic cells in the DG, and result in progressive increases in the neuronal cell number of the DG (Sun et al. 2004; Kuhn et al. 2005). Gene array analysis (Ophir et al. 2005) reveals that, under standard conditions, the basal levels of the pro-apoptotic genes bad and bid and of the apoptosome genes apaf1 and casp-9 are higher in the hippocampi of apoE4 mice than of apoE3 mice (not shown). Pharmacological cell culture experiments suggest that the pro-apoptotic effects of apoE4 in vitro are mediated by the low-density lipoprotein-related receptor protein (Hashimoto et al. 2000; Ji et al. 2006) and that low-density lipoprotein-related receptor protein mediates both the basal and the NMDA-driven stimulatory effects of apoE4 on neuronal uptake of Ca2+ (Ohkubo et al. 2001; Qiu et al. 2003). Further studies are needed to unravel the mechanisms underlying the pro-apoptotic and pro-neurogenesis effects of apoE4 under different environmental conditions and the extent to which they are meditated by Ca2+ and share common constituents.
AD is associated with enhanced hippocampal neurogenesis, which gives rise to new cells that can potentially replace neurons lost as a result of the disease (Jin et al. 2004 Greenberg and Jin 2006; Jin et al. 2006). Nevertheless, the homeostasis between neurogenesis and neuronal cell death is impaired in AD and is tilted towards the latter. The present finding in an animal model that environmental enrichment stimulates neurogenesis in carriers of the apoE3 genotype but triggers apoptosis in apoE4 carriers shows that the balance between neurogenesis and apoptosis can be markedly affected by the crosstalk between the apoE genotype and environnmental conditions. The recently reported finding of an inverse relationship between physical activity and the risk of dementia in human subjects who do not express apoE4 and that, in contrast, the risk for dementia in apoE4 carriers is not diminished by physical activity (Podewils et al. 2005), suggests that similar interactions might play a role in AD. Accordingly, protective measures such as environmental enrichment and physical activity might be beneficial for apoE3-positive AD patients but not for patients who carry apoE4. Conversely, anti-apoptotic treatments might be more effective in apoE4 than in apoE3 patients.
We thank Dr Allen Roses (Duke University, Durham, NC, USA) and Glaxo Wellcome for kindly providing the transgenic mice. This work was supported in part by grants from the US–Israel Binational Science Foundation and the Eichenbaum Foundation. DMM is the incumbent of the Myriam Lebach Chair in Molecular Neurodegeneration (Tel Aviv University).