Generation and characterization of APP_751SL/PS1_M146L (APP/PS1) mice have been previously reported (Blanchard et al., 2003; Ramos et al., 2006; Caballero et al., 2007; Jimenez et al., 2008, 2011; Moreno-Gonzalez et al., 2009; Baglietto-Vargas et al., 2010, 2017; Sanchez-Varo et al., 2012; Torres et al., 2012; Trujillo-Estrada et al., 2014; Gomez-Arboledas et al., 2018; Sanchez-Mejias et al., 2020). The transgenic mice were obtained by crossing heterozygotic Thy1-APP_751SL [Swedish (KM670/671NL) and London (V717I) FAD mutations] with homozygous PS1_M146L mice (Charles River, France). Both lines were maintained on a C57BL/6J background. Age-matched wild-type (WT) mice of the same genetic background were used as controls. All animal experiments were performed in accordance with the Spanish and the European Union regulations (RD53/2013 and 2010/63/UE) and approved by the Animal Research Committee from the University of Malaga and Seville (Spain). Experiments and procedures with animals were designed to minimize the animal suffering and reduce the number of animals used.

The 4.5-month-old mice were submitted to motor function and cognitive tests. Behavioral studies were conducted on age-matched male littermates WT (n = 23) and APP/PS1 (n = 20). Animals were housed in groups of five and maintained on a 12 h light/dark cycle (lights on at 8.00) at a constant temperature (21 ± 1°C). Behavioral tests were performed during the light period of the light/dark cycle, and the experimenter was blind to the mice genotypes. Habituation and open-field locomotion test (Filali et al., 2012) were performed as described earlier by our group (Trujillo-Estrada et al., 2013; Fernandez-Valenzuela et al., 2020). Monitoring of behavior was done using the software Ethovision XT 7.0 (Noldus, Netherlands).

After deep anesthesia with sodium pentobarbital (60 mg/kg), 2-, 4-, 6-, 12-, and 18-month-old APP/PS1 and WT male mice were perfused transcardially with 4% paraformaldehyde, 75 mM lysine, 10 mM sodium metaperiodate in 0.1 M phosphate buffer (PB), pH 7.4. Brains were 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. Hippocampus-containing sections between bregma coordinates −1.58 and −3.64 mm (Paxinos and Franklin, 2013) were used. For electron microscopy, 4.5-month-old WT and APP/PS1 male 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 PB, pH 7.4 (Sanchez-Varo et al., 2012). Brains were post-fixed in the same fixative solution overnight at 4°C, washed several times in 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, 100 μm sections were fixed in 2% osmium tetroxide in 0.1 M PB, dehydrated, and embedded in Araldite (EMS, United States). Tissue blocks were serially sliced into semithin (1.5 μm) with a diamond knife in a Leica ultramicrotome (EM UC6), placed on slides, stained with 1% toluidine blue, and scanned under the light microscope for amyloid plaques. Finally, selected areas from semithin samples were cut in ultrathin sections that were placed on Formvar-coated grids and stained with uranyl acetate and lead citrate before being examined with an electron microscope (JEOL JEM 1400). In addition, fixed brains from 6-month-old mice were selected to be washed in PBS 0.1 M and directly sectioned at 50 or 250 μm in the coronal plane on a vibratome (Leica VT1000M). Then, semithin sections (1 μm thickness) were serially collected from embedded tissue blocks using a Leica ultramicrotome (EM UC6) and mounted on gelatine-coated slides. Hippocampus-containing sections between bregma coordinates −1.58 and −3.64 mm (Paxinos and Franklin, 2013) were used.

Human AD brain tissue samples from medial temporal lobe (hippocampal/parahippocampal regions) were obtained from the Institute of Neuropathology Brain Bank IDIBELL-Hospital Universitari de Bellvitge (Barcelona, Spain) at autopsy (n = 5, Braak V–VI stage, and clinically classified as demented AD cases) with a post mortem delay between 3 and 16 h (Supplementary Table 1). The utilization of brain samples was approved by the corresponding ethics committees following Spanish legislation. Samples of the hippocampal cortex were fixed in 4% paraformaldehyde in 0.1 M PB for 24–48 h, cryoprotected in 30% sucrose, stored at −80°C, sectioned (30 μm thickness) on a freezing microtome, and serially collected in wells containing 0.1 M phosphate buffer saline (PBS) and 0.02% sodium azide. Following a neuropathological examination, AD stage was defined according to Braak and Braak (1991). Brain samples were used for light microscopy studies.

The following primary antibodies were used for immunohistochemistry: anti-human amyloid precursor protein (hAPP) rabbit polyclonal antibody (1:20,000, Sigma-Aldrich), anti-amyloid oligomers (OC) rabbit polyclonal (1:5,000, Millipore), anti-Aβ mouse monoclonal 6E10 (1:1,500, Sigma-Aldrich), anti-Aβ42 rabbit polyclonal (1:40, Biosource 44-344), anti-microtubule-associated protein 1 light chain 3 (LC3b) goat polyclonal (1:1,000, Santa Cruz Biotechnology), anti-MAP2 rabbit polyclonal (1:5,000, Chemicon), monoclonal anti-oligomeric Aβ antibody NU-4 (1:2,000; kind gift from Dr. Klein; Northwestern University, IL, United States; Lambert et al., 2007), anti-PSD95 rabbit polyclonal (1:1,000, Invitrogen), anti-synaptophysin (SYN) rabbit polyclonal (1:1000, Abcam), and anti-VGLUT1 polyclonal guinea-pig (1:10,000, Chemicon). The following antibodies were used for molecular studies: anti-beta actin mouse monoclonal (1:5,000, Sigma-Aldrich), anti-NeuN mouse monoclonal (1:1,000, Chemicon International), anti-PSD95 rabbit polyclonal (1:1,000, Cell Signaling), anti-Synaptophysin rabbit polyclonal (1:1,000, Synaptic Systems), anti-VGLUT1 rabbit polyclonal (1:2,000, Synaptic Systems), and anti-VGAT rabbit polyclonal (1:5,000, Synaptic Systems).

The immunolabeling procedures were done as described earlier (Sanchez-Varo et al., 2012). Serial sections were processed in parallel using the same batches of solutions to minimize variability during immunostaining processing. 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) for 30 min to block endogenous avidin, biotin, and biotin-binding proteins. For single immunolabeling, sections were incubated with one 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), followed by streptavidin-conjugated horseradish peroxidase (90 min, 1:2,000, Sigma–Aldrich). The peroxidase reaction was visualized with 0.05% 3-3-diaminobenzidine tetrahydrochloride (DAB, Sigma–Aldrich), 0.03% nickel ammonium sulfate, and 0.01% hydrogen peroxide in PBS. Some sections immunolabeled for SYN were incubated 3 min in a solution of 20% of Congo red. Immunohistochemistry in semithin sections was performed as follows: 50-μm-thick sections were dehydrated, flat embedded in Araldite, and sectioned at 1–1.5 μm thickness using an ultramicrotome (Leica VT1000M). After etching the resin, mounted sections were immunolabeled for OC using the same procedure described before, and then were counterstained with Harry’s hematoxylin (Panreac).

Double immunoperoxidase MAP2/6E10 labeling was performed as described previously (Moreno-Gonzalez et al., 2009). Sections were first incubated in anti-MAP2, and after the DAB-nickel reaction (dark blue end product), sections were incubated in 6E10 antibody. The second immunoperoxidase reaction was develop with DAB only (brown reaction end product). Sections were 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 labeling, sections were first sequentially incubated with the indicated primaries antibodies followed by the corresponding Alexa488/568 secondary antibodies (1:1,000; Invitrogen). OC-immunolabeled sections were stained with 0.02% Thioflavin-S (Sigma-Aldrich) in 50% ethanol. Human sections were previously incubated in autofluorescence eliminator reagent (Millipore), following the manufacturer’s recommendations, to eliminate fluorescence emitted by intracellular lipofuscin accumulation. Finally, all sections were coverslipped with 0.01 M PBS containing 50% glycerine and 3% triethylenediamine and examined under a confocal laser microscope (Leica SP5 II).

The immunogold procedure was done as we previously described (Sanchez-Varo et al., 2012; Gomez-Arboledas et al., 2018). Sections (50 μm) from 4.5- and 6-month-old APP/PS1 mice were first washed in PBS and incubated in a 50 mM glycine solution 10 min, or cryoprotected in a 25% sucrose and 10% glycerol solution, followed by freezing at −80°C, to increase the antibody-binding efficiency. After following the standard protocol, the tissue was incubated overnight in primary anti-oligomeric OC rabbit polyclonal (1:5,000, Millipore), mouse monoclonal anti-oligomeric Aβ (12–24 mers) NU-4 (1:2,000), or anti-Aβ42 rabbit polyclonal (1:5,000, Abcam) antibodies in a PBS 0.1 M/1% BSA solution at room temperature. Then, sections were washed in PBS and incubated with 1.4 nm gold-conjugated secondary antibody (1:100, Nanoprobes) overnight at room temperature. After post-fixing with 1% glutaraldehyde and washing with 50 mM sodium citrate, the labeling was enhanced with the HQ Silver^TM Kit (Nanoprobes). Finally, the immunolabeled sections were processed as we previously described by the osmium fixation, dehydration, and embedding steps. The primary antibody was omitted in negative control experiments.

Plaque loading was defined as the percentage of total hippocampal area stained with the anti-Aβ mouse monoclonal antibody 6E10 (1:1,500, Sigma). Quantification of extracellular Aβ content was performed similarly to the previously reported method (Moreno-Gonzalez et al., 2009; Baglietto-Vargas et al., 2010, 2017; Trujillo-Estrada et al., 2014). Sections were examined under a Nikon Eclipse 80i microscope, and images were acquired with a Nikon DS-5M high resolution digital camera using the ACT-2U imaging software (Nikon Corporation, Minato, Tokyo, Japan). The camera settings were adjusted at the start of the experiment and maintained for uniformity. Digital 4× images (2560 × 1920, 300 ppp) from 2- to 18-month-old APP/PS1 mice (5 sections/mouse; n = 4 mice/age) were analyzed using Visilog 6.3 analysis program (Noesis, France). The hippocampal area in each image was manually outlined. Then, 6E10-positive plaque area within the hippocampus was identified by level threshold which was maintained throughout the experiment for uniformity. The color images were converted to binary images with plaques. The plaque loading (%) for each transgenic mouse was estimated and defined as (sum plaque area measured/sum hippocampal area analyzed) × 100. The sums were taken over all slides sampled, and a single plaque burden was computed for each mouse. The mean and standard deviation (SD) of the plaque loading were determined using all the available data (mean hippocampal area analyzed/animal 1901893 ± 80315.51 μm²; 618430.92 ± 42359.93 μm² for CA1 region). Quantitative comparisons were carried out on sections processed at the same time with same batches of solutions.

The quantification of synaptophysin-immunonegative area (SIA) was performed over hippocampal sections (between −0.94 mm anterior and 3.64 mm posterior to Bregma) from 2- to 18-month-old APP/PS1 stained with anti-SYN (n = 4 mice/age). The SIA analysis was done using the nucleator method with isotropic probes (n = 5 radii) over 992 immunonegative areas by the NewCast software (Olympus). The different hippocampal layers were manually outlined at low magnifications [stratum oriens (SO), stratum radiatum (SR), and stratum lacunosum moleculare (SLM) from CA1, stratum lucidum (SL) from CA3, and stratum moleculare (SM) from dentate gyrus (DG)] following the atlas of Paxinos and Franklin (2013) to establish the corresponding boundaries. Then, they were analyzed using a counting frame of 8840.7 and 7072.6 μm², respectively, and a step length of 194.02 × 194.02 μm. Individual immunonegative-area measurements were performed under a 40×/0.95 objective; then, the summation was calculated for each layer. SIA was defined as the percentage of the non-immunoreactive area related to the total surface analyzed. To estimate plaque size, digital panoramic ultramicrophotographs (JEOL JEM 1400) of extracellular deposits (n = 15) were analyzed similarly to SIA estimation, using the nucleator method with isotropic probes (n = 5 radii).

Hippocampal ultrathin sections (60–70 nm) with evident amyloid plaques were selected among serial semithins sections from 4.5-month-old APP/PS1 mice (n = 3). Taking the plaque border (defined as the point from which there were no detectable amyloid fibers extending from the deposits in the cerebral parenchyma) as a point of reference, six consecutive ultramicrophotographs were captured in 3–4 different radii (step length ∼4.7 μm; 21.7 μm²/image). Moreover, 30 ultramicrophotographs/animal were captured in areas far from plaques (at least, 70 μm distance) to be used as control interplaque regions. For WT mice, three starting points were randomly selected at low magnifications (4,000×) from each analyzed region (SO, SM, and SR). Then, 10 consecutive ultramicrophotographs were captured following the same parameters defined for transgenic mice. Thus, 30 ultramicrophotographs (3 × 10) were taken from each region/animal. Sampling was thoroughly performed to avoid overlapping images (DeFelipe et al., 1999). The density of synaptic boutons within each periplaque (distance of 0–5, 5–10, 10–15, 15–20, 20–25, and 25–30 μm) or free-of-plaque regions (either interplaque or control areas) was estimated considering only those synapses displaying a visible presynaptic element (with more than three synaptic vesicles), the synaptic cleft, and a post-synaptic density. Then, using the same digital images, the SVD measurement was undergone over the synaptic boutons showing a minimum section surface of 0.5 μm² and counted in the area that included the high electron dense postsynaptic zone, as it has been previously published by others (Parodi et al., 2010). For this assay, we established three range of distances from plaque border (0–10, 10–20, and 20–30 μm). Each analysis was done by a single examiner blinded to sample identities.

The S1 fractions were prepared as described previously (Jimenez et al., 2008; Sanchez-Mico et al., 2021). Primary neuronal cultures were done essentially as described by Jimenez et al. (2011). Briefly, embryonic E18–20 or postnatal P1 brains were dissected and treated, for 5 min, with trypsin-DMEM-EDTA medium (Biowhittaker, Cambrex, Belgium). The treatment was stopped using complete DMEM plus 10% FBS and the cells were mechanically dissociated. Then, the debris were eliminated by filtration (40 μm, BD Falcon) and the cells were cultured (at a density of 60,000 cells/ml) in Neurobasal medium plus B27 supplemental (containing glutamine, 1% penicillin-streptomycin, and gentamycin) on poly-D-lysine (Sigma-Aldrich) treated Nunc 12-well plates. The cells were cultured at 37°C in humidified 5% CO₂/95% atmosphere. Medium was half-replaced every 4 days. After 13–15 days in culture, the cultures were treated with 10 μg of S1 protein/100 μl of medium. The cells were incubated for 24 h and rinsed with PBS, and the proteins were extracted (see below).

Hippocampal samples from 2-, 4-, and 6-month-old APP/PS1 and WT male mice were used (mean sample size 5/group: APP/PS1 mice n = 6/age, WT n = 4/age). The protein pellets, obtained using the Tripure Isolation Reagent and isopropanol-mediated precipitation, were resuspended in 4% SDS and 8 M urea in 40 mM Tris–HCl, pH 7.4, and rotated overnight at room temperature to get complete protein solubilization (Ramos et al., 2006). Western blots were performed as described (Araujo et al., 1996). Briefly, 0.5–15 μg of proteins from the different samples were loaded on 10%-SDS-tris-glycine-PAGE and transferred to nitrocellulose (Hybond-C Extra; Amersham). After blocking, using 5% non-fat milk TPBS, membranes were incubated overnight, at 4°C, with the appropriate primary antibody. Membranes were then incubated with the corresponding horseradish-peroxidase-conjugated secondary antibody (Dako, Denmark) at a dilution of 1:10,000. Each blot was developed using the ECL-plus detection method (Amersham) and quantified using ImageQuant Las 4,000 mini gold (GE Healthcare Bio-Sciences). For normalization purposes, proteins were first estimated by Lowry and protein loading corrected by beta-actin.

Data were expressed as mean ± SD. Cognitive data were expressed as mean ± SEM. Parametric and non-parametric statistical analyses were used when appropriate. The comparison between two groups was done by two-tailed t-test. To compare several groups, we used one-way ANOVA, repeated-measures (RM) one-way and two-way ANOVA tests, or Kruskal–Wallis, followed by Tukey’s post hoc multiple comparison test or Dunn test, respectively. The significance was set at 95% of confidence. Statistical analysis was performed using the GraphPad 7.0 (Prism) software.

First, the motor and cognitive status of young (4.5-month-old) APP/PS1 mice were assessed. Spatial memory was checked using the Morris water maze (MWM) test, a hippocampal-dependent task (Morris, 1984). Previously, the presence of visual or motivational defects was discarded with the visible platform task (Figure 1A). This test was successfully performed by both genotypes, showing differences across the training trials in both latency (time to reach the platform) and distance traveled. Thus, these results demonstrated that both WT and APP/PS1 mice were able to find the platform, progressively diminishing the time or distance to reach it (Figure 1A1). Respect to the swim speed (Figure 1A2), no genotype, trial, or interaction effects were found.

During acquisition phase, mice learned to locate the hidden platform by spatial navigation, as demonstrated by the progressive decrease in the distance traveled along training days. Statistical analysis (Figure 1B) did not reveal a genotype effect, since there were no significant differences in either distance (Figure 1B1) or swim speed (Figure 1B2) between groups, neither interaction effect in distance or velocity. The distance traveled [F_{(3, 123)} = 9.72, p < 0.0001) and the swim speed (F_{(3, 123)} = 11.48, p < 0.0001] were changing along the course of the training. Importantly, the last training day (D4), there were significant differences between WT and APP/PS1 groups in the traveled distance to reach the platform, without differences in swim speed. Therefore, APP/PS1 mice spent more time to reach the target, indicating learning impairment.

During the probe trial (Figure 1C), significant differences were detected regarding the time spent within the quadrants by WT or APP/PS1 mice (Figure 1C1): control group stayed longer in the target zone (quadrant C) than in the opposite (A) or adjacent (B, D) quadrants. By contrast, intra-group APP/PS1 analysis did not yield statistically significant differences among the four quadrants. In addition, APP/PS1 mice spent significantly less time in the target quadrant than in the others (C, Figure 1C2), and more time within the quadrant A, in comparison to control group (Figure 1C3). Overall, these data indicated that double transgenic mice displayed memory deficiencies since early ages.

To check whether the detected cognitive deficits correlate with an early synaptic dysfunction, the amount of synaptic proteins was measured by Western blots in hippocampal protein extracts obtained from 2-, 4-, and 6-month-old APP/PS1 and WT mice (representative blots are shown in Figure 2A). Levels of the postsynaptic marker PSD95 (Figure 2B), and the presynaptic synaptophysin (SYN; Figure 2C), and the GABA vesicular transporter VGAT (Figure 2D) were already significantly lower in APP/PS1 than in WT animals at 4 months of age (−27.04 ± 6.09%, −21.29 ± 9.12%, and −27.68 ± 9.31%, respectively). Moreover, the level of these synaptic markers was even more reduced at 6 months (−36.66 ± 9.17%, −29.97 ± 17.97%, and −37.12 ± 9.15%, respectively), along with a decrease in the vesicular glutamate transporter 1 (VGLUT1; −26.94 ± 6.53%; Figure 2E), only detected from 6 months. In addition, SYN- and VGLUT1-immunostainings demonstrated that the presynaptic terminals were directly affected by the presence of Aβ deposits from early ages (Supplementary Figures 1A–D,G,H,M–P). SYN immunoreactivity was virtually absent in some neuropil areas, which appeared to correspond with the location of amyloid plaques, as confirmed by both Congo red staining (Supplementary Figures 1E,F). Similarly, MAP2 (dendrites) and PSD95 (dendritic spines) immunostainings showed the loss of postsynaptic elements because of the plaques (Supplementary Figures 1I–L, double MAP2/6E10 labeling evidence plaque impact in J). All these data together with the ultrastructural inspection using transmission electron microscopy (TEM) indicate that the areas of brain parenchyma occupied by fibrillary plaques were devoid of synaptic contacts. Therefore, the memory impairment observed in young APP/PS1 mice correlates with a reduction of critical synaptic proteins. Our results suggest that this decrease in synaptic markers may be (at least partly) due to extracellular amyloid deposition (hippocampal plaque loading in APP/PS1 with 6E10 antibody: 0 ± 14.26% at 2 m; 5.14 ± 5.7% at 4 m; 7.15 ± 3.65% at 6 m; 13.73 ± 4.9% at 12 m, and 21.35 ± 4.92% at 18 m; see graphics from Supplementary Figures 1M–Q).

The presence of dystrophic neurites (axons and presynaptic terminals) surrounding Aβ plaques within the hippocampal formation of this model has been extensively reported (Blanchard et al., 2003; Ramos et al., 2006; Moreno-Gonzalez et al., 2009; Baglietto-Vargas et al., 2010; Sanchez-Varo et al., 2012; Torres et al., 2012; Trujillo-Estrada et al., 2013, 2014; Fernandez-Valenzuela et al., 2020; Sanchez-Mejias et al., 2020). Moreover, the ultrastructural analysis of hippocampus from young (4–6 months of age) APP/PS1 mice revealed the close spatial relationship between amyloid plaques and aberrant neuronal processes (Sanchez-Varo et al., 2012; Gomez-Arboledas et al., 2018). Our ultrastructural analysis indicates that these abnormally swollen neurites displayed a round/oval shape and a giant size (Figure 3A), compared to other normal neuronal processes in the nearby neuropile. Moreover, Aβ fibrils were found to be closely associated with dystrophic neurites, which exhibit vesicular content of autophagic/synaptic nature (Figures 3B,C). In fact, some small presynaptic-like profiles seemed to be mixed with the fibrillar Aβ at the plaque edge. In this sense, our group has previously demonstrated that periplaque aberrant presynaptic terminals are filled with autophagic vesicles containing amyloid precursor protein (APP) and Aβ (Sanchez-Varo et al., 2012). Numerous periplaque SYN-positive dystrophies are found within the hippocampus (Supplementary Figures 1D,F, 2A–C). Double confocal immunofluorescence labeling demonstrated the colocalization of SYN with either the autophagic marker LC3b (Supplementary Figures 2A1–A3) or APP (Supplementary Figures 2B1–B3) within these enlarged structures. The prevalence of SYN, LC3b, or APP was found to be variable among dystrophic neurites (see insets in Supplementary Figures 2A3,B3). Interestingly, SYN colocalized with Aβ42 in swollen neuronal processes both around and within the plaques (Supplementary Figures 2C1–C3). TEM analysis demonstrated the scarce presence of synaptic contacts around the amyloid plaques as soon as 4.5 months of age in APP/PS1 mice. Seemingly, some of them were established between apparently healthy postsynaptic elements and abnormal presynaptic terminals. These elements were filled with many autophagic vesicles and, conversely, very few (if any) synaptic vesicles close to the active zones (Figures 3C–E). Furthermore, some of the presynaptic terminals with vesicular build-up displayed an electron dense cytoplasm (Figures 3F–G), a typical hallmark of degenerating processes. Finally, these terminals were found to establish synaptic contacts with either normal (Figures 3E,F) or abnormal (Figure 3G) postsynaptic components. These periplaque disturbed synaptic boutons may represent the morphological correlate leading to the initial cognitive failure, preceding synaptic and neuronal loss.

In order to determine the impact of amyloid deposition over synaptic density, we performed an ultrastructural quantification in the hippocampus of 4.5 months APP/PS1 mice. Starting from the edge of the plaques (Figure 4A), six consecutive photomicrographs were taken, covering 3–4 different radii per plaque (30-μm-length radii). Identifiable synaptic contacts (only those displaying clearly visible pre- and post-synaptic elements, and the synaptic cleft) were manually counted within each field. Electron-photomicrographs from interplaque zones and age-matched WT mice were used as control. The analysis showed scarce or none synaptic contacts at the plaque perimeter (Figure 4B, zone between 0 and 5 μm from plaque) in comparison to more distant areas (Figure 4C, 20–25 μm), which were progressively increasing the synaptic density reaching values more similar to control zones. Therefore, we found a significant synaptic loss around the plaques within all the hippocampal regions studied (stratum oriens, Figure 4D; stratum moleculare, Figure 4E; stratum radiatum, not shown). When data were arranged according to plaque size, we observed that the largest amyloid deposits (> 500 μm²) provoked a higher synaptotoxic impact, especially close to the plaque border (Figure 4F; one-way ANOVA, p = 0.0027). Overall, these data indicated that there is a tight relationship between synaptic damage or loss and extracellular amyloid deposition.

Next, we focused on the effect of plaques on SVD near to the active zones (releasable pool), a parameter more related with the correct functioning of synapses. To that purpose, we analyzed the SVD in the remaining presynaptic terminals from the same samples. To avoid underestimation of terminals quantification, only presynaptic terminals over 0.5-μm²-sized were considered (Parodi et al., 2010). A comparison among several presynaptic elements, showing the counting area, is depicted in Figures 5A–C. The quantification revealed that the periplaque areas (APP/PS1p) displayed a reduced proportion of presynaptic boutons with high vesicular density, compared to the control or interplaque (APP/PS1ip) areas (Figure 5D). Moreover, among APP/PS1p terminals, we found a group of synapses showing a low-density category (<10 vesicles/μm²). In addition, the closer to the plaques, the lower SVD within the presynaptic compartments (Figure 5E). Interestingly, the drop in the vesicular density was usually associated with the build-up of autophagic vesicles (Figure 5C). Therefore, these data contribute to shed light on the negative impact of Aβ deposits on the presynaptic function, which was inversely correlated with the distance to plaques.

Given that Aβ oligomeric species have been recognized as synaptotoxic agents, we intended to investigate their location within the plaques in the hippocampus from both APP/PS1 mice and AD patients. We performed double-fluorescence staining with Thioflavin-S (ThioS) and the OC antibody in human AD samples (Braak VI stage), showing that the oligomeric Aβ was located at the plaque edge, surrounding the fibrillar core (Figures 6A,B). Interestingly, the plaques from 6-month-old APP/PS1 mice exhibited the same OC/Thio-S relative distribution as human deposits did (Figure 6C). In addition, we performed OC immunohistochemistry on semithin sections from APP/PS1 hippocampus, which was mainly detected at the plaque periphery (detail at higher magnification of CA1 deposit in inset on Figure 6D). The ultrastructural analysis of Aβ immunogold-labeled sections using two different conformational antibodies (Figures 6D,E, OC antibody; Figures 6F–H, NU-4 antibody) corroborated the location of these specific Aβ species at the plaque periphery, bordering the Aβ fibrils. All these data demonstrated the preferential disposition of Aβ oligomers in the periphery of extracellular plaques.

Importantly, immunogold staining in 6-month-old APP/PS1 mice evidenced the presence of Aβ42 isoform, not only in the plaques, but also in the membrane of periplaque dystrophic neurites (Figures 7A–H; the inset in H depicts amyloid fibers intermingled with a dystrophic process). These Aβ42-positive swollen axonal/presynaptic neuronal processes were filled with vesicular compartments related to autophagy-lysosomal pathway (Figures 7I–L) as previously described in this model hippocampus, supporting the idea of dystrophic neurites as a place of APP processing, amyloid generation, and release (Sanchez-Varo et al., 2012).

To confirm the presence of soluble synaptotoxic molecules in APP/PS1 mice at early ages, primary neuronal cultures were incubated with soluble fractions (S1) from 6-month-old APP/PS1 hippocampus for 48 h. S1 from age-matched WT mice or PBS were used as controls. The quantitative analysis (Figure 8A) demonstrated that the levels of the presynaptic markers VGLUT1 (Figure 8B) and SYN (Figure 8C) were significantly decreased (−26.29 ± 23.26% and −20 ± 4.5%, respectively), after incubating cells with S1 from APP/PS1 hippocampus, but not in PBS or WT S1 fractions. As supporting evidence of specific synaptic protein loss, we analyzed the neuronal marker NeuN (Figure 8D) which remained unchanged (4.84 ± 9.13%), and rule out neuronal death as a direct cause of this synaptic drop.