|Home | About | Journals | Submit | Contact Us | Français|
Although β-amyloid (Aβ) plaques and tau neurofibrillary tangles are hallmarks of Alzheimer’s disease (AD) neuropathology, loss of synapses is considered the best correlate of cognitive decline in AD, rather than plaques or tangles. How pathological Aβ and tau aggregation relate to each other and to alterations in synapses remains unclear. Since aberrant tau phosphorylation occurs in amyloid precursor protein (APP) Swedish mutant transgenic mice, and since neurofibrillary tangles develop in triple transgenic mice harboring mutations in APP, tau and presenilin 1, we utilized these well-characterized mouse models to explore the relation between Aβ and tau pathologies. We now report that pathological accumulation of Aβ and hyperphosphorylation of tau develop concomitantly within synaptic terminals.
AD neuropathology is traditionally characterized by the abnormal deposition of Aβ in extracellular plaques and tau in intracellular tangles. More recently, early intraneuronal accumulation of Aβ42, the most pathogenic Aβ species, has been described in AD (Alafuzoff et al., 2008; Cataldo et al., 2004; D'Andrea et al., 2001; Gouras et al., 2000; Ohyagi et al., 2005), Down syndrome (Busciglio et al., 2002; Cataldo et al., 2004; Gouras et al., 2000; Gyure et al., 2001; Mori et al., 2002), and transgenic AD mouse models (Lord et al., 2006; Oakley et al., 2006; Oddo et al., 2003; Sheng et al., 2003; Shie et al., 2003; Stokin et al., 2005; Takahashi et al., 2002; Van Broeck et al., 2008; Wirths et al., 2001; Zerbinatti et al., 2006). Further, transgenic AD mice develop physiological and behavioral abnormalities prior to plaques (Chapman et al., 1999; Holcomb et al., 1998; Moechars et al., 1999) but concomitant with intraneuronal Aβ peptide accumulation (Bayer and Wirths, 2008; Billings et al., 2005; Cruz et al., 2006; Echeverria et al., 2004; Knobloch et al., 2007; Lord et al., 2006; Oddo et al., 2003), supporting that intraneuronal Aβ peptides are involved in the initiation of AD pathogenesis (Gouras et al., 2005).
Evidence supports that Aβ accumulation precedes and promotes tau pathology. Crossbreeding of mutant amyloid precursor protein (APP) transgenic mice with (Lewis et al., 2001) or intracerebral injection of Aβ into tau mutant transgenic mice (Gotz et al., 2001) led to enhanced tau pathology. In human brains with early AD changes or Down syndrome, intraneuronal Aβ42 accumulation in CA1 pyramidal cell bodies preceded hyperphosphorylation of tau (Gouras et al., 2000; Gyure et al., 2001). In the 3xTg-AD mouse harboring mutations in APP, tau and presenilin, intraneuronal Aβ accumulation in cell bodies preceded tau hyperphosphorylation, and Aβ antibodies reduced both Aβ and tau pathologies (Oddo et al., 2004; 2003). Recent evidence that behavioral deficits in transgenic mouse models of AD can be attenuated by reduction in tau (Roberson et al., 2007) further underscores the relevance in elucidating the biological mechanism(s) linking Aβ and tau. Here we analyze the relation between Aβ42 and phosphorylated tau in two well established transgenic mouse models of AD and utilize the anatomy of the hippocampus to co-localize both early Aβ42 accumulation and tau phosphorylation to synapses.
Aβ42 antibody AB5078P (Chemicon, Temecula, CA) is a rabbit polyclonal antibody directed against the C-terminus of Aβ42 that was previously biochemically characterized (Kamal et al., 2001). The specificity of this Aβ42 antibody was additionally shown by absence of immunofluorescence in cultured neurons derived from well-established APP knockout mice (Zheng et al., 1995) compared to wild-type mice (Almeida et al., 2006). The well-established antibody AT8 (Endogen, Rockford, IL) detects tau phosphorylated at serine 202 and threonine 205. MC1 antibody recognizes a conformational tau epitope in paired helical filaments (Jicha et al., 1997). Tau antibody 12E8 detects tau phosphorylated at Ser 262 and 356 (Litersky et al., 1996). AT180 tau antibody (Endogen, Rockford, IL) is directed against the phosphorylated Thr231 residue.
All mouse experiments were performed in compliance with the institutional guidelines of the Institutional Animal Care and Use Committee of Weill Cornell Medical College in accordance with the National Institutes of Health guidelines. Brain sections from Tg2576 mice, wild-type mice, and heterozygous 3xTg-AD mice at 5 and 13 months of age, were analyzed with at least n=2–3 mice per group. Preparation of tissue sections from Tg2576 mice (Hsiao et al., 1996) and 3xTg-AD mice (Oddo et al., 2003) was similar to that described previously (Takahashi et al., 2002). Mice were anesthetized with sodium pentobarbital (150 mg/kg, i.p.) and perfused via the ascending aorta with 3.75% acrolein (Polyscience, Warrington, PA) and 2% paraformaldehyde in 0.1 M phosphate buffer (PB), pH 7.4. Vibratome-cut 40 µm tissue sections were kept in storage buffer composed of 30% sucrose and 30% ethylene glycol in PB at −20°C.
Vibratome-cut 40 µm tissue sections of human cortical brain biopsy tissue kept in storage buffer at −20°C was from neurologically normal controls (n=2; ages 44 and 54) and subjects with AD (n=2; ages 54 and 62) originally obtained from the Department of Pathology, Weill Medical College of Cornell University, as a result of neurosurgical procedures unrelated to this study (Takahashi et al., 2004; 2002).
Immunolabeling for light microscopy was performed as previously described (Milner et al., 1998; Takahashi et al., 2004). Free-floating sections were incubated in primary antibodies for 24 h at room temperature and then for 40–48 h at 4°C (Chemicon Aβ42 antibody, 1:150; AT8, AT180, 12E8, 1:500; MC1, 1:10). The sections were incubated in biotinylated horse anti-mouse immunoglobulin (IgG) secondary antibody (1:400, Vector Laboratories, Burlingame, CA) for 30 min, followed by the peroxidase-avidin complex (Vectastain ABC kit, Vector) for 30 min. The secondary antibody was diluted in 0.1 M Tris-saline (pH 7.6) containing 0.1% bovine serum albumin (BSA). The reaction product with the ABC kit was visualized after incubation of sections with 3, 3’-diaminobenzidine (Aldrich Chemical, Milwaukee, WI) and hydrogen peroxide. The sections were observed using a system consisting of a Nikon Eclipse E600 microscope (Morrell Instrument Co., Melville, NY) equipped with a computer-controlled LEP BioPoint motorized stage (Ludl Electronic Products, Hawthorne, NY), a DEI-750 video camera (Optronics, Goleta, CA), and a Dell Dimension 4300 computer (Dell, Round Rock, TX).
For immunofluorescence, free-floating sections were first incubated in primary antibodies (Chemicon Aβ42 antibody, 1:150; AT8, 1:500) for 24 h at room temperature and for 40–48 h at 4°C, followed by appropriate fluorescent secondary antibodies Alexa 488 goat anti-rabbit IgG (green) and/or Alexa 546 goat anti-mouse IgG (red) (1:500; Molecular Probes, Eugene, OR) for 1 h at 37°C. Nuclei were stained using 1 µg/ml Hoechst dye in PB (Bisbenzimide H33258, Sigma-Aldrich, St Louis, MO). Images were taken using either a Leica DM IRB microscope or an Olympus IX70 microscope with a Hamamatsu digital camera.
Immuno-EM localization was performed as previously described (Takahashi et al., 2002). Free-floating sections for dual-labeling immuno-EM with Aβ42 (1:50) and AT8 (1:500) antibodies were first processed for immunoperoxidase localization with AT8 antibodies as described above. Sections then were processed by the immunogold-silver method for localization of Aβ42 antibody. For this, sections were incubated with goat anti-rabbit IgG conjugated to 1 nm gold particles (AuroProbe One; Amersham Biosciences, Arlington Heights, IL) in 0.01% gelatin and 0.08% BSA in PBS, pH 7.4, for 2 hours at room temperature. Sections were rinsed in PBS, postfixed in 2% glutaraldehyde in PBS for 10 minutes, and rinsed in PBS and 0.2 M sodium citrate buffer (pH 7.4). Conjugated gold particles were enhanced by treatment with silver solution (Amersham Biosciences). Sections were fixed in 2% osmium tetroxide in PB, dehydrated, embedded in EMBed 812, sectioned (65- to 76-nm thick), and counterstained with uranyl acetate and Reynolds’ lead citrate. Final preparations were examined using a Philips CM10 electron microscope. Illustrations were generated from a high-resolution digital imaging CCD camera system (Advanced Microscopy Techniques, Danvers, MA) and processed using Adobe Photoshop 7.0 (Adobe System, Mountain View, CA). Aβ42 or AT8 immunolabeled profiles were classified as previously described (Takahashi et al., 2002). Somata were identified by the presence of a nucleus. Dendrites contained regular microtubule arrays and were usually postsynaptic to axon terminal profiles.
Since Aβ42 accumulates with aging especially in distal neurites of AD transgenic mouse brains (Takahashi et al., 2004; 2002), and since hyperphosphorylated tau localizes to dystrophic neurites around plaques in human subjects, as well as APP mutant transgenic mice (using antibodies AT8, PHF1, R27, R32 and Alz50) (Moechars et al., 1999; Otth et al., 2002; Sturchler-Pierrat et al., 1997), we investigated whether there was a relation between Aβ42 accumulation and hyperphosphorylated tau in dystrophic neurites of Tg2576 mice harboring Swedish mutant APP. Remarkably, hyperphosphorylated tau (using antibody AT8) co-localized with Aβ42 within dystrophic neurites around plaques (Fig. 1A). We next examined alterations in tau, considered normally to be an axonal protein, in relation to Aβ42 in dystrophic neurites around plaques in Tg2576 mice at an ultrastructural level, utilizing dual-labeling immuno-EM. Tau was hyperphosphorylated and mislocalized, near to sites of Aβ42 within dendritic profiles, including postsynaptic compartments (Fig. 1B,C). Hyperphosphorylated tau was evident on tubular-filamentous structures and in clusters associated with the microtubule network near to Aβ42 immuno-gold particles. Similar localization of hyperphosphorylated tau near sites of Aβ42 in neurites was also evident in human AD biopsy brain tissue (Fig. 1D).
To further elucidate the relation between Aβ and tau, we next examined brain sections from 3xTg-AD mice, which develop neurofibrillary tangles as well as prominent early intraneuronal Aβ and subsequent plaques (LaFerla et al., 2007). Notably, prior to plaque formation, Aβ42 immunoreactivity in the CA1 region of the hippocampus of 5 month old 3xTg-AD mice was not only prominent in CA1 pyramidal neuron somata, but was also pronounced in the stratum lacunosum-moleculare (SLM) (Fig. 2A). The SLM contains the postsynaptic compartments of distal CA1 apical dendrites. In fact, band-like plaque deposition in the SLM of the CA1 region is often evident in images of the hippocampus in published studies on transgenic mouse models of AD. Remarkably, we noticed that hyperphosphorylated tau (AT8) was especially prominent in the SLM of CA1 in these 5-month-old 3xTg mice (Fig. 2A). A similar pattern of immunoreactivity in the SLM was also evident using additional phospho-tau specific antibodies (antibodies AT180 and 12E8; data not shown) and the abnormal tau conformational specific MC1 antibody (Supplementary Fig. 1). Such SLM staining is also evident, although not specifically commented on, in the original study characterizing tau phosphorylation in 12-month-old 3xTg mice using antibodies AT8, PHF1 and MC1 (Oddo et al., 2003). At 13 months of age, when 3xTg-AD mice deposit further plaques and intraneuronal Aβ42 declines in CA1 pyramidal somata, the SLM remained the most prominent site of both Aβ42 and hyperphosphorylated tau in the CA1 region, evident as the yellow staining (overlap of Aβ42, red, and AT8, green) of the SLM (Fig. 2B,C).
To determine whether Aβ42 and hyperphosphorylated tau within SLM of 3xTg-AD mice were intraneuronal and to investigate their potential subcellular relation, we again used dual-labeling immuno-EM. Remarkably, in the SLM, hyperphosphorylated tau specifically localized to CA1 dendritic profiles and their postsynaptic compartments, and in particular those demonstrating Aβ42 accumulation (Fig. 3A, B). In some Aβ42 accumulating neurites, AT8-positive filaments were tangled (Fig. 3C, inset).
We quantified the number of identifiable dendritic terminals in a total area of 1406 µm2 in the Aβ42/AT8 dual-immuno-EM images of the SLM of 13 month-old 3xTg mouse brain. We stratified them as Aβ42-gold and AT8-peroxidase negative/negative, positive/negative, negative/positive and positive/positive. Of a total of 72 identifiable dendritic terminals, the majority (78%) were either −/− or +/+. Specifically, 48.6% were −/−, while 29.2% were +/+. Considerably fewer dendritic terminals were only positive for Aβ42 (13.9%) or only positive for AT8 (8.3%).
We noted that dendritic terminals positive for AT8 were positive throughout the length of their course within the plane of the EM section. In contrast, Aβ42 gold localized to discrete locations within post-synaptic compartments. This discrete labeling pattern could lead to an underestimation of Aβ42 relative to AT8-labeled tau. There should not be false negatives for AT8, while false negatives are expected for Aβ if gold labeling is either above or below the plane of section in a given dendritic terminal. Additional factors resulting in underestimation of Aβ42 relative to AT8 include: (1) immuno-gold labeling identifies less antigen than immuno-peroxidase using our dual-immuno EM method. (2) Aβ42 antibodies have conformational specificity, so that not all Aβ species are detected. We previously showed that another Aβ42 end-specific antibody predominantly reacts to Aβ42 monomers (Takahashi et al., 2004). The Chemicon Aβ42 specific antibody similarly reacts preferentially to monomers on Western blot (data not shown).
Overall, the Aβ42/AT8 dual immuno-EM supports the conclusion of co-occurrence of Aβ and tau pathologies in dendritic terminals of the SLM. In contrast to labeling of dendrites, Aβ42 and AT8 did not co-occur in axons/pre-synaptic terminals in the SLM of 13-month-old 3xTg-AD mice.
Decline in the level of intracellular Aβ42 in neuron somata with plaque pathology (Gouras et al., 2000; Mori et al., 2002; Oddo et al., 2003) has been a critique regarding the importance of intraneuronal Aβ (Duyckaerts et al., 2008). It was proposed that intraneuronal Aβ42 declines from the emergence of extracellular Aβ plaques (LaFerla et al., 2007). Alternatively, peptide competition from abundant Aβ42 within plaques was hypothesized to lead to the erroneous appearance of reduced intraneuronal Aβ42 (Gouras et al., 2005). To test this latter possibility, we now co-processed CA1 hippocampal sections of 3xTg-AD mice with marked intraneuronal Aβ42 either with sections with or without plaques. Co-incubation with plaque-containing compared to non-plaque-containing sections had no influence on intraneuronal Aβ42 (Supplementary Fig. 2). This new data argue against peptide competition and confirm the age-related reduction of intraneuronal Aβ42 in cell bodies. Indeed, immuno-EM studies emphasized age-related intraneuronal Aβ42 accumulation within dystrophic neurites and synaptic compartments rather than in cell bodies of Tg2576 mice with aging (Takahashi et al., 2002). In contrast to Aβ42 accumulation, hyperphosphorylated tau (AT8) labeling is present throughout the cytoskeleton in vulnerable CA1 pyramidal neuron cell bodies and their apical dendrites by immuno-EM in older 3xTg-AD mice (Fig. 3D).
The relationship between Aβ and tau is a central question in AD research, and synaptic alterations are the best correlate of cognitive dysfunction in AD (Coleman and Yao, 2003). Although plaques and tangles are not obviously linked to synapses, our work shows that Aβ and tau pathologies co-occur at synapses. Taking advantage of the welldefined anatomy of the hippocampus, we found that Aβ42 accumulation and the mislocalization and hyperphosphorylation of tau in the CA1 region occurred early and prominently in distal dendrites and postsynaptic compartments of pyramidal neuron apical dendrites in the SLM of 3xTg-AD mice. Ultrastructural analysis of the SLM was required to show that these events were in fact intracellular. Our data emphasize synapses not only as early sites of Aβ42 accumulation but now also link the initiation of tau pathology to Aβ42 accumulation in postsynaptic terminals.
We highlight that with age, intraneuronal Aβ42 decreases in cell bodies but not in dystrophic neurites and synapses. In contrast, tau phosphorylation is maintained throughout cell bodies and apical dendrites in 3xTg-AD mice. Our data provide new insights into the apparent anatomical dissociation of plaque and tangle pathologies in AD. Aβ42 accumulates progressively in distal processes whereas after initial increases, Aβ42 declines in cell bodies, with plaques developing preferentially in distal processes and synapses. In contrast, hyperphosphorylated tau develops throughout processes and cell bodies, with tangles eventually developing in neuron cell bodies and neuropil threads in neurites. We hypothesize that extracellular release of Aβ from degenerating neurites might up-regulate Aβ42 (Glabe, 2001) within adjacent synaptic compartments, leading to the synaptic spread of AD. Our present data emphasizes pathology in postsynaptic CA1 apical dendrites, because of the well-defined anatomy of the hippocampus, although post- and also pre-synaptic compartments with intraneuronal Aβ42 accumulation develop at other sites.
In summary, we show that tau alterations are spatially associated with intraneuronal Aβ accumulation, providing a link between plaques and tangles, the two neuropathologic hallmarks of AD. Aβ and tau pathologies develop in synapses, a key observation given the importance of synaptic alterations in cognitive dysfunction in AD. The spatial association of Aβ accumulation with tau pathology within neurites and synaptic compartments reinforces the importance of intraneuronal Aβ accumulation in AD.
Sections of 3xTg-AD mouse hippocampus show abnormal tau conformation with MC1 antibody already in 5-month-old 3xTg-AD compared to wild type mice (A, B). Notably, abnormal tau conformation immunoreactivity was not only prominent in CA1 pyramidal neuron somata, but was also pronounced in the SLM (plus sign). Progressively increased MC1 immunoreactivity was observed at 10 month of age. (C). No positive labeling was observed in 5 or 10 month-old wild-type mice (B, D). Insets: Higher magnification of the SLM. Abbreviation: SLM, stratum lacunosum-moleculare. Scale bar: 100 µm.
Sections of AD transgenic mouse CA1 hippocampus with high intraneuronal Aβ42 staining (5 month old 3xTg-AD mouse; A and C) were co-processed either with sections without plaques (2 month old Tg2576 mouse; B) or with high plaque burden (23 month old Tg2576 mouse, D). Co-incubation with the high plaque-containing section did not reduce levels of intraneuronal Aβ42 in CA1 pyramidal neurons of 3xTg-AD mouse (C) compared to co-incubation with the no plaque-containing section (A), arguing against peptide competition from Aβ42 within plaques as a cause of reduced intraneuronal Aβ42 immunoreactivity. Note also the age-related reduction of intraneuronal Aβ42 in the CA1 neurons in the section from the 23 month-old compared to the 2 month-old Tg2576 mouse. Scale bar: 100 µm.
Supported by an Alzheimer’s Association Zenith award and National Institutes of Health grants NS045677, AG027140, AG028174 (GKG), and HL18974 (TAM). We thank Drs. Claudia Almeida, Jordi Magrane and Davide Tampellini for helpful discussions. We thank Drs. Frank LaFerla, University of California at Irvine, and Karen H. Ashe, University of Minnesota, for the 3xTg-AD and Tg2576 mice, respectively. We thank Dr. Peter Davies, Albert Einstein University School of Medicine for providing tau antibody MC1 and Dr. Peter Seubert, Elan Pharmaceuticals, for tau antibody 12E8.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
All authors disclose that there are no actual or potential conflicts of interest including any financial, personal or other relationship with other people or organizations.