Introduction
Alzheimer’s disease (AD) gradually damages the function and structure of particularly vulnerable brain areas, as those used for memory and cognition. Accumulation of aggregated proteins at the extracellular (amyloid-beta, Aβ) and intracellular (hyperphosphorylated tau) levels is one of the major abnormalities found in the brain of AD patients (revised in [
13,
26,
50]). Another key pathological feature, that defines this disease, is the early appearance of amyloid plaque-associated neuritic changes in the form of dystrophic neurites, together with a selective loss of connections and neuronal groups. Dystrophic neurites, defined as thickened or irregular neuronal processes, are considered to be an expression of a widespread alteration of the neuronal cytoskeleton. In AD, dystrophic axons are particularly abundant in the hippocampal fiber systems originating from the subiculum, CA1, and the entorhinal cortex [
57]. Synaptic loss in both the neocortex and the hippocampus is, so far, the best pathological correlate of early cognitive decline [
19,
41,
52‐
54,
59,
61]. The initial neuritic degenerative changes may represent an early manifestation of axonal damage that precede the appearance of synaptic loss and, therefore, a promising disease-modifying morphological target for early intervention strategies to reverse the damage and rescue the deteriorating neurons. Supporting the idea that these changes could be potentially reversible, a recent study has reported that dystrophic axons surrounding amyloid plaques remain connected to viable neuronal bodies over a relatively long period of time [
1].
The expression of the human amyloid precursor protein (APP) with single or double mutations in transgenic mouse lines leads to the formation of neuritic plaques with clusters of dystrophic neurites and glial recruitment that resembles the amyloid pathology seen in AD brains [
4,
28]. After plaque formation, neuritic abnormalities progressively develop as shown by in vivo multiphoton imaging in an AD model [
42]; however, the cellular dysfunction underlying the neuritic pathology is not well understood. Numerous autophagic vacuoles accumulate within dystrophic neurites in the brains of humans with AD and AD models [
46,
47,
66] and several lines of investigation support the notion that defects in the autophagy process, a cellular catabolic mechanism essential for the degradation of aggregated proteins and organelles, significantly contributes to AD pathogenesis [
11,
37,
46,
48]. Interestingly, it has been reported that autophagic compartments participate in APP processing and Aβ peptides production [
65,
66] suggesting a possible causal relationship between plaque formation and neuritic dystrophy. Remarkably, restoring the intracellular autophagy pathway ameliorates disease progression and cognition deficits in a transgenic model [
65] proving the potential therapeutic value of autophagy induction in early stages of the disease for neuronal function recovery.
Here, in this work, we have characterized the morphological and subcellular abnormalities associated with dystrophic neurites around plaques in the hippocampus of our PS1/APP model at 4–6 months of age. Such neuritic abnormalities may result in defects in maintaining axonal and synaptic terminals structure and function. As we have reported previously, this bigenic model reproduces major amyloid-induced pathogenic steps seen in humans. The most relevant feature of our model is that, unlike many other transgenic mice in which neuronal loss is not observed, the selective neurodegenerative phenotype with specific subsets of interneurons and pyramidal neurons is affected in hippocampus and entorhinal cortex, following a regional and temporal pattern [
5,
29,
44,
51]. This neuronal loss was found to be associated with a neurotoxic inflammatory response induced by soluble oligomeric Aβ peptides [
29]. A better understanding of the pathological basis of the neuritic changes, prior to neuronal loss in this model, will provide valuable insights into the potential causes of early axonal damage and synaptic dysfunction and will further improve the accuracy of preclinical evaluation of novel therapeutic agents intended to reverse axonal damage.
Materials and methods
Transgenic mice
The generation and characterization of PS1/APP transgenic (tg) mice has been reported previously [
5,
10,
12,
29,
30,
44,
51]. These double transgenic mice (C57BL/6 background) were obtained by crossing homozygotic PS1M146L transgenic mice with heterozygotic Thy1-APP751SL (Swedish: K670N, M671L and London: V717I FAD mutations) mice (Charles River, France). Mice represented F6–F10 offspring of heterozygous transgenic mice. Non-transgenic mice of the same genetic background and ages were used as controls. All animal experiments were carried out in accordance with the European Union regulations and approved by the committee of animal use for research at Malaga University.
Antibodies
The following primary antibodies were used in this study: anti-human amyloid precursor protein (hAPP) rabbit polyclonal (1:20,000, Sigma A8717); anti-Aβ (clone 6E10) mouse monoclonal (1:5,000, Sigma A1474); anti-oligomer A11 (recognizes Aβ42 oligomers but not monomers or fibrils) rabbit polyclonal (1:500; Millipore AB9234); anti-Aβ42 rabbit polyclonal (1:40; Biosource 44-344); anti-phospho-PHF-tau pSer202/Thr205 mouse monoclonal (clone AT8) (1:500; Pierce MN1020); anti-cofilin rabbit polyclonal (1:2,000, Cytoskeleton ACFLO2); anti-microtubule-associated protein 1 light chain 3 (LC3) goat polyclonal (1:1,000; Santa Cruz Biotechnology Sc16755); anti-neurofilament 150 kDa rabbit polyclonal (1:5,000; Millipore AB1981); anti-microtubule-associated protein 2 (MAP-2); rabbit polyclonal (1:5,000; Chemicon Ab5622); anti-synaptophysin rabbit polyclonal (1:1,000; Abcam ab14692); anti-vesicular GABA transporter (VGAT) guinea pig polyclonal (1:5,000; Calbiochem 676780); anti-vesicular glutamate transporter (VGLUT1) guinea pig polyclonal (1:10,000; Millipore AB5905); anti-human Aβ, N terminus (clone 82E1) mouse monoclonal (1:1,000, IBL 10323); anti-kinesin heavy chain (clone KN-01) mouse monoclonal (1:1,000, Abcam AB9097); anti-dynein, 74 kDa (clone 74.1) mouse monoclonal (1:1,000, Millipore MAB1618).
Tissue preparation
After deep anesthesia with sodium pentobarbital (60 mg/kg), 2-, 4- and 6-month-old control (WT), PS1 and PS1/APP tg mice were perfused transcardially with 0.1 M phosphate buffered saline (PBS), pH 7.4 followed by 4% paraformaldehyde, 75 mM lysine, 10 mM sodium metaperiodate in 0.1 M phosphate buffer (PB), pH 7.4. Brains were then removed, post-fixed overnight in the same fixative at 4°C, cryoprotected in 30% sucrose, sectioned at 40 μm thickness in the coronal plane on a freezing microtome and serially collected in wells containing cold PBS and 0.02% sodium azide.
For electronic microscopy, 4.5-month-old PS1/APP tg mice were perfused transcardially with 0.1 M phosphate buffered saline (PBS)/1% heparin, pH 7.4 followed by 2.5% glutaraldehyde–2% paraformaldehyde in 0.1 M phosphate buffer (PB), pH 7.4. After being removed, the brains were post-fixed in the same fixative overnight at 4°C, washed several times with PB, sectioned at 50 or 100 μm thickness in the coronal plane on a vibratome (Leica VT1000M) and serially collected in wells containing cold PB and 0.02% sodium azide. Then, the 100-μm sections were fixed in 2% osmium tetroxide in 0.1 M PB and dehydrated, to be finally embedded in Araldite (EMS, USA). Tissue blocks were cut serially into semithin (1.5 μm) with a diamond knife in a Leica ultramicrotome (EM UC6), placed on slides, stained with 1% toluidine blue and explored with the light microscope for amyloid plaques. Then, selected areas from semithins were cut in ultrathin sections. Ultrathin sections were placed on Formvar-coated grids and stained with uranyl acetate and lead citrate before being examined with an electron microscope (FEI Tecnai Spirit, OR, USA).
Light microscopy immunohistochemistry
Serial sections from control (WT) and both tg mice (PS1 and PS1/APP) were processed in parallel for light microscopy immunostaining using the same batches of solutions to minimize variability in immunohistochemical labeling conditions. Free-floating sections were first treated with 3% H2O2/10% methanol in PBS, pH 7.4 for 20 min to inhibit endogenous peroxidases, and with avidin–biotin Blocking Kit (Vector Labs, Burlingame, CA, USA) for 30 min to block endogenous avidin, biotin and biotin-binding proteins. Sections were immunoreacted with one or two of the primary antibodies over 24 or 48 h at room temperature. The tissue-bound primary antibody was then detected by incubating for 1 h with the corresponding biotinylated secondary antibody (1:500 dilution, Vector Laboratories), and then followed by incubating for 90 min with streptavidin-conjugated horseradish peroxidase (Sigma–Aldrich) diluted 1:2,000. The peroxidase reaction was visualized with 0.05% 3-3-diaminobenzidine tetrahydrochloride (DAB, Sigma–Aldrich), 0.03% nickel ammonium sulphate and 0.01% hydrogen peroxide in PBS. After DAB, sections immunolabeled for APP, MAP-2, neurofilament or synaptophysin were incubated 3 min in a solution of 20% of Congo red. Sections were then mounted on gelatine-coated slides, air dried, dehydrated in graded ethanol, cleared in xylene and coverslipped with DPX (BDH) mounting medium. Specificity of the immune reactions was controlled by omitting the primary antisera.
For double immunofluorescence labelings, sections were first sequentially incubated with the indicated primaries antibodies followed by the corresponding Alexa488/568 secondary antibodies (1:1,000; Invitrogen). APP-immunolabeled sections were stained with 0.02% thioflavine-S in 50° ethanol for 5 min. Sections processed for immunofluorescence were mounted onto gelatin-coated slides, coverslipped with 0.01 M PBS containing 50% glycerin and 3% triethylenediamine and then examined under a confocal laser microscope (Leica SP5 II).
Immunoelectron microscopy
Sections of 50 μm from 4.5-month-old PS1/APP mice were first washed with PBS and incubated in a 50 mM glycine solution 10 min in order to increase the antibody-binding efficiency. Following the standard immunohistochemical protocol, the tissue was incubated overnight in primary rabbit polyclonal antibodies anti-hAPP or anti-Aβ42 in a PBS 0.1 M/0.01% Tx-100/1% BSA solution at room temperature. Then, they were washed in PBS, and incubated with 1.4 nm gold-conjugated secondary antibody goat anti-rabbit IgG (1:100; Nanoprobes) for one night at room temperature. After postfixing with 1% glutaraldehyde and washing with 50 mM sodium citrate, the labeling was enhanced with the HQ Silver™ Kit (Nanoprobes). In negative control experiments, primary antibody was omitted. Then, the slices were processed by the standard fixation, dehydration and embedding steps.
Stereological analysis
Density and size of 6E10-positive amyloid plaques were obtained by stereology-based quantification in the hippocampal formation of PS1/APP at 6 and 18 months of age (
n = 4/age; 5 sections per animal) according to the optical fractionator method as previously described [
44]. Briefly, an Olympus BX61 microscope and the NewCAST software package (Olympus, Glostrup, Denmark) were used. In order to obtain the plaque density, the number of plaques was quantified in five sections through the antero-posterior extent of the hippocampus and then divided between the sampled areas. CA1 subfields were defined using a 10× objective and the number of plaques was counted using a 40× objective. The number of counting frames varied with the hippocampal region or subfield layer analyzed. We used a counting frame of 7,154.7 μm
2 with step lengths of 84.58 × 84.58 μm. Neurite and plaque sizes were estimated by the nucleator application with isotrophic probes (
n = 5 radii). The number of APP-positive dystrophic neurites per plaque was quantified over Congo red stained Aβ deposits. Each analysis was done by a single examiner blinded to sample identities.
Total protein extraction and Western blots
The protein pellets, obtained using the Tripure™ Isolation Reagent, were resuspended in 4% SDS and 8 M urea in 40 mM Tris–HCl, pH 7.4 and rotated overnight at room temperature. The protein content was evaluated using Lowry.
Western blots were performed as described previously [
12,
29,
51]. Briefly, 10–20 μg of protein from the different samples were loaded on 16% SDS-Tris-Tricine-PAGE and transferred to nitrocellulose (Hybond-C Extra, Amersham, Sweden). After blocking, the membranes were incubated overnight, at 4°C, with the appropriate antibody. The membranes were then incubated with anti-mouse horseradish-peroxidase-conjugated secondary antibody (Dako, Denmark) at a dilution of 1/8,000. The blots were developed using the ECL-plus detection method (Amersham, Sweden). For quantification, the scanned (Epson 3200) images were analyzed using PCBAS program. In each experiment, the intensity of bands from WT mice and/or experimental condition were averaged and considered as 1 relative unit. Data were always normalized by the specific signal observed in 6-month-old WT group.
Synaptosomes and microsomes preparation, soluble fractions isolation and immunoprecipitation
The synaptosomal fractions were obtained basically as described previously [
62]. Briefly, the tissue was homogenized (using a Dounce homogenizer) in 0.32 M Sucrose, 10 mM Tris–HCl (pH 7.4) buffer (buffer A) containing complete protease and phosphatase inhibitor cocktails (Sigma). After homogenization, the crude synaptosomal fraction (synaptosomes plus mitochondria) was isolated by two sequential centrifugations (1,500×
g, 10 min followed by 12,500×
g, 20 min; at 4°C). The crude synaptosomes were resuspended in 13% (final concentration) Ficoll 400 (in buffer A) and layered on the bottom of a discontinuous gradient, composed by buffer A and 7% Ficoll (in buffer A). The gradients were centrifuged at 100,000×
g (45 min at 4°C) and the synaptosomes were isolated at the 7.5–13% interface. After washing (twice with buffer A), the protein content of the synaptosomal fractions was quantified by Lowry.
The soluble and microsomal fractions (supernatant and pellet, respectively) from PS1/APP and WT mice were obtained after centrifugation at 100,000×
g (1 h, 4°C) as described previously [
29,
30].
The A11 or 6E10 immunoprecipitation experiments were also performed as described in detail previously [
29,
30]. Since the epitope recognized by A11 was sensitive to detergents, synaptosomes and microsomes were disturbed by sonication (4 pulses at 100 W, 30 s at 4°C). After sonication, the synaptosomes and microsomes were centrifuged (30,000×
g, 30 min at 4°C) and soluble proteins were used for immunoprecipitation. A11 and 6E10 immunoprecipitation was done using 50 μg of soluble protein.
Statistical analysis
Data was expressed as mean ± SD. The comparison between two mice groups (WT and PS1/APP mice or PS1 and PS1/APP tg mice) was done by two-tailed t test, and for comparing several groups (WT, PS1 and PS1/APP mice) and ages we used one-way ANOVA, followed by Bonferroni post hoc multiple comparison test (SigmaStat® 2.03, SPSS Inc). For both tests, the significance was set at 95% of confidence.
Discussion
Here we report data showing that amyloid plaques are associated with dystrophies of axonal origin (loaded with autophagic vesicles) and constitute a very early pathological event in the hippocampus of PS1
M146L/APP
751SL mice. The presence of morphologically disrupted presynaptic terminals may be one of the initial stages for synaptic loss and dysfunction, so far the best correlate for early symptoms in AD patients [
3,
9,
19,
41,
45,
52,
53,
59,
61] and, therefore, a pre-clinical manifestation of progression of the disease.
Amyloid accumulation is considered a key event in AD pathology by causing glial activation, neuritic alteration, synaptic damage and neuronal death [
34,
35,
42]. All the fibrillar amyloid deposits in the hippocampus of our transgenic mice were identified as human-like neuritic plaques with dystrophic neurites and reactive gliosis. Dystrophic neurites are classically associated with compacted plaques in AD tissue [
21,
24,
25,
39,
40,
57,
58] and these neuritic plaques have been considered to be a pathological correlation of dementia in AD patients [
43]. Though neuritic dystrophy may apply to both dendritic and axonal morphological changes, in our model the predominant axonal nature of the dystrophies was demonstrated by colocalization with common axonal/synaptic (neurofilament, synaptophysin, VGLUT1 and VGAT) but not dendritic (MAP-2, α1GABA
AR) markers. Furthermore, assessment of the dystrophies using electron microscopy also confirmed their axonal structure. Notably, a thorough examination of dystrophic neurites at the electron microscopic level revealed that they were not postsynaptic to any presynaptic bouton (not shown in results). Although we cannot rule out the existence of dendritic alterations, in our model most dystrophies displayed an axonal origin. The presence of axonal dystrophies has been reported in AD patients in both early and late stages of the disease [
7,
21,
22,
57] and also in another AD models [
1,
8,
14,
20,
49,
64].
As found in the brains of people with AD [
47,
66] and in other transgenic models [
65,
66], in our PS1/APP model the axonal dystrophies were seen to have a large accumulation of a great variety of vesicles in the process of autophagy. Autophagy, an efficient cellular degradation and maintenance pathway for multiple components, is not usually observed in normal brains. Recently, it has been demonstrated that PS1 mutations accelerate the pathogenesis of AD by impairing autophagy and organelle turnover, since PS1 is essential for autolysosome acidification and maturation [
37]. The critical contribution of defective autophagy proteolytic clearance to the Aβ pathology has been recently demonstrated in TgCRND8 transgenic mice [
65]. Restoring the autophagy-lysosomal pathway by deletion of cystatinB in these AD mice reduced intracellular and extracellular amyloid load, and rescued memory performance. The existence of autophagic pathology in hippocampal dystrophic neurites in our PS1/APP model was evidenced at early ages (4–6 months) by increased protein levels of the marker of autophagosome formation LC3-II, by specific immunolocalization of LC3 in the axonal dystrophies, surrounding plaques, and by electron microscopy morphological identification of the AVs that were completely filling the axon and causing the local axonal swelling. This agreement between different mice models and AD cases strongly support the involvement of autophagy in the development of AD.
Intact microtubules are needed for AVs transport and their clearance, facilitating fusion between autophagosomes and late endosomal and lysosomal compartments thus preventing the accumulation of immature AVs in neurites [
11,
33,
36]. Supporting the possible involvement of microtubule destabilization in AD pathogenesis, it has been recently reported [
31] that natural oligomeric Aβ, isolated from AD patients, disrupted the microtubule cytoskeleton and caused neuritic dystrophy (see also [
67]). This effect seemed to be mediated by tau hyperphosphorylation at Ser202/Ser205 (AT8 epitope). In fact, tau reduction prevents Aβ-induced impairment of axonal transport [
63]. Interestingly, it has been just reported that disease-associated changes in tau conformation inhibit kinesin-dependent axonal transport by modulating axonal phosphotransferases [
32]. In line with this our data indicated the existence of early defective axonal transport within dystrophic axons in our transgenic model. These dystrophic neurites displayed tau phosphorylation (AT8-positive) and actin–cofilin rods, and both alterations are known to alter cytoskeletal dynamics in neurons (revised in [
6]). Moreover, both the kinesin-1 and dynein levels were significantly reduced and, although non-significantly at 6 months of age, neurofilaments (H and M chains) were phosphorylated (not shown). Taken together, our current results suggest that early tau phosphorylation produced axonal transport defects leading to AV accumulation and neurite pathology.
A relevant issue is the pathogenic signal behind this tau and transport dysregulation. The density and size of the neuritic plaques in our transgenic model increased in an age-dependent manner while the number of the associated dystrophic neurites was strongly related to the plaque size. The relationship between formation of extracellular amyloid deposits and their associated dystrophies remains elusive. Whereas some authors have proposed that the appearance of dystrophic neurites precedes plaque formation [
2,
23,
27,
56], a recent study using in vivo multiphoton imaging showed that dystrophies develop following amyloid deposition [
42]. Supporting the later notion, in our AD model the onset of dystrophic neurites occurred in parallel to the formation of amyloid plaques, and most (if not all) dystrophic neurites were associated with Aβ plaques. Furthermore, we have also observed plaque-associated axonal dystrophies in GABAergic and cholinergic neurons ([
5,
44,
51]; this work, see Fig.
4g, and unpublished results). These neurons did not express the transgenic APP and, consequently, did not accumulate intracellular Aβ [
5,
44,
51]. Thus, the formation of the dystrophic neurites in these neurons might be induced by extracellular Aβ, in close proximity to plaques. In contrast, the high content of APP within the dystrophic neurites suggests a possible direct role of such malformations in plaque formation. Increasing evidence implicates axons as an important source of extracellular amyloid deposits [
68] and synaptic activation promotes amyloid secretion, whereas chronic reduction of synaptic activity was found to reduce plaque loading in an AD transgenic mouse model [
15,
16,
60]. Interestingly, in our transgenic model, hippocampal plaque deposition is tightly linked to axonal pathways, as seen by simultaneous Congo red staining and neurofilament immunostaining. This agrees with the idea of Aβ released by axonal terminals and the consequent formation of extracellular deposits. In any case, the two processes, plaque induction of neuritic changes and a contribution of dystrophic neurites to Aβ deposition are not mutually exclusive and could occur concomitantly.
The soluble forms of oligomeric amyloid peptides are of special interest to play a significant role in AD pathology. Soluble Aβ oligomers have been shown to induce an increase in tau hyperphosphorylation [
17,
31] as well as severe axonal transport failure and disruption of organelle trafficking through GSK3-beta signalling (manuscript in preparation;[
18,
31]). Furthermore, we have recently reported [
30] the involvement of natural and synthetic Aβ oligomers in the activation of GSK3-beta and tau phosphorylation. Thus, it is tempting to speculate that Aβ oligomers, acting through a yet unidentified mediator, caused the interruption of axonal transport, accumulation of vesicles and axonal dystrophy.
The extracellular or intracellular origin of these soluble amyloid oligomers has not yet been well defined. In relation to this, as recently reported in AD cases [
55], the results we report here demonstrate that PS1/APP synaptosomes accumulate Aβ peptides (Aβ
42 the most abundant). However, these Aβ peptides were in monomeric form and no oligomers were detected at this age. On the other hand, we also identified, by immunogold labeling, the presence of APP and Aβ
42 in AVs which agrees with the participation of the autophagic compartments in APP processing and Aβ production, as described by others [
65,
66]. We also found that AVs were mainly accumulated in the microsomal fractions and, although the Aβ seemed to be produced principally in the synaptosomes, Aβ oligomers were identified in these microsomal fractions. These results provide evidence for the idea that the AVs, within the axonal dystrophies, are a possible compartment for Aβ oligomerization. On the other hand, taken together, the close temporal and spatial association between amyloid plaques and dystrophic neurites, the presence of A11-positive Aβ oligomers in the plaque periphery and the Aβ oligomers in the soluble fractions suggested that plaques also might be a source of the Aβ oligomers that could induce neuritic damage. In support of this idea, amyloid fibrils, major component of amyloid plaques, can be destabilized and easily reverted to soluble and highly toxic Aβ aggregates by biological lipids that are present in the brain [
38] and, as mentioned above, GABAergic cells (that did not express hAPP and did not accumulated Aβ peptides) also displayed axonal dystrophies. In any case, further experiments are needed to clarify this point.
A further significant result was the identification by electron microscopy of dystrophic axon terminals that were making contact with morphologically normal postsynaptic elements. These abnormal presynaptic boutons contained numerous AVs and were observed to have a low content in synaptic vesicles. These results were confirmed by LC3-II Western blots using synaptosomes isolated from the hippocampus of 6-month-old PS1/APP mice. Although we cannot rule out a direct effect of soluble Aβ on these presynaptic terminals, their relative distance from Aβ plaques (between 10 and 30 μm) together with the low soluble Aβ content at this age, suggested that AV accumulation might reflect the axonal transport defects at dystrophies. These altered synaptic terminals may represent one of the initial pathogenic steps of synaptic loss leading to early deficits in synaptic transmission and plasticity.
As early memory loss in AD is increasingly attributed to synaptic failure, we can conclude that this PS1/APP transgenic model shows, at young ages and in absence of pyramidal degeneration, a presynaptic pathology progression that may closely resemble the pre-clinical or early stages of human AD.