AETA peptide drives Alzheimer’s disease signature of synapse dysfunction

doi:10.21203/rs.3.rs-8192959/v1

Download PDF

Research Article

AETA peptide drives Alzheimer’s disease signature of synapse dysfunction

https://doi.org/10.21203/rs.3.rs-8192959/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Alzheimer’s disease (AD), the leading cause of dementia, is characterized by early synaptic dysfunction that precedes overt cognitive decline. While amyloid-β and Tau remain central to AD pathogenesis, molecular triggers of synapse weakening remain unclear. Here, we investigated AETA, a novel brain-secreted peptide derived from amyloid precursor protein (APP), as a potential mediator of synapse dysfunction in AD. We previously identified AETA as a unique modulator of NMDA receptor activity in the healthy brain; however, its role in AD etiology was yet to be explored. Post-mortem analyses of human hippocampal and prefrontal cortex tissues revealed significantly elevated AETA levels in AD patients, particularly in females. To further explore the contribution of AETA to AD synaptic pathology, we analyzed a new mouse model, the AETA-m mouse, exhibiting chronically increased brain AETA expression. RNA sequencing of AETA-m hippocampi revealed synaptic gene expression alterations that closely paralleled those observed in vulnerable human AD brain regions, most notably in the hippocampus. Functionally, hippocampal neurons from AETA-m mice displayed impaired NMDA receptor signaling, dendritic spine loss, and memory deficits especially in females, mirroring early AD-associated synaptic dysfunction. Together, these findings identify AETA as a novel key contributor of synaptic vulnerability in AD. Targeting AETA signaling may therefore offer new therapeutic avenues for preventing or mitigating synaptic and cognitive decline in AD.

synapse

hippocampus

memory

Alzheimer’s disease

Amyloid precursor protein

Alzheimer’s disease (AD), the most prevalent form of dementia, is clinically characterized by progressive memory decline [3]. Post-mortem examination of AD brains consistently reveals hallmark pathological features: extracellular amyloid-β (Aβ) plaques and intracellular neurofibrillary tangles composed of hyperphosphorylated Tau proteins [35]. While Aβ and Tau remain central to AD pathogenesis, growing evidence suggests that they act synergistically within a broader landscape of molecular disruptions, particularly during the early, preclinical ‘cellular phase’ of AD [10].

Aβ peptides originate from the proteolytic cleavage of the amyloid precursor protein (APP) through the amyloidogenic processing pathway. In contrast, non-amyloidogenic processing does not yield Aβ. A third APP cleavage route, the η-secretase pathway, was identified more recently [49] (Fig. 1a). This pathway involves cleavage by η-secretase at the N-terminal extracellular domain of APP, generating soluble APPη and a membrane-bound C-terminal fragment (CTFη). Subsequent cleavage of CTFη by α- or β-secretases produces distinct Aη peptides (Aη-α and Aη-β), collectively termed AETA.

We previously demonstrated that AETA impairs synaptic function by inhibiting long-term potentiation (LTP) at CA3-CA1 synapses and suppressing hippocampal calcium dynamics in vivo [26, 49]. More recently, we elucidated the molecular mechanism of the acute action of AETA at excitatory synapses [11]. AETA partially inhibits the ionotropic activity of NMDA receptors (NMDARs), by competing with the co-agonists glycine and D-serine. Concurrently, AETA promotes a non-canonical, ion flux-independent signaling mode of NMDARs, ultimately leading to a rapid reduction of synaptic strength and dendritic spine retraction. This AETA-dependent regulation of NMDARs is necessary in the healthy brain as, when AETA production is prevented, synapses cannot weaken when submitted to a long-term depression protocol [11].

Synaptic dysfunction and dendritic spine loss in the hippocampus are among the earliest and most predictive pathological changes in AD, tightly associated with cognitive decline and linked mechanistically to NMDAR signaling abnormalities [16, 20, 22, 38, 39, 43]. Several culprits have been previously identified as contributors to this synapse dysfunction including Aβ, Tau, mitochondrial dysfunction and microglia and astrocytes [16, 43]. However, the precise molecular triggers of this dysfunction remain to be fully elucidated, hindering therapeutic development for treatment of AD. Notably, how AETA contributes to AD is currently unknown. Here we show that AETA increases in human AD patient brains. Mimicking this elevation in the rodent brain leads to human AD-like molecular and functional signatures of synapse dysfunction associated with hippocampus-dependent memory decay, particularly in female mice. These findings position AETA as a critical mediator of synaptic pathology in AD and highlight its potential as a novel therapeutic target.

Human brain samples

All experiments were performed in accordance with relevant guidelines and regulations. The case–control study of post-mortem human brain was approved by ministère de l’enseignement supérieur, de la recherche et de l’innovation (CODECOH N° DC-2022-5317). Human hippocampi were obtained from The Netherlands Brain Bank (NBB), Netherlands Institute for Neuroscience, Amsterdam, Netherlands (open access: www.brainbank.nl). Human prefrontal cortex samples were obtained from the Neuro-CEB Brain bank, Hôpital de la Pitié-Salpêtrière, Paris, France (open access: www.neuroceb.org). All Material was collected from donors for or from whom a written informed consent for a brain autopsy and the use of the material and clinical information for research purposes had been obtained by the NBB and Neuro-CEB. Patients had ante-mortem evidence of clinical dementia, whereas controls did not. Controls were selected by matching for age and sex (see Supplemental Tables S1 and S2 for details).

Mice

All experiments and protocols on mice were performed in accordance with specific European laws described in the European Communities’ Council Directive 2010/63/EU. Protocols used in this study were approved by the committee for the Care and Use of Laboratory Animal in the framework of project authorizations APAFIS#6856-2016091610462338 and APAFIS#37493-2022050311352580s. Mice had ad libitum access to tap water and standard chow. Mice were maintained under constant environmental conditions (12h:12 h light/dark cycle, 23 ± 2°C and humidity of 55%). They were housed in groups of 5 or 6 animals by sex in standard mouse cages (542 cm²) under Specific Pathogen Free (SPF) status. Sex used is reported for each experiment.

Generation and maintenance of AETA-m mouse line

Generation of AETA-m mouse line was briefly described previously [11]. Human AETA (the long alpha form) was first cloned into the multiple cloning site of pSecTag2 A plasmid downstream of the Ig k-chain leader sequence for secretion (Invitrogen Life Technologies, France) and then cloned into cDNA inserted in the mouse thy-1 gene cassette at the XhoI site of the pTSC21k plasmid [2, 27]. This expression cassette was then linearized and injected into male pronuclei of fertilized zygotes of C57Bl6/NCrl donor mice. Embryos were transferred into Crl:CD1(ICR) pseudopregnant recipient female mice (ca. 20 embryos per recipient). The recipient mice were mated with sterile males (vasectomized) Crl:CD1(ICR). The AETA-m mouse line was chosen amongst four founder lines as it displayed moderate expression of recombinant AETA in the hippocampus and cortex (data not shown). Mice were generated under the license 24-9168.11-9/2012-5. All animal experiments were performed in accordance with the European Communities Council Directive (86/609/EEC) and were approved by the local ethics committee (Government of Saxony, Germany). The line was backcrossed regularly (at least 8 generations prior to use of mice for experiments) on C57/BL6J background strain (Charles River, France). Mouse genotype was confirmed by genomic DNA extraction from tissue and PCR as follows: primers sequences: CGCGCCATGATTAGTGAA and GACCTCTGCAGAGGAAGGA; PCR protocol: 5 min 95°C, 10 cycles of 94°C-30s/62°C-30s/72°C-30s, 24 cycles of 94°C-30s/52°C-30s/72°C-30s, 7 min 72°C. PCR samples were run on 2% agarose gel containing ethidium bromide and visualized by UV light.

Immunoblotting of APP peptides in mouse and human tissues

Mouse and human brain homogenates for analysis of APP processing products (AETA for human brain; AETA, APP-FL, sAPPa/b and CTF-η for mouse brains) were essentially prepared as described previously[49]. In brief, DEA lysates (0,2% Diethylamine in 50 mM NaCl, pH 10) and RIPA lysates (20 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM Na2EDTA, 1% NP-40, 0,5 % sodium deoxycholate, 0,05 % Triton X-100) with protease inhibitors (Sigma-Aldrich, P8340) were prepared from brain samples using the Precellys system (Bertin) for homogenisation followed by ultracentrifugation. For the DEA and RIPA samples we used the Bradford (Biorad) to measure protein concentration, which was adjusted equally to all samples before immunoblotting. AETA (human and mouse) and sAPPa/b (mouse) and GAPDH (human) were quantified in DEA fraction and CTF-η (mouse) and full-length APP (mouse) were quantified in RIPA fraction. For detection by immunoblotting, proteins were separated on 8% Tris-Glycine gels or alternatively on Tris-Tricine (10-20%, Thermo Fisher Scientific) gels, transferred to nitrocellulose membranes (0.22 μm, GE Healthcare) which were boiled for 5 min in phosphate buffer saline (PBS) and subsequently incubated with the blocking solution containing 0.2% I-Block (Thermo Fisher Scientific) and 0.1% Tween 20 (Merck) in PBS for 1 hour, followed by overnight incubation with antibody in the blocking solution. Antibody detection was performed using the corresponding anti-rat/mouse/rabbit-IgG-HRP conjugated secondary antibody (Thermo Fisher Scientific) and chemiluminescence detection reagent ECL (Thermo Fisher Scientific). Antibodies used for immunoblotting were: anti-GAPDH (#G9545, Sigma-Aldrich, rabbit IgG, 1/1000 dilution), M3.2 (Biolegend, #805701; mouse IgG, 1/5000 dilution) for detection of mouse AETA and CTF-h, 2D8 (Sigma-Aldrich, #MABN2273, rat IgG; 2 mg/mL) or 2E9 (Sigma-Aldrich, #MABN2295, rat IgG, 2 mg/mL) for detection of human AETA, Y188 (Abcam, ab32136; rabbit IgG, 1/1000 dilution) for detection of human and mouse APP, and 22C11 (Merck, #MAB348; mouse IgG, 1/5000 dilution) for detection of mouse sAPPa/b. For the loading control, when necessary, we used an antibody specific to β-actin (Sigma-Aldrich, #A5316, mouse IgG, 1/5000 dilution) or ponceau staining (as stated in results description, see also Supplemental file with full blots). For quantification, peptide levels were normalized to WT average for each immunoblot. For quantification of recombinant human AETA in hippocampi of AETA-m mice, the standard curve was made using the following synthetic peptide (Peptide Specialty Laboratories; PSL GmbH; Heidelberg, Germany):

MISEPRISYGNDALMPSLTETKTTVELLPVNGEFSLDDLQPWHSFGADSVPANTENEVEPVDARPAADRGLTTRPGSGLTNIKTEEISEVKMDAEFRHDSGYEVHHQK.

The peptide was dissolved in dimethyl sulfoxide (DMSO) at 100 mM and placed at −80°C for long term storage and diluted to final concentration on day of experiment.

Immunohistochemistry

Brains were quickly removed and maintained in 4% paraformaldehyde (PFA) for 24h, then washed in PBS. Brain sections (40µm) were sliced with a vibratome (Leica VT 1000S). Cresyl violet staining to label Nils bodies was performed on slices just after slicing procedure as follows. Slices were mounted on superfrost slides (VWR) and allowed to dry. Dried slices were stained for 3 min in cresyl violet solution (5g cresyl violet/3ml acetic acid/ 1L ddH₂O, filtered), then well rinsed in ddH₂O. Slices were further washed in successive baths 70, 90 and 100% Ethanol for 1 min each. Slices were dried again at room temperature and mounted in Entellan (Sigma-Aldrich). Pictures were taken under Olympus BX41 microscope at 10X magnification. Cell body layer thickness was calculated manually with Image J software with the experimenter blinded to mouse genotype.

RNA sequencing

Mouse hippocampi were rapidly dissected and snap frozen in liquid nitrogen. Messenger RNAs (mRNA) from tissues were prepared following the Illumina® Stranded mRNA Prep protocol. They were then sequenced on an Illumina® NextSeq 2000 sequencer, each library receiving at least a 40M reads sequence coverage. Generated FASTQ files were used as input of the nf-core/rna-seq nextflow pipeline (v3.8.1), which performed quality control, trimming and alignment to the mouse reference genome GRCm39 with default parameters (alignment with STAR v2.7.10a followed by quantification with Salmon v1.5.2). Raw counts matrix was then loaded and processed in R (v4.3.1) to perform differential expression analysis following the standard DESeq2 framework (v1.40.2). Genes with a log2FC > 0.2 or log2FC < -0.1, an adjusted P-value < 0.05 and a minimal base mean > 10 in either AETA-m or WT were designated as differentially expressed (DESeq2 analysis output provided in Supplemental Tables S4 and S5). Functional annotation of genes identified as differentially expressed was performed using the Ingenuity Pathway Analysis (IPA) tool (QIAGEN Inc., https://digitalinsights.qiagen.com/IPA). For comparison with the human NBB-HPC-AD cohort, we extracted the RNA-seq data from the supplemental tables provided by Van Rooij et al. [36]. To compare the neuron function-linked signature obtained in the AETA-m mouse model with the other human AD RNA-seq data, we extracted Supplemental tables provided by Marques-Coelho et al. [24], which correspond to the differential expression analysis between disease and control RNA-seq samples from MSBB BM36 and MSBB BM22 cohorts [46], and the ROSA cohort [9], performed with DESeq2. For comparison with the NBB-PD cohort [7], we ran the differential expression analysis between Braak Lewis Body stages 5 versus 0 (no pathology) samples with the same workflow used for the analysis of our mouse data, starting from the raw count matrix extracted from GSE216281.

qPCR

RNA was extracted from human hippocampi obtained from the NBB (Supplemental Table S1) using the RNeasy Lipid Tissue kit (Qiagen) according to the manufacturer's instructions. The samples were retrotranscribed using Promega GoScript Reverse Transcriptase kit according to the manufacturer's instructions in a Biometra TAdvanced thermal cycler according to the program: 5min at 25°C, 50min at 42°C and 15min at 70°C. The qPCR amplification of genes was performed on a mix consisting of 1 μL of RT product diluted at 900ng/µL and 19 μL of mix (containing 10 μL of LightCycler 480 SYBR green I master, 7 μL of UltraPure water, 1μL of forward primer and 1 μL reverse primer). The qPCR was performed with the LightCycler 480 (Roche) (95°C for 5 minutes, (95°C for 10 seconds, 60°C for 30 seconds, 72°C for 30 seconds) x 40 cycles). The melting curve was obtained by increasing the temperature by 1°C every 30 seconds from 70°C to 95°C. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a reference gene. Primers used for this qPCR analysis are described in Supplemental Table S3.

Mouse hippocampi were dissected and snap frozen in liquid nitrogen. The samples were retrotranscribed in a Biometra gradient T-type thermal cycler (Applied Biosystems VeritiPro) according to the program: 10sec at 25°C, 120min at 37°C, 5sec at 85°C. The qPCR amplification of genes was performed using SYBR green (Applied Biosystems; #4367659). The samples were diluted to 1/10 (for an RT of 500 ng/20 μL). qPCR was performed on a mix consisting of 2 μL of RT product diluted and 8 μL of mix (containing 5 μL of SYBR green, 2.8 μL of RNase-free water, 0.1 μL of forward primer and 0.1 μL reverse primer). The qPCR was performed with the StepOnePlus Real-Time PCR System (Biosystems) (50°C for 2 minutes, 95°C for 10 minutes, (95°C for 15 seconds, 60°C for 25 seconds) x 40 cycles, 95 ° C for 15 seconds). The melting curve was obtained by increasing the temperature by 1°C every minute from 60°C to 95°C. Peptidylprolyl isomerase A (PPIA, also known as Cyclophilin A) was used as a reference gene. Primers used for this qPCR analysis are described in Supplemental Table S3.

Immunoblotting of p38 and tau in mouse hippocampi

For p38 immunoblotting, freshly dissected mouse hippocampi were homogenized with a potter and then passage through a 1ml syringe with needle (26G) in 150 μL Tris buffer (pH 7.5) containing 2% SDS, 1% triton X100, 10% glycerol, protease inhibitors (#P5726, Sigma-Aldrich) and phosphatase inhibitors (#P8340, Sigma-Aldrich). Lysates were centrifuged for 10 min at 9000 rpm. Proteins in the supernatant were quantified by BCA assay (Pierce) and stored at -80°C until further use. 10 or 20 μg of samples were loaded on precast Novelx 4-20% Tris-tricin gels, in quadruplicates, and transferred to nitrocellulose membranes (0.22 μm, GE Healthcare) which were subsequently incubated with the blocking solution containing 0.2% I-Block (Thermo Fisher Scientific) and 0.1% Tween 20 (Merck) in PBS for 1 hour, followed by overnight incubation with primary antibody in the blocking solution. One set of gels was first immunoblotted for P38, then membranes were stripped, reblocked as above and then immunoblotted for P-P38. Another set of gels was first immunoblotted for P-P38 and then P38. The stripping procedure consisted of incubating membranes in ddH₂0 containing 1.5g/L glycine, 0.1% SDS, 1% tween 20 for 15 min at room temperature followed by rinsing two times for 5 min with PBS. Primary antibodies used were rabbit anti-MAPK p38 (#8690, Ozyme; diluted 1/1000) and mouse anti-phospho-p38 (#9216, Ozyme; diluted 1/400). Antibody detection was performed using the corresponding anti-mouse/rabbit-IgG-HRP conjugated secondary antibody (Thermo Fisher Scientific) and chemiluminescence detection reagent ECL (Thermo Fisher Scientific). Quantification of immunoblots to calculate P-p38/p38 ratio was performed using ImageJ software.

For phosphorylated tau immunoblotting, previously snap-frozen mouse hippocampi were homogenized in 200 μl Tris buffer (pH 7.4) containing 10% sucrose and protease inhibitors (Complete; Roche Diagnostics) and sonicated. Homogenates were kept at -80°C until use. Protein concentrations of the samples were quantified using the BCA assay (Pierce), diluted in lithium dodecyl sulphate buffer supplemented with reducing agents (Invitrogen) and then separated on 4-12% Criterion Bis-Tris Gels (Invitrogen). Proteins were transferred to nitrocellulose membranes, which were then saturated with 5% non-fat dried milk in TNT (Tris 15mM pH 8, NaCl 140mM, 0.05% Tween) and incubated at 4°C for 24h or 48h with the primary antibodies. Rabbit polyclonal anti-phospho-Tau (Ser199) (home-made, 24h incubation) was diluted at 1/2000. Rabbit polyclonal anti-phospho-tau (Ser396) (Invitrogen #44-752G, 48h incubation) was diluted at 1/10000. Appropriate HRP-conjugated secondary antibodies were incubated for 45 min at RT. Signals were visualized using chemiluminescence kits ECL (#RPN2106; Amersham) or ECL Prime (#RPN2232; Amersham) and an Amersham ImageQuant 800 imaging system. Quantifications were performed using ImageJ software and results normalized to β-actin.

Electrophysiology in hippocampal slices

For ex vivo electrophysiology recordings, 6-10 months old males and females AETA-m and WT littermates were used. Mice were culled by cervical dislocation. Hippocampi were dissected and sliced (250 μm for patch-clamp recordings, 350 μm for field recordings) on a vibratome (Microm HM600V, Thermo Scientific, France).

Different external solutions for slicing, recovery and recording were used in this study as follows. For most LTP recordings (except experiment in Figure S4c-d), after dissection, hippocampi were incubated for 5 min and then sliced in ice-cold oxygenated (95 % O₂/ 5 % CO₂) cutting solution (in mM): 234 sucrose, 2.5 KCl, 1.25 NaH2PO4, 10 MgSO4, 0.5 CaCl2, 26 NaHCO3, 11 glucose (pH 7.4). For recovery (1h at 37°C) and recordings (at 27-29°C), slices were then incubated in artificial cerebral spinal fluid (aCSF; in mM): 119 NaCl, 2.5 KCl, 1.25 NaH2PO4, 26 NaHCO3, 1.3 MgSO4, 2.5 CaCl2 and 11 D-glucose, oxygenated with 95 % O2 and 5 % CO2, pH 7.4 for 1h at 37 ± 1°C. For field LTD recordings, after dissection, hippocampi were incubated for 5 min, sliced in ice-cold oxygenated (95 % O₂/ 5 % CO₂) and then incubated for recovery (1h at 37°C) in the same condition as described above for LTP. To optimize LTD recordings, slices were recorded (at 27-29°C) in another aCSF (in mM): 119 NaCl, 2.5 KCl, 1.25 NaH2PO4, 26 NaHCO3, 2 MgSO4, 4 CaCl2 and 11 D-glucose, oxygenated with 95 % O2 and 5 % CO2, pH 7.4 for 1 h at 37 ± 1 °C, with picrotoxin (50 M; Sigma-Aldrich, France) supplementation to block GABA_A receptors. For experiment in Figure S4c-d, after dissection, hippocampi were incubated for 5 min and then sliced in ice-cold oxygenated (95 % O₂/ 5 % CO₂) cutting solution (in mM): 206 sucrose, 2.8 KCl, 1.25 NaH2PO4, 2 MgSO4, 1 CaCl2, 26 NaHCO3, 0.4 sodium ascorbate, 10 glucose (pH 7.4). For recovery (1h at 37°C) and recordings (at 27-29°C), slices were then incubated in aCSF (in mM): 124 NaCl, 2.8 KCl, 1.25 NaH2PO4, 26 NaHCO3, 2 MgSO4, 3,6 CaCl2, 0.4 sodium ascorbate and 10 D-glucose, oxygenated with 95 % O₂ and 5 % CO₂, pH 7.4 for 1h at 37±1°C.

For all whole-cell patch-clamp recordings, slicing, recovery and recordings were made in same aCSF as for most LTP experiments described above. Recordings were performed at 30-31°C. The recording aCSF was supplemented with picrotoxin (50 mM; Sigma-Aldrich, France). For NMDAR spontaneous excitatory post-synaptic current (sEPSC) recordings, the aCSF was also supplemented with NBQX (10 μM; Tocris, England) to block AMPAR activity. For AMPAR sEPSC recordings, the aCSF was also supplemented with APV (50 μM; Tocris, England) to block NMDAR activity.

All recordings were made in a recording chamber on an upright microscope with IR-DIC illumination (SliceScope, Scientifica Ltd, UK) using a Multiclamp 700B amplifier (Molecular Devices, San Jose, CA, USA), under the control of pClamp10 software (Molecular Devices, San Jose, CA, USA). Data analysis was executed using Clampfit 10 software. Field excitatory post-synaptic potentials (fEPSPs) were recorded in the stratum radiatum of the CA1 region (using a glass electrode filled with 1 M NaCl and 10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), pH 7.4) and the stimuli were delivered at 0.1 Hz to the Schaffer collateral pathway by a monopolar glass electrode filled with aCSF. fEPSP response was set to approximately 30% of the maximal fEPSP response i.e. approx. 0.2–0.3 mV, with stimulation intensity 10 μA ± 5 μA delivered via a stimulation box (ISO-Flex, A.M.P.I. Inc., Israel). A stable baseline of 20 min was first obtained before induction for long-term plasticity recordings. The time courses were obtained by normalizing each experiment to the average value of all points constituting a 20 min stable baseline before induction. fEPSP magnitude was measured during the last 15 min of recording (45–60 min after induction) and calculated as % change fEPSP slope from baseline average.

For patch clamp recordings, after obtaining a tight seal (>1GΩ) on the cell body of the selected neuron, whole-cell patch clamp configuration was established, and cells were left to stabilize for 2-3 min before recordings began. Holding current and series resistance were continuously monitored throughout the experiment, and if either of these two parameters varied by more than 20%, the cell was discarded.

For paired-pulse ratios (PPR), EPSCs were obtained and two stimuli were delivered at 100, 200, 300 and 400 ms inter-stimulus interval (ISI). PPR was calculated as EPSC2 slope/EPSC1 slope (10 sweeps average per ISI). Recordings of WT and AETA-m mice were interleaved between days. Internal solution was a CS-gluconate solution: 117.5 mM Cs-gluconate, 15.5 mM CsCl, 10 mM TEACl, 8 mM NaCl, 10 HEPES, 0.25 mM EGTA, 4 mM MgATP and 0.3 NaGTP (pH 7.3; osmolarity 290-300 mOsm).

AMPAR sEPSCs were recorded at -65 mV using the same CS-gluconate internal solution detailed above. NMDAR sEPSCs were recorded at -65mV using a Cesium-methanesulfonate internal solution (mM): 143 Cesium-methanesulfonate, 5 NaCl, 1 MgCl2, 1 EGTA, 0.3 CaCl2, 10 Hepes, 2 Na2ATP, 0.3 NaGTP and 0.2 cAMP (pH 7.3 and 290-295 mOsm). sEPSCs were recorded in gap-free mode for 5 min for individual neurons of each genotype (WT and AETA-m). The Clampfit 10.6 software (Axon instruments) was used for analysis of sEPSCs to compare frequencies and amplitudes. sEPSCs were detected manually by following criteria of peaks with a threshold 2xSD of baseline noise level and a faster rise time than decay time. Analyses were performed blind to experimental condition. In the NMDAR sEPSC condition, we had previously ensured that the recorded currents were NMDAR current as they were absent in presence of APV [11].

Golgi Cox staining for spine density analysis

Brains were rapidly collected and impregnated in a Golgi-Cox solution (1% potassium dichromate, 1% mercuric chloride, 0.8% potassium chromate) for 3 weeks at room temperature according to manufacturer instruction (FD rapid GolgiStain kit, FD Neurotechnologies, USA). Brains were sectioned coronally (80 µm) using a vibratome, and stained and mounted according to protocol. Images were acquired under white light on a DMD108 Leica microscope (60X magnification). Spine density (number of spines per 1 µm length) in selected segments (minimum of 30 per mouse) of secondary dendrites of CA1 pyramidal neurons in stratum radiatum was estimated using Image J and manual count by an experimenter blinded to mouse genotype.

Morris Water Maze

AETA-m (6-10 months old) were submitted to the Morris water maze task in either a 90 cm or a 150 cm pool. The apparatus consisted in a circular tank (Ø 90 cm or 150 cm) filled with water (temperature 25 ± 1 °C) made opaque with the addition of white opacifier (Viewpoint, France). For the 90 cm pool protocol, the test consisted in 3 phases: (1) Cue task training (2 days), (2) Spatial learning training (7 days) and (3) Long-term reference memory Probe test (at 24h and at 7 days). An escape platform (Ø 8 cm) was submerged 1 cm below the water surface for the cue task training and the spatial learning training. The animals performed 4 trials/day with a maximum trial duration of 90s (+30s on the platform at the end of each trial) and with an inter-trial interval of 10 min. During the cue task, a visible flag was placed on the top of the submerged platform and the tank was surrounded by opaque curtains. The mice were allowed to find the visible platform. The platform emplacement was changed for the second day of the cue task (Day 2). For the spatial learning training phase, extra-maze cues were placed on the walls surrounding the maze to allow the animals to create a spatial map and find the submerged platform. A new platform placement was designed and kept in the same position for the 7 training days. Probe test was run to assess the strength of the memory for the platform location: the platform was removed, and the test mouse was allowed to search for it for 90s. For the 150 cm pool protocol, the test consisted in 3 phases: (1) Cue Task training (1 day), (2) Spatial Learning training (8 days) and (3) long reference memory Probe test (at 24h and at 7 days (for males only)). Similarly, an escape platform (Ø 8 cm) was submerged 1 cm below the water surface for the cue task training and the spatial learning training. During Cue Task training, a visible flag was placed on top of the submerged platform and the tank was surrounded by opaque curtains. Mice were allowed to find the visible platform. The mice performed the test along 2 sessions (T1 and T2) of 3 trials each with a maximum trial duration of 90s (+30s on the platform at the end of each trial) and with an inter-trial interval of 10 min. For the Spatial Learning training phase, extra-maze cues were placed on the walls surrounding the maze to allow the animals to create a spatial map to find the submerged platform and the tank was surrounded by white and patterned curtains. A new platform placement was designed and kept in the same position for 8 training days until the WT group reached the performance criteria. The probe test was run at 24h and 7 days (for males only) to assess memory strength at which time the platform was removed, and the test mouse was allowed to search for it for 60s. For both protocols, the training phase was stopped once the WT mice exhibited an average latency that was statistically different from the first day of training and not statistically different from 20s. All the trials were video recorded and tracked using ANYmaze software (Stoelting, Wood Dale, USA) to analyze escape latency.

Statistical analysis

Results are shown as mean ± s.e.m. Numbers and their correspondence are given in each figure. Statistical analysis was performed with GraphPad Prism 9 software (Dotmatics). Statistical analyses are described in brief in figure legends and are presented in detail in Supplemental Tables S7-S15 (one table per figure). Statistical significance was set at p < 0.05; *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Use of AI and AI-assisted technologies

ChatGPT (OpenAI, USA) was used in order to improve readability and language. After using this tool, the lead author reviewed and edited the content as needed and takes full responsibility for the content of the publication.

Availability of data and materials

Raw and processed RNA-seq data will be deposited on in the Gene Expression Omnibus.

AETA levels are increased in AD patient brain.

AETA levels were quantified in hippocampi and prefrontal cortices (PFC) of AD patient brains (Braak stage V/VI) against age-matched control subjects (Supplemental Table S1 and S2) using protocols and antibodies largely characterized in mouse and human samples as described previously [49]. We extracted the soluble protein fraction from these brains (DEA fraction). Control and AD brain DEA fractions displayed preserved protein integrity as exemplified by GAPDH protein content (Supplemental Figure S1a). We observed a significant increase in AETA in both hippocampi and PFC of AD brains as compared to controls (Figure 1b; Supplemental Figure S1b). As expected, AETA was represented as a multiband signal [49]. This increase in AETA levels was not due to changes in protein or mRNA expression of APP (Figure 1c; Supplemental Figure S1c). We also compared AETA levels between both genders in hippocampi in either control subjects or AD patients. While AETA levels were similar in hippocampi of male and female control subjects (Figure1 Supplemental Sd), they were increased in AD female when compared to AD males (Figure 1d). Together, these data show that AETA is increased in brains of human AD patients, with a stronger increase in women.

Chronic increase in AETA brain levels drives human AD-like molecular signature of synapse dysfunction.

Considering this increase in AETA levels in brains of AD patients, we asked if a chronic increase in brain AETA levels could contribute to AD pathology. To address this question, we used a new mouse model, named AETA-m, which harbors a transgene allowing expression of a secreted form of human AETA (long alpha form) under the Thy1.2 neuronal promoter (Figure 2a). In this model, human AETA expression was observed at similar levels between 1 and 12 months of age, albeit with some inter-individual differences. Notably, no accumulation of human AETA was evident with age in this model (Figure 2b). We quantified the average amount of human AETA being expressed in hippocampi of these mice by comparing to known amount of synthetic human AETA peptide (Supplemental Figure S2a). Human AETA concentration was calculated as approximately 0.27 ng/mg of protein. We also analyzed this new mouse model for other parameters to check for general health. Body weights at weaning and in adult AETA-m mice were normal (Supplemental Figure S2b-c). The general cytoarchitecture of the hippocampus was normal as measured by the thickness of the CA1 cell body layer in hippocampi slices stained with cresyl violet (Supplemental Figure S2d), suggesting normal neuron density. Human AETA expression did not alter levels of endogenous mouse AETA or endogenous Ab_1-40, as published in a preliminary analysis of hippocampal protein content from these AETA-m mice [11]. We further analyzed the levels of endogenous full-length APP and other peptides associated to APP processing (CTFη and sAPPα/β). Levels of these endogenous peptides were unchanged (Supplemental Figure S2e) showing that increasing AETA levels does not alter endogenous APP processing in these mice. We also tested the neuroinflammatory status (levels of mRNA of Trem2, C1qa, ccl4 and Vim genes, Supplemental Table S3) and Tau phosphorylation in hippocampi of these mice to seek for these two hallmarks of AD pathology [1, 17, 25]. We could not detect any evidence of neuroinflammation or tau hyperphosphorylation (Supplemental Figure S3).

To broadly address the molecular changes promoted by the production of AETA in the hippocampus, we performed RNA sequencing of hippocampi from adult AETA-m mice and WT littermates. We identified 567 differentially expressed genes between both conditions (Supplemental Table S4 and S5). Within this dataset, we focused our attention to a major hallmark of AD that is neuronal dysfunction, notably of synaptic pathways linked to NMDA receptor activity [20, 22, 38]. Functional annotation of this molecular signature with IPA revealed modifications in diseases & molecular functions, signaling pathways, and upstream regulators tightly linked to neuron function (Figure 2c, Supplemental Table S6), with all these pathways showing mild but significant downregulated expression in AETA-m mice, except for ‘ion channel transport’ which was selectively upregulated. Notably, most genes associated to ‘SNARE signaling pathway’ and ‘Assembly and cell surface presentation of NMDA receptors’ with IPA, which are particularly relevant to synapse function, were significantly down-regulated in hippocampi of AETA-m mice (Figure 2d).

Next, we sought to determine if this molecular synaptic signature was pertinent to human AD pathology, particularly the hippocampus and para-hippocampal regions, which are most vulnerable in AD. We compared the synaptic transcriptional signature obtained in AETA-m hippocampi (Figure 2d) with publicly available transcriptomic data of Van Rooij et al. [36] where they compared bulk RNAseq data of hippocampi of 20 AD cases and 10 age- and sex- matched cognitively healthy controls (NBB-HPC-AD). We observed a strong correlation with most mRNA being up and down-regulated in both set of data (Figure 2e). Indeed, 11 out of the above-mentioned 13 genes associated to the to ‘SNARE signaling pathway’ and ‘Assembly and cell surface presentation of NMDA receptors’ pathways also showed significant expression differences compared with control brains and displayed identical up (two genes) or down (nine genes) regulation (Figure 2e), the other two genes (SYN2 and NEFL) not being reported in the human RNAseq data analysis. Using publicly available transcriptomic data of the Mount Sinai AD cohort (MSBB) [46], we also compared the synaptic signature obtained in AETA-m mice with data bulk RNAseq data obtained from the human Brodmann area 36 (BM36). This parahippocampal area is highly relevant to formation, consolidation and retrieval of hippocampus-linked memories [8] and particularly vulnerable, even amongst temporal lobe regions, in human AD brains showing the most profound molecular alterations as previously evidenced with microarray profiling [47]. A molecular signature similar to that observed in AETA-m hippocampi was also evident in this region. We also compared the AETA-m molecular signature to another temporal lobe area of the MSBB cohort (Brodmann area 22 (BM22)), which was defined as a region less vulnerable than BM26 in AD [47]. This molecular signature was slightly less pronounced, although 9 out of 13 genes exhibited a similar pattern (Figure 2e). We also tested if this molecular signature could be evidenced in another publicly available dataset from a different cohort, The Religious Order Study and Memory and Aging Project (ROSMAP) dataset, where the transcriptome profiling was performed on the dorsolateral prefrontal cortex of control and AD post-mortem tissue [9]. This brain region is highly implicated in the control of executive functions, including working memory and cognitive flexibility, both of which are impaired during AD progression [19]. The molecular signature was less obvious in this dataset, but still 7 out of the 13 genes displayed a similar pattern to the molecular signature of AETA-m hippocampi (Figure 2e). To confirm a specific link of this molecular signature to AD, we also compared our mouse data to a publicly available dataset from frontal cortex samples of Parkinsonian human patients (NBB-PD) [7]. In this dataset, we used the differential profiling observed between Braak Lewis Body stages 0 (no pathology) and 5 (moderate Lewis Body pathology in frontal cortex). This molecular signature was completely absent in this PD dataset (Figure 2e). As the NBB-HPC-AD dataset exhibited the strongest similarity to the AETA-m synaptic signature, we extended the comparison to all genes identified as significantly differentially expressed in the AETA-m model (567 genes on which 540 have a known human ortholog, Supplemental Tables S4 and S5). We extended the comparison to all genes identified as significantly differentially expressed in the AETA-m model (567 genes on which 540 have a known human ortholog, Supplemental Tables S4 and S5). By correlating the log2FC obtained in both analyses, we obtained a Pearson’s correlation coefficient of 0.52. By selecting only the genes contributing to the neuron-relevant enriched pathways highlighted in Figure 2c and in Supplemental Table S6, this correlation reached a coefficient of 0.65 (Figure 2f). Together, these data demonstrate that the neuronal molecular signature, and particularly the synaptic signature, observed in hippocampi of AETA-m mice is reminiscent of human AD pathology with the strongest correlation observed in memory-encoding regions, which are highly vulnerable in AD.

Increased brain AETA levels modify NMDA receptor-linked signaling.

We next tested if, in AETA-m mice, these molecular alterations correlated with hippocampal synapse dysfunction by analyzing NMDAR-dependent long-term potentiation (LTP) and long-term depression (LTD) at the CA3-CA1 synapse in acutely dissected hippocampal slices of adult (5-10 months) AETA-m and WT littermates. We observed that LTP was reduced in both female and male AETA-m mice (Figure 3a and Supplemental Figure S4a). We observed enhanced LTD at this same synapse in male AETA-m mice (Figure 3b). LTD in WT female mice could not be induced at this age using this NMDAR-dependent LTD induction protocol (data not shown). Plasticity deficits were postsynaptic as presynaptic plasticity, measured by the paired-pulse ratio, was normal at this synapse (Supplemental Figure S4b). We tested both AMPA receptor (AMPAR) and NMDAR signaling, the two receptors required for these plasticity mechanisms [21], by recording spontaneous excitatory synaptic currents (sEPSC) in CA1 pyramidal neurons. AMPAR signaling, as measured by AMPAR sEPSC frequency and amplitude, was not affected (Figure 3c), while NMDAR sEPSC frequency, but not amplitude, was reduced (Figure 3d). Lower LTP and enhanced LTD could be explained by compromised NMDAR signaling, possibly due to inhibition of calcium entry via NMDARs in presence of AETA [11]. To test this possibility, we tested LTP again at this synapse using a different extracellular medium containing higher calcium [49]. We observed that, in this condition, LTP was normal (Supplemental Figure S4c-d). This suggests that, although NMDAR-dependent signaling is compromised, it might be due to over-activation of AETA-linked signaling that shunts calcium entry via partial inhibition of NMDAR ionotropic activity [11]. To test this further, we examined p38 MAP kinase activation in hippocampi of these mice. This activation was shown to be important for LTD and spine shrinkage via the unconventional non-ionotropic function of NMDARs and favored in presence of AETA [11, 29, 31]. P38 phosphorylation was enhanced in AETA-m hippocampi (Figure 3e). This was linked to a reduction of spine density in these neurons (Figure 3f and Supplemental Figure S4e). Together these data suggest that, upon chronic increase of AETA levels, NMDAR signaling is particularly compromised favoring synapse weakening both functionally and structurally.

AETA-m mice exhibit memory decay reminiscent of prodromal AD

As hippocampi of AETA-m mice display and AD-like molecular signature of synapse dysfunction and NMDAR associated synapse weakening, we would expect hippocampus-dependent memory alterations in AETA-m mice. We thus submitted these mice to the Morris water maze task. Mice were first tested in a small Morris water maze pool (90 cm diameter). Two days of cue task was performed to ensure that AETA-m mice exhibited normal locomotor activity and could see visual cues adequately (Figure 4a, Supplemental Figure S5a). Mice were trained until WT mice reached the criterion of less than 20s delay to reach the platform. AETA-m mice exhibited normal learning (Figure 4a, Supplemental Figure S5a). After a 24h delay, memory retrieval was measured as time spent in target quadrant. Memory retrieval was normal in AETA-m mice (Figure 4b, Supplemental Figure S5b). To test for persistence of memory consolidation, we submitted these mice to a second probe test 7 days later. No significant difference was observed between the two genotypes, although, in AETA-m females, the 7-day memory was slightly weaker than WT females when compared to chance level (25%) (Figure 4c and Supplemental Figure S5c). We also analyzed the accuracy of this memory measured as time spent in platform zone and number of platform crossings. While WT mice conserved a stable ability to locate this position between 24h and 7 days, AETA-m mice, most notably female AETA-m mice, exhibited a decay of this accuracy (Figure 4d-e and Supplemental Figure S5d-e). We also challenged AETA-m mice in a larger pool (150 cm), more demanding in terms of memory accuracy, asking if we could exacerbate this memory phenotype. Again, during this new protocol, AETA-m mice performed the cue task correctly and training of AETA-m mice was normal (Figure 4f and Supplemental Figure S5f). After a 24h delay WT and AETA-m females correctly located the target quadrant (their scores were significantly above chance level), but performance of AETA-m females was significantly lower compared to WT females in this larger pool (Figure 4g). Male mice performed as well as WT mice, even for the 7 days probe test (Supplemental Figure S5g-j). These results indicate that female AETA-m mice exhibit weaker consolidation of long term memory, suggesting prodromal human AD-like cognitive deficits [28, 45].

Ab has been long suspected to be the main culprit of synapse dysfunction in AD. There is strong evidence supporting this claim, although the exact mechanism by which it impairs synapse signaling remains unclear. Yet, an increasing number of studies suggest that other amyloid-independent mechanisms could also lead to synapse dysfunction in AD, beyond Ab actions [4, 10, 18, 33, 42] . Our study provides the first evidence that AETA could represent an important additional factor of this AD hallmark. Here, we show that AETA levels are elevated in the hippocampus and PFC of in AD patients, with sex-dimorphism characterized by a stronger increase in AD female compared to AD male. Increasing AETA levels in the mouse brain leads to hippocampal transcriptomic alterations, particularly in genes associated with synaptic functions, which closely resemble alterations observed in human brain regions vulnerable to AD, especially those related to memory processes. These transcriptomic changes correlate with NMDAR-mediated excitatory synapse weakening and to hippocampus-dependent memory deficits, particularly in females.

Previous work on AETA’s precursor CTF-h suggested that AETA might be increased in human AD patients because accumulation of CTF-h was reported in dystrophic neurites [49]. Here we show significant accumulation of AETA in the hippocampus and PFC, both memory-linked regions of the brain. This increase seems to be due to post-transcriptional mechanisms since APP protein and mRNA were not altered in AD brains. As AETA was just identified as a physiological modulator of NMDARs with strong consequences on NMDAR function [11], we would expect fine regulation of its availability with efficient clearance around synapses. Increased processing or reduced clearance of AETA could be at play. Interestingly, we recently reported that AETA production is increased upon enhanced neuronal activity[11]. As AD brains are known to exhibit hyperexcitability and even epileptic activity, especially during the early phase of the disease [40, 44], this could contribute to enhanced AETA levels. Since AETA was shown to dampen synaptic activity [11, 49], this activity-dependent increase in AETA levels might represent a physiologically-encoded control of network excitability that becomes over-solicited in the AD brain leading to abnormal synapse weakening. Why female AD brains exhibit more AETA than male AD brains has also to be further addressed, but a more complete understanding of this could help explain the prevalence of AD in women.

Our first analysis of the new AETA-m mouse model provides several important insights into the contribution of AETA in AD etiology. First, we did not observe any aggregation of AETA even following chronic over-expression. This could partly explain why this APP fragment was not easily detected in human AD brains remaining under the radar of neuropathologists for so long. Second, increasing its brain levels does not lead to clear neuroinflammation nor tau hyperphosphorylation, suggesting that AETA does not contribute to the rise of these molecular hallmarks typical of late AD. Third, a neuron-centric analysis of RNA-seq data from AETA-m hippocampi, compared with human RNA-seq datasets, reveals that elevated AETA expression alters the hippocampal transcriptome, particularly affecting synaptic gene expression, resulting in a molecular profile that closely resembles that observed in human AD. As expected in a mouse model without overt neuroinflammation nor tau pathology, these molecular modifications were subtle but sufficient to provoke synapse dysfunction and could thus reflect early AD-like synaptopathy. Indeed, functionally, this translates to synapse weakening (less LTP, more LTD and spine loss), phenotypes also observed in human AD brains and in animal models of AD [38, 43]. Of note, in AD mouse models overexpressing APP, there will be de facto more AETA, as shown for the APPPS1 line [49], which might contribute to NMDAR-dependent synapse weakening observed in these models [23].

In our RNA-seq analysis, most genes that were down-regulated belong to pathways relevant for synapse function, consistent with our functional assessment of synapse weakening. Yet, we also observed a strong correlation of up-regulated genes when comparing the AETA-m and MSBB BM36 RNA-seq datasets. We could not find evidence of previous association of some of these up-regulated genes (FZD5, PLXNB2, CELSR1, FLNB) with AD pathology. However, there is previous evidence of the implication of the other genes (LEPR, ANO6, SGK3, FNBP1L, ATP11A and ADAM17) and their related signaling pathways in AD [6, 12, 13, 32, 34, 41, 48]. Although beyond the scope of the current study, analyzing further these upregulated genes, as well as the RNA-seq data beyond the neuronal signature, may yield additional insights into AETA-dependent brain perturbations. Fourth, the synaptic phenotypes observed in AETA-m mice upon chronic expression of AETA are similar to the acute action of AETA, most prominently affecting NMDAR signaling [11]. This latter observation provides key information. On one hand, it suggests that AETA does not out pass its primary physiological function at synapses, which is to weaken synapses via modulation of NMDAR ionotropic and ion flux-independent activities, although it is constantly available in brains of AETA-m mice. On the other hand, as we recently classified AETA as an endogenous modulator of NMDARs [11], it is likely to be dynamically regulated with specific production and clearance mechanisms. There is strong evidence that NMDAR dysfunction is a hallmark of AD [20, 22] and should be further explored as a therapeutic target in AD beyond the current FDA-approved treatment with the NMDAR antagonist memantine [14]. Once better identified, exploiting this endogenous regulation if AETA production and clearance or directly interfering with AETA’s action on NMDARs could represent innovative therapeutic strategies to normalize synapse function in AD patient brains.

Finally, AETA-m females exhibited weaker hippocampus-dependent memory retrieval compared to WT. This deficit emerged when the delay of retention was increased (24h vs 7days) or when the task was more demanding (small pool vs larger pool). These memory deficits may reflect consolidation deficits reminiscent of prodromal AD [28, 45]. Surprisingly, while male AETA-m mice exhibited similar synapse weakening, they did not display significant memory deficits in these tasks. Sex-specific memory deficits, prevailing in females, have been observed in several mouse models of AD [5, 15, 30, 37, 50]. Explaining this sexual dimorphism, which currently remains enigmatic, might help provide insights as to why AD is more prevalent in women.

Our study provides the first compelling evidence that AETA contributes to synaptic dysfunction in AD both at the transcriptional and functional levels, particularly modifying NMDAR signaling. Our findings align with and expand upon emerging views that synaptic failure in AD arises from multiple, potentially interacting molecular pathways beyond Aβ. Targeting AETA production, clearance or its interaction with NMDARs may represent a promising therapeutic strategy to restore synaptic function in prodromal AD [39], before irreversible neurodegeneration occurs.

Author Contributions: JD performed immunoblotting, field and patch-clamp electrophysiology, behavior, mouse breeding and genotyping. CG performed immunoblotting, immunohistochemistry, Golgi Cox staining and analysis. SM, SMY and HM performed field electrophysiology. SM and HM performed patch camp experiments. AL, DB and LB performed qPCR on mouse tissue and tau immunoblotting. BA performed behavior, mouse breeding and genotyping. GP performed Golgi Cox staining and analysis, behavior and qPCR on human tissue. MT, KL, HL, HM and BM performed bioinformatics analysis of RNA-seq. PP and HM supervised and analyzed patch-clamp electrophysiology work. IB and HM supervised and analyzed behavior work. MW characterized AETA-m model and supervised human and mouse immunoblotting. HM piloted all aspects of study and manuscript writing with JD and contribution of other authors.

Disclosure of potential conflicts of interest: The authors report no biomedical financial interests or potential conflicts of interest.

Acknowledgments

We thank the Neuro-CEB Neuropathology network, which includes: Dr Franck Letournel (CHU Angers), Dr Marie-Laure Martin-Négrier (CHU Bordeaux), Dr Maxime Faisant (CHU Caen), Pr Catherine Godfraind (CHU Clermont-Ferrand), Pr Claude-Alain Maurage (CHU Lille), Dr Vincent Deramecourt (CHU Lille), Dr Mathilde Duchesne (CHU Limoges), Dr David Meyronnet (CHU Lyon), Dr Clémence Deltei (CHU Marseille), Pr Valérie Rigau (CHU Montpellier), Dr Fanny Vandenbos-Burel (Nice), Pr Danielle Seilhean (CHU PS, Paris), Dr Susana Boluda (CHU PS, Paris), Dr Isabelle Plu (CHU PS, Paris), Dr Dan Christian Chiforeanu (CHU Rennes), Dr Florent Marguet (CHU Rouen), Dr Béatrice Lannes (CHU Strasbourg), for providing the human PFC samples. We thank patient associations and research foundations that support the Neuro-CEB Biobank: ARSLA, CSC, France DFT, Fondation France SEP, Fondation Vaincre Alzheimer, France Parkinson, PSP France, for providing the human hippocampi samples. The authors also thank the technical support of the UCA GenomiX. The authors acknowledge the flow cytometry or microscopy facility from the « Institut de Pharmacologie Moléculaire et Cellulaire » part of the « Microscopie Imagerie Côte d’Azur » GIS IBiSA labeled platform. We thank Dr Camilla Giudici, German Center for Neurodegenerative Diseases (DZNE) Munich, Germany, for help with AETA-m mouse characterization. We thank Julien Fassy, IPMC, for help with qPCR of human tissues.

Funding sources

This work was supported by the Centre National de la Recherche Scientifique (CNRS); the French Association France Alzheimer (AAP SM 2018 Dossier 1795 and APP PFA 2024 Dossier 6479) to BA, IB and HM; the French Fondation Alzheimer (AAP 2015 AETAPHYS) to HM; the Agence Nationale de la Recherche (ANR-21-CE16-0032 APPYSYNAPSE) to HM; the Flag ERA JTC 2019 (MILEDI Project) to SM and HM; the French government, through the UCAJEDI Investments in the Future project managed by the National Research Agency (ANR) with the reference number ANR-15-IDEX-01 to KL and MT. JD was recipient of a PhD fellowship from the Ministère de la Recherche, de l’Enseignement Supérieur et de l’Innovation. AL was supported by PhD grants from Fondation pour la Recherche Médicale (ECO202106013670) and Vaincre Alzheimer (FR-24054T). UCA GenomiX is supported by the France 2030 framework managed by the Agence Nationale de la Recherche (4D-OMICs: ANR-21-ESRE-0052, National Infrastructure France Génomique: ANR-10-INBS-09-02, ANR-10-INBS-09-03).

Abey A, Davies D, Goldsbury C, Buckland M, Valenzuela M, Duncan T (2021) Distribution of tau hyperphosphorylation in canine dementia resembles early Alzheimer’s disease and other tauopathies. Brain Pathol 31:144–162. doi: 10.1111/bpa.12893
Andrä K, Abramowski D, Duke M, Probst A, Wiederhold KH, Bürki K, Goedert M, Sommer B, Staufenbiel M (1996) Expression of APP in transgenic mice: a comparison of neuron-specific promoters. Neurobiol Aging 17:183–190. doi: 10.1016/0197-4580(95)02066-7
Arvanitakis Z, Shah RC, Bennett DA (2019) Diagnosis and Management of Dementia: A Review. JAMA 322:1589–1599. doi: 10.1001/jama.2019.4782
Barthet G, Mulle C (2020) Presynaptic failure in Alzheimer’s disease. Prog Neurobiol 194:101801. doi: 10.1016/j.pneurobio.2020.101801
Britz J, Ojo E, Dhukhwa A, Saito T, Saido TC, Hascup ER, Hascup KN, Tischkau SA (2022) Assessing Sex-Specific Circadian, Metabolic, and Cognitive Phenotypes in the AβPP/PS1 and APPNL-F/NL-F Models of Alzheimer’s Disease. J Alzheimers Dis 85:1077–1093. doi: 10.3233/JAD-210629
Bruno F, Aceto MA, Paparazzo E, Arcuri D, Vozzo F, Mirante S, Greco BM, Serra Cassano T, Abondio P, Canterini S, Malvaso A, Grecucci A, Citrigno L, Geracitano S, Spadafora P, Puccio G, Frangipane F, Curcio SM, Ferrise F, Laganà V, Colao R, Passarino G, Bruni AC, Maletta R, Cavalcanti F, Montesanto A (2025) Genetic variability in ADAM17/TACE is associated with sporadic Alzheimer’s disease risk, neuropsychiatric symptoms and cognitive performance on the Rey Auditory Verbal Learning and Clock Drawing Tests. PLoS One 20:e0309631. doi: 10.1371/journal.pone.0309631
Cappelletti C, Henriksen SP, Geut H, Rozemuller AJM, van de Berg WDJ, Pihlstrøm L, Toft M (2023) Transcriptomic profiling of Parkinson’s disease brains reveals disease stage specific gene expression changes. Acta Neuropathol 146:227–244. doi: 10.1007/s00401-023-02597-7
Danieli K, Guyon A, Bethus I (2023) Episodic Memory formation: A review of complex Hippocampus input pathways. Progress in Neuro-Psychopharmacology and Biological Psychiatry 126:110757. doi: 10.1016/j.pnpbp.2023.110757
De Jager PL, Ma Y, McCabe C, Xu J, Vardarajan BN, Felsky D, Klein H-U, White CC, Peters MA, Lodgson B, Nejad P, Tang A, Mangravite LM, Yu L, Gaiteri C, Mostafavi S, Schneider JA, Bennett DA (2018) A multi-omic atlas of the human frontal cortex for aging and Alzheimer’s disease research. Sci Data 5:180142. doi: 10.1038/sdata.2018.142
De Strooper B, Karran E (2016) The Cellular Phase of Alzheimer’s Disease. Cell 164:603–615. doi: 10.1016/j.cell.2015.12.056
Dunot J, Moreno S, Gandin C, Pousinha PA, Amici M, Dupuis J, Anisimova M, Winschel A, Uriot M, Petshow SJ, Mensch M, Bethus I, Giudici C, Hampel H, Wefers B, Wurst W, Naumann R, Ashby MC, Laube B, Zito K, Mellor JR, Groc L, Willem M, Marie H (2024) APP fragment controls both ionotropic and non-ionotropic signaling of NMDA receptors. Neuron S0896-6273(24)00404–5. doi: 10.1016/j.neuron.2024.05.027
Florentinus-Mefailoski A, Bowden P, Scheltens P, Killestein J, Teunissen C, Marshall JG (2021) The plasma peptides of Alzheimer’s disease. Clin Proteomics 18:17. doi: 10.1186/s12014-021-09320-2
Flores-Cordero JA, Pérez-Pérez A, Jiménez-Cortegana C, Alba G, Flores-Barragán A, Sánchez-Margalet V (2022) Obesity as a Risk Factor for Dementia and Alzheimer’s Disease: The Role of Leptin. Int J Mol Sci 23:5202. doi: 10.3390/ijms23095202
Folch J, Busquets O, Ettcheto M, Sánchez-López E, Castro-Torres RD, Verdaguer E, Garcia ML, Olloquequi J, Casadesús G, Beas-Zarate C, Pelegri C, Vilaplana J, Auladell C, Camins A (2018) Memantine for the Treatment of Dementia: A Review on its Current and Future Applications. J Alzheimers Dis 62:1223–1240. doi: 10.3233/JAD-170672
Ganesh A, Choudhury W, Coutellier L (2024) Early spatial recognition memory deficits in 5XFAD female mice are associated with disruption of prefrontal parvalbumin neurons. Brain Res 1841:149122. doi: 10.1016/j.brainres.2024.149122
Griffiths J, Grant SGN (2023) Synapse pathology in Alzheimer’s disease. Seminars in Cell & Developmental Biology 139:13–23. doi: 10.1016/j.semcdb.2022.05.028
Heneka MT, van der Flier WM, Jessen F, Hoozemanns J, Thal DR, Boche D, Brosseron F, Teunissen C, Zetterberg H, Jacobs AH, Edison P, Ramirez A, Cruchaga C, Lambert J-C, Laza AR, Sanchez-Mut JV, Fischer A, Castro-Gomez S, Stein TD, Kleineidam L, Wagner M, Neher JJ, Cunningham C, Singhrao SK, Prinz M, Glass CK, Schlachetzki JCM, Butovsky O, Kleemann K, De Jaeger PL, Scheiblich H, Brown GC, Landreth G, Moutinho M, Grutzendler J, Gomez-Nicola D, McManus RM, Andreasson K, Ising C, Karabag D, Baker DJ, Liddelow SA, Verkhratsky A, Tansey M, Monsonego A, Aigner L, Dorothée G, Nave K-A, Simons M, Constantin G, Rosenzweig N, Pascual A, Petzold GC, Kipnis J, Venegas C, Colonna M, Walter J, Tenner AJ, O’Banion MK, Steinert JR, Feinstein DL, Sastre M, Bhaskar K, Hong S, Schafer DP, Golde T, Ransohoff RM, Morgan D, Breitner J, Mancuso R, Riechers S-P (2024) Neuroinflammation in Alzheimer disease. Nat Rev Immunol. doi: 10.1038/s41577-024-01104-7
Herrup K (2015) The case for rejecting the amyloid cascade hypothesis. Nat Neurosci 18:794–799. doi: 10.1038/nn.4017
Kumar S, Zomorrodi R, Ghazala Z, Goodman MS, Blumberger DM, Cheam A, Fischer C, Daskalakis ZJ, Mulsant BH, Pollock BG, Rajji TK (2017) Extent of Dorsolateral Prefrontal Cortex Plasticity and Its Association With Working Memory in Patients With Alzheimer Disease. JAMA Psychiatry 74:1266–1274. doi: 10.1001/jamapsychiatry.2017.3292
Liu W, Li Y, Zhao T, Gong M, Wang X, Zhang Y, Xu L, Li W, Li Y, Jia J (2023) The role of N-methyl-D-aspartate glutamate receptors in Alzheimer’s disease: From pathophysiology to therapeutic approaches. Prog Neurobiol 231:102534. doi: 10.1016/j.pneurobio.2023.102534
Malenka RC, Bear MF (2004) LTP and LTD: an embarrassment of riches. Neuron 44:5–21
Malinow R (2012) New developments on the role of NMDA receptors in Alzheimer’s disease. Curr Opin Neurobiol 22:559–563. doi: 10.1016/j.conb.2011.09.001
Marchetti C, Marie H (2011) Hippocampal synaptic plasticity in Alzheimer’s disease: what have we learned so far from transgenic models? RevNeurosci 22:373–402. doi: 10.1515/RNS.2011.035
Marques-Coelho D, Iohan L da CC, Melo de Farias AR, Flaig A, Brainbank Neuro–CEB Neuropathology Network, Lambert J-C, Costa MR (2021) Differential transcript usage unravels gene expression alterations in Alzheimer’s disease human brains. NPJ Aging Mech Dis 7:2. doi: 10.1038/s41514-020-00052-5
Maurage C-A, Sergeant N, Ruchoux M-M, Hauw J-J, Delacourte A (2003) Phosphorylated serine 199 of microtubule-associated protein tau is a neuronal epitope abundantly expressed in youth and an early marker of tau pathology. Acta Neuropathol 105:89–97. doi: 10.1007/s00401-002-0608-7
Mensch M, Dunot J, Yishan SM, Harris SS, Blistein A, Avdiu A, Pousinha PA, Giudici C, Busche MA, Jedlicka P, Willem M, Marie H (2021) Aη-α and Aη-β peptides impair LTP ex vivo within the low nanomolar range and impact neuronal activity in vivo. Alzheimers Res Ther 13:125. doi: 10.1186/s13195-021-00860-1
Moechars D, Dewachter I, Lorent K, Reversé D, Baekelandt V, Naidu A, Tesseur I, Spittaels K, Haute CV, Checler F, Godaux E, Cordell B, Van Leuven F (1999) Early phenotypic changes in transgenic mice that overexpress different mutants of amyloid precursor protein in brain. J Biol Chem 274:6483–6492. doi: 10.1074/jbc.274.10.6483
Moulin CJA, James ,Niamh, Freeman ,Jayne E., and Jones RW (2004) Deficient Acquisition and Consolidation: Intertrial Free Recall Performance in Alzheimer’s Disease and Mild Cognitive Impairment. Journal of Clinical and Experimental Neuropsychology 26:1–10. doi: 10.1076/jcen.26.1.1.23940
Nabavi S, Kessels HW, Alfonso S, Aow J, Fox R, Malinow R (2013) Metabotropic NMDA receptor function is required for NMDA receptor-dependent long-term depression. Proc Natl Acad Sci U S A 110:4027–4032. doi: 10.1073/pnas.1219454110
Navakkode S, Gaunt JR, Pavon MV, Bansal VA, Abraham RP, Chong YS, Ch’ng TH, Sajikumar S (2021) Sex-specific accelerated decay in time/activity-dependent plasticity and associative memory in an animal model of Alzheimer’s disease. Aging Cell 20:e13502. doi: 10.1111/acel.13502
Park DK, Stein IS, Zito K (2022) Ion flux-independent NMDA receptor signaling. Neuropharmacology 210:109019. doi: 10.1016/j.neuropharm.2022.109019
Prokopenko D, Morgan SL, Mullin K, Hofmann O, Chapman B, Kirchner R, Alzheimer’s Disease Neuroimaging Initiative (ADNI), Amberkar S, Wohlers I, Lange C, Hide W, Bertram L, Tanzi RE (2021) Whole-genome sequencing reveals new Alzheimer’s disease-associated rare variants in loci related to synaptic function and neuronal development. Alzheimers Dement 17:1509–1527. doi: 10.1002/alz.12319
Puzzo D (2024) Beyond Amyloid-β: Reevaluating Its Physiological Role to Find Safe Therapies Against Alzheimer’s Disease. J Alzheimers Dis 100:783–786. doi: 10.3233/JAD-240406
Qi W, Ying Y, Wu P, Dong N, Fu W, Liu Q, Ward N, Dong X, Zhao RC, Wang J (2024) Inhibition of miR-4763-3p expression activates the PI3K/mTOR/Bcl2 autophagy signaling pathway to ameliorate cognitive decline. Int J Biol Sci 20:5999–6017. doi: 10.7150/ijbs.103225
Querfurth HW, LaFerla FM (2010) Alzheimer’s disease. N Engl J Med 362:329–344. doi: 10.1056/NEJMra0909142
van Rooij JGJ, Meeter LHH, Melhem S, Nijholt DAT, Wong TH, Netherlands Brain Bank, Rozemuller A, Uitterlinden AG, van Meurs JG, van Swieten JC (2019) Hippocampal transcriptome profiling combined with protein-protein interaction analysis elucidates Alzheimer’s disease pathways and genes. Neurobiol Aging 74:225–233. doi: 10.1016/j.neurobiolaging.2018.10.023
Schmid S, Rammes G, Blobner M, Kellermann K, Bratke S, Fendl D, Kaichuan Z, Schneider G, Jungwirth B (2019) Cognitive decline in Tg2576 mice shows sex-specific differences and correlates with cerebral amyloid-beta. Behav Brain Res 359:408–417. doi: 10.1016/j.bbr.2018.11.022
Selkoe DJ (2002) Alzheimer’s disease is a synaptic failure. Science 298:789–791
Taddei RN, E Duff K (2025) Synapse vulnerability and resilience underlying Alzheimer’s disease. EBioMedicine 112:105557. doi: 10.1016/j.ebiom.2025.105557
Targa Dias Anastacio H, Matosin N, Ooi L (2022) Neuronal hyperexcitability in Alzheimer’s disease: what are the drivers behind this aberrant phenotype? Transl Psychiatry 12:257. doi: 10.1038/s41398-022-02024-7
Tian Y, Fopiano KA, Buncha V, Lang L, Suggs HA, Wang R, Rudic RD, Filosa JA, Bagi Z (2023) The role of ADAM17 in cerebrovascular and cognitive function in the APP/PS1 mouse model of Alzheimer’s disease. Front Mol Neurosci 16:1125932. doi: 10.3389/fnmol.2023.1125932
Tse K-H, Herrup K (2017) Re-imagining Alzheimer’s disease - the diminishing importance of amyloid and a glimpse of what lies ahead. J Neurochem 143:432–444. doi: 10.1111/jnc.14079
Tzioras M, McGeachan RI, Durrant CS, Spires-Jones TL (2023) Synaptic degeneration in Alzheimer disease. Nat Rev Neurol 19:19–38. doi: 10.1038/s41582-022-00749-z
Vossel KA, Tartaglia MC, Nygaard HB, Zeman AZ, Miller BL (2017) Epileptic activity in Alzheimer’s disease: causes and clinical relevance. Lancet Neurol 16:311–322. doi: 10.1016/S1474-4422(17)30044-3
Wagner M, Wolf S, Reischies FM, Daerr M, Wolfsgruber S, Jessen F, Popp J, Maier W, Hüll M, Frölich L, Hampel H, Perneczky R, Peters O, Jahn H, Luckhaus C, Gertz H-J, Schröder J, Pantel J, Lewczuk P, Kornhuber J, Wiltfang J (2012) Biomarker validation of a cued recall memory deficit in prodromal Alzheimer disease. Neurology 78:379–386. doi: 10.1212/WNL.0b013e318245f447
Wang M, Beckmann ND, Roussos P, Wang E, Zhou X, Wang Q, Ming C, Neff R, Ma W, Fullard JF, Hauberg ME, Bendl J, Peters MA, Logsdon B, Wang P, Mahajan M, Mangravite LM, Dammer EB, Duong DM, Lah JJ, Seyfried NT, Levey AI, Buxbaum JD, Ehrlich M, Gandy S, Katsel P, Haroutunian V, Schadt E, Zhang B (2018) The Mount Sinai cohort of large-scale genomic, transcriptomic and proteomic data in Alzheimer’s disease. Sci Data 5. doi: 10.1038/sdata.2018.185
Wang M, Roussos P, McKenzie A, Zhou X, Kajiwara Y, Brennand KJ, De Luca GC, Crary JF, Casaccia P, Buxbaum JD, Ehrlich M, Gandy S, Goate A, Katsel P, Schadt E, Haroutunian V, Zhang B (2016) Integrative network analysis of nineteen brain regions identifies molecular signatures and networks underlying selective regional vulnerability to Alzheimer’s disease. Genome Med 8:104. doi: 10.1186/s13073-016-0355-3
Wang Y, Li P, Liang Y, Wang D (2025) ANO6 Targets TMEM30A to Regulate Endoplasmic Reticulum Stress-Induced Lipid Peroxidation and Ferroptosis in Alzheimer’s Cells. Cell Biochem Biophys. doi: 10.1007/s12013-025-01748-9
Willem M, Tahirovic S, Busche MA, Ovsepian SV, Chafai M, Kootar S, Hornburg D, Evans LDB, Moore S, Daria A, Hampel H, Müller V, Giudici C, Nuscher B, Wenninger-Weinzierl A, Kremmer E, Heneka MT, Thal DR, Giedraitis V, Lannfelt L, Müller U, Livesey FJ, Meissner F, Herms J, Konnerth A, Marie H, Haass C (2015) η-Secretase processing of APP inhibits neuronal activity in the hippocampus. Nature 526:443–447. doi: 10.1038/nature14864
Yang J-T, Wang Z-J, Cai H-Y, Yuan L, Hu M-M, Wu M-N, Qi J-S (2018) Sex Differences in Neuropathology and Cognitive Behavior in APP/PS1/tau Triple-Transgenic Mouse Model of Alzheimer’s Disease. Neurosci Bull 34:736–746. doi: 10.1007/s12264-018-0268-9

No competing interests reported.

Download PDF

Reviews received at journal
16 Dec, 2025
Reviews received at journal
04 Dec, 2025
Reviewers agreed at journal
01 Dec, 2025
Reviewers agreed at journal
26 Nov, 2025
Reviewers agreed at journal
26 Nov, 2025
Reviewers invited by journal
25 Nov, 2025
Editor assigned by journal
25 Nov, 2025
Submission checks completed at journal
24 Nov, 2025
First submitted to journal
24 Nov, 2025

You are reading this latest preprint version

AETA peptide drives Alzheimer’s disease signature of synapse dysfunction

Status:

Version 1

Abstract

Figures

Introduction

Materials and Methods

Results

Discussion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1