Amyloid plaque-associated dystrophic neurites display a massive accumulation of autophagic vesicles from early ages
This PS1/APP transgenic model exhibited extracellular Aβ deposits throughout the hippocampus from a very early age as illustrated in Fig. 1a with Congo red staining at 4 months. The number and size of the amyloid deposits progressively increased with age (Fig. 1b). In young mice (4- to 6-month old), the most abundant plaques were those less than 500 μm2 (70.53 ± 9.74%), whereas in older mice (18 months) the vast majority of plaques (69.41 ± 11.73%) were medium to large (>500 μm2).
Double labeling APP/Congo red (Fig. 1a–c) and APP/thioflavin-S (Fig. 1d) experiments demonstrated that, at every age examined, almost all (91.61 ± 0.14%, the percentage was practically identical at 4, 6 and 18 months of age) fibrillar amyloid deposits were decorated with clusters of APP-positive dystrophic neurites (APP is a well-reported marker for dystrophies) from the time of the appearance of amyloid plaques. The number of dystrophic neurites per plaque increased with age in parallel with the size of the plaque (Fig. 1e). Results showed that, in fact, the number of these dystrophic neurites correlated with the size of the plaque and was independent of the age of the mice. Thus, neuronal pathology in the form of dystrophic neurites occurred very early in this transgenic model. These pathological structures were not found in wild-type (WT) or PS1 transgenic mice of the same age (data not shown). Therefore, plaque-associated abnormal swelling of neuronal processes represented an early indicator of disease development and might compromise neuronal integrity and hippocampal function in young PS1/APP mice. No dystrophic neurites were found in areas remote from Aβ plaques or in 2-month-old PS1/APP mice (before the Aβ deposition).
Transmission electron microscopy analysis of the hippocampus of 4.5-month-old PS1/APP mice revealed a close spatial association between amyloid plaques and neuronal dystrophies (Fig. 2a, b). No dystrophic neurites were found in areas remote from plaques. These abnormal swollen neurites had a round/oval profile and were giant-sized, compared to normal neuronal processes in the adjacent neuropil. Ultrastructural morphometric analysis (100 aberrant neurites; n = 3) revealed that the predominant size of these dystrophic structures was between 10 and 50 μm2 (63.74%), followed by those ranging from 50 to 100 μm2 (20.43%). Notably, 5% of the dystrophic neurites measured over 100 μm2, only 9.68% of neurites were in the range 5–10 μm2 and just 1% under 5 μm2. In contrast, normal non-dystrophic neurites had an average size of 1.42 ± 0.77 μm2.
Dystrophic neurites were massively filled with collections of vacuolar structures of putative autophagic nature with different morphologies and heterogeneous intraluminal contents (Fig. 2c, e). The most common morphology corresponded to autophagic vesicles (AVs) consisting of double membrane-bound vesicles with densely compacted amorphous or multilamellar contents named as autophagosomes (Fig. 2c, d). These AVs represent the initial stages of autophagy which contain undigested compacted organellar material. In addition, there were also single or double membrane vesicles with translucent or amorphous electron-dense material in some dystrophic neurites (Fig. 2e, f), and these might presumably represent autophagosomes with partially digested material and/or the mature degradative forms of AVs (autophagolysosomes). Overall, a substantial accumulation of early and, to a less extent, late AVs within hippocampal aberrant neurites, surrounding amyloid plaques, occurred at very early ages in these PS1/APP mice.
LC3-positive autophagic vesicles within dystrophic neurites are implicated in the amyloidogenic pathway
To corroborate the autophagic nature of the heterogeneous vesicles accumulated within the dystrophic neurites, we immunostained PS1/APP hippocampal sections with the anti-LC3 antibody, a marker of autophagy (Fig. 3a, b). At 4 months (Fig. 3a), LC3 immunoreactivity was mainly found in pyramidal somata and their apical dendrites, as well as in punctate structures resembling dystrophic neurites around plaques (see insets in Fig. 3a). At 6 months (Fig. 3b), the immunoreactivity for LC3 around plaques was markedly increased while, in parallel, the staining of somata and apical dendrites decreased.
To more specifically determine the proportion of LC3 that was in the LC3-II form, which migrates faster than LC3-I on SDS-PAGE and is the form associated (by lipidation) with the autophagosomal structures, we performed quantitative immunoblot analysis of LC3-I and LC3-II forms in the hippocampus of 6-month-old PS1/APP and WT mice (Fig. 3c). Significantly higher levels of LC3-II were observed in PS1/APP mice than in age-matched WT mice (2.15 ± 0.35 fold, n = 6, p < 0.05).
Confocal imaging of double APP/LC3 immunolabeling (Fig. 3d1–d3) revealed the punctate nature of the LC3 labeling (see inset in Fig. 3d2) and the colocalization of LC3 in both APP-positive (glutamatergic) and non-APP (likely GABAergic or cholinergic) dystrophic neurites (Fig. 3d3). Considering the colocalization of APP and LC3 within dystrophic neurites in our AD model, we next wanted to assess the early implication of AVs in APP processing, and in turn likely involvement in Aβ production. To that end, we performed immunoelectron microscopy for APP and Aβ42 in 4.5-month-old PS1/APP hippocampus. Silver-enhanced immunogold labeling revealed that APP localized preferentially to the AVs within plaque-associated dystrophic neurites (Fig. 3e–g), as well as to the Golgi and endoplasmic reticulum (ER) membranes in the neuronal cell somata (Fig. 3h). No APP-labeling was found in other organelles or plasma membrane. Immunoelectron microscopy detection of Aβ was much less stronger than of APP, since optimal intracellular labeling with the antibodies for Aβ forms in our model requires pre-treatment with formic acid which is not compatible with EM processing. Nevertheless, as expected and in contrast to APP immunolabeling, Aβ42 label was mainly associated with plaques (asterisk in Fig. 3i). Interestingly, some autophagic vesicles within dystrophic neurites were also positive for the Aβ42 antibody (inset in Fig. 3i).
Dystrophic neurites represent axonal structures with cytoskeletal abnormalities
To determine the dendritic and/or axonal nature of the plaque-associated dystrophic neurites in our PS1/APP model, we have performed light and electron microscopy studies in 4- to 6-month-old mice. Immunolabeling for the MAP-2 protein (a marker of dendritic processes) (Fig. 4a, b) and for the postsynaptic marker α1GABAAR (not shown) revealed no positive dystrophic neurites around plaques at the early ages investigated. Moreover, confocal double MAP-2/APP immunofluorescence labeling confirmed the lack of colocalization for the dendritic marker in APP-positive dystrophic neurites (Fig. 4c1–c3). On the other hand, the close spatial relationship between amyloid plaques and axonal fibers tracts in the hippocampus, as revealed by neurofilament (NF) immunolabeling and Congo red staining (Fig. 4d), along with the presence of swollen NF-positive neurites (insets in Fig. 4d) indicated a possible axonal/synaptic origin of these dystrophic structures.
In fact, the labeling of the dystrophic neurites was very patent with the presynaptic marker synaptophysin (Fig. 4e). Numerous synaptophysin-positive punctated structures were observed around amyloid plaques. To determine the neurochemical nature of these synaptophysin-positive dystrophic neurites, we performed double immunofluorescence labeling for APP and the two major neurotransmitter vesicular transporters, VGLUT1 for glutamate (Fig. 4f1–f3) and VGAT for GABA (Fig. 4g1–g3). As shown in Fig. 4f3 most APP-positive dystrophic neurites contained VGLUT1 indicating the glutamatergic nature of the abnormal axons surrounding amyloid plaques. Consistent with the exclusive expression of the human mutated transgene for APP by principal cells, many enlarged inhibitory GABAergic dystrophic neurites, immunonegative for APP, were also identified around the plaques (Fig. 4g3).
In addition, electron microscopy in the hippocampus of 4.5-month-old PS1/APP mice confirmed the presence of some dystrophic myelinated axons around/near plaques (Fig. 4h–j). These axonal dystrophies had a severe (Fig. 4h, i) to moderately (not shown) pathological number of autophagic vesicles. The enlarged size of an aberrant axon (110.22 μm2) compared to adjacent normal ones (1.57 ± 0.63 μm2) is shown in Fig. 4j.
In order to identify possible early microtubule-associated axonal transport deficits in the PS1/APP hippocampus, which might lead to vesicle accumulation (autophagic, synaptic, etc.) along axons and the consequent development of dystrophy, we first assessed tau abnormalities by quantitative Western blots experiments with the AT8 antibody (which detects tau phosphorylated at both serine 202 and threonine 205 residues, one of the first to be phosphorylated) (Fig. 5a). Immunoblotting revealed a significantly higher level of expression in young PS1/APP mice (1.75 ± 0.15 fold) compared to age-matched controls. We have also confirmed by AT8 immunohistochemistry the presence of phospho-tau positive neurites surrounding amyloid plaques in 4- to 6-month-old transgenic animals (Fig. 5b). To determine whether phospho-tau was present within APP-positive dystrophic neurites, we performed double APP/AT8 immunofluorescence labeling (Fig. 5c1–c4 and d1–d4). The presence of AT8 was found in some, but not all, APP-positive dystrophic neurites (Fig. 5c3 and detail in c4; Fig. 5d3 and detail in d4).
Tau could also induce changes in the organization and stability of neuronal actin filaments, and it is known that the formation of cofilin/actin pathological bundles occludes neurites and vesicle transport. To assess possible early alterations of the actin cystoskeleton in our AD model we analyzed actin and cofilin immunolabeling in the hippocampus of young PS1/APP mice. Results showed numerous rod-like inclusions around amyloid plaques as shown for cofilin in Fig. 5e. Moreover, double APP/cofilin labeling (Fig. 5f1–f3) showed colocalization of both markers in some, particularly small, APP-positive neurites.
Finally, to further explore whether microtubule vesicular transport was compromised in young PS1/APP mice we measured the levels of kinesin-1 and dynein, two microtubule-associated motor proteins, in hippocampal protein extracts prepared from 6-month-old PS1/APP and WT mice (Fig. 5g). We found significantly lower levels of both kinesin-1 heavy chain (−24.71 ± 11.80%, n = 8) and dynein (−38.06 ± 14.40%, n = 8) in PS1/APP than age-matched WT animals.
These microtubule, actin cytoskeletal and molecular motor defects are early pathogenic events in our AD model and might lead to transport abnormalities and the accumulation of organelles (synaptic vesicles, autophagosomes, mitochondria, lysosomes) within axonal neurites promoting the dystrophic process.
The early axonal pathology includes morphologically disrupted presynaptic terminals
The axonal defects in our AD model could also affect presynaptic terminals. Therefore, in order to investigate the possible early synaptic pathology we examined the hippocampus of 4.5-month-old PS1/APP mice using electron microscopy (Fig. 6a–e). Our ultrastructural study showed that, near to amyloid plaques, there were presynaptic elements that displayed pathological changes including large diameter with a considerate number of AVs and, in contrast, fewer synaptic vesicles (Fig. 6a, b; presynaptic terminals outlined with a white line). The presence of presynaptic terminals, at the beginning of the dystrophy process, with few and early stage AVs formation, as well as with synaptic-like vesicles were also detected (Fig. 6c–e; see also the presynaptic terminal outlined with a black line in Fig. 6b). However, these altered presynaptic elements were making synaptic contacts with morphologically normal dendrites or dendritic spines with postsynaptic density (see Fig. 6a–d). These morphologically altered presynaptic terminals may represent the initial stages of synaptic disruption and loss.
Moreover, we compared the LC3-II accumulation in synaptosomal and microsomal fractions isolated from 6-month-old WT and PS1/APP animals (Fig. 6f). As expected, a low percentage of LC3-II was observed in isolated synaptosomes from WT mice (15.42 ± 2.25%, n = 4, of microsomal fractions in WT mice). Further, in agreement with our electron microscopy studies, the amount of LC3-II in PS1/APP mice was higher in both synaptosomal and microsomal fractions. Although the relative abundance of LC3-II presented in PS1/APP-derived synaptosomes was still low (22.22 ± 3.84% of PS1/APP microsomal LC3-II), the level was consistently higher (2.54 ± 0.54, n = 5) than in WT synaptosomes.
These data demonstrated the existence of an early autophagy-associated axonal/synaptic pathology in the hippocampus of this AD mouse model.
Extracellular periplaque oligomeric Aβ spatially correlates with axonal/synaptic dystrophy
Taking into account the close spatial relationship between axonal dystrophies and Aβ plaques, we next examined the possible intracellular and/or extracellular origin of the pathogenic Aβ agent. We first investigated the possible intracellular accumulation of oligomeric Aβ in isolated synaptosomes and microsomes by immunoprecipitation experiments using the monoclonal antibody 6E10 (Fig. 7a). Results demonstrated the presence of a relatively large accumulation of monomeric Aβ in synaptosomes, whereas lower levels were detected in microsomes (Fig. 7a1). Furthermore, within the different Aβ peptides, the Aβ42 was the major form observed in these synaptosomal fractions (Fig. 7a2). On the other hand, no Aβ oligomers were observed with this approach.
We have also used the anti-oligomeric Aβ antibody A11 in immunohistochemistry and immunoprecipitation experiments. Our results demonstrated that most Aβ plaques were immunopositive for the A11 antibody (Fig. 7b), with a preferential immunolabeling at the plaque periphery (see inset). Furthermore, A11-immunoprecipitation using the soluble (S1) fractions demonstrated the presence of minute amount of extracellular oligomeric Aβ (Fig. 7c) in the hippocampus of the PS1/APP mice at this early age (6 months). These data were consistent with previous experiments (see also [29, 30]). However, in spite of the relative abundance of monomeric Aβ in synaptosomes, under the present experimental conditions, no Aβ oligomers were detected in this fraction (Fig. 7c). On the other hand, in the microsomal fractions, in the A11 immunoprecipitation experiments many non-specific bands were observed. The presence of this non-specific immunoprecipitation precluded the identification of putative oligomers. It is possible that the relative high abundance of AVs in this fraction could interfere with the immunoprecipitation experiments. Aiming to overcome this problem, the presence of the possible oligomers was also assayed using Western blots with 6E10 and 82E1 antibodies. The 6E10 antibody produced no specific signal (not shown); the 82E1 antibody, however, produced a clear specific (as compared with age-matched WT results) identification of Aβ oligomers (Fig. 7d). Taken together, these data indicate the presence of Aβ monomers in synaptosomes and Aβ oligomers in extracellular plaques and the microsomal fractions.