Aging alters the vulnerability pattern to amyloid-beta oligomers in wild-type mice: a behavioral and neurobiological study Authors: Ahmad Allouche, Julie Colin, Catherine Birck, Henri Schroeder, Valentin Tallandier, Marion Baldoni, Christophe Muller, Mohamed Afrassi, Nicolas Violle Published: 2026 ABSTRACT AbstractBackgroundAging is the primary risk factor for sporadic Alzheimer’s disease (AD). While amyloid-beta oligomers (AβOs) accumulation is a key neuropathological process in AD, their specific effects in aged brains and how aging modulates brain response to AβOs remains poorly understood. We investigated how aging contributes to AβO-induced neurotoxicity and cognitive deficits in mice.MethodsAfter biochemical and in vitro characterizations on primary cultures of cortical neurons, AβOs or their vehicle were intracerebrally injected into both 3- and 18-month-old wild-type mice. A broad spectrum of assays including synaptic markers, neuroinflammation, apoptosis and cognitive functions was used to establish a preliminary characterization of the interplay between age and AβOs. In vivo data were analyzed using a multifactorial design (Treatment × Age), with two-way ANOVA or other appropriate statistical models.ResultsOld mice had significantly reduced synaptic proteins SNAP-25 and PSD-95, elevated neuroinflammatory markers, and increased neuronal apoptosis in hippocampus and cortex, despite showing cognitive performances similar to young mice. All brain biomarkers were worsened after AβO injection in both young and old mice. Age and AβO effects either accumulated or interacted to promote neuroinflammation and apoptosis, depending on brain areas, whereas their effects on synaptic proteins were strictly additive. Moreover, AβO injection induced only mild spatial memory deficits in young mice, in contrast with those observed in old mice in both episodic and spatial memory tests.DiscussionWhereas the young brain showed resilience to maintain memory performances after AβO injection, the coping capacities of the aging brain were exceeded by AβO effects. At the neurobiological level, age and AβO effects were mainly additive, but also acted synergistically in a brain region-dependent vulnerability pattern. This study highlights the value of incorporating aging into preclinical models to improve their translational validity and enhance their relevance for drug testing targeting early stages of sporadic AD.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13195-026-02051-2.Keywords: Aging, Alzheimer’s disease, Amyloid-beta oligomers, Neurodegeneration, Neuroinflammation, Memory Background Sporadic Alzheimer's disease (AD) progresses through a long asymptomatic stage characterized by the accumulation of amyloid-beta (Aβ), before neurodegeneration and clinical symptom onset. After decades of debate regarding the significance of Aβ aggregates, recent clinical successes with antibodies targeting Aβ oligomers (AβOs) and protofibrils [1–3] have highlighted the role of small and soluble Aβ species as key pathological drivers of the disease. However, the disease mechanisms triggered by AβO accumulation in the aging brain remain under investigation. The Aβ1–42 peptide plays a major role in AD owing to its prevalence in senile plaques, its strong propensity to aggregate [4], its heightened neurotoxicity compared with other Aβ species [5], and its use as a reliable biomarker for the diagnosis of both familial and sporadic AD [6]. Aβ1–42 progresses from monomers to β-sheet-rich fibrils through oligomeric intermediates, which are now widely regarded as key drivers of synaptic dysfunction and cognitive decline [7–10]. However, due to their polymorphic, transient, and heterogeneous nature, AβOs remain technically challenging to isolate, stabilize, and characterize [9, 11]. These difficulties underscore the importance of using rigorously defined and well-characterized AβO preparations in experimental systems to ensure reproducibility and relevance. AβOs accumulate in an age-dependent manner within specific brain regions critical for learning and memory, notably the hippocampus and entorhinal cortex, areas that exhibit early pathological changes in AD [12]. At the synaptic level, AβOs exert significant neurotoxic effects, contributing to synaptic dysfunction that precedes widespread neuronal loss and cognitive decline [13, 14]. AβOs alter dendritic spine morphology, leading to the formation of aberrant spines and a reduction in spine density, both of which compromise synaptic plasticity and lead to memory deficits even before significant neuronal death occurs [14, 15]. Furthermore, AβOs trigger mitochondrial dysfunction and promote further Aβ production, thereby establishing a self-sustaining cycle of cellular injury [16, 17]. These oligomers also induce neuronal apoptosis through the activation of stress-related pathways, including oxidative stress and lipid signaling cascades, indicating their key role in mediating neurodegeneration in AD [18]. Under normal conditions, the aging brain maintains endogenous mechanisms to tightly regulate AβO clearance, preventing them from interacting with synapses and thereby damaging neural networks. For reasons that remain unclear, these regulatory mechanisms become impaired in AD, leading to excessive AβO accumulation in synapses – a feature that distinguishes patients from healthy elderly individuals [19–21]. Beyond their direct neuronal effects, AβOs also instigate early activation of astrocytes and microglia, initiating chronic neuroinflammation [22, 23]. While glial activation initially participates in AβO phagocytosis and clearance, chronic exposure results in the release of proinflammatory cytokines, chemokines, reactive oxygen species, and nitric oxide, amplifying immune responses and contributing to neuronal damage [24–26]. Although age remains the principal risk factor for sporadic AD, preclinical drug development frequently relies on pathological models that do not incorporate aging processes and the associated progressive biological changes. Indeed, most animal models used in AD research are transgenic mouse models, which are valuable for studying familial AD but rely on accelerated aging or non-physiological Aβ overproduction (i.e., overexpression of human mutant proteins) that is active from embryonic or early postnatal stages [27, 28]. Such early overexpression introduces developmental perturbations and gene–development interactions that modify neuronal maturation and circuit formation long before aging begins [29, 30]. As a consequence, these models fail to recapitulate the gradual, age‑dependent progression of sporadic AD or to distinguish the respective contributions of aging and AβOs to disease evolution. The present study aims to investigate whether aging modulates AβO-induced synaptic dysfunction, neuroinflammation, neuronal death, and cognitive impairments. We employed a single intracerebral injection of AβO preparations in wild-type mice aged 3 and 18 months, and generated preliminary data using a broad and comprehensive approach that includes behavioral, biochemical, and histological analyses. In vivo administration of AβOs has become an important experimental approach for investigating the early pathogenic mechanisms of AD in young adult rodents [31–33]. By bridging the gap between aging and AβO pathology, this work provides a translational framework for testing AβO-induced toxicity across multiple AD-relevant pathways and for therapeutic development in sporadic AD. Methods Chemicals All chemicals were purchased from Sigma Aldrich unless otherwise stated. Preparation of oligomers Oligomers were prepared from human Aβ1–42 peptide obtained from Bachem. Briefly, Aβ peptide was dissolved at a concentration of 1 mM in 100% hexafluoroisopropanol (HFIP). Then, HFIP was removed under a nitrogen stream and the peptide was resuspended in dimethyl sulphoxide (DMSO) to a final concentration of 5 mM. F-12 culture medium (without phenol red) was added to bring the peptide to a final concentration of 500 µM and the solution was incubated at 22 °C for 18 h. Characterization of oligomers Coomassie blue staining The mixture of oligomers was separated using Sodium Dodecyl Sulfate–Polyacrylamide Gel Electrophoresis (SDS-PAGE), followed by Coomassie blue staining. Human oligomer preparations (20 µg) were mixed at room temperature (RT) with an equal volume of 2X Laemmli sample loading buffer. Oligomers were subjected to 15% Tris–HCl gels. Dot-blot assay An Immobilon-P polyvinylidene difluoride membrane (Millipore) was used for the dot blot assay. Aβ monomers (AβM, 4 µg), AβO (1 and 4 µg) and Aβ fibrils (AβF, 4 µg) were spotted within a pre-marked grid. The membrane was left for 2 h at RT to fix the proteins, blocked in Tris-buffered saline (pH 7.4) and 0.1% (v/v) Tween 20 (TBST) with 2% (w/v) bovine serum albumin (BSA) for 30 min (min) at RT, then incubated 1 h at RT in TBST containing 2% BSA and one of the following antibodies: anti-A11 (1:5000, rabbit, Invitrogen), Aducanumab, Gantenerumab or Lecanemab (3.75 µg/ml, human, MedChem Express). After being washed with TBST, the membranes were incubated with the appropriate anti-IgG conjugated to Europium 1024 (ImmunoReagents) at a dilution of 1:10,000 for 1 h at RT. Membranes were washed with TBST before signal detection using SpectraMax® i3x Multi-Mode Microplate Reader (Molecular Devices). Spectroscopic studies Samples (10 µM) were incubated at 37 °C, in a black 96-well fluorescence plate with an optical bottom (Greiner Bio-One), with PBS (Phosphate Buffered Saline) or amyloid oligomerization buffer and Thioflavin T (ThT; 20 µM, Thermo Fisher Scientific) or 8-Anilino-1-naphthalenesulfonic acid assay (ANS; 20 µM, Thermo Fisher scientific). Epigallocatechin gallate (EGCG, 50 µM) and phenazopyridine hydrochloride (PH) were used to inhibit fibrils and oligomer formation respectively. Fluorescence was recorded using SpectraMax® i3x Multi-Mode Microplate Reader (Molecular Devices) at λex = 440 nm and λem = scan from 465 to 595 nm (ThT assay) and at λex = 365 nm and λem = scan from 430 to 600 nm (ANS assay) with a bandwidth of 5 nm for both assays. Stability The presence of oligomers and the absence of fibrils in the AβO preparations were confirmed by monitoring the oligomeric content using the previously described spectroscopic assays and dot-blot test over several days of incubation at 37 °C. Sedimentation Velocity Analytical Ultracentrifugation (SV-AUC) Samples stored at −80°C were thawed and centrifuged at 20,000g for 5 min before use. SV-AUC experiments were carried out in a Beckman-Coulter Optima XL-I (Beckman-Coulter, Brea, CA) at 20°C using an An60-Ti rotor. AUC cell assemblies with 2-sector centerpiece and sapphire windows were filled with solutions containing AβM (2 different batches, AβM-1 and AβM-2) or AβOs and reference buffer (15 mM Hepes pH 7.5, 121 mM NaCl, 15 mM NaHCO3, 18 mM glucose). After 3 h of temperature equilibration, absorbance data and interference data were collected at 60,000 rpm and 3 min intervals for 18 h. Data were analyzed in SEDFIT using c(s) distribution models [34] with 68% regularization. The sedimentation coefficient distributions were normalized to the total signal in each distribution and to the standard conditions of 20 °C and pure water solvent (s20,w), and rendered in GUSSI. The partial specific volume of Aβ (0.7365 mL/g), buffer density (1.0066 g/mL) and viscosity (0.01031 p) were calculated using the software SEDNTERP. Cell cultures and treatments Cytotoxicity assays The 3-(4,5-dimethylthiazolyl-2)−2,5-diphenyltetrazolium bromide (MTT) assay was performed at DIV5. After 24 h of treatment, the medium was removed and replaced with fresh medium containing MTT (0.5 mg/ml). Cells were incubated for 1 h at 37 °C. The resulting MTT formazan crystals were dissolved in DMSO, and absorbance was measured at 550 nm using SpectraMax® i3x Multi-Mode Microplate Reader (Molecular Devices). The Live/Dead assay was performed at DIV5, after 24 h of treatment, according to the manufacturer's instructions, in black 96-well fluorescence plate with optical bottom (Greiner Bio-One). Briefly, cortical neurons were incubated in the dark for 30 min at RT with the LIVE/DEAD™ viability/cytotoxicity kit containing calcein-AM and the DNA stain ethidium homodimer-I (Thermo Fisher Scientific). Images were acquired using an Operetta CLS High-Content Analysis System (HCAS) (Revvity) with a 10 × objective and analyzed using the Harmony 5.1 software (Revvity). Immunocytochemistry and automated imaging and analysis Primary cortical neurons were fixed at DIV14 using a 3% v/v glyoxal fixative solution (pH 4.0) for 1 h at RT. Fixed cells were rinsed with PBS, then permeabilized and blocked using 2.5% BSA and 0.1% Triton X‐100 in PBS for 1 h at RT. Primary antibodies were diluted in blocking solution and incubated for 1 h at RT: anti-PSD95 (1:500, mouse, Abcam), anti-Bassoon (1:1000, guinea pig, Synaptic Systems) and anti-MAP2 (1:500, rabbit, Cell Signaling). After three PBS washes, samples were incubated for 1 h at RT with the appropriate Alexa Fluor 488– or 647–conjugated anti-IgG (Cell Signaling), or Alexa Fluor 555 (Abcam) at a dilution of 1:2000. Cells were counterstained with DAPI (4',6-diamidino-2-phenylindole, 1:5000, Santa Cruz Biotechnology) for 5 min at RT. To avoid biases inherent to manual quantification, an automated imaging-and-analysis workflow was used to quantify neuronal nuclei, neurite network and synapse density. Imaging was performed using Operetta CLS HCAS (Revvity) equipped with a 40 × water-immersion objective (NA 1.1). Imaging was conducted in non-confocal mode with an image resolution of 2160 × 2160 pixels. A set of 3–4 excitation/emission filters was used to detect DAPI, Alexa Fluor 647, Alexa Fluor 488 and Alexa Fluor 555. Z-stack images were acquired from 25 fields of view per well, focused on the central portion of the well. Data analysis was carried out using Harmony 5.1 (Revvity). Analysis was performed on maximum-intensity projections over a 1.5 µm height of selected planes. Nuclei were first identified using the DAPI and Alexa Fluor 647 channels to detect neuronal nuclei. Total neurite length, number of roots, extremities, and segments per well, as well as the number of nodes per neuronal cell were automatically quantified. PSD95 puncta were identified in the Alexa Fluor 488 channel using a machine-learning-based detection method to enhance accuracy of selection. All image-based measurements were summed across all acquired fields per well. For colocalization analysis, Bassoon puncta were first identified in the TRITC channel using the same machine-learning approach, and Bassoon served as the mask to detect PSD95/Bassoon overlapping puncta. Animals All procedures were performed in compliance with the 3Rs ethical principles for animal experimentation and in accordance with the European Directive of 22 September 2010 concerning animal experimentation (2010/63/EU). All experimental procedures were reviewed and approved by the local ethics committee, and an authorization was obtained from the French Ministry for Higher Education and Research. Male C57BL/6J mice of two ages (3-month-old “young” (n = 48) and 18-month-old “old” (n = 66), Janvier Lab) were used. Mice were housed individually in cages containing bedding material with food and water ad libitum and maintained under standard husbandry conditions with a reversed light–dark cycle (12 h light/dark cycle, lights on at 8:00 pm), an ambient temperature of 24 ± 2 °C and an average humidity of 50%. Body weight and general follow-up Body weight was recorded from day 0 (D0) to D3, then twice weekly. During the entire study period, each animal was observed daily and evaluated to assess its health and well-being. Surgery and stereotaxic injection On D0, a mixture of ketamine/xylazine (VIRBAC/ELANCO) (100 and 10 mg/kg respectively) was administered intraperitoneally (i.p.). During the anesthesia, rectal temperature was maintained at 37.5 ± 0.5 °C using a feedback-controlled heating pad. Mice were positioned in a sphinx position in a stereotaxic frame. A lubricant ophthalmic gel was applied to both eyes to prevent keratitis during surgery. The right side of the skull was shaved and the skin was cleaned using iodine-based antiseptic (Betadine). Under aseptic conditions, a straight incision through the skin was made. Gently, the connective tissue on top of the skull was pushed aside as needed for a clear view. Soluble AβOs (1 nmol) or Vehicle were injected into the right ventricle, with stereotaxic coordinates from the bregma being, in mm, anteroposterior –0.22, lateral –1.0 and depth 2.5. Injections were made using a 5-μL glass microsyringe fitted with a 26-gauge needle (Hamilton Company), using a microinjection pump (UMP3 UltraMicropPump, World Precision Instruments) at a rate of 1 µL/min. When the injection was completed, the needle was left in place for a minimum resting time of 2 min before withdrawing the syringe from the brain. Then, the edges of the skin were pulled together and sutured. Mice were placed individually in their home cage and carefully monitored until full recovery. The procedure ended with a single subcutaneous injection of buprenorphine (Buprecare) for general analgesia (0.05 mg/kg). Pharmacological agents and administration Freshly prepared donepezil (DPZ: 1 mg/kg) or equivalent volume of sterile saline solution were administered daily (i.p.) 1 h prior to behavioral testing. Behavioral evaluation Open field and novel object recognition test The novel object recognition (NOR) test was performed from D3 to D5 in an open area measuring 60 × 60 × 45 cm. During habituation (D3 and D4), animals were allowed to freely explore an empty arena (three trials; 10 min/trial). The distance traveled by each animal was recorded and automatically analyzed using Anymaze software (Ugo Basile). Twenty-four hours after habituation (i.e., D5), animals were exposed for 15 min to the familiar arena containing two identical objects (familiar objects, acquisition trial). Objects in each pair differed from one another in material, appearance, and texture. None of the objects used in this experiment evoked innate preference (data not shown). Five minutes later, one familiar object was replaced with a novel object for a retention test. Object type and location were counterbalanced between mice within the same experimental group to avoid object and place preference biases. All trials were video-recorded using Streampix software (Norpix). Object recognition times were scored manually by a blinded investigator. The recognition index (RI) was calculated as (novel object exploration / total exploration) × 100. Total exploration was defined as the sum of novel and familiar object exploration. Exploration was counted when the mouse directed its nose toward an object within ∼2 cm or less, and actively investigated it (sniffing, touching). Mouse motivation was assessed by total exploration time. Animals with total exploration ≤ 5 s were excluded from RI calculation and statistical analyses. Morris water maze test The Morris water maze (MWM) test was performed from D6 to D15 using a well‐established protocol, providing valuable insights into spatial learning and memory capabilities. Data were collected and analyzed using Anymaze software (Ugo Basile) and behavioral parameters (swim speed, latency, percentage of time in target quadrant, number of platform area crossings) were automatically calculated. Training consisted of four trials per day over five consecutive days. Start positions varied pseudorandomly among the four cardinal points. Each trial ended when the animal reached the platform; a 60-s cut-off was applied, after which mice were guided to the platform. Once on the platform, animals were given a 20-s rest before being returned to their cage. Twenty-four hours after the final training trial (D6), a probe trial was performed in which the platform was removed to assess retention. Euthanasia and tissue processing On D17, mice were euthanized via an overdose of pentobarbital (Exagon®, Axience). Within each treatment group, brains were collected either for ELISA or histological analyses. For ELISA, mice were perfused transcardially with heparinized saline (10 IU/mL; 6 mL/min rate). Brains were removed, hemispheres separated along the midline, and the ipsilateral hemispheres were dissected to isolate hippocampus and cortex. For histological processing, mice were perfused transcardially first with heparinized saline (10 IU/ml; 6 mL/min rate) then with 4% paraformaldehyde in 0.1 M PBS. Brains were removed and cryoprotected by incubation in a 30% sucrose/0.1 M PBS solution. Bioassays Preparation of hippocampal homogenates Isolated structures were homogenized (20 mg/ml) at 4 °C in ice-cold RIPA buffer containing protease inhibitor cocktail (Roche), 1 mM phenylmethylsulfonyl fluoride and 1 mM sodium orthovanadate, added freshly, just before use. Lysates were centrifuged at 800 g for 15 min at 4 °C. The supernatant was recovered, aliquoted and stored at − 80 °C for further analysis. Lysate protein concentrations were quantified by Pierce BCA protein assay kit (Thermo Scientific). Data were recorded at 570 nm using SpectraMax® i3x Multi-Mode Microplate Reader (Molecular Devices). ELISA ELISAs were performed for the determination of the concentrations of five biomarkers, namely PSD-95, SNAP-25, Bcl-2, Bax and TNFα, using commercially available kits according to the manufacturer’s recommendations (Cloud-Clone Corp.). Data (in duplicate) were recorded at 490 nm using SpectraMax® i3x Multi-Mode Microplate Reader (Molecular Devices) and expressed as pg or ng protein per mg of total protein. Immunohistochemistry on brain slices Frozen brain samples were embedded in optimal cutting temperature (OCT) compound (CellPath). Ipsilateral sagittal Sects. (14-µm-thick) were cut using a CM 3050S Cryostat (Leica Biosystems) at − 20°C. Sections were collected on Superfrost+ slides (Epredia™) and stored at − 80 °C until use for immunofluorescence. Co-staining of microglial (Iba1) and astrocyte (GFAP) activation was performed on one slice set to evaluate inflammation. PBS (1X, pH 7.4) was used for all washes. The slices were thawed and washed 3 times for 5 min. Brain tissue sections were permeabilized and blocked using a 0.2% Triton X-100 (Thermo Scientific), 1% bovine serum albumin (Fisher BioReagents™) and 10% fetal bovine serum (Gibco™) solution prepared in PBS 1X for ~ 1 h 30 min at RT. Then, sections were incubated at 4°C overnight with mixed antibodies in the blocking solution. The mixture was composed of anti-GFAP (1: 200, rabbit, Agilent) and anti-Iba1 (1: 200, goat, Novus-Bio). After 3 washes, the sections were incubated with the appropriate Alexa Fluor 488– or 647–conjugated anti-IgG (Cell Signaling) at a dilution of 1:200 for 2 h at RT. The staining of apoptotic cells was performed with the ApopTag® Red In Situ Apoptosis Detection Kit (EMD Millipore) following the indicated protocol. PBS (1X, pH 7.4) was used for all washes. Briefly, the slices were thawed and washed 3 times for 5 min. Brain tissue sections were incubated in equilibration buffer for 10 min and then incubated with terminal deoxynucleotidyl transferase for 1h at 37 °C. Then, sections were incubated for 10 min in Stop buffer to stop the activity of the enzyme. After 3 washes, sections were incubated for 30 min at RT with anti-Digoxigenin-Rhodamine antibody. All sections were counterstained with DAPI (1:10,000, Santa Cruz Biotechnology) for 10 min, then mounted in Fluoromount-G (InvitrogenTM), and left at 4 °C for 48 h, protected from light. Fluorescence images were digitally captured using an Olympus BX51WI epifluorescence microscope-coupled ProgRes camera (MF Cool, Jenoptik). This system was controlled by ProgRes® CapturePro software (Jenoptik). Images were captured using a 20 × objective in cortex and five different regions of the hippocampus: namely the Dentate Gyrus (DG), the three Ammon’s Horn areas (CA1, CA2 and CA3) and the Stratum Lacunosum-Moleculare (SLM). Whole-hippocampus counts were obtained by summing the five regions. Images were then analyzed by ImageJ® software (NIH). Statistical analysis Regarding in vivo data, behavioral outcomes and related biomarkers were summarized by treatment (DPZ, AβO and sham) and age (3 vs 18 months) in a two-way factorial design (3 × 2). All analyses were performed using R (v4.4.2) and JMP (v18.0.1) with α = 0.5 (two-sided). In all two-factor designs, a consistent hierarchical interpretation strategy was applied. The Treatment × Age interaction term was examined first. If the interaction was statistically significant (p < 0.05), post hoc simple-effect analyses were conducted within each age group to decompose the interaction. If the interaction term was not significant, interpretation was restricted to main effects only, and within-group comparisons were not performed or used to imply age-specific differential effects. When a significant omnibus F-test was obtained — either for the interaction term or, in its absence, for a main effect — pairwise comparisons were conducted using Tukey's HSD post hoc test, which controls for alpha inflation due to multiple comparisons. Prior to parametric testing, normality of residuals was assessed using the QQplot and Shapiro–Wilk test and homogeneity of variance was evaluated using Residuals vs Fitted graph and Levene's test. For continuous endpoints meeting model assumptions, we applied a two-way ANOVA (Type II) to test main effects and the age × treatment interaction. Post hoc pairwise comparisons were conducted using Tukey’s HSD. For count data, analyses were performed using generalized linear models (GLMs) with a Poisson distribution and log link. Post-hoc comparisons were based on estimated marginal means (EMMs) with Tukey adjustment. For non-normal continuous data or when ANOVA assumptions were violated, we used the Scheirer–Ray–Hare non-parametric two-way test. Post-hoc analyses employed Dunn’s procedure with false discovery rate (FDR) correction. For repeated-measures outcomes (e.g., body weight), we fitted mixed-effects models with animal as a random effect to account for within-subject correlation. For in vivo experiments, each mouse was considered as an independent biological replicate. For in vitro experiments, biological replicates corresponded to independent cell preparations, with technical replicates ranging from 4 to 8 per biological replicate averaged prior to statistical analysis. Single-factor in vitro comparisons were analyzed using a linear mixed-effects model. Post hoc pairwise comparisons were conducted using Tukey’s HSD. Statistical analyses were performed on biological replicate means (n = 2 per group). We acknowledge that given n = 2 biological replicates per group, formal evaluation of the validity assumptions underlying the mixed-effects model — including normality of residuals and homogeneity of variance — was not feasible, and these results should therefore be interpreted with appropriate caution. Sample size was determined a priori based on historical data and power analyses targeting 80% power at α = 0.05. Calculations assumed an effect size of Treatment (Cohen’s f ≈ 0.345) and Age (Cohen’s d ≈ 0.619) for in vivo experiments. The sample size for the various sets of data is mentioned in the corresponding figure legends. Results were expressed as mean ± standard error of mean (SEM); the distribution of experimental points is also shown. Significance levels are indicated as follows in graphs: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 and n.s., not significant. Results Biochemical characterization of AβOs The presence of soluble oligomers was assessed by Coomassie blue staining of proteins in SDS-PAGE gels. Along with traces of remaining monomers, mostly trimers and tetramers (low molecular weight, LMW) were detected as well as high molecular weight (HMW) oligomers (Fig. 1A). However, SDS-PAGE is known to create artefactual assemblies, introducing potential bias in our data. Thus, we performed SV-AUC experiments that confirmed the presence of small oligomers detected at ~ 0.87 S (59% of total signal) and 1.15 S (34% of total signal) in AβOs compared to AβM samples (Fig. 1F). These s-values translate into molecular masses of 13.8 kDa and 19.7 kDa, based on a frictional coefficient ratio (shape factor, f/f0) of 2.19, and can thus be related to trimeric and tetrameric species, confirming the results obtained with SDS-page. Larger oligomeric species ranging from 2 to 15 S are also detected but in very small quantities compared to the total sample concentrations (Fig. 1G). Estimated molecular mass distribution ranged from 64 to 814 kDa which may be related to 14- to 180-mer species. Biophysical and biochemical characterization of AβO preparations. A Coomassie blue staining of the AβOs after SDS-PAGE. B Dot blot protein-binding assay was performed using the anti-oligomer antibody A11 on AβM (4 µg), various amounts (1 µg and 4 µg) of AβOs and on AβF (4 µg). C Fluorescence spectra of ThT at excitation wavelength 440 nm. ThT-binding assays were performed in Vehicle solution (Aβ oligomerisation’s buffer) or in solutions containing AβM or AβOs ± 50 µM of EGCG, used as an inhibitor of the Aβ aggregation. The intensity of ThT fluorescence emission (relative fluorescence units) depends on the fibril concentration of each sample. Each spectrum is the average of three individual scans (data shown 24 h post incubation). D and E Fluorescence spectra of ANS at excitation wavelength 365 nm. ANS-binding assays were performed in (D) Vehicle solution (Aβ oligomerisation’s buffer) or in solutions containing AβM, AβOs or AβFs and in (E) Vehicle solution (Aβ oligomerisation’s buffer) or in solutions containing AβOs ± Phenazopyridine Hydrochloride (PH, from 0.2 µM to 2 mM), compound used to inhibit Aβ oligomerization. D and E The intensity of ANS fluorescence emission (relative fluorescence units) depends on the oligomer concentration of each sample. Each spectrum is the average of (D) four or (E) three individual scans. (F and G) SV-AUC of AβM and AβOs. Sedimentation coefficient distributions analyzed in SEDFIT using data collected at 280 nm showing the predominant higher s-value species in AβOs compared to AβM (F). The smaller peak at 0.1 S is either an optical artefact or a small contaminant molecule. Estimated masses are in line with the main presence of monomers in AβM and trimers in AβOs (G) and the magnification of c(s) distributions to highlight the larger oligomeric species detected in AβM-2 and AβOs Then, we performed further characterizations to ensure that our preparations were representative of oligomeric conformations. Dot blot assay confirmed the presence of oligomers using the anti-oligomer antibody A11 (Fig. 1B). Moreover, we used ThT as a probe to monitor amyloid fibril formation and demonstrate the absence of fibrils in oligomer aggregates. Immediately after mixing the compounds, the spectra showed no fluorescence emission peak for any of the preparations. However, a time-dependent increase in fluorescence emission was observed exclusively in the AβM condition, suggesting that new fibrils were formed from monomeric peptides (data not shown). Interestingly, both AβM + EGCG (aggregation inhibitor) and AβOs showed no fluorescence emission peak, indicating that fibril formation was prevented under these conditions (Fig. 1C), and this persisted for up to 4 days of incubation at 37°C for the AβO preparations. To further characterize AβOs, ANS, a useful tool for providing additional structural and kinetic information of the oligomeric aggregates, was used. The fluorescence spectra displayed a significant increase in fluorescence intensity upon binding to AβOs but not monomer or fibrils (Fig. 1D). Interestingly PH, used as an antiaggregant, exerted an inhibitory effect on oligomer formation in a dose-dependent manner (Fig. 1E). AβO-specific neurotoxicity and inhibition Cytotoxicity tests were performed to assess the specific neurotoxicity of oligomers. The effects of oligomers were compared with those of monomers and fibrils on mitochondrial activity (MTT assay) in primary cortical neurons. Cells were exposed to the different Aβ forms (1 µM) for 24 h (Fig. 2). The colorimetric assay showed that exposure to AβOs significantly reduced metabolic activity at 1 µM, indicating decreased cell viability compared with healthy control neurons (Vehicle), while no significant impairment was observed with the other Aβ forms. Indeed, incubation of neurons with AβOs reduced mitochondrial activity relative to Vehicle (65.8 ± 0.7% in AβO group vs 100.0 ± 1.3% in Vehicle group; P = 0.003; Fig. 2A). This disruption was significantly higher than that induced by insoluble AβF (96.5 ± 1.3% in AβF; P = 0.005 vs AβO; Fig. 2A) or AβM (104.0 ± 1.7%; P = 0.002 vs AβO; Fig. 2A) at the same concentration. No difference was detected between Vehicle and AβM or AβF (P = 0.746 and P = 0.809 respectively, Fig. 2A). An additional set of experiments demonstrated that AβOs caused a dose-dependent decrease in mitochondrial activity starting at 0.01 µM (78.2 ± 0.7% vs 100.0 ± 0.4% in Vehicle group; P < 0.001; Fig. 2B). The Live/Dead viability/cytotoxicity assay confirmed a dose-dependent increase in cell death, with a significant reduction in the viability ratio starting from 2 µM (76.0 ± 2.1% vs 100.0 ± 3.3% in Vehicle group; P = 0.013; Fig. 2C). AβOs induce dose-dependent neurotoxicity in primary cortical neurons. A MTT assay performed after 24 h of AβM (1 µM), AβO (1 µM) or AβF (1 µM) treatments (n = 12; N = 2). B MTT assay performed after 24 h of AβO (from 0.0001 to 1 µM) treatments (n = 18–48; N = 3–8). C Cytotoxicity assessed by live–dead assay in the control (Vehicle) and in AβO preparations (AβOs from 0.1 to 8 µM) (n = 12; N = 2). D MTT assay performed after 24 h of AβO (1 µM) treatment ± BDNF (15 nM) (n = 68; N = 11). E MTT assay performed after 24 h of AβO (0.1 µM) treatment ± BDNF (15 nM), ALZ-801 (10 µM) and PH (10 µM) (n = 4; N = 1). F Dot blot protein-binding assay performed using anti-Aβ antibodies (Aducanumab, Gantenerumab and Lecanemab) on various amounts (2 µg and 6 µg) of AβOs. G MTT assay performed after 24 h of AβO (0.1 µM) treatment ± anti-Aβ antibodies ratio 1:5 (Aducanumab, Gantenerumab, Lecanemab), AβOs and antibodies were mixed at RT for 2 h before neurons challenge (n = 8; N = 1). Data from separate experiments were pooled and individual data (technical replicates) are plotted (the distinct symbols for each condition represent biological replicates). Results are expressed as percentages relative to the Vehicle, untreated cultures. Data are presented mean ± SEM. Statistical significance level: *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 (B and C, only significances vs Vehicle are shown for clarity)) We then evaluated several compounds with distinct mechanisms of action for their potential neuroprotective effects against AβOs. First, BDNF was tested because it plays key roles in neuroplasticity, neuroprotection, and memory. BDNF is a growth factor belonging to the neurotrophin family that influence the proliferation, differentiation, survival and death of neuronal and non-neuronal cells. Moreover, alterations in neurotrophin levels have been implicated in neurodegenerative disorders, such as AD and in particular deficits in BDNF signaling contribute to the pathogenesis of several major diseases such as AD [35]. Compared with AβO-exposure alone, concomitant treatment with BDNF limited the decrease in metabolic activity (63.9 ± 0.5% in AβO vs 89.4 ± 0.7% in BDNF; P < 0.0001; Fig. 2D), although values did not return to Vehicle levels (100.0 ± 0.2%; P < 0.0001 vs BDNF and P < 0.0001 vs AβO; Fig. 2D). Next, compound known to inhibit AβO formation or block their toxicity is also the subject of much ongoing research. Several other disease-modifying treatments for AD targeting the amyloid cascade have been described, are on the market or are in clinical trials. Among them, ALZ-801 [1], PH [36], Aducanumab [1], Gantenerumab [2] and Lecanemab [3] have been tested in our primary cortical neuron cultures for their neuroprotective properties against AβO-induced toxicity (these experiments were carried out only once for indicative purposes). Compared to AβO-exposure, concomitant treatment with BDNF, ALZ-801 or PH limited the decrease in mitochondrial activity (66.6 ± 0.3% in AβO vs 83.8 ± 1.7% in BDNF; P < 0.0001; vs 78.0 ± 1.5% in ALZ-801; P < 0.0001; vs 8358 ± 1.3% in PH; P < 0.0001; Fig. 2E) without completely restoring it compared with the Vehicle group (100.0 ± 0.3%; P < 0.0001 vs AβO; P < 0.0001 vs BDNF; P < 0.0001 vs ALZ-801; P < 0.0001 vs PH; Fig. 2E). Dot-blot assays confirmed that all three antibodies recognized between the antibodies (Aducanumab, Gantenerumab and Lecanemab) and our AβO preparations. This assay confirmed that AβOs were recognized by all three antibodies (Fig. 2F). Compared to the physiological conditions in Vehicle treatment (i.e., 100.0 ± 1.0%), after AβO-exposure, mitochondrial activity decreased to 64.7 ± 0.4% (P < 0.0001; Fig. 2G). Aducanumab partially neutralized AβO-induced toxicity (82.3 ± 0.8%; P < 0.0001; Fig. 2G), as did Gantenerumab (84.0 ± 1.4%; P < 0.0001; Fig. 2G), and Lecanemab (85.0 ± 1.0%; P < 0.0001; Fig. 2G), without recovering the physiological conditions (P < 0.0001 vs Vehicle for all three antibodies; Fig. 2G). Altogether, these data indicate that AβOs have a direct impact on cortical neurons, impairing cell viability. This cytotoxicity is specific to oligomers and can be partially attenuated by molecules that promote cell survival, inhibit oligomer formation, or neutralize oligomers. We next examine in more detail the toxic effects induced by oligomers on cortical neurons, including alteration in the neurite network and synapse density. AβOs decreased neurite network and synaptic density in primary neurons To evaluate the effects of AβOs on DIV14 primary cortical neurons obtained from embryonic rodents, cells were exposed to varying concentrations of the oligomers (0.1, 0.5, 1, 2, and 3 µM) for 72 h (Fig. 3). Automated high-content imaging analysis was employed to quantify neuronal nuclei, neurite network and synapse density using specific neurite marker MAP2 and post-synaptic marker PSD95. Representative images of neurons exposed to a high concentration of AβOs (3 µM) revealed a striking reduction in the number of nuclei, neurite network integrity, and PSD95 puncta compared to healthy control neurons (Vehicle). Notably, reductions in post-synaptic puncta were evident even at mid-toxic concentrations of AβOs (1 µM) (Fig. 3A). AβOs induce dose-dependent morphological neurodegenerative impairments in primary cortical neurons. A Representative images of DIV14 primary cortical neurons labelled with MAP2 neurite network and PSD95 markers after a 72-h exposure to AβOs. The concentrations of AβO exposure are indicated at the top of the images. Scale bar = 50µm. Areas delineated by a white square were further magnified; scale bar = 20µm. The number of neuronal nuclei (B), neurite length (C) and PSD-95 synapse density (D) were quantified using a high-content system after 72 h treatment with Vehicle and AβOs (from 0.1 µM to 3 µM). Data from separate experiments were pooled and individual data (technical replicates) are plotted (the distinct symbols for each condition represent biological replicates), n = 16; N = 2. Results are expressed as percentages relative to the Vehicle, untreated cultures. Data are presented mean ± SEM. Statistical significance level: *P < 0.05 and ***P < 0.001 (only significances vs Vehicle are shown for clarity) Quantitative analysis indicated that exposure to AβOs tended to reduce all three morphological features in a dose-dependent manner. The number of nuclei was significantly reduced compared to Vehicle group at 3 µM AβO exposure (reduction of 50.9%; P = 0.047; Fig. 3B), while the reduction in neurite length did not reach statistical significance (−73%, P = 0.089; Fig. 3C). The number of PSD95 puncta were significantly reduced from 1 µM AβO exposure, with a reduction of 51.3% (P = 0.004; Fig. 3D). Similarly, AβO exposure appeared to affect various parameters of the neurite network, including the number of roots (neurites originating from the soma), extremities (terminal points of neurites), and individual segments (Supplementary Fig. 1B-C). A dose-dependent decline in the number of nodes was also apparent from 0.5 µM AβO exposure (reduction of 27.7%, P = 0.043; Supplementary Fig. 1E). The largest reduction compared to Vehicle after exposure to 3 µM AβOs was observed in PSD95 puncta counts (92.2% reduction; Fig. 3D), followed by neurite network measurements (73.1% reduction; Fig. 3C) and neuronal nuclei counts (50.8% reduction; Fig. 3B). In our model systems, preliminary experiments showed that AβOs induced significant reductions in PSD95 puncta, Bassoon puncta, and PSD95/Bassoon colocalized puncta, suggesting synaptic loss (Supplementary Fig. 2). Taken together, these observations suggest that quantifying PSD95 puncta may provide a more sensitive readout than neurite network or nuclei-based measurements for detecting low-dose AβO‑associated changes in primary cortical neurons, although additional studies would be required to reproduce and confirm this relative sensitivity. We next examined the alterations induced by AβOs in the normal brain of wild-type mice and the impact of aging on the response to these alterations. To this end, we conducted the study in the 3-months-old (3-mo) and in the 18-months-old (18-mo) C57BL/6J wild-type mice. Aging impairs memory performances and promotes AβO-induced cognitive decline in wild-type mice To decipher the events underlying AβO-induced pathophysiology, we injected AβOs into the i.c.v. region of 3- and 18-month-old wild-type mice, hereafter referred to as “young” and “old” mice. After a 2-day recovery period, behavioral and cognitive functions were assessed. All animals were tested in the open field task to determine baseline activity, in the NOR test (D5) to evaluate recognition and episodic memory, and in the MWM (D10-D15) to examine learning, spatial memory and executive function. These tasks were selected to evaluate behaviors supported by different brain regions involved in cognitive dysfunction in AD models. DPZ was administered daily following AβO exposure (Fig. 4A). This compound is approved by health authorities for the symptomatic treatment of mild to moderately severe forms of AD. Although its efficacy is modest and varies across patients, it has been demonstrated in clinical use. DPZ increases acetylcholine availability in the brain, thereby temporarily limiting memory loss and confusion. More recently, efforts have been made to modify the structure of DPZ in order to develop new bioactive chemical entities for AD treatment [37]. Cognitive impairments induced by AβO injection on young versus old mice. A Overview of study design. B Animal follow-up and body weight during study (n = 16–20 mice per group). C Exploration time and D RI in test trials of NOR (n = 14–16 mice per group). E Locomotor activity measured by the distance travelled in the open field test (n = 16–22 mice per group). F The mean daily escape latency calculated from four trials per day in spatial acquisition test of MWM (n = 12–17 mice per group). G Cumulative escape latency calculated from total five-day trials in spatial acquisition test of MWM (n = 12–17 mice per group). H Swimming speed measured during MWM test (n = 12–15 mice per group). I Escape latency, J percentage time of mice spent in the target quadrant and (K) the number of times that the mice crossed the original platform location in the probe trial of MWM (n = 12–15 mice per group). L The heat map of mice in the probe trial were shown (n = 12–15 mice per group). Data are presented mean ± SEM. Individual data are plotted. Statistical significance level: ×P < 0.05, ××P < 0.01 and ××××P < 0.0001, significant main effect of treatment; ¤¤P < 0.01 and ¤¤¤¤P < 0.0001, significant main effect of age; *P < 0.05, **P < 0.01 and ****P < 0.0001, interactions. #P < 0.05, ##P < 0.01, ###P < 0.001 and.####P < 0.0001 (comparison to chance, (D) RI = 50% and (J) Target quadrant = 25%) Daily monitoring showed that neither oligomer nor DPZ affected the animals' welfare. Body weight was measured every 2–3 days (Fig. 4B). The Time × Treatment × Age interaction was statistically significant (t826 = 54.29, P < 0.0001). After an initial weight loss likely due to surgery, body weight increased over time in all groups. Old mice remained significantly heavier than young mice throughout the procedure, regardless of treatments (Time × Age: t826 = −4.74; P < 0.0001). Moreover, AβO injection did not alter body-weight evolutions compared with the Sham group (t826 = 1.70; P = 0.089) or the DPZ group (t826 = 0.95; P = 0.342). Firstly, in the NOR test, the memory task relies on the innate preference of mice to spend more time exploring a novel object than a familiar one. Recognition of novel objects involves recalling previously encountered stimuli and distinguishing them from new ones [38]. During the familiarization phase, neither treatment group showed a preference for either of the identical objects (data not shown). First, total exploration time was analyzed using a two-way ANOVA. The Treatment × Age interaction was statistically significant and large (F2, 85 = 8.80, P < 0.001; Fig. 4C), indicating that the effect of treatment on total exploration differed as a function of age. Consequently, interpretation focused on simple effects within each age group. Post hoc Tukey's HSD tests revealed that in old mice, total exploration time was significantly increased in the DPZ group (69.0 ± 9.0 s) compared to the Sham group (40.6 ± 6.3 s; P = 0.026) and the AβO group (39.2 ± 6.4 s; P = 0.017), while AβO injection had no significant effect relative to the Sham group (P = 0.999). No statistically significant differences were observed among the 3-month-old treatment groups (all adjusted P > 0.05). Overall, old mice spent significantly less time exploring than young mice in the Sham group (P = 0.026) and in the AβO group (P = 0.009). Although the Treatment × Age interaction was not significant (P = 0.399), we examined the simple effects of each treatment group within each age category in order to describe treatment-related differences at each age (Treatment: F2, 85 = 3.65, P = 0.030). This approach is recommended for interpreting interactions, even when they are not significant, as it allows potentially relevant local trends to be identified. Comparisons were adjusted for multiple testing (Tukey method). Estimated marginal means post-hoc analyses showed that AβO injection significantly decreased RI in old mice (Sham: 64.6 ± 2.8% vs AβO: 54.9 ± 2.6%; P = 0.031; Fig. 4D), but not in young mice (Sham: 65.2 ± 2.4% vs AβO: 60.3 ± 2.9%; P = 0.423). Despite the increase in RI to a mean of 61.5 ± 2.9% in old mice treated with DPZ, the statistical significance was not reached (P = 0.206). Moreover, RI did not significantly differ from chance (i.e. 50%) in AβO old mice (P = 0.082; Fig. 4D), whereas young ABO mice were able to learn (P < 0.0001; Fig. 4D). Old DPZ-treated mice displayed an RI of 61.4 ± 2.9%, significantly different from chance (P = 0.002; Fig. 4D). Overall, these results demonstrate that AβO injection produced learning deficits only in old mice in the NOR test, and that these deficits were partially reversed by DPZ. Lastly, it is important to note that animals showed normal locomotor behavior. Distance covered was analyzed using a two-way ANOVA and revealed that the Treatment × Age interaction was small and not statistically significant (Treatment × Age: F2, 107 = 1.55, P = 0.217; Fig. 4E). In accordance with the pre-specified hierarchical strategy, interpretation was therefore restricted to main effects. The main effect of Treatment was small and not statistically significant (Treatment: F2, 107 = 0.71, P = 0.496), indicating that treatment group did not influence distance covered regardless of age (55.0 ± 2.8 m vs 57.8 ± 2.6 m vs 53.0 ± 2.6 m respectively in Sham-, AβO- or DPZ-treated young mice and 38.6 ± 2.0 m vs 40.9 ± 1.8 m vs 43.2 ± 1.7 m respectively in Sham-, AβO- or DPZ-treated old mice). The main effect of Age was statistically significant and large (Age: F1, 107 = 61.93, P < 0.0001). Post hoc Tukey's HSD test confirmed that young mice covered significantly greater distances than old mice (P < 0.0001), regardless of treatment group. These results indicate that distance covered was driven by age rather than treatment, with younger animals exhibiting higher locomotor activity than older mice independently of the experimental condition. Mice were then tested for spatial learning and reference memory using the MWM, a widely used test known to require intact hippocampal function. The MWM is a robust and reliable assay, strongly correlated with hippocampal synaptic plasticity. One of its major advantages is that it minimizes differences in motivation between animals, as escape from water is the primary and uniform objective for all subjects [39]. During MWM training, mice were required to locate a hidden platform positioned beneath the water surface. The Time × Treatment × Age interaction was statistically significant (t426 = 17.07, P < 0.001), indicating that all young animals exhibited good learning abilities, as did Sham and DPZ old mice, whereas AβO old mice showed significant impairment in their ability to learn this task (Time × Treatment [AβO] × Age [18-mo]: t426 = 2.57, P = 0.011; Fig. 4F). Similar treatment effects were observed when the cumulated escape latency across the 5 days of learning was analyzed. Although the Treatment × Age interaction was not significant (P = 0.414), we examined the simple effects of each treatment group within each age category in order to describe treatment-related differences at each age (Treatment: F2, 82 = 3.63, P = 0.031; Fig. 4G). Comparisons were adjusted for multiple testing (Tukey method). Estimated marginal means post-hoc analyses showed that AβO injection significantly increased cumulated escape latency in old mice (Sham: 152.6 ± 6.0% vs AβO: 185.2 ± 7.5%; P = 0.012), but not in young mice (Sham: 151.9 ± 10.4% vs AβO: 162.9 ± 7.9%; P = 0.624). Despite the decrease in cumulated escape latency to a mean of 168.5 ± 8.1% in old mice treated with DPZ, statistical significance was not reached (P = 0.304). Average swimming speed was analyzed using a two-way ANOVA and revealed that the Treatment × Age interaction was small and not statistically significant (Treatment × Age: F2, 78 = 1.66, P = 0.197; Fig. 4H). In accordance with the pre-specified hierarchical strategy, interpretation was therefore restricted to main effects. Average swimming speed was comparable across all age-specific treatment groups; the main effect of Treatment was small and not statistically significant (Treatment: F2, 78 = 1.95, P = 0.148; Fig. 4H), ruling out potential motor abilities disturbances due to AβO exposure. However, there was a significant difference in swimming speed between young and old mice; the main effect of Age was large and statistically significant (Age: F1, 78 = 54.83, P < 0.0001; Fig. 4H), which did not interfere with learning ability. After the fifth hidden-platform session, we assessed reference memory using the probe trial. The platform was removed from the pool, and animals were allowed to explore 60 s. The maze was virtually divided into four quadrants, and the time spent in each quadrant was calculated. Performance was evaluated based on the percentage of time spent in the target quadrant – where the platform had been located during training – as well as on escape latency from the platform area and the number of crossings over the former platform location. Escape latency in the probe test analyses showed the Treatment × Age interaction was statistically significant (β = 1.79, P < 0.001; Fig. 4I). Consequently, interpretation focused on simple effects within each age group. Escape latency was significantly longer in old Sham mice compared to young Sham mice (P = 0.008). Conversely, when escape performance was assessed using distance-based parameter instead of a time-based parameter, a significant difference was also observed between old and young Sham mice (Supplementary Fig. 3). Consistent with the reduction in swimming speed, the difference in escape latency during the probe test between old and young Sham mice was not attributable to learning performance, but to locomotor skills. AβO young mice showed impairment reference memory, as indicated by the significantly longer escape latency (16.5 ± 3.2 s vs 6.0 ± 1.1 s; P < 0.0001; Fig. 4I) compared with Sham young mice. This reference memory impairment was rescued by DPZ treatment, which restored performance in AβO young mice to the level of Sham mice (11.6 ± 3.3 s; P = 0.007; Fig. 4I). As expected, AβO old mice showed impaired reference memory, with higher escape latency (20.4 ± 4.8 s vs 9.9 ± 1.3 s; P < 0.0001; Fig. 4I) compared with Sham old mice, and this impairment was also reversed by DPZ treatment (8.4 ± 0.8 s; P < 0.0001; Fig. 4I). The analysis of the distance traveled to escape during the probe test (Supplementary Fig. 3) indicated that the observed impairment reflects cognitive deficits rather than a locomotor-related bias in old mice. In the analysis of the percentage of time spent in the target quadrant, the Treatment × Age interaction was small and not statistically significant (Treatment × Age: F2, 78 = 0.29, P = 0.749; Fig. 4J). Overall, AβO mice spent significantly less time in the target quadrant than Sham mice (P = 0.003; Fig. 4J), irrespective of the age (Age: F1, 78 = 0.93, P = 0.339). The effects of AβO were partially reversed by DPZ treatment (P = 0.053; Fig. 4J) in both old and young mice. It should also be noted that both young and old mice explored the target quadrant at levels significantly different from chance (Fig. 4J). Similarly, the Treatment × Age interaction was small and not statistically significant for platform crossings, and AβO mice crossed the platform area significantly less often than Sham mice (P < 0.0001; Fig. 4K), whereas DPZ treatment did not significantly improve this parameter (P = 0.505 vs AβO; Fig. 4K). Finally, old mice crossed the platform area significantly less than young mice (P = 0.002), with the effects of treatment and age being cumulative. A visible-platform trial did not reveal any differences in the time needed to reach the platform among the six groups, confirming preserved visual capacities of both young and old mice (data not shown). These data clearly demonstrate that AβOs induced memory impairments, which were more pronounced in old mice. Moreover, DPZ rescued spatial learning, reference memory, and recognition memory in AβO-treated mice. To determine whether the observed memory deficits were related to a biological process such as AβO-mediated synaptic loss or neuronal death, the expression of the synaptic proteins PSD95 and SNAP25, as well as the pro-/anti-apoptotic markers Bcl2/Bax, was evaluated in the hippocampus. Apoptosis was also assessed by immunofluorescence in the hippocampal regions and adjacent cortex of a few representative animals from each group. Aging and AβOs reduce synaptic protein level and promote both neuroinflammation and apoptosis in wild type mice The levels of pre-synaptic (SNAP-25), post-synaptic (PSD-95) and cytokine (TNFα) were quantified in the hippocampus of young and old mice, with or without AβO injection (Fig. 5A-C). The Treatment × Age interaction was not statistically significant for either synaptic protein (Treatment × Age: P = 0.464 for SNAP-25 and P = 0.923 for PSD-95) nor for the TNFα cytokine (Treatment × Age: P = 0.553). In accordance with the pre-specified hierarchical strategy, interpretation was therefore restricted to main effects. Neuroinflammation and neurodegeneration induced by AβO injection into the hippocampus of young versus old mice. The levels of proteins in the HPC tissues are determined by ELISA (n = 9–15 mice per group): levels of (A) SNAP-25 proteins, (B) PSD-95 proteins, (C) TNFα proteins, and (D) ratio of levels of Bcl-2 and Bax proteins. Astrogliosis and microglial activation, as well as apoptosis, are assessed by immunofluorescence (n = 6 mice per group): fluorescence detection of (E) GFAP, a marker of astrocytes in dentate gyrus, (F) quantification of the total number of GFAP-positive cells in dentate gyrus, and (G) in total hippocampus, (H) quantification of the percentage of GFAP-reactive cells to the total GFAP-positive cells in dentate gyrus, and (I) in total hippocampus. (J) Fluorescence detection of Iba-1, a marker of microglia in dentate gyrus, (K) quantification of the number of Iba-1-positive cells in dentate gyrus, and (L) in total hippocampus. (M) Fluorescence detection of ApopTag® TUNEL, a marker of apoptosis in dentate gyrus, (N) quantification of the number of apoptosis-positive cells in dentate gyrus, and (O) in total hippocampus. Scale bar = 100 µm. Data are presented mean ± SEM. Individual data are plotted. Statistical significance level: ×P < 0.05 and ××××P < 0.0001, significant main effect of treatment; ¤P < 0.05, ¤¤¤P < 0.001 and ¤¤¤¤P < 0.0001, significant main effect of age; **P < 0.01, ***P < 0.001 and ****P < 0.0001, interactions Although a decrease in SNAP-25 protein levels was observed in the hippocampus of AβO-injected young mice compared to Sham young mice, the effect was not statistically significant (Treatment: P = 0.239). In contrast, age induced a significant decrease in SNAP-25 levels (7.6 ± 0.5 vs 3.5 ± 0.2 ng/mg total protein in young and old Sham mice, respectively; 5.7 ± 0.2 vs 3.4 ± 0.2 ng/mg total protein in young and old AβO-injected mice, respectively; Age: P < 0.0001; Fig. 5A). AβOs and age significantly decreased PSD-95 protein levels (17.2 ± 0.6 vs 13.5 ± 0.8 ng/mg total protein in young and old Sham mice, respectively; 15.3 ± 0.5 vs 11.2 ± 0.9 ng/mg total protein in young and old AβO-treated mice, respectively; Treatment: P = 0.013; Age: P < 0.001; Fig. 5B). These results indicate that the effects of age and AβOs were cumulative for both synaptic proteins. Old mice showed lower TNFα levels compared to young mice irrespective of treatment (73.0 ± 4.1 pg/mg total protein vs 63.8 ± 1.7 pg/mg total protein in Sham mice and 75.4 ± 2.3 pg/mg total protein vs 62.7 ± 1.6 in AβO mice; Age: P < 0.0001; Fig. 5C). TNFα levels were not significatively modified by AβOs (Treatment: P = 0.660; Fig. 5C) whatever the age of the mice. Immunofluorescence analyses revealed that neither the Treatment × Age interaction (Treatment × Age: F1, 20 = 0.39, P = 0.541; Fig. 5E and F), nor age (Age: F1, 20 = 0.004, P = 0.950; Fig. 5E and F) nor AβO injection (Treatment: F1, 20 = 2.83, P = 0.108; Fig. 5E and F) significantly affected the total number of GFAP-positive astrocytes in the DG. Similar results were observed across the other hippocampal regions (DG, CA1, CA2, CA3 and SLM), which were averaged and shown in Fig. 5G. The effects of age and AβOs were cumulative in the DG (Treatment × Age: F1, 20 = 2.43, P = 0.13). Indeed, both aging (Age: F1, 20 = 85.79, P < 0.0001; Fig. 5E and H) and AβOs (Treatment: F1, 20 = 192.11, P < 0.0001; Fig. 5E and H) led to a significant increase in reactive astrocytes. However, the effect of AβOs on reactive astrocytes was significantly greater in old mice compared with young mice in the whole hippocampus (Treatment × Age: F1, 20 = 4.54, P = 0.046; Fig. 5E and I). Reactive astrocytes proportions were higher in old AβO-injected mice compared to old Sham mice (P < 0.001), and also higher than in young AβO-injected mice (P < 0.001). In the adjacent cortex, immunofluorescence analyses of astrocytes did not reveal a significant Treatment × Age interaction for either parameter. Only AβO-injected mice showed a significant increase in total astrocytes (Treatment: F1, 20 = 4.97, P = 0.037; Supplementary Fig. 4A and B), irrespective of age. However, no effect of age or AβO was observed on reactive astrocytes in the adjacent cortex (Supplementary Fig. 4A and C). Interestingly, old mice exposed to AβOs were more susceptible to astrogliosis than young mice (Fig. 5I) and this astrogliosis was accompanied by microglial activation in the hippocampus and cortex. Although the Treatment × Age interaction was not significant (P = 0.079) in the DG, we examined the simple effects of each treatment group within each age category in order to describe treatment-related differences at each age. This approach is recommended for interpreting interactions, even when they are not significant, as it allows potentially relevant local trends to be identified. Comparisons were adjusted for multiple testing (Tukey method). Estimated marginal means post-hoc analyses showed that the effect of AβOs on microglia was significantly greater in old mice compared with young mice in the DG (P = 0.001; Fig. 5K). The Treatment × Age interaction was not significant in the whole hippocampus (P = 0.189) or the adjacent cortex (P = 0.224). Irrespective of AβO injection, aging was associated with a significant increase in Iba-1-positive microglia counts in the whole hippocampus (Age: F1, 20 = 5.91, P = 0.025; Fig. 5L) and in the cortex (Age: F1, 19 = 33.03, P < 0.0001; Supplementary Fig. 4D and E). Moreover, AβO administration induced significant microglial activation in the whole hippocampus (Treatment: F1, 20 = 5.38, P = 0.031; Fig. 5L), whereas it did not significantly promote microglial activation in the cortex (Treatment: F1, 19 = 2.22, P = 0.152; Supplementary Fig. 4E). Although the difference did not reach statistical significance, it is worth noting that AβO injection produced an increase of approximately 35% in the mean number of reactive microglial cells in the hippocampus of aged mice compared with young mice. The levels of pro-/anti-apoptotic proteins (Bcl-2 and Bax) were quantified in the hippocampus of young and old mice, with or without AβO injection (Fig. 5D). The analysis revealed significant main effect of age and no statistically significant effect of Treatment or interaction between treatment and age. The levels of pro-/anti-apoptotic proteins (Bcl-2 and Bax) were quantified in the hippocampus of young and old mice, with or without AβO injection (Fig. 5D). The analysis revealed no statistically significant Treatment × Age interaction. Aging resulted in a significant disruption of the anti-/pro-apoptotic balance, with an increase in the pro-apoptotic Bax protein level and a decrease in the anti-apoptotic Bcl-2 protein level. Consequently, the Bcl-2/Bax ratio was significantly reduced in old mice compared to young mice in all groups (7.3 ± 0.3 vs 3.5 ± 0.2 in Sham mice and 8.3 ± 0.3 vs 3.0 ± 0.1 in AβO mice; Age: P < 0.0001; Fig. 5D). However, AβO injection had no significant effect on the Bcl-2/Bax ratio (Treatment: P = 0.487; Fig. 5D). Overall, these findings suggest that AβOs trigger greater neuroinflammation, particularly microglial activation, in aged mice compared to young mice. Nevertheless, analyses revealed that the Treatment × Age interaction significantly affected apoptosis, as assessed by TUNEL immunofluorescence, in the DG (Treatment × Age: F1, 20 = 5.08, P = 0.036; Fig. 5M and N) and in the cortex (Treatment × Age: F1, 18 = 96.35, P < 0.001; Supplementary Fig. 4F and G). Both aging and AβOs induced significant increases in the number of TUNEL-positive neurons (P < 0.0001 for Sham young vs Sham old mice; P < 0.0001 for AβO young vs Sham young mice; P < 0.0001 for AβO young vs Sham young mice and P < 0.0001 for AβO old vs Sham old mice). However, the synergistic effects of aging and AβOs on apoptosis varied across the brain regions. In the whole hippocampus, the Treatment × Age interaction was small and not statistically significant (P = 0.457; Fig. 5O) but the effects of aging (Age: F1, 19 = 745.7, P < 0.0001) and AβOs (Treatment: F1, 19 = 152.0, P < 0.0001) were cumulative. Taken together, these findings reveal increased neuronal apoptosis related to aging and AβO-induced toxicity, accompanied by disruptions in pro- and anti-apoptotic signaling, and reductions in key synaptic protein levels. Discussion The study was designed to evaluate whether the neurotoxic effects of AβOs are modulated by aging in wild-type mice, using well-characterized AβO preparations. First, we successfully generated stable AβOs that exhibit dose-dependent toxicity in primary neurons, progressing through distinct pathological features. Metabolic and membrane-trafficking disruptions were observed at sublethal concentrations (0.001 µM), synaptic protein loss emerged from 1 µM, followed by neurite density reduction and ultimately neuronal death at 3 µM. These findings align with previous reports using synthetic AβOs in primary neurons [40]. Our AβOs resolved into LMW species, mainly dimers/trimers, and tetramers. While oligomer toxicity does not strictly correlate with size [15, 41–44], several studies suggest that smaller assemblies tend to be more neurotoxic. LMW oligomers isolated from the buffer-soluble fraction of AD brains are strongly associated with high toxicity [45, 46]. Moreover, Aβ dimer assemblies derived from AD cerebral tissue are recognized to play a critical role in pathogenesis [47]. Given the transient and heterogeneous nature of AβOs, accurate characterization is essential to ensure biological relevance and experimental reproducibility. We rigorously characterized our AβO preparations in terms of size, aggregation state, stability, and antibody-binding profiles, including reactivity with FDA-approved antibodies. This level of characterization is crucial, as even minor variations in preparation can lead to different biological outcomes, confounding cross-study comparisons. Overall, this extensive characterization demonstrates the biological relevance of our AβOs for modelling the neurotoxic effects of human LMW AβOs in biological systems. Building on these findings, we developed a mouse model to distinguish the impact of aging – by comparing 3- and 18-month-old C57BL/6J mice, considered mature and aged respectively [48] – after a single AβO exposure on cognitive performance and underlying brain alterations. Consistent with prior work [49–52], healthy old mice exhibited significant reductions in both pre‑ and postsynaptic proteins, indicating age-related synaptic alterations, although these bulk measurements do not distinguish between actual synaptic elimination and other processes (e.g. changes in synaptic protein content). Additionally, we observed a pronounced shift in the Bax/Bcl-2 ratio toward pro-apoptotic signaling, along with an increase in TUNEL-positive cells in the hippocampus and cortex. Despite these clear alterations, healthy old mice maintained intact memory performance, although locomotor activity was reduced. This is consistent with previous reports demonstrating that non-spatial working memory remains intact in C57BL/6J mice up to 18 months of age, with impairments emerging around 22–24 months [53–56]. Thus, the 18-month-old mouse represents a relevant model for healthy brain aging, exhibiting the main hallmarks of age-related synaptic plasticity impairment without a loss in cognitive performance. In young mice, a single AβO injection (1 nmol) had no effect in the NOR test but induced a slight deficit in memory recall during the MWM probe test. These differences may reflect the timing of assessments – 5 versus 15 days post-injection for NOR and MWM respectively – suggesting a progressive increase in cognitive impairment. However, prior studies have reported impairments as early as 4 days post-injection [57, 58], indicating that methodological factors may be involved, such as differences in the AβO species used and their respective biological activities [59]. Another explanation is that AβOs may differentially affect memory subtypes: the MWM is a spatial memory test whose performances strongly correlates with hippocampal synaptic plasticity and critically depends on hippocampal integrity and functionality [60], whereas NOR assesses episodic-like memory and does not rely on spatial learning. The involvement of the hippocampus in NOR performance has been debated for years. Numerous studies have shown that hippocampal lesions do not impair NOR performance when spatial or contextual cues are not required [38, 60]. Moreover, Gaskin et al. demonstrated that hippocampal lesions induce retrograde amnesia (inability to recognize objects explored before the lesion), but not anterograde amnesia (recognition of newly explored objects after injury) in the NOR test, suggesting that NOR can be supported by extrahippocampal circuitry when the hippocampus is not engaged during initial learning phase [61]. Thus, we hypothesize that AβO injection in young mice did not impair NOR performance due to the greater resilience of neuronal circuits underlying episodic-like memory compared with those supporting spatial memory. The dissociation between preserved NOR and impaired MWM performance in young mice suggests that hippocampal circuits are more vulnerable to AβO toxicity than cortical networks. This observation aligns with human imaging data showing that hippocampal atrophy precedes cortical thinning in early AD [62]. Finally, widespread and long-lasting synaptic and behavioral deficits are poorly reproduced in studies using young mice after a single AβO injection, whereas repeated direct AβO infusions into the hippocampus have been shown to induce transient memory deficits [15, 45, 63–65]. In contrast, old mice showed robust memory deficits in both NOR and MWM following AβO injection. This difference between young and old mice has been consistently replicated in our laboratory, and AβO injection in old mice has proven to be reliable for behavioral pharmacology. For instance, we provide data showing that chronic treatment with DPZ, a drug indicated for the symptomatic treatment of mild to moderate AD [37], alleviates cognitive deficits in AβO-treated old animals. It is important to emphasize that the observed DPZ rescue reflects an acute functional effect consistent with its mechanism as an acetylcholinesterase inhibitor and the dosing regimen used (administration 1 h prior to testing). This result does not constitute evidence of disease modification. Instead, it supports the model’s validity by showing that the deficits are pharmacologically tractable and related, at least in part, to a disruption of the cholinergic system – a hallmark of AD in humans that is already present from early stages [33, 66, 67]. The marked exacerbation of AβO-related cognitive impairments in old mice underscores the age-dependent vulnerability of the brain. The behavioral deficits observed therefore reflect AβO-driven pathological processes in the absence of confounding plaque pathology, neurodevelopmental bias, or genetic overexpression artifacts, allowing focused investigation of relevant biomarkers. To define the synaptic correlates of these behavioral outcomes, we assessed the effects of AβOs on two key synaptic proteins: SNAP-25 and PSD-95. Both proteins are elevated in cerebrospinal fluid of AD patients and serve as predictors of mild cognitive impairment (MCI) symptoms evolving toward AD [68, 69]. PSD-95 has been consistently reported to decrease across various brain regions in AD patients, showing a strong correlation with the degree of cognitive impairment [70–72]. However, studies measuring SNAP-25 expression have yielded mixed results, reporting either stable protein levels or mild decreases compared to the general aged population, with poor correlation to cognitive impairment [70–74]. Notably, SNAP-25 and other presynaptic proteins abnormally accumulate in swollen axons and neuropil in the brain of AD patients, suggesting that presynaptic protein assays may overestimate synaptic density, thereby explaining the discrepancy between SNAP-25 levels and cognitive impairments [75]. Abnormal presynaptic neuronal sprouting – probably as a mechanism to compensate for synaptic loss – has also been proposed to explain the maintenance of presynaptic protein levels in AD patients, particularly at early stages with MCI [70]. In our study, AβOs produced a clear additive effect with aging on PSD-95 expression in the hippocampus of old mice, in line with our behavioral findings. In old mice, SNAP-25 levels were not further reduced beyond the effect of aging alone. Intriguingly, AβOs induced a non-significant decrease in SNAP-25 expression in young mice (−25%) while they did not alter expression in old mice (−3%), suggesting that abnormal presynaptic accumulation may also occur in aged mouse brain. Overall, these preliminary data indicate that AβO-induced synaptic alterations in mice mainly accumulate with aging. Additional studies incorporating mRNA profiling and an expanded panel of synaptic markers will be required to determine whether the observed reductions in protein levels truly reflects synapse loss. Besides synaptic alterations, we also observed pronounced age-related and AβO-induced neuroinflammation in the hippocampus, as shown by increased microglial activation and astrogliosis. This aligns with growing evidence that glial dysfunction contributes to AD pathogenesis [76, 77]. Our data suggest that the specific interplay between aging and AβOs may depend on the cell type and the hippocampal subregion considered. While AβO-induced astrocyte activation was potentiated by age in the whole hippocampus, the effect of AβOs and aging were only additive when focusing on the DG. Conversely, the effect of AβOs on microglia was potentiated by age in the DG, whereas AβOs and aging were strictly additive in the whole hippocampus. Overall, the effects of age and AβOs were cumulative, as old mice injected with AβOs consistently displayed much more neuroinflammation than the other groups. While we noted gliosis, the specific inflammatory cascades driving AβO toxicity remain unclear. With aging, microglia may shift from an AβO-clearing phenotype to one dominated by cytokine release, particularly IL-1β, which exacerbates synaptic loss [78]. An important next step will be to determine whether AβOs trigger inflammasome activation in old brains, as this pathway links AβO recognition to inflammatory damage and cell death. In our study, the AβO-induced neuroinflammatory response coincided with an exacerbation of apoptotic signaling. We observed that a single administration of AβO increased apoptosis in both young and old mice, irrespective of the brain region. In AD patients, hallmarks of apoptosis are well described, including DNA fragmentation detected by TUNEL staining, reduced Bcl-2 expression, and increased Bax expression in neurons. Neuronal Bcl-2 protein levels are affected by age and dysregulation of Bcl-2-family members have been associated with the AD progression. The balance between pro-and anti-apoptotic Bcl-2 family proteins shifts towards the pro-apoptotic side during disease progression, increasing neuronal vulnerability [79, 80]. Interestingly, the pronounced pro-apoptotic shift in the Bax/Bcl-2 ratio observed in the old mouse hippocampus did not translate into a synergistic interaction with AβO exposure to further enhance apoptosis as quantified by TUNEL staining. Indeed, apoptotic cell counts increased twice as much in young mice as in old mice in the DG, whereas the effects were strictly additive when measured across the whole hippocampus. In contrast, the effects of AβOs on apoptosis in the cortex were clearly potentiated by age. Overall, old mice injected with AβOs consistently displayed higher levels of apoptosis than all the other groups. Taken together, our results demonstrate that the effects of AβOs and aging are at least cumulative, based on convergent alterations in synaptic, neuroinflammatory, and apoptotic biomarkers. The brains of 3-month-old mice injected with AβOs exhibit histopathological changes similar to those of healthy 18-month-old mice, suggesting an accelerated aging-like phenotype. In 18-month-old mice, AβO injection combined with existing age-related changes to produce clear functional impairments, suggesting that the combination exceeded a critical threshold for the brain’s coping capacities. Although preliminary, our data indicate that age synergizes with AβOs to promote neuroinflammation and apoptosis, but this interaction is region‑dependent and does not occur uniformly across brain areas. Physiological aging is driven by gene expression alterations, as demonstrated in mice, rat, monkey and humans. These changes affect key cellular signaling pathways involved in brain cell function and memory processes. The GABAergic, glutamatergic, cholinergic, and dopaminergic systems notably exhibit progressive dysfunction, impairing synaptic transmission and plasticity. Such impairments arise, for example, from disruptions in glutamate‑, GABA- or acetylcholine‑dependent signaling pathways that are essential for maintaining the excitation/inhibition balance and for long‑term potentiation [81–86]. In parallel, several fundamental cellular processes deteriorate with age, particularly autophagy (LC3, Beclin‑1, ATG proteins), a key mechanism for the removal of damaged proteins and organelles. Its decline compromises the clearance of toxic proteins and interferes with signaling pathways governing cellular homeostasis, thereby promoting neuronal senescence and the disorganization of memory networks. Furthermore, systemic dysregulations affecting intercellular communication, genomic stability (PARPs, p53, p21), and mitochondrial metabolism (mTOR, AMPK, sirtuins) contribute to metabolic stress and reduce the efficiency of adaptive responses in aged neurons, increasing the risk of brain function deterioration in both rodent models and humans [83–86]. In this framework, the hypothesis that misfolded proteins such as tau or Aβ aggregates may produce distinct outcomes in young versus old brains has long been proposed, but has never been directly tested. To the best of our knowledge, we provide the first demonstration of synergistic interactions between aging and AβOs, thereby introducing a novel experimental paradigm suited to investigate the underlying mechanisms that drive this interaction. We also propose that intracerebral AβO injection in aged wild-type mice offers a physiologically relevant, non-transgenic tool for drug testing targeting early sporadic AD. Indeed, this model recapitulates several key features of human disease, including some age-dependent vulnerabilities and cognitive deficits, pre-plaque synaptic alteration, and responsiveness to symptomatic treatment, without the confounding effects of genetic overexpression. This study has several limitations. First, we did not assess the effects of a chronic exposure to AβOs, nor whether our model recapitulates tau pathology, an important hallmark of AD. A related limitation is that our model does not evaluate potential secondary effects of exogenous AβOs on other AD- or neurodegeneration-associated proteins (e.g., endogenous Aβ, tau, TDP-43, or α-synuclein). While AβOs may indirectly influence tau phosphorylation or neuroinflammatory priming of co-pathologies, our experimental scope and tissue analyses were restricted to AβO-specific endpoints. Future studies could extend this model to examine whether aging amplifies AβO-induced tauopathy or other proteinopathic cascades. Finally, our study is limited by the use of only male mice. Future studies incorporating both sexes are needed to determine whether the age-related susceptibility to AβOs is sex-dependent and to develop sex-tailored therapeutic strategies. Conclusions Our findings demonstrate that aging and AβOs either accumulate or act synergistically in mice to increase neuroinflammation and apoptosis in a region‑dependent manner, while their impact on synaptic protein content appears strictly additive. In our study, young mice exhibited relative resilience, showing only mild memory impairments following AβO injection. In contrast, AβO injection in old mice exceeded the coping capacities of the aged brain resulting in impaired episodic and spatial memory performances. In conclusion, our study provides the first evidence that the aging brain is more susceptible to AβO-induced neurotoxicity, paves the way for further investigation into the underlying mechanisms, and highlights the value of incorporating aging into preclinical models to improve their translational validity and enhance their relevance for drug testing targeting early stages of sporadic AD. Supplementary Information Supplementary Material 1: Supplementary Figure 1. AβOs induce dose-dependent neurite impairments in primary cortical neurons. (A) Schematic representation illustrating the neurite quantified parameters: the number of extremities defined as the terminal points of individual neurites (marked by numbered grey circles), the number of roots corresponding to the neurites originating from the soma (marked by numbered black circles), the number of individual segments within each detected neurite tree (represented using three distinct colors within a single neurite tree) and the number of primary nodes corresponding to the primary bifurcation points along neurites (marked by orange circles). (B-E) The number of neuronal roots (B), neurite extremities (C), and neurite segments (D) and nodes (E) were quantified using a high-content system after 72 h treatment with Vehicle and AβOs (from 0.1 µM to 3 µM). Data from separate experiments were pooled and individual data (technical replicates) are plotted (the distinct symbols for each condition represent biological replicates), n = 16; N = 2. Results are expressed as percentages relative to the Vehicle, untreated cultures. Data are presented mean ± SEM. Individual data are plotted. *P < 0.05, **P < 0.01 and ****P < 0.0001 (only significances vs Vehicle are shown for clarity). Supplementary Material 2: Supplementary Figure 2. AβOs induce dose-dependent synaptic impairments in primary cortical neurons. (A) Representative images showing colocalization of PSD95 (postsynaptic) and Bassoon (presynaptic) along MAP2-positive neurites. Scale bar = 10 µm. (B) Representative images of DIV14 primary cortical neurons labelled with MAP2 neurite outgrowth and PSD95 markers after a 72-h exposure to AβOs. The concentrations of AβO exposure are indicated at the left and markers are shown at the top. Scale bar = 20µm. (C–E) Quantification of PSD95 puncta, Bassoon puncta, and colocalized PSD95/Bassoon puncta using a high-content imaging system after 72 h treatment with vehicle or AβOs (1 µM and 3 µM). Results are expressed as percentages relative to the Vehicle, untreated cultures. Data are presented mean ± SEM. Individual data are plotted; n = 6; N = 1. Statistical significance level: ****P < 0.0001 (only significances vs Vehicle are shown for clarity). Supplementary Material 3: Supplementary Figure 3. Memory performance evaluated based on distance travelled during the probe trial of the MWM. (A) Escape latency based on distance travelled in the probe test of MWM (n = 12-15 mice per group). Data are presented mean ± SEM. Individual data are plotted. ××××P < 0.0001, significant main effect of treatment; ¤¤¤¤P < 0.0001, significant main effect of age. Supplementary Material 4: Supplementary Figure 4. Neuroinflammation and neurodegeneration induced by AβO injection into the cortex of young versus old mice. (A) Fluorescence detection of GFAP, (B) quantification of the total number of GFAP-positive cells, and (C) quantification of the percentage of GFAP-reactive cells to the total GFAP-positive cells in cortex. (D) Fluorescence detection of Iba-1, (E) quantification of the number of Iba-1-positive cells in cortex. (F) Fluorescence detection of ApopTag® TUNEL, and (G) quantification of the number of apoptosis-positive cells in cortex. Scale bar = 100 µm. Data are presented mean ± SEM (n = 6 mice per group). Individual data are plotted. Statistical significance level: ×P < 0.05, significant main effect of treatment; ¤¤¤¤P < 0.0001, significant main effect of age; ****P < 0.0001, interactions. Acknowledgements The authors wish to thank the Integrated Structural Biology platform of the Strasbourg Instruct-ERIC center IGBMC-CBI supported by FRISBI (ANR-10-INBS-0005) and the UMR Inserm 1256 nGERE. Associated data This section collects any data citations, data availability statements, or supplementary materials included in this article. Abbreviations Amyloid-beta Fibrils Amyloid-beta Monomers Amyloid-beta Oligomers Alzheimer’s disease 8-Anilino-1-naphthalenesulfonic acid assay Brain-Derived Neurotrophic Factor Bovine Serum Albumin Three Ammon’s Horn areas 4',6-Diamidino-2-Phenylindole Dentate Gyrus Days In Vitro Dimethyl Sulphoxide Donepezil Day with x as the number of days from surgery Epigallocatechin Gallate Hour(s) Enzyme-Linked Immunosorbent Assay High-Content Analysis System Hexafluoroisopropanol Higher Molecular Weight Intracerebroventricular injection Intraperitoneally Low Molecular Weight Mild Cognitive Impairment Minute(s) 3-(4, 5-dimethylthiazolyl-2)-2, 5-diphenyltetrazolium bromide Morris Water Maze Novel Object Recognition Phosphate Buffered Saline Phenazopyridine Hydrochloride Recognition Index Room Temperature Sodium Dodecyl Sulfate–PolyacrylAmide Gel Electrophoresis Standard Error of Mean Stratum Lacunosum-Moleculare Sedimentation Velocity Analytical Ultracentrifugation Tris-Buffered Saline Tween 20 Thioflavin T Authors’ contributions Ahmad Allouche and Julie Colin contributed equally to this work. Funding This work was supported by grants from the “Region Grand-Est” and the European Union through the operational programme “FEDER-FSE Lorraine et Massif des Vosges 2014–2020”. Data availability The authors confirm that the data supporting the findings of this study are available within the article and its supplementary material or are available from the corresponding author, upon reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Ahmad Allouche and Julie Colin contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1: Supplementary Figure 1. AβOs induce dose-dependent neurite impairments in primary cortical neurons. (A) Schematic representation illustrating the neurite quantified parameters: the number of extremities defined as the terminal points of individual neurites (marked by numbered grey circles), the number of roots corresponding to the neurites originating from the soma (marked by numbered black circles), the number of individual segments within each detected neurite tree (represented using three distinct colors within a single neurite tree) and the number of primary nodes corresponding to the primary bifurcation points along neurites (marked by orange circles). (B-E) The number of neuronal roots (B), neurite extremities (C), and neurite segments (D) and nodes (E) were quantified using a high-content system after 72 h treatment with Vehicle and AβOs (from 0.1 µM to 3 µM). Data from separate experiments were pooled and individual data (technical replicates) are plotted (the distinct symbols for each condition represent biological replicates), n = 16; N = 2. Results are expressed as percentages relative to the Vehicle, untreated cultures. Data are presented mean ± SEM. Individual data are plotted. *P < 0.05, **P < 0.01 and ****P < 0.0001 (only significances vs Vehicle are shown for clarity). Supplementary Material 2: Supplementary Figure 2. AβOs induce dose-dependent synaptic impairments in primary cortical neurons. (A) Representative images showing colocalization of PSD95 (postsynaptic) and Bassoon (presynaptic) along MAP2-positive neurites. Scale bar = 10 µm. (B) Representative images of DIV14 primary cortical neurons labelled with MAP2 neurite outgrowth and PSD95 markers after a 72-h exposure to AβOs. The concentrations of AβO exposure are indicated at the left and markers are shown at the top. Scale bar = 20µm. (C–E) Quantification of PSD95 puncta, Bassoon puncta, and colocalized PSD95/Bassoon puncta using a high-content imaging system after 72 h treatment with vehicle or AβOs (1 µM and 3 µM). Results are expressed as percentages relative to the Vehicle, untreated cultures. Data are presented mean ± SEM. Individual data are plotted; n = 6; N = 1. Statistical significance level: ****P < 0.0001 (only significances vs Vehicle are shown for clarity). Supplementary Material 3: Supplementary Figure 3. Memory performance evaluated based on distance travelled during the probe trial of the MWM. (A) Escape latency based on distance travelled in the probe test of MWM (n = 12-15 mice per group). Data are presented mean ± SEM. Individual data are plotted. ××××P < 0.0001, significant main effect of treatment; ¤¤¤¤P < 0.0001, significant main effect of age. Supplementary Material 4: Supplementary Figure 4. Neuroinflammation and neurodegeneration induced by AβO injection into the cortex of young versus old mice. (A) Fluorescence detection of GFAP, (B) quantification of the total number of GFAP-positive cells, and (C) quantification of the percentage of GFAP-reactive cells to the total GFAP-positive cells in cortex. (D) Fluorescence detection of Iba-1, (E) quantification of the number of Iba-1-positive cells in cortex. (F) Fluorescence detection of ApopTag® TUNEL, and (G) quantification of the number of apoptosis-positive cells in cortex. Scale bar = 100 µm. Data are presented mean ± SEM (n = 6 mice per group). Individual data are plotted. Statistical significance level: ×P < 0.05, significant main effect of treatment; ¤¤¤¤P < 0.0001, significant main effect of age; ****P < 0.0001, interactions. Data Availability Statement The authors confirm that the data supporting the findings of this study are available within the article and its supplementary material or are available from the corresponding author, upon reasonable request.