Aβ Immunotherapy Protects Morphology and Survival of Adult-Born Neurons in Doubly Transgenic APP/PS1 Mice

Abstract

The hippocampus is heavily affected by progressive neurodegeneration and β-amyloid pathology in Alzheimer's disease (AD). The hippocampus is also one of the few brain regions that generate new neurons throughout adulthood. Because hippocampal neurogenesis is regulated by both endogenous and environmental factors, we determined whether it benefits from therapeutic reduction of β-amyloid peptide (Aβ)-related toxicity induced by passive Aβ immunotherapy. Aβ immunotherapy of 8–9-month-old mice expressing familial AD-causing mutations in the amyloid precursor protein and presenilin-1 genes with an antibody against Aβ decreased compact β-amyloid plaque burden and promoted survival of newly born neurons in the hippocampal dentate gyrus. As these neurons matured, they exhibited longer dendrites with more complex arborization compared with newly born neurons in control-treated transgenic littermates. The newly born neurons showed signs of functional integration indicated by expression of the immediate-early gene Zif268 in response to exposure to a novel object. Aβ immunotherapy was associated with higher numbers of synaptophysin-positive synaptic boutons. Labeling dividing progenitor cells with a retroviral vector encoding green fluorescent protein (GFP) showed that Aβ immunotherapy restored the impaired dendritic branching, as well as the density of dendritic spines in new mature neurons. The presence of cellular prion protein (PrP^c) on the dendrites of the GFP⁺ newly born neurons is compatible with a putative role of PrP^c in mediating Aβ-related toxicity in these cells. In addition, passive Aβ immunotherapy was accompanied by increased angiogenesis. Our data establish that passive Aβ immunotherapy can restore the morphological maturation of the newly formed neurons in the adult hippocampus and promote angiogenesis. These findings provide evidence for a role of Aβ immunotherapy in stimulating neurogenesis and angiogenesis in transgenic mouse models of AD, and they suggest the possibility that Aβ immunotherapy can recover neuronal and vascular functions in brains with β-amyloidosis.

Introduction

The pathology of Alzheimer's disease (AD) is characterized by the accumulation of β-amyloid peptides (Aβ), neurofibrillary tangles, reactive astrocytes, and activated microglia, leading to cognitive decline and dementia (Haass and Selkoe, 2007; Marcello et al., 2008). Mouse models expressing familial AD-causing mutations reproduce some of these signs, allowing for the exploration of treatments designed to reduce the pathology (Holcomb et al., 1998; Oddo et al., 2003; Kobayashi and Chen, 2005). Aβ immunotherapy can reduce brain β-amyloid and restore cognition in such transgenic mouse models (Schenk et al., 1999; Bard et al., 2000; Morgan et al., 2000; DeMattos et al., 2001). In patients with AD, Aβ vaccination cleared brain Aβ and slowed cognitive decline in a subcohort of patients (Hock et al., 2003; Nicoll et al., 2003). Although both Aβ clearance and stabilized cognition correlated with titers of anti-Aβ antibodies, the initial clinical trial was prematurely interrupted because of side effects, and it is currently unclear whether reducing brain Aβ is sufficient to halt disease progression and whether therapeutic Aβ clearance should start before the onset of neurodegeneration (Hock et al., 2003; Holmes et al., 2008). Adult neurogenesis continues throughout life in two restricted brain regions (Altman and Das, 1965; Eriksson et al., 1998). In the dentate gyrus, new granule cells contribute to the maintenance of hippocampal functions, including mood regulation, learning, and memory (Shors, 2008; Zhao et al., 2008). Enriched environment, exercise, pharmacological compounds, as well as stroke, epilepsy, and neurodegeneration, affect adult neurogenesis and enhanced neurogenesis may restore functions in neurodegenerative diseases including AD (Bengzon et al., 1997; Kempermann et al., 1997; Parent et al., 1997; van Praag et al., 1999; Arvidsson et al., 2002; Curtis et al., 2003; Santarelli et al., 2003; Lie et al., 2004; Kuhn et al., 2007; Abdipranoto et al., 2008). Initial signs for increased neurogenesis in postmortem brains obtained from patients with AD were not confirmed by subsequent studies suggesting impaired maturation of new neurons (Jin et al., 2004; Li et al., 2008). Because aggregated oligomeric forms of Aβ are neurotoxic in vivo (Lesne et al., 2006; Shankar et al., 2008), we explored whether reducing brain Aβ by Aβ immunotherapy has a protective role in adult neurogenesis in doubly transgenic amyloid precursor protein/presenilin-1 (APP/PS1) mice. In a prior experiment, active vaccination increased neurogenesis in young transgenic mice (Becker et al., 2007). It is unknown, however, whether passive Aβ immunotherapy can rescue adult neurogenesis in aged transgenic mice with β-amyloid-related impairments of neurogenesis. We addressed this question by treating 8–9-month-old APP/PS1 mice characterized by pre-existing β-amyloid pathology with weekly injections of an antibody against Aβ over a 3.5 month period. We found that Aβ immunotherapy increased numbers of new neurons, these showed morphological signs of synaptic activity, improved dendritic branching, and restored numbers of dendritic long-thin, stubby, and mushroom spines. Moreover, Aβ immunotherapy increased angiogenesis in the dentate gyrus.

Materials and Methods

Animals and treatments.

Heterozygous doubly transgenic mice expressing both human mutant APP (Tg 2576; Hsiao et al., 1996) and human mutant PS1 (mutation: M146L) were generated by crossing heterozygous Tg2576 mice on a hybrid C57BL6/J × FVB/N × SJL/J background, with no obvious signs of embryonic lethality, with PS1-transgenic mice on a hybrid Swiss Webster × B6D2 background. The APP/PS1-transgenic mice were compared with age-matched littermates or nontransgenic (non-tg) mice with the same hybrid background. Genotypes were determined by PCR. Both males and females were housed in standard cages with females in groups of two to four and males single-caged under 12 h light/dark conditions with ad libitum access to water and food. The α-Aβ antibody used throughout these experiments is a chimeric monoclonal antibody consisting of human variable domains and mouse IgG2a constant regions. It selectively recognizes with high-affinity (low nm range) aggregated forms of Aβ1-40 and Aβ1-42, as well as β-amyloid plaques on tissue sections obtained from human or transgenic mouse brains. The antibody is specific for aggregated Aβ and does not cross-react with the unprocessed full-length APP.

In a first experiment, at 8 months of age, 14 APP/PS1 mice were either treated with α-Aβ (n = 7, six males and one female) or an irrelevant control antibody with the same isotype (IgG2) raised against bovine herpes virus (clone 2H6-C2; European Collection of cell cultures, Ecacc) (ct ab) (n = 5; 4 males and 1 female). The antibodies were administered once a week at a dose of 5 mg/kg i.p. for 3.5 months before perfusion. In a second experiment, 15 mice were injected with green fluorescent protein (GFP)-expressing retrovirus into the dentate gyrus at 3 months of age, 1 month after the beginning of either passive Aβ immunization or vehicle treatment; n = 5 (2 males and 3 females) APP/PS1 mice with α-Aβ treatment (5 mg/kg i.p.) and n = 5 (2 males and 3 females) APP/PS1 and n = 5 (2 males and 3 females) non-tg mice with vehicle treatment (PBS, 100 μl/10 g body weight, i.p.) for 2 months before perfusion, starting at 2 months of age. In general, after a 5 min transport from the animal facility within the same building, mice were placed in their home cages under a hood for 30 min before anesthesia and transcardial perfusion. This procedure caused no increases in expression of the immediate early gene Zif268 in the brains. To experimentally stimulate immediate early gene expression, mice were exposed for 10 min to a novel object by placing a quarter of an apple wrapped in tin foil into their home cages. All mice intensively explored the novel object. The novel object exposure was followed by a 10 min interval until anesthesia was initiated and followed by an additional 15 min interval until brains were perfused. This experimental setup resulted in the consistent appearance of Zif268 protein in discrete foci in cells throughout the brain cortex and hippocampus, including the nuclei of bromodeoxyuridine-positive (BrdU⁺) newly born neurons in the SVZ/GCL. The presence of Zif268 protein as early as 35 min after the onset of novel object exposure is compatible with the known time dependence of immediate early gene expression in brain following stimulation (Richardson et al., 1992; Guzowski et al., 1999; Kubik et al., 2007).

To exclude nonspecific effects of Aβ immunotherapy on neurogenesis, we used non-tg, wild-type mice that do not express human APP or have β-amyloid pathology. These mice were treated with either vehicle (PBS, 100 μl/10 g body weight, i.p.; n = 5; 2 males and 3 females) or α-Aβ (5 mg/kg i.p.; n = 4; 1 male and 3 females) for 10 weeks before perfusion, starting at 2 months of age.

In the different experiments, there was a good balance between numbers of male and female mice, and results were, therefore, pooled. Because of low numbers of either male or female animals, a reliable statistical comparison could not be made, but no obvious influence of gender on any parameter of neurogenesis was detected. All experimental procedures followed guidelines set by the Swiss veterinary cantonal office for the use and care of laboratory animals (License Nr 48/08).

Bromodeoxyuridine labeling.

Mice were injected with the thymidine analog BrdU (50 mg/kg i.p.; dissolved in potassium PBS (KPBS)) twice daily for 2 weeks, for labeling of mitotic cells (Dolbeare, 1995), starting 1 month before perfusion. In non-tg mice subjected to either vehicle or Aβ immunotherapy, BrdU (50 mg/kg i.p.; twice daily for 3 d) was injected 1 month after the beginning of the treatment.

Retrovirus-GFP labeling.

An oncoretroviral vector derived from the Moloney sarcoma virus and expressing GFP under control of the Rous sarcoma virus promoter (MolRG) was used. Transfer vector and helper vectors were gifts from Wolfgang Kelsch and Carlos Lois (Massachusetts Institute of Technology, Cambridge, MA). The viral solution was prepared by cotransfection of HEK293FT cells (Invitrogen) using calcium–phosphate precipitation. After 48 h, conditioned media were concentrated by two sequential ultracentrifugations in sucrose gradients. Viral particles were resuspended in sterile PBS, aliquoted, and stored at −80°C until use. Viral concentrations (10^8–9 cfu/ml) were determined by serial dilutions on HEK293FT cells, and the number of GFP⁺ cells was counted 48 h after infection using flow cytometry. For labeling of the new neurons, mice were deeply anesthetized with ketamin/xylaxine and given unilateral injections of the retroviral vector (1.5 μl at 0.2 μl/min) into the dentate gyrus (coordinates: 2 mm posterior and 1.5 mm lateral from bregma and 2.3 mm ventral from skull).

Immunohistochemistry.

Mice were anesthetized with ketamin/xylaxine and transcardially perfused with 50 ml of PBS, followed by 100 ml of ice-cold 4% paraformaldehyde in 0.1 m PBS, pH 7.4. Brains were removed, post-fixed overnight in the same medium, and put in 20% sucrose in 0.1 m phosphate buffer for 24 h. Coronal sections (30 or 40 μm for the analyses following retrovirus injections) were cut on a sliding microtome and stored in cryoprotective solution. For immunohistochemistry, preincubation with appropriate normal sera was carried out for 1 h at room temperature. For epitope retrieval of BrdU, phospho-histone H3 (p-H3) and proliferating cell nuclear antigen (PCNA), free-floating sections were denatured in 1 m HCl for 30 min at +65°C. Primary antibodies were as follows: rabbit anti-ionized calcium-binding adapter molecule 1 (Iba1) [a marker for microglia/macrophages (Imai et al., 1996); 1:1000; Wako Chemicals]; rat anti-CD11b (a marker for activated microglia; 1:200; BD Pharmingen); mouse anti-human β-amyloid (6E10) (1:1000; Covance); rat anti-BrdU (1:100; Oxford Biotec); mouse anti-neuronal nuclei marker (NeuN) (1:100; Millipore Bioscience Research Reagents); rabbit anti-S100β (a marker for astrocytes; 1:5000; SWANT); rabbit anti-C terminus of the transcription factor Zif268 [a marker for synaptic activity (Beckmann and Wilce, 1997); 1:250 (Santa Cruz Biotechnology)]; mouse anti-polysialic acid-neural cell adhesion molecule (PSA–NCAM) [a marker for immature neurons (Rougon et al., 1982); 1:2000; AbCys]; mouse anti-synaptophysin (SYN) (a marker for synapses; 1:200; Sigma); rabbit anti-mouse glucose transporter 1 (Glut1), a marker for glucose transporter1 highly expressed in endothelial cells of barrier tissues such as the blood–brain barrier (1:500; Alpha Diagnostic), goat anti-prion protein [(Hantman and Perl, 2005); 1:400; Millipore Bioscience Research Reagents], and biotinylated–isolectin IB4 (1:200; Invitrogen). Isolectin–IB4 binds to microglia, group B erythrocytes, perivascular cells, and endothelial cells and can be used to stain adult brain vasculature (Ernst and Christie, 2006).

Free-floating sections were incubated overnight at +4°C. Secondary antibodies for detection were Cy3-conjugated donkey-anti-rabbit/rat, FITC-conjugated goat anti-mouse/rabbit, Cy5-conjugated donkey anti-mouse, biotinylated anti-goat, Cy5-conjugated streptavidin and for lectin-staining FITC-conjugated streptavidin (all 1:200; Jackson ImmunoResearch Laboratories) with incubation for 2 h at room temperature in the dark. Rinsing in 0.25% Triton X-100 was carried out between each incubation. To detect β-amyloid plaques decorated with the antibody used for passive immunotherapy, sections were incubated with Cy3-conjugated donkey-anti-mouse (1:200; Jackson ImmunoResearch Laboratories). For single-staining with rabbit anti p-H3, specifically expressed only from late G₂ interphase until anaphase, thus providing the actual number of dividing cells (Hendzel et al., 1997) (1:400; Upstate Biotechnology) or with mouse anti-PCNA [expressed in early G₁ and S phases (Bravo and Macdonald-Bravo, 1987); 1:200; Santa Cruz Biotechnology], avidin–biotin peroxidase complex (Elite ABC kit; Vector Laboratories), 3,3′-diaminobenzidine and hydrogen peroxide were used. Chromogenic visualization included pretreatment with blocking of endogenous peroxidase activity with 3% H₂O₂ and 10% methanol. Sections were mounted on chrom-gelatin-coated microscope slides (Super-frost-plus; Menzel) and coverslipped with aqueous anti-fading mounting medium (PVA–DABCO). Thioflavin-S (ThioS) staining for fibrillar Aβ was done as described previously (Schmidt et al., 1995). In short, floating sections from APP/PS1 mice were washed in KPBS and mounted on chrom-gelatin-coated glass slides before being processed. After treatment for 10 min with 0.25% potassium permanganate, sections were washed in KPBS and incubated in 1% potassium metabisulfite and 1% oxalic acid until they appeared white. Sections were then floated 3 s in 0.25% acetic acid, washed in water and stained for 5 min with a solution of 0.015% Thio-S in 50% ethanol. Finally, sections were washed in 50% ethanol and in water, then dried, and coverslipped with PVA–DABCO. For Prussian blue staining for hemosiderin, whose presence is an indication of a past microhemorrhage event, sections were mounted on coated slides and then stained with equal solutions of 20% HCl and 10% potassium ferrocyanide [K₄Fe(CN)₆*3H₂O] in water for 30 min at 37°C. All chemicals were from Sigma. Control stainings included omission of the primary antibody and/or use of different tissues known to be positive for the specific antibody (e.g., tonsils for p-H3 and hippocampus from young mice exposed to novel enviroment for Zif268).

Stereology and microscopic analyses.

All measurements were performed by an observer blind to animal identifications. Immunohistochemical stainings were examined with an Inverted Leica DM IRE2 fluorescence and light microscope. Numbers of immunoreactive cells were counted bilaterally in the granule cell layer (GCL), and within one cell diameter below this region in the subgranular zone (SGZ), throughout the entire dentate gyrus. Stereological estimations of the total number of BrdU⁺ and PSA–NCAM⁺ in the SGZ/GCL, as well as numbers of Glut1⁺ and IsolectinB4⁺ blood vessels in the dentate gyrus, were performed using the optical fractionator method (Gundersen and Jensen, 1987; West et al., 1991). Six to eight coronal sections were analyzed per animal using the Leica DM4000B microscope with a 100× oil objective, Olympus DP71 color digital camera, and newCAST software (Visiopharm). For systematic sampling, the frame area was set to 3590 μm² with a sampling interval of 226 μm at the x and y level, and the optical dissector constituting a 15-μm-thick fraction of the total section thickness (measuring 22–25 μm after processing). Neither the area for counting nor the thickness of the analyzed sections differed between groups, which allowed for comparisons of cell counts. The SGZ/GCL and dentate hilus volumes were measured using the same stereological equipment in the same sections. The intersections of NeuN-stained dentate gyri with a digitally over imposed counting grid were counted in 8–10 coronal sections with a 40× objective. A maximum of 150–200 intersecting points (area per point: 0.078 mm²) were counted per animal, and the volume was estimated applying Cavalieri's principle (Gundersen et al., 1988). Because of their small numbers, all p-H3⁺, BrdU⁺, and PSA–NCAM⁺ cells in the SGZ/GCL were counted in each section. Colocalization of up to 50 double- or triple-stained BrdU⁺ cells with NeuN/S100β/Zif268 was validated using a confocal scanning microscope (Leica TCS/SP2; Leica) with Argon laser 488, HeNE laser, 568 and 633 excitation filters. Numbers of blood vessels bilaterally in the entire dentate gyrus of antibody-treated APP/PS1 mice were estimated as described elsewhere (Lee et al., 2005) using the same stereological settings as described above. Briefly, the total number of branching points (defined as the point in which a single capillary branches into two, forming a Y) was counted using the optical dissector. If these points are set as the counting targets, twice this number is equal to the number of capillary segments. Since thick vertical uniform random sections were used, the length of blood vessels was estimated via counting the intersections between capillaries and “virtual” isotropic planes projected through the section by computer-assisted stereology (Larsen et al., 1998). SYN-positive presynatic terminals were counted bilaterally in the molecular layer (ML) of three hippocampal sections using a Leica DM4000B microscope with a 100× oil objective, Olympus DP71 color digital camera, and newCAST software (Visiopharm) with frame area set to 250 μm², with 200 μm sampling intervals at the x and y levels, and with the optical dissector constituting 2.5-μm-thick fractions of the total section thickness.

The morphology of Iba1⁺ microglia in the dentate gyrus and septum was characterized as ramified, intermediate, amoeboid or round profile, as described elsewhere (Thored et al., 2009), and the percentage of the respective phenotype was calculated over the total number of Iba1⁺ cells considered (∼90 cells).

Morphological analysis of new immature neurons.

To determine dendrite length and branching, PSA–NCAM⁺ cells (n = 40 from each group) were analyzed using a Leica DM4000 microscope with a 63× water objective and a digital zoom of 2. On average, 50–60 Z-series of 0.25 μm were merged for analysis. The measurements were done by using NeuronJ (http://www.imagescience.org/meijering/software/neuronj/).

NeuronJ is an ImageJ plugin module developed for tracing and quantification of elongated structures in two-dimensional images, in particular neurites in fluorescence microscopy images.

Morphological analysis of new mature neurons.

Following retrovirus injection, the dendritic length and branching of 5-week-old GFP⁺ new granule cells (n = 15/group) were analyzed using the Leica DM4000 microscope with 20× water objective and a digital zoom of 2 (1024 × 1024 pixel). On average, 50–70 pictures with 0.5 μm steps were merged for analyses. High-magnification pictures for spine density analysis were also taken using the Leica DM4000 microscope with 63× water objective and a digital zoom of 6 (512 × 512 pixel). On average, 100–130 pictures of 0.120 μm steps were merged for analysis. The autofluorescence of the labeled neurons was sufficiently strong so that anti-GFP staining was unnecessary. Z-stacks were deconvoluted, a process that filters the signal to improve the clarity of images by increasing resolution, removing out-of-focus blur and eliminating noise, using the open source Huygens remote manager (http://hrm.sourceforge.net/). Three-dimensional reconstructions of deconvoluted z-stacks were performed with Imaris 6.1 (Bitplane). After three-dimensional reconstructions, Scholl analysis, in which intersections between dendrites and concentric circles were counted, each one with a 10 μm increment from the previous and centered at the soma, to determine the total dendrite length and branching density, or spine classification (stubby, long-thin and mushroom according to the original classification of Peters and Kaiserman-Abramof, 1970) along 40 μm segments (n = 50/group) were done using the aforementioned software and its MatLab extensions (ImarisXT). Because spine density increases proportionally to the distance from the cell body, for each GFP⁺ neuron used in the Scholl analysis, one to two dendritic segments were analyzed in the medial and external part of the outer ML, respectively.

Image analysis.

All images were subjected to the following transformations: color pictures were first converted to grayscale images (8-bit), and the resulting images were then further adjusted to black and white by defining a grayscale cutoff point (threshold), and finally, all pictures within each experimental group were adjusted to the same threshold to allow for comparison of results. Quantifications of diffuse plaque load (6E10 staining), compact plaque burden (ThioS), and cerebral amyloid angiopathy (CAA) measured in the ThioS staining, were performed as described elsewhere (Wilcock et al., 2006). Briefly, ImageJ was used for measuring the area fraction (% area), defined as the area positive for the staining in a constant and determined area, in cortex, hippocampus, and thalamus of both hemispheres. Similarly, ImageJ was used to quantify Iba1⁺ immunoreactivity: in the dentate gyrus, the number of Iba1⁺ cells was electronically counted, whereas in broader cerebral regions, e.g., lateral septal nuclei, fornix, and thalamus, the area fraction positive for Iba1 staining was evaluated. The same analysis was used to quantify CD11b⁺ areas in the hippocampus and septum as a measure of microglia activation. Assessment of the SYN levels was performed in the frontal neocortex and the hippocampal outer ML (Buttini et al., 2005; Priller et al., 2006). For each zone of interest, four confocal images were taken at magnification 63× in water with digital zoom 2, with z-stacks of 0.25 μm. After maximum intensity projection of z-stacks, the digitized images were analyzed for the signal intensity (average pixel intensity) of SYN-labeled structures by using ImageJ. For the histological analysis of the colocalization of cellular prion protein (PrP^c) on dendrites of GFP⁺ newly born neurons, high-magnification stacks (63× z8; 512 × 512 pixel; 0.16 z-step) were analyzed by using ColocImaris software. This software specifically identifies voxels that are positive in both fluorescent channels, thereby dramatically reducing artifacts generated by overlays in z-levels.

Aβ ELISA.

Blood was withdrawn by heart puncture after deep anesthesia with ketamin/xylazine directly before perfusion and separated from erythrocytes by centrifugation. Plasma Aβ_1-42 was measured with the INNOTEST β-Amyloid (1-42) ELISA kit (Innogenetics) using the “high sensitivity” conjugate.

Statistical analyses.

Comparisons between animal groups were performed using unpaired Student's t test or one-way ANOVA, followed by Bonferroni's or Fisher's least significant difference post hoc tests. Data are presented as mean ± SEM. Differences are considered significant at p < 0.05. Calculations were made with Statview 5.0.

Results

Passive Aβ immunotherapy attenuated β-amyloid pathology in APP/PS1-transgenic mice

After 3.5 months of treatment of 8–9-month-old doubly transgenic APP/PS1 mice with weekly i.p. injections of α-Aβ, the highest accumulation of antibody was present in septum, fornix, and thalamus, as indicated by specific Cy3 anti-mouse IgG antibody staining (Fig. 1A,B). Antibody was also detected at lower concentrations within the hippocampus. The reasons for differences in antibody concentrations throughout the brains may include region-specific variations of extraction rates or varying amounts of CAA influencing both delivery and extraction rates of the antibody. Another possibility is a concentration gradient of antibody surrounding potential sites of entry into the brain, including the choroid plexus and the lateral ventricles. Fixation artifacts are likely excluded, because all brains were perfused with fixative before immunohistochemical analyses. In general, the antibody was associated with amyloid deposits, if present.

Figure 1.

Aβ immunotherapy attenuated β-amyloid pathology. A, B, Immunohistochemical staining of brain tissue sections obtained from APP/PS1 mice treated with Aβ immunotherapy (α-Aβ) or a control antibody (ct ab) with a Cy3-conjugated antibody directed against mouse IgG revealed binding of the therapeutic antibody to β-amyloid plaques (B), an indication that the antibody had crossed the blood–brain barrier and reached the brain. CX, Cortex; CC, corpus callosum; LSI, lateral septal nuclei. C, D, Representative ThioS staining of compact β-amyloid plaques. CX, Cortex; DG, dentate gyrus; the dotted lines depict the corpus callosum (CC). There was a significantly lower area fraction of ThioS-positive β-amyloid plaques in APP/PS1 mice treated with Aβ immunotherapy (1.14 ± 0.14% area) compared with control antibody-treated mice (1.75 ± 0.16% area; p < 0.05; unpaired t test). Scale bars: (in A and B) 500 μm; (in C and D) 250 μm.

At ∼1 year of age, the APP/PS1 mice treated with the control antibody presented severe β-amyloid deposits throughout their brains including the dentate gyrus as well as some deposits in the GCL. In addition, these mice developed CAA in cortex and thalamus. Although the antibody treatment did not change the 6E10-immunopositive Aβ burden (data not shown), the amount of β-amyloid plaques detected by ThioS staining in the whole brain was significantly reduced by 37.8% in α-Aβ compared with ct ab-treated mice (APP/PS1+ct ab: 1.75 ± 0.16 vs APP/PS1+α-Aβ: 1.14 ± 0.14% area; p < 0.05) (Fig. 1C,D). In the hippocampus, the area fraction covered with ThioS⁺ β-amyloid plaques in antibody-treated APP/PS1 mice was >60% lower, from 1.32 ± 0.29% in vehicle-treated APP/PS1 animals to 0.48 ± 0.21% (p = 0.05) in α-Aβ-treated mice. Progressive accumulation of β-amyloid in the vicinity of blood vessels can result in microhemorrhages, and some studies have linked immunotherapy to an initial increase in CAA and microhemorrhages (Pfeifer et al., 2002; Wilcock et al., 2004a; Racke et al., 2005), followed by decrease during longer treatment periods (Schroeter et al., 2008). To exclude that the α-Aβ antibody treatment used here decreased parenchymal Aβ plaques but increased vascular Aβ deposits, we quantified CAA visualized by ThioS staining. However, no differences were detected between treatment and control groups (APP/PS1+ct ab: 0.21 ± 0.09 vs APP/PS1+α-Aβ: 0.17 ± 0.09% area). The presence of microhemorrhages was detected by Prussian blue staining of hemosiderin. We observed microhemorrhages in 26% of the immunotherapy-treated APP/PS1-transgenic mice compared with 60% in ct ab-treated mice (χ²: NS) indicating the absence of increased microhemorrhages during Aβ immunotherapy in our experimental setting. Reductions in β-amyloid plaque load were not accompanied by increased plasma levels of Aβ₄₂ (APP/PS1+ct ab: 1413.5 ± 96.6 pg/ml vs APP/PS1+α-Aβ: 1753 ± 193.9 pg/ml). To explore whether microglia could be involved in the reduction of β-amyloid plaques (Schenk et al., 1999), we evaluated the morphology and counted numbers of Iba1-positive cells. The α-Aβ- and ct ab-treated mice showed similar numbers of Iba1⁺ cells in the dentate gyrus (SGZ/GCL: APP/PS1+ct ab: 50.5 ± 8.6 vs APP/PS1+α-Aβ: 57.3 ± 6.4 cells; hilus: APP/PS1+ct ab: 50.8 ± 14.4 vs APP/PS1+α-Aβ: 55.9 ± 7 cells; septum-fornix-thalamus: APP/PS1+ ct ab: 7.4 ± 1.02 vs APP/PS1+α-Aβ: 8.9 ± 1.06% area). There was no difference between the groups in the percentage of ramified-, intermediate-, amoeboid- or round-shaped Iba1⁺ microglia in the dentate gyrus or in the septum (Table 1). Furthermore, measurement of areas in the hippocampus and in the septum immunoreactive (CD11b⁺) for activated microglia did not reveal any differences between groups (Table 1). Microglia can increase their proliferation rate during changes in activation state, but numbers of BrdU⁺/Iba1⁺ cells in the SGZ/GCL did not differ between α-Aβ and ct ab-treated mice (APP/PS1+ct ab: 20.2 ± 2.1 vs APP/PS1+α-Aβ: 21.6 ± 2.7). Also, no significant differences were observed in SGZ/GCL or hilar volumes (SGZ/GCL: APP/PS1+ct ab: 2.6 ± 0.6 vs APP/PS1+α-Aβ: 3.6 ± 0.2 mm³; hilus: APP/PS1+ct ab: 3.3 ± 0.2 vs APP/PS1+α-Aβ: 3.4 ± 0.2 mm³).

Table 1.

Morphological phenotypes of Iba1 plus microglia and area fraction for CD11b plus activated microglia in APP/PS1 doubly transgenic mice subjected to Aβ immunotherapy

Region	Microglia type/Area fraction	Control antibody	Aβ Immunotherapy
SGZ/GCL	% ramified Iba1+	36.4 ± 22.5	24.4 ± 4.9
	% intermediate Iba1+	49.3 ± 3.3	57 ± 4.8
	% amoeboid Iba1+	13 ± 2.9	15.5 ± 2.7
	% round Iba1+	1.3 ± 0.5	3.2 ± 1.4
Hilus	% ramified Iba1+	20.5 ± 2.9	12.4 ± 3.0
	% intermediate Iba1+	57.7 ± 7.2	59.8 ± 3.4
	% amoeboid Iba1+	13.7 ± 4.2	21.5 ± 4.7
	% round Iba1+	3.3 ± 1.1	6.3 ± 1.1
Septum	% ramified Iba1+	18.3 ± 3.3	17.7 ± 3.4
	% intermediate Iba1+	57.7 ± 7.2	49.8 ± 3.7
	% amoeboid Iba1+	15.4 ± 5.3	21.4 ± 2.9
	% round Iba1+	8.6 ± 3.2	11.1 ± 2.7
Hippocampus	% area CD11b+	0.50 ± 0.10	0.65 ± 0.11
Septum	% area CD11b+	0.84 ± 0.20	0.98 ± 0.31

Morphological subtypes of Iba1 plus microglia: ramified and intermediate represent quiescent and amoeboid and round represent activated microglia. No significant differences were observed between groups. Values are means ± SEM; n = 5 (control antibody) and n = 7 (Aβ immunotherapy).

Passive Aβ immunotherapy increased neurogenesisand dendritic arborization of new immature neurons inAPP/PS1-transgenic mice

In the APP/PS1 mice used in these experiments, the pathology associates with ∼50% more proliferating cells (p-H3⁺ and BrdU⁺) than non-tg controls. However, in the same animals, this initial increase does not result in higher numbers of mature (BrdU⁺/NeuN⁺) nor immature (PSA–NCAM+) neurons, suggesting that the progenitors may not survive the Aβ-related toxicity (B. Biscaro, unpublished observations; Chen et al., 2008). We first tested whether the antibody-mediated reduction in β-amyloid pathology had reversed these impairments of neurogenesis. Interestingly, both the numbers of BrdU⁺/NeuN⁺ new mature neurons (APP/PS1+ct ab: 306 ± 31 vs APP/PS1+α-Aβ: 452 ± 42 cells; p < 0.05) and PSA–NCAM⁺ young neurons (APP/PS1+ct ab: 219 ± 79 vs APP/PS1+α-Aβ: 842 ± 275 cells; p = 0.06) were higher in the α-Aβ- compared with the ct ab-treated APP/PS1 mice. Moreover, the PSA–NCAM⁺ neurons in the α-Aβ-treated APP/PS1 mice had higher dendritic complexity with more and longer dendrites (number of dendrites per cell: APP/PS1+ct ab: 2.45 ± 0.3 vs APP/PS1+α-Aβ: 4 ± 0.4, p < 0.01; and length of dendrites: APP/PS1+ct ab: 54.4 ± 7 vs APP/PS1+α-Aβ: 85.5 ± 10 μm, p < 0.01) (Fig. 2A–D). The number of BrdU⁺/S100β⁺ astrocytes did not differ between the groups (APP/PS1+ct ab: 56 ± 12.13 vs APP/PS1+α-Aβ: 43 ± 12.72). Also the numbers of p-H3⁺ and BrdU⁺, proliferating cells in the SGZ/GCL were similar in the two groups (APP/PS1+ct ab: 174.4 ± 92 vs APP/PS1+α-Aβ: 210.3 ± 45.4 p-H3⁺ cells and APP/PS1+ct ab: 547.2 ± 57.5 vs APP/PS1+α-Aβ: 676.6 ± 83.4 BrdU⁺ cells). Synaptic activity in the dentate gyrus induces robust expression of the immediate early gene Zif268 (or Erg-1), which is essential for the formation of long-term memories (Jones et al., 2001). If not exposed to stimulation such as novel environment or LTP, Zif268 is not constitutively transcribed in the GCL (Davis et al., 2003; Bruel-Jungerman et al., 2006). We analyzed Zif268 expression in the SGZ/GCL following a physiological stimulus induced by a novel object (“activation”) (Fig. 3). Without exposure to the activation stimulus, there was no Zif268 expression in APP/PS1 mice (Fig. 3A–D). The activation stimulus induced Zif268 expression throughout the neuronal population in the SGZ/GCL both in vehicle-treated (Fig. 3E–H) and Aβ immunotherapy-treated mice (Fig. 3I–L). We established that also new mature neurons in Aβ immunotherapy-treated mice responded to the activation stimulus as a sign of functional integration by showing stimulus-evoked Zif268 expression in BrdU⁺/NeuN⁺ cells (Fig. 3I–L).

Figure 2.

Aβ immunotherapy increased neurogenesis and dendritic arborization of newly born neurons in 11–12-month-old APP/PS1 mice. Aβ immunotherapy significantly increased numbers of mature BrdU⁺/NeuN⁺; this increase was accompanied by a trend for higher numbers of immature PSA-NCAM+ neurons (APP/PS1+ct ab: 306 ± 31 vs APP/PS1+α-Aβ: 452 ± 42 BrdU⁺/NeuN⁺ cells; p < 0.05; and APP/PS1+ct ab: 219 ± 79 vs APP/PS1+α-Aβ: 842 ± 275 PSA-NCAM⁺ cells; p = 0.06). A, B, Representative low-magnification images of PSA-NCAM⁺ neurons in the dentate gyrus of control-treated (A) and Aβ immunotherapy-treated (B) mice. Aβ immunotherapy significantly increased the number of dendrites of PSA-NCAM⁺ neurons in APP/PS1 mice and the length of dendrites of PSA-NCAM⁺ neurons in APP/PS1 mice (number of dendrites per cell: APP/PS1+ct ab: 2.45 ± 0.3 vs APP/PS1+α-Aβ: 4 ± 0.4; p < 0.01; and length of dendrites: APP/PS1+ct ab: 54.4 ± 7 vs APP/PS1+α-Aβ: 85.5 ± 10 μm; p < 0.01). C, D, High-magnification confocal details, including corresponding computer renderings resulting from quantitative NeuronJ analyses of the cells shown in the yellow squares in A and B, respectively, taken from comparable regions within the dentate gyrus. Scale bars: A, 60 μm; C, 20 μm. N = 40 cells were analyzed per group. Unpaired t test.

Figure 3.

Newborn BrdU⁺ neurons expressed the activity-dependent immediate early gene Zif268 during Aβ immunotherapy as a sign of functional integration in pre-existing hippocampal synaptic circuits. Immunohistochemical stainings for the immediate-early gene product Zif268 (green), BrdU⁺ nuclei (red), NeuN⁺ neuronal nuclei (blue), overlay of the three channels (merge), and confocal orthogonal projections of merged Zif268 and BrdU images of selected granular neurons (ortho). A–D, Background expression of Zif268 in untreated APP/PS1 control mice without exposure to novel object stimulation: absent localization of Zif268 in nuclei of BrdU⁺ granular neurons. E–H, Presence of Zif268 in BrdU⁺ nuclei of granular neurons within the SGZ/GCL in response to novel object exposure in vehicle-treated APP/PS1 mice. I–M, Presence of Zif268 in BrdU⁺ nuclei of granular neurons within the SGZ/GCL during Aβ immunotherapy and in response to novel object exposure. Circles depict selected BrdU⁺ cells to facilitate their identification in the Zif268 staining. Scale bar: A, 10 μm.

Both active and passive Aβ-immunizations were reported to prevent synaptic degeneration in the frontal cortex and hippocampus of transgenic mice expressing APP mutants (Buttini et al., 2005). Consistent with these findings, α-Aβ-treated mice showed increased numbers of SYN-positive boutons detected by stereological counting in the ML of the dentate gyrus, in which granule cells extend their dendrites toward the perforant path (APP/PS1+ct ab: 1.4 ± 0.3 vs APP/PS1+α-Aβ: 2.8 ± 0.2 SYN⁺ boutons (× 10⁶), p < 0.01) (Fig. 4A–D). Also, computer-assisted analysis of SYN intensity in the ML revealed a significant increase following the Aβ immunotherapy (APP/PS1+ct ab: 21.8 ± 2.0 vs APP/PS1+α-Aβ: 35.8 ± 3.0 staining intensity; p < 0.05).

Figure 4.

Aβ immunotherapy increased the number of SYN-positive boutons in the ML of the hippocampus. A, B, SYN-positive synapses in the hippocampal outer ML from controls (ct ab; A) and Aβ immunotherapy-treated APP/PS1 mice (B). Dotted lines depict the area analyzed stereologically for quantification of SYN-positive boutons. Arrowheads show the presence of β-amyloid plaques. C, D, High magnification details of the yellow squares in A and B, respectively. The arrowhead in C depicts dystrophic boutons around a β-amyloid plaque. Stereological quantification of SYN-positive boutons in the ML revealed higher SYN levels in Aβ immunotherapy-treated mice (APP/PS1+ct ab: 1.4 ± 0.3 vs APP/PS1+α-Aβ: 2.8 ± 0.2 SYN⁺ boutons (× 10⁶); p < 0.01; unpaired t test). Scale bars: A, 120 μm; C, 20 μm.

To exclude nonspecific effects of the Aβ immunotherapy on neurogenesis, we administered the antibody to 4 months old non-tg wild-type mice. As expected, the numbers of new cells were substantially higher than in the older mice. The Aβ immunotherapy did not induce any significant changes in numbers of proliferating or differentiating cells in wild-type mice (Table 2). These data indicate that binding of the antibody to human Aβ aggregates, rather than nonspecific effects of IgG2, mediated the effects of Aβ immunotherapy on neurogenesis in the dentate gyrus. Likewise, quantification in the dentate gyri of markers for gliogenesis, proliferation, and neuroinflammation did not reveal any differences between antibody and vehicle treatments of non-tg wild-type mice (Table 2).

Table 2.

Effects of Aβ immunotherapy on cell genesis in dentate gyrus of 4 month-old non-tg, wild-type mice

Staining	Vehicle	Aβ Immunotherapy
BrdU/NeuN	2466 ± 260	2231 ± 657
BrdU/S100β	29.5 ± 21	61 ± 52
PSA–NCAM	5698 ± 1689	5702 ± 681
PCNA	49.2 ± 6.4	38.4 ± 4.8
GFAP	8797 ± 1964	7433 ± 1194
Iba1	3097 ± 497	3818 ± 355
CD11b	0	0

Markers: BrdU/NeuN and PSA–NCAM for neurogenesis; BrdU/S100β for gliogenesis; PCNA for cell proliferation; and GFAP, Iba1, and CD11b for neuroinflammation. Values represent stereological cell counts and are means ± SEM; n = 5 (vehicle) and n = 4 (Aβ immunotherapy). No significant differences between groups were observed.

Passive Aβ immunotherapy restored dendritic branching and spine density and morphology of new mature neurons in APP/PS1-transgenic mice

To investigate whether Aβ immunotherapy affects the dendritic morphology of the new granule cells also when they have matured, we repeated the α-Aβ treatment in mice that received retroviral injections into the dentate gyrus to label dividing cells with GFP. The GFP⁺ granule cells were analyzed 5 weeks after the injection, a survival time considered to be sufficient for the newly born neurons to mature (van Praag et al., 2002; Zhao et al., 2006). We observed that the dendritic arborizations of the GFP⁺, new, and mature granule cells were significantly fewer in dentate gyri of vehicle-treated APP/PS1 mice, whereas the numbers in α-Aβ antibody-treated mice resembled those of new neurons born in non-tg mice (Fig. 5A–D). Spines are the principal sites of excitatory synaptic transmission and can be classified according to their shapes and appearance. We classified spines as proposed by Peters and Kaiserman-Abramof (1970). “Long-thin” spines have a total length greater than the neck diameter; “stubby” spines are short and wide, whereas “mushroom” spines have a thick neck and a large irregular head. We found significant reduction in all three spine subgroups on GFP⁺ cells in ct ab-treated APP/PS1 mice compared with non-tg mice. The α-Aβ treatment reversed this loss of spines by ∼50% for long-thin, stubby, and mushroom spines (Fig. 5E–H).

Figure 5.

Aβ immunotherapy increased dendritic branching and spine densities of mature retrovirally labeled newly born neurons. A–C, Representative confocal pictures of GFP⁺ mature new neurons in the three groups of the study. Scale bar, 60 μm. D, Scholl analysis of new mature granule cells labeled with retrovirus expressing GFP revealed lower dendritic complexity in APP/PS1 mice compared with non-tg wild-type mice and restored dendritic complexity in APP/PS1 mice during Aβ immunotherapy compared with vehicle-treated APP/PS1 mice. The graph represents the mean number of intersections between dendrites and concentric radii, centered at the cell body, as a function of distance from the soma; n = 15 cells/group. Error bars represent SEM *p < 0.05 non-tg versus APP/PS1, and APP/PS1 versus APP/PS1 treated with α-Aβ. E, Representative high-magnification segments from dendrites of new mature neurons in non-tg, vehicle-treated APP/PS1, and α-Aβ-treated APP/PS1 mice. Scale bar, 10 μm. F–H, Computer-assisted classification of spines along 40 μm segments detected differences in the number of mushroom, long-thin, and stubby spines. F, Significant reduction of numbers of mushroom spines in APP/PS1 compared with non-tg mice and significant rescue during Aβ immunotherapy. G, Significant reduction of numbers of long-thin spines in APP/PS1 compared with non-tg mice and significant rescue during Aβ immunotherapy. H, Significant reduction of numbers of stubby spines in APP/PS1 compared with non-tg mice and a trend (p = 0.07) toward rescue during Aβ immunotherapy. N = 50 segments/group. Error bars represent SEM *p < 0.05, **p < 0.01, ***p < 0.001. ANOVA, followed by Bonferroni post hoc tests.

Passive Aβ immunotherapy increased angiogenesis in dentate gyrus of APP/PS1-transgenic mice

To explore the previously described close links between neurogenesis and angiogenesis in the stem cell niche (Palmer et al., 2000), we determined whether the enhancement of neurogenesis in APP/PS1 treated with passive Aβ immunotherapy was associated with increased angiogenesis. Following the treatment, many BrdU⁺ cells were located adjacent to blood vessels, and numbers of blood vessels, quantified by endothelial stains for lectin and Glut1, were significantly higher in α-Aβ-treated APP/PS1 mice compared with control antibody-treated mice (APP/PS1+ct ab: 5991 ± 384 vs APP/PS1+α-Aβ: 8978 ± 766 blood vessels; p < 0.01) (Fig. 6D). Stereological estimation of vessel length detected a trend toward an increase following antibody treatment (APP/PS1+ct ab: 0.17 ± 0.03 vs APP/PS1+α-Aβ: 0.31 ± 0.06 m; p = 0.07). Combining the ThioS staining of CAA with the Glut1 staining of endothelial cells often showed disruption of Glut1 immunoreactivity in the presence of ThioS-stained CAA around vessels (Fig. 6A–C), indicating that progressive accumulation of compact β-amyloid leads to the deterioration of endothelial tissues. Preventing or delaying this deposition by passive Aβ immunotherapy may reverse this effect and result in higher numbers of blood vessels, as described above. Together, these findings indicate that passive immunotherapy can lead to increased angiogenesis as well as neurogenesis.

Figure 6.

Aβ immunotherapy increased angiogenesis in the dentate gyrus. Stereological estimation of the number of blood vessels indicated significant increases in the number of blood vessels in APP/PS1 mice during Aβ immunotherapy (APP/PS1+ct ab: 5991 ± 384 vs APP/PS1+α-Aβ: 8978 ± 766 blood vessels; p < 0.01; unpaired t test). A, B, Confocal images of ThioS⁺-Aβ-deposits along the wall of blood vessels in the cortex. C, Disruption of Glut1 expression in the presence of congophilic amyloid angiopathy. Arrow depicts colocalization of Aβ with Glut1, excluding the possibility that Glut1 was present but not detected because of epitope-masking by β-amyloid. D, Glut1-stained blood vessels (blue) and BrdU (red) in the dentate gyrus: many BrdU⁺ cells are located in the vicinity of blood vessels. Glut1 receptor is expressed at highest levels in the endothelial cells of barrier tissues such as the blood-brain barrier. Dotted lines in A depict SGZ. Scale bars: A, 40 μm; D, 70 μm.

Localization of cellular prion protein on dendrites of newly born neurons

A possible role for PrP^c in mediating synaptic toxicity of Aβ oligomers was described recently (Lauren et al., 2009). To explore whether PrP^c is present on retrovirus-labeled, GFP⁺ new neurons, we analyzed colocalization of PrP^c with GFP by using ColocImaris software on high-magnification images to specifically identify colocalized proteins and to exclude overlay artifacts. As expected, PrP^c was present ubiquitously on cells throughout the brain. PrP^c was also present, at equal levels compared with pre-existing brain cells, on the dendrites of retrovirus-labeled, GFP⁺ newly born neurons. PrP^c staining on dendrites of newly born neurons had a patchy appearance consistent with its known cellular distribution (Hantman and Perl, 2005), and ∼5% of the analyzed dendritic surface of newly born neurons was costained for PrP^c (Fig. 7). PrP^c was present at equal levels on dendrites of newly born neurons in both non-tg, wild-type, and APP/PS1 mice, suggesting that the expression of the APP and PS1 transgenes under the control of the PrP^c and PDGFβ promoters, respectively, did not affect the colocalization of PrP^c with dendrites of newly born neurons. Aβ immunotherapy had no obvious effect on the colocalization of PrP^c with dendrites of newly born neurons.

Figure 7.

Expression of cellular prion protein (PrP^c) on the dendrites of newly born neurons. A, Colocalization analysis of PrP^c (red) with the GFP⁺ dendrites (green) and overlay (yellow) by ColocImaris software. Inset, Maximum projection of this image stack. B, Three-dimensional reconstruction using Imaris Surpass of both PrP^c and GFP channels shows the broad presence of PrP^c on brain cells along with its colocalization on the dendrites of newly born neurons. Scale bar, 4 μm.

Discussion

The results of this study establish that reductions in compact β-amyloid plaque load in adult APP/PS1 doubly transgenic mice by passive Aβ immunotherapy rescue impairments in adult hippocampal neurogenesis. Following Aβ immunotherapy, more new neurons survived, and their morphological deficits were reversed. Although the untreated transgenic APP/PS1 mice exhibited abnormalities in dendritic branching and lower numbers of all spine subtypes in the new hippocampal neurons, the mice treated with Aβ immunotherapy had longer dendrites, and more complex dendritic arborization with restored stubby, mushroom, and long-thin spines. The morphology of the newly born hippocampal neurons in transgenic mice subjected to Aβ immunotherapy resembled that in untreated non-tg mice consistent with the possibility of rescued synaptic structures. These results were likely related to the specific interaction of the antibody with human Aβ aggregates deposited in brains of APP/PS1 mice—and possibly to the lowering of β-amyloid pathology—rather than to nonspecific effects of the antibody, as indicated by the absence of stimulated neurogenesis in the dentate gyrus of antibody-treated non-tg, wild-type mice.

The stimulation of neurogenesis and the improved dendrite and spine morphology of the new neurons could be related to a more favorable microenvironment created by antibody-related neutralizing events and the reduced neurotoxicity. In line with this interpretation, passive Aβ-immunotherapy ameliorates synaptic deficits on already formed, mature hippocampal and cortical neurons (Buttini et al., 2005). Moreover, the formation of new hippocampal neurons in APP-transgenic mice is increased by active Aβ vaccination (Becker et al., 2007). In these mice, preventive active vaccinations with Aβ-related antigens were initiated before the onset of β-amyloid plaque formation, and the immune response was then boosted >8–9 months. Compared with the study of Becker et al. (2007), we report here the following novel observations: First, we demonstrate that Aβ-immunotherapy improves neurogenesis in adult animals with substantial pre-existing β-amyloid pathology, more closely resembling the stage of pathology in human patients with a clinical diagnosis of probable AD. Second, we show that active vaccination associated with stimulation of cellular immunity is not necessary to increase neurogenesis in vivo but that humoral immunity induced by passive antibody transfer is sufficient. Third, we provide the first experimental evidence that Aβ immunotherapy not only enhances hippocampal neurogenesis but also reverses deficits in the morphological development of the new neurons. Because we found that the effects on neurogenesis were achieved in the presence of considerable amounts of Aβ in the brain, our data indicate that already the initial phase of antibody-mediated β-amyloid clearance is associated with beneficial therapeutic effects.

The involvement of adult hippocampal neurogenesis in learning and memory is still unclear, but a growing body of evidence supports a role of the new neurons in memory acquisition (Snyder et al., 2005; Kee et al., 2007; Jessberger et al., 2009). If the birth of new granule cells in the SGZ/GCL leads to their recruitment into preexisting neuronal circuits contributing to hippocampal functions related to learning (Kempermann, 2008), it is conceivable that the up-regulation of neurogenesis reported here may be related to the reversal of cognitive deficits. In support, restored behavior in transgenic mice treated with Aβ immunotherapy has been observed after both active vaccination and passive antibody transfer (Janus et al., 2000; Morgan et al., 2000; Dodart et al., 2002; Chen et al., 2007). We found that new granule cells exhibited expression of the activity-dependent immediate early gene Zif268 (Egr-1). Because Zif268 expression is up-regulated by neuronal activity, this result is consistent with functional integration of the new neurons into pre-existing neural circuitries.

Dendritic spines are the major sites for excitatory synaptic input. Although large stubby spines and mushroom spines are stable and required for the maintenance of synapses, small, long-thin spines are unstable and thought to be required for memory acquisition (Kasai et al., 2003). The numbers of all types of dendritic spines were increased during Aβ immunotherapy. Together with the longer dendrites and more ramified dendritic arborizations, these findings are indicative of better functional integration of the new neurons in treated compared with untreated transgenic mice. In support of this possibility, stereological estimation of SYN-positive boutons in the area in which the dendrites of the new neurons receive synaptic inputs from the perforant path, the dentate outer ML, revealed increases following Aβ immunotherapy. Spine densities as well as dendritic arborization and length of newly born neurons were already decreased in 4-month-old APP/PS1 before the onset of β-amyloid plaque deposition in the dentate gyrus consistent with the possibility that smaller, oligomeric Aβ species mediated the impairment. As a corollary, the rescue during Aβ immunotherapy may also be related to neutralization of oligomeric Aβ species by the antibody reacting with Aβ aggregates.

Our data also show that Aβ immunotherapy increased the numbers of blood vessels in APP/PS1 doubly transgenic mice compared with control-treated mice. More numerous blood vessels may be related to increased neurogenesis by creating niches for the new cells. It is known that increased angiogenesis can stimulate neurogenesis via vascular endothelial growth factor, and that it can be associated with improved functions. In addition, it has been suggested that angiogenesis links external stimuli including exercise or learning to neurogenesis (Jin et al., 2002; Cao et al., 2004; van Praag et al., 2005; Udo et al., 2008).

We obtained no evidence for changes in numbers or morphological phenotypes of Iba1⁺ microglia or in the expression of CD11b after 3.5 months of Aβ immunotherapy. This finding raises the possibility that microglia were either not activated further over levels in non-tg controls or that they were only transiently activated during immunotherapy and have returned to levels of control-treated transgenic mice after 3.5 months of treatment as described previously (Wilcock et al., 2004b; Dickstein et al., 2006). The fact that there were no significant changes in plasma Aβ₄₂in α-Aβ-treated mice argues against a significant role of peripheral Aβ sink in our experiments (DeMattos et al., 2001).

The mechanism linking Aβ-related toxicity to impaired neurogenesis and to defective morphology of newly born neurons identified in our experiments is unknown. A possible role for PrP^c in mediating Aβ-related impairments of synaptic plasticity involves PrP^c functions as a cell surface receptor that binds oligomeric Aβ aggregates to a domain of PrP^c that is associated with neurodegeneration and with impaired hippocampal LTP in response to Aβ oligomer binding (Lauren et al., 2009). Interestingly, we found that PrP^c was present on the dendrites of newly born neurons at approximately equal levels compared with pre-existing brain cells, covering 5% of the dendritic surface of newly born neurons. Dendritic PrP^c had a patchy appearance consistent with a possible presence of PrP^c on dendritic spines of the newly born neurons, which may explain the observed vulnerability of dendritic spines in our APP/PS1 mice as well as the recovery of spine density and dendritic morphology during Aβ immunotherapy.

In conclusion, our study shows that passive Aβ immunotherapy can restore hippocampal neurogenesis in APP/PS1-transgenic mice. In particular, Aβ immunotherapy rescued the length and arborization of dendrites of newly born neurons and restored the density of mushroom and long-thin spines. The newly born neurons showed signs of functional activity indicated by expression of the immediate early gene Zif268 in BrdU⁺/NeuN⁺ cells. Aβ immunotherapy increased the numbers of SYN-positive boutons, and it increased angiogenesis in APP/PS1 mice with pre-existing β-amyloid pathology. The presence of PrP^c on the dendrites of GFP⁺ newly born neurons provides suggestive evidence that it may mediate Aβ-related toxicity in these cells. The rescued neurogenesis and morphological development of the newly born neurons constitute a novel mechanism that could explain part of the functional improvements during passive Aβ immunotherapy. Our findings provide the first evidence that antibody-related stimulation of endogenous stem cells and the development of their progeny, mediated through clearance or neutralization of the pathological microenvironment, may become a new therapeutic strategy for the rehabilitation of functions compromised by the toxicity of Aβ-related peptides in AD.