Mitigation of synaptic and memory impairments via F-actin stabilization in Alzheimer’s disease

Abstract

Background

Synaptic dysfunction, characterized by synapse loss and structural alterations, emerges as a prominent correlate of cognitive decline in Alzheimer’s disease (AD). Actin cytoskeleton, which serves as the structural backbone of synaptic architecture, is observed to be lost from synapses in AD. Actin cytoskeleton loss compromises synaptic integrity, affecting glutamatergic receptor levels, neurotransmission, and synaptic strength. Understanding these molecular changes is crucial for developing interventions targeting synaptic dysfunction, potentially mitigating cognitive decline in AD.

Methods

In this study, we investigated the synaptic actin interactome using mass spectrometry in a mouse model of AD, APP/PS1. Our objective was to explore how alterations in synaptic actin dynamics, particularly the interaction between PSD-95 and actin, contribute to synaptic and cognitive impairment in AD. To assess the impact of restoring F-actin levels on synaptic and cognitive functions in APP/PS1 mice, we administered F-actin stabilizing agent, jasplakinolide. Behavioral deficits in the mice were evaluated using the contextual fear conditioning paradigm. We utilized primary neuronal cultures to study the synaptic levels of AMPA and NMDA receptors and the dynamics of PSD-95 actin association. Furthermore, we analyzed postmortem brain tissue samples from subjects with no cognitive impairment (NCI), mild cognitive impairment (MCI), and Alzheimer’s dementia (AD) to determine the association between PSD-95 and actin.

Results

We found a significant reduction in PSD-95-actin association in synaptosomes from middle-aged APP/PS1 mice compared to wild-type (WT) mice. Treatment with jasplakinolide, an actin stabilizer, reversed deficits in memory recall, restored PSD-95-actin association, and increased synaptic F-actin levels in APP/PS1 mice. Additionally, actin stabilization led to elevated synaptic levels of AMPA and NMDA receptors, enhanced dendritic spine density, suggesting improved neurotransmission and synaptic strength in primary cortical neurons from APP/PS1 mice. Furthermore, analysis of postmortem human tissue with NCI, MCI and AD subjects revealed disrupted PSD-95-actin interactions, underscoring the clinical relevance of our preclinical studies.

Conclusion

Our study elucidates disrupted PSD-95 actin interactions across different models, highlighting potential therapeutic targets for AD. Stabilizing F-actin restores synaptic integrity and ameliorates cognitive deficits in APP/PS1 mice, suggesting that targeting synaptic actin regulation could be a promising therapeutic strategy to mitigate cognitive decline in AD.

Introduction

Alzheimer’s disease (AD), the most common form of dementia, is characterized by the presence of two pathological (protein aggregates) hallmarks – extracellular deposition of amyloid β plaques and hyperphosphorylated tau protein [1]. AD causes a progressive decline in memory and other cognitive functions in elderly individuals, which is attributed to brain atrophy resulting from neuronal loss and synapse degeneration. Among all the pathological changes observed in AD brains, the loss of synapses most strongly correlates with cognitive decline [2,3,4]. Soluble Aβ and tau are identified as deleterious species that cause synaptic impairment and contribute to cognitive deficits in AD [5]. Studies on the post-mortem brain of AD patients and mouse models with amyloid plaque deposition suggest the accumulation of oligomeric Aβ within synapses [6]. In the early stages of AD, dendritic spines undergo gradual and progressive loss in number and morphology, indicating substantial alterations in synaptic integrity. Recent studies using cultured neurons and AD mouse models have demonstrated that F-actin is selectively lost from synapses as early as one month of age [7]. Soluble forms of Aβ from the AD brain inhibit long-term potentiation (LTP), and augments long-term depression (LTD), and the loss of dendritic spine density in the hippocampus of normal rodents [8, 9].

Dendritic spines are highly dynamic structures that undergo activity-dependent modifications to control synaptic function and cognitive processes such as learning and memory [10]. They are highly enriched with the actin cytoskeleton as branched or linear actin filaments and the dynamic equilibrium between globular and filamentous actin modulates the spine morphology and plasticity [11]. Maintaining the ratio of F-actin and G-actin is essential for actin cytoskeleton reorganization during activity dependent changes in dendritic structures. Actin dynamics during structural plasticity are regulated by postsynaptic actin binding proteins (ABPs), which include cofilin/actin depolymerizing factor (ADF), Aip1, α-actinin, CaMKII, Arp2/3, profilin and drebrin. In addition, ABPs involved in receptor trafficking and LTP maintenance are implicated in learning and memory processes [11]. Given the critical role of the actin cytoskeleton in synaptic plasticity, any perturbation in actin dynamics has the potential to exacerbate AD pathology, resulting in impaired synaptic neurotransmission. F-actin, the major constituent of the cytoskeleton, undergoes depolymerization to G-actin when exposed to amyloid beta (Aβ) and contributes to synapse loss [7]. In an AD mouse model, studies revealed that the depolymerization of synaptosomal F-actin occurs as early as one month of age, preceding the onset of pathological hallmarks [7]. Reduced synaptic F-actin levels are also noted in the postmortem brains of individuals with mild cognitive impairment and AD [12, 13]. Loss of spines mediated by F-actin depolymerization contributes directly to memory deficits and shows an inverse correlation with AD pathology [7]. In addition, dysregulation of several actin-binding proteins is linked to synaptic deficits associated with AD.

Actin cytoskeleton which serves as the structural backbone of the synaptic architecture, orchestrating various aspects of synaptic function, including structural integrity, vesicle trafficking, dendritic spine morphology, and synaptic plasticity [14]. Actin filaments provide the structural scaffold necessary for synaptic vesicle docking, release, and recycling, facilitating efficient neurotransmission. Actin filaments extend into the postsynaptic density (PSD), where they engage with several proteins [15], potentially including scaffolding proteins such as PSD-95, a membrane-associated guanylate kinase. Actin dynamics mediate the remodeling of the underlying scaffold, finely regulating the spatial rearrangement of glutamatergic receptors within PSDs [16]. Changes in actin dynamics elicit distinct responses in the localization of glutamate receptors, such as AMPA and NMDA receptors at the postsynaptic site. Latrunculin, an actin-depolymerizing agent, decreases GluA1–AMPA receptor clusters in the hippocampal neurons by 40%, whereas NMDA receptors remain clustered at synapses [17, 18]. In addition, the AMPA receptor content increases rapidly in response to LTP aligned with changes in F-actin and spine size, resulting in a sustained increase in synaptic strength [19]. Conversely, NMDAR-dependent LTD necessitates actin depolymerization, accompanied by a decrease in spine size and F-actin [20]. Over time, F-actin not only determines spine size but also influences PSD size and AMPAR content. However, the molecular association of F-actin at synapses remains unclear, posing a crucial question for comprehending synaptic strength and plasticity. To further elucidate the molecular association of actin at the synapse in AD, employing proteomic strategies to assess actin-interacting proteins can offer valuable insights. This approach has the potential to identify alterations in protein interactions in AD, providing a better understanding of the molecular dynamics involved. This research article advocates for the application of proteomic techniques to uncover novel insights into the molecular underpinnings of the role of actin in synaptic dysfunction within the context of AD.

We utilized the AD mouse model APP/PS1 to characterize the actin interactome in synaptosomes. APPswe/PS1ΔE9 mice express a chimeric mouse/human amyloid precursor protein (Mo/HuAPP695Swe) and a mutant human presenilin 1 (PS1ΔE9), linking them to early-onset AD. Our findings reveal several actin-binding proteins crucial for synaptic structure and function and implicated in the progression of AD pathology. Notably, PSD-95 was found to be associated with actin. However, we observed a significant decrease in PSD-95-actin association in middle-aged APP/PS1 mice compared to that of WT. To investigate the functional implications of this interaction, we administered an F-actin stabilizing agent via intrathecal injection, which effectively rescued fear memory deficits and restored the reduced association and synaptic F-actin levels in APP/PS1 mice. Primary cortical neurons from APP/PS1 mice displayed reduced levels of synaptic AMPA and NMDA receptors, as well as decreased F-actin levels and dendritic spine density, all of which were restored by F-actin stabilization. Fluorescence recovery after photobleaching (FRAP) experiments further supported the association between actin and PSD-95, as evidenced by the reduced fluorescence recovery of actin when coexpressed with PSD-95. Postmortem human tissue from subjects with no cognitive impairment (NCI), mild cognitive impairment (MCI), and AD revealed a disrupted PSD-95 actin interaction, highlighting its relevance to humans. Our findings highlight the pivotal role of PSD-95 as a crucial mediator facilitating crosstalk between glutamatergic receptors and F-actin. We concluded that alterations in the PSD-95-actin interaction significantly affect glutamatergic neurotransmission, synaptic strength, and plasticity in AD.

Materials and methods

Experimental design

The experimental design involved isolating synaptosomes from the brains of six-month-old WT and APP/PS1 mice for synaptic actin interactome analysis (Fig. 1a). This was followed by employing immunoprecipitation techniques to quantify PSD-95-actin association levels in one- and nine-month-old mice. In vivo experiments were conducted using age-matched wild-type and APP/PS1 mice. Additionally, postmortem brain tissue samples were obtained from no cognitive impairment (NCI), mild cognitive impairment (MCI), and Alzheimer’s dementia (AD) subjects from participants in the Religious Orders Study Project (ROS), for determining PSD-95-actin association. Behavioral assessments were conducted using contextual fear conditioning paradigms to assess memory recall and an actin stabilizing agent was introduced to APP/PS1 mice, evaluating its effects on memory recall, PSD-95-actin association, and synaptic F-actin levels. Synaptic receptor levels and colocalization studies were conducted using immunofluorescence staining and confocal microscopy in primary cortical neurons from APP/PS1 mice to assess AMPA and NMDA receptor levels, PSD-95 colocalization with actin and receptors. Fluorescence Recovery After Photobleaching (FRAP) assay was utilized to quantify the dynamics of PSD-95 and actin association in neuronal cultures. Data were analyzed using statistical methods such as ANOVA or t-tests to compare results between experimental groups, aiming to elucidate the role of altered PSD-95 actin association in synaptic dysfunction and cognitive decline in AD.

Fig. 1

Quantitative LC-MS/MS analysis of synaptic actin-interactome in WT and APP/PS1 mice. (a) Graphical illustration of the workflow for the characterization of synaptic actin interactome. (b) Venn diagram representing the overlap and unique proteins identified in the synaptosomes of wild-type (WT) and APP/PS1 mice (n = 3). The numbers within each section indicate the total proteins identified in WT and APP/PS1 mice, and the overlapping region represents proteins found in both groups. (c) Gene ontology (GO) analysis of functionally enriched pathways of proteins identified in both the groups

Reagents

Acti-stain 555 (Cat. No. PHDH1), acti-stain 670 (Cat. No. PHDN1) and G-Actin / F-actin In Vivo Assay Biochem Kit (Cat. No. BK037) were purchased from Cytoskeleton. Jasplakinolide (Cat. No. J7473) was purchased from Invitrogen. DNase I (Cat. No. D-4513) and Papain (Cat. No. P4762), Brilliant Blue G-Colloidal Concentrate (Cat. No. B2015) were sourced from Sigma-Aldrich. Poy-D-lysine (Cat. No. 150175) was from MP biomedicals, LLC. Dynabeads Protein G (Cat. No. 10004D), Dynabeads Protein A (Cat. No. 10002D), Neurobasal-A medium (Cat. No. 10888022), Penicillin-Streptomycin (Cat. No. 15140122), GlutaMAX (Cat. No. 35050061), B27-supplement (Cat. No. 175040444), Lipofectamine 2000 (Cat. No. 11668027) and Pierce BCA Protein Assay kits (Cat. No. 23227) were purchased from Thermo Fisher Scientific Inc. TGX Stain-Free Fast Cast Acrylamide Kit (Cat. No. 1610185), Precision Plus Protein Dual Color Standards (Cat. No. 1610394), Clarity Western ECL Substrate (Cat. No. 1705061), Clarity Max Western ECL Substrate, (Cat. No. 1705062) were purchased from Bio-Rad Laboratories, Inc. Q5 High-Fidelity PCR Kit (Cat. No. E0555L) was purchased from New England Biolabs. T4 DNA ligase (Cat. No. 2011 A) was purchased from Takara Bio Inc. All remaining chemicals and reagents were of analytical grade and obtained from Sigma-Aldrich.

Antibodies

Rabbit anti-beta actin antibody (Cat. No. ab16039, RRID: AB_956497), mouse anti-PSD-95 (Cat. No. ab2723, RRID: AB_303248), rabbit anti-PSD-95 (Cat. No. ab18258, RRID: AB_444362), mouse anti-GluN1 (Cat. No. ab134308, RRID: AB_2818983), rabbit anti-GluN2A (Cat. No. ab133265, RRID: AB_11158532), mouse anti-GluN2B (Cat. No. ab28373, RRID: AB_776810), rabbit anti-synaptophysin (Cat. No. ab14692, RRID: AB_301417) were purchased from Abcam. Rabbit anti-GluA1 (Cat. No. AB1504, RRID: AB_2113602), rabbit anti-GluA2 (Cat. No. AB1768-I, RRID: AB_2922404), mouse anti-GluR2 (Cat. No. MAB397, RRID: AB_2113875) and mouse anti-β-tubulin (Cat. No. T8328, RRID: AB_1844090) were purchased from Sigma. Mouse anti-GluN1 (Cat. No. 114 011, RRID: AB_887750) and rabbit anti-homer (Cat. No. 160003, RRID: AB_887730) was purchased from Synaptic system. Mouse anti-β-actin was from MP Biomedicals (Cat. No. 0869100, RRID: AB_2920628). Secondary antibodies, (anti-mouse IgG antibody (H + L), peroxidase (Cat. No. PI-2000, RRID: AB_2336177), anti-rabbit IgG antibody (H + L), peroxidase (Cat. No. PI-1000, RRID: AB_2916034)) were purchased from Vector Laboratories, Inc. Goat anti-Rabbit IgG (H + L) Alexa Fluor 488 (Cat. No. A-11034, RRID: AB_2576217), anti-Mouse IgG (H + L) Alexa Fluor 594 (Cat. No. A-21203, RRID: AB_2535789) were purchased from Thermo Fisher Scientific.

Animals

The APPSwe/PS1ΔE9 (B6;C3-Tg(APPswe, PSEN1dE9)85Dbo/Mmjax) double transgenic mice with a C57BL/6J background were acquired from the Jackson Laboratory (https://www.jax.org/strain/004462#; RRID: MMRRC_ Stock_ No: 034829-JAX). WT and APP/PS1 mice were bred and maintained within the Institutional Central Animal Facility, where they resided in controlled environments with a 12-hour light/12-hour dark cycle, ensuring sterile and pathogen-free conditions. The animal experiments followed ARRIVE (Animal Research: Reporting of In Vivo Experiments), guidelines, and the experimental protocols were approved by the Institutional Animal Ethics Committee. following the guide for the care and use of laboratory animals. We confirmed the presence of Aβ aggregates in brain sections from six- and nine-month-old APP/PS1 mice by thioflavin-S and 6E-10 immunostaining (Fig. 5, Additional file 1). The levels of synaptic markers, homer and synaptophysin, were analyzed in one- and nine-month-old APP/PS1 mice, along with total Aβ₁₋₄₂ levels (Fig. 6, Additional file 1). Synaptic markers (synaptophysin, PSD-95 and homer), 6E10 intensity (antibody reactive to amino acid 1 to 16 of β-amyloid), and total Aβ₁₋₄₂ levels were also measured in primary cortical neurons from WT and APP/PS1 mice at DIV7 and DIV21 (Figs. 7 and 8, Additional file 1).

Postmortem brain tissues

Frontal neocortical tissue was obtained from participants involved in the Religious Orders Study (ROS) conducted by the Rush Alzheimer’s Disease Center (Chicago, IL). Participants were persons evaluated for possible dementia as home visits across the United States. All participants agreed to an annual clinical evaluation and brain donation at the time of death. The study was approved by an Institutional Review Board of the Rush University Medical Center [21]. All participants signed informed and repository consents and an Anatomic Gift Act. The diagnosis of AD required a history of cognitive decline and evidence of impairment in memory and other cognitive abilities according to the National Institute of Neurological and Communicative Disorders and Stroke/Alzheimer’s Disease and Related Disorders Association (NINDCS/ADRDA) criteria [21]. At autopsy, brains from all participants were removed in a standard fashion as described previously [21, 22]. We have included no cognitive impairment (NCI), mild cognitive impairment (MCI), and Alzheimer’s dementia (AD) participants, as previously described [21, 23, 24], for our experimental analysis. A detailed description of the cases has been provided in the supplementary text (Table 3, Additional file 4). All experiments involving human postmortem tissues were conducted according to institutional guidelines and after obtaining approval from the Institutional Human Ethics Committee. We analyzed a total of 30 brains, specifically including 10 cases each of NCI, MCI, and AD, for comparative analysis. The brains were also assessed for Braak Stage, β-amyloid load, and PHF tau tangle density as previously described [25, 26]. Synaptosomes were prepared from frontal cortical tissues from each postmortem brain as described below.

Synaptosome isolation

WT and APP/PS1 mice were euthanized using carbon dioxide (CO₂) inhalation method in their home cage placing inside the CO₂ euthanasia chamber followed by decapitation. The mouse brain was dissected to remove the cortex and hippocampus, which were then frozen in liquid nitrogen and stored at -80 °C until needed. Synaptosomes were prepared as mentioned earlier [27]. Briefly, Potter–Elvehjem homogenizer was used to homogenize the brain tissue in 10 volumes of homogenization buffer containing 0.32 M sucrose, 5mM HEPES buffer (pH 7.4), 1mM sodium orthovanadate, 50mM sodium fluoride, aprotinin (2 µg/ml), pepstatin A (7 µg/ml), leupeptin (10 µg/ml), 100 µg/ml of phenyl methane sulfonyl fluoride (PMSF), and protease inhibitor cocktail (10 µl/ml). After homogenization, we centrifuged the homogenate at 1200 ×g at 4 °C for 10 min to collect the post-nuclear supernatant (PNS). Subsequently, we centrifuged the PNS again at 15,000 ×g at 4 °C for 15 min, and then resuspended the resulting pellet in the homogenization buffer. This resuspended pellet was added on to a discontinuous sucrose gradient (0.85–1.0–1.2 M) and centrifuged at 85,000 ×g for 2 h at 4 °C using table-top ultracentrifuge (Optima MAX-XP ultracentrifuge-Beckman-Coulter). The pellet obtained at the interface of the 1 and 1.2 M sucrose gradient was collected as the synaptosomal fraction and washed twice in 5 mM HEPES buffer by centrifuging at 20,000 ×g for 20 min at 4 °C and then resuspended it in homogenization buffer. The bicinchoninic acid protein assay was used to measure the protein concentration.

Immunoprecipitation

Mass spectrometry: The synaptosomes were resuspended in lysis and F-actin stabilization buffer and incubated with anti-β-actin antibody (2 µg/ml) at 4 °C with continuous end-to-end rotation overnight. Subsequently, 50 µl of Dynabeads Protein A was added to the solution and incubated for 4 h at 4 °C with continuous end-to-end rotation. The immunocomplexes were extensively washed with Nonidet P-40 lysis buffer to remove non-specific binding. After the final wash, the pellet was resuspended in 2× SDS-PAGE sample buffer and boiled for 10 min. The immunoprecipitated proteins were then subjected to SDS-PAGE, fixed, stained with Brilliant Blue G-Colloidal Concentrate, distained, and the protein bands were excised. The peptides extracted from the bands were subjected to LC-MS/MS.

Association of PSD-95 with actin, AMPA, and NMDA receptor subunits

To demonstrate the association of PSD-95 with actin and glutamatergic receptors, a similar protocol was followed as described above. Synaptosomes were incubated with mouse anti-PSD-95 antibody (3 µg/ml) at 4 °C, followed by Dynabeads Protein G incubation for 4 h at 4 °C. Subsequently, the immunoprecipitated proteins were then subjected to SDS-PAGE and western blotting, followed by probing with respective antibodies.

SDS-PAGE and immunoblotting

The synaptosomes and immunoprecipitated samples were subjected to SDS-PAGE. After electrophoresis, the proteins were transferred from the gel to a PVDF membrane using the electroblotting technique [28]. The membranes were blocked in 5% (w/v) bovine serum albumin (BSA) in tris-buffered saline (TBS; 10 mM Tris, pH 8.0, 150 mM NaCl) for 1 h at room temperature before being immunoblotted with respective primary antibodies and incubated overnight at 4 °C. Immunoblots were washed with TBST (10 mM Tris, pH 8, 150 mM NaCl, 0.05% Tween 20) and incubated with secondary antibodies (5% (w/v) skimmed milk powder in TBS) for 1 h at room temperature. The washed membranes and immunoreactive bands were visualized using enhanced chemiluminescence (Clarity Western ECL blotting substrate, Bio-Rad) and captured using the Bio-Rad Chemidoc-XRS imaging system, with subsequent analysis conducted using Image Lab software [29].

Contextual fear conditioning

The experiments involved male mice aged 9 months. The contextual fear conditioning (cFC) training context was a rectangular shape, with its identity maintained by a distinct odor (2% acetic acid, vol/vol). Before and after each session, the conditioning chamber was cleaned with 70% ethanol. Mice were single-housed and handled for 5 min for 3 days prior to training. On the training day, mice explored the context for 1 min and then received 3-foot shocks (2 s each, 0.6 mA, 30 s intertrial interval). Immediately after training, Jasplakinolide, freshly dissolved in 3% DMSO in saline (10 µL), was injected intrathecally at a dose of 0.025 µg/g body weight. Contextual fear memory was assessed 24 h later by returning mice to the training context for a 2-minutes test period, where freezing was measured as a complete absence of somatic mobility other than respiratory movements. Immediately after recall, brain tissues were dissected and subsequently processed to assess synaptic F-actin levels and the association between PSD-95 and actin.

Isolation of F-actin fraction from synaptosome

Synaptosomes were resuspended in a lysis buffer containing 1 mM ATP and a protease inhibitor mixture to stabilize F-actin. G-actin and F-actin fractions were isolated using the G-Actin/F-Actin In Vivo Assay Kit following the guidelines provided by the manufacturer. Prior to immunoblotting, protein concentrations were measured using the Pierce BCA protein assay kit. The TGX Stain-Free Fast Cast Acrylamide Kit (12%; Bio-Rad) was used to separate all samples and transfer them onto a PVDF membrane for immunoblotting. Before antibody incubation, stain-free blots were imaged using the Bio-Rad Chemidoc-XRS and analyzed with Image Lab software (Bio-Rad). The stain-free detection method was preferred as a loading control instead of β-tubulin for normalization [30]. Immunoreactive bands were visualized using enhanced chemiluminescence (Clarity Western ECL blotting substrate; Bio-Rad) and analyzed using Image Lab software.

Preparation of primary cortical neuronal cultures

Primary cortical neuronal cultures were prepared from postnatal day 0 to 1 (P0-P1) WT and APP/PS1 mouse brain cortex [31]. The cerebral cortex from the mouse brain was dissected out after removing the meninges. Cortical neurons were dissociated by incubating in a solution of 0.1% DNase and 0.25% papain in HHGN (1 × Hanks’ Balanced Salt Solution, 12.5 mM HEPES-buffered sterile saline, D-glucose) followed by mechanical trituration. Cortical neurons were seeded onto poly-D-lysine (0.1 mg/ml) precoated cover slips in 12-well plates. The cells were cultured in neurobasal medium supplemented with 1 × B27, 2 mM L-GlutaMAX, and 100 µg/ml penicillin/streptomycin under serum-free conditions. The cultures were maintained at 37⁰C in 5% CO₂ for a period of 3 weeks.

Jasplakinolide treatment and immunofluorescence staining of primary cortical neuronal cultures

Primary cortical cultures from WT and APP/PS1 at DIV 21 were treated with vehicle and jasplakinolide (10 nM) for 20 mints at 37⁰C in 5% CO₂ [32]. Following the treatment, the cultures were rinsed, fixed with a solution of 4% paraformaldehyde and 4% sucrose, and permeabilized using 0.2% Triton X-100. The neurons were blocked with 5% BSA (w/v) and separately incubated with mouse anti-GluA2 antibody (specific to the large N-terminal extracellular domain of GluA2) and mouse anti-GluN1 antibody (suitable for surface staining) for 1 h (synaptic AMPA and NMDA receptor staining as described) [33]. Following this, the primary antibodies were removed, and the samples were incubated with a mouse Alexa Fluor 594 secondary antibody for 1 h. Subsequently, a co-staining was performed using a rabbit anti-PSD-95 antibody for 2 h at room temperature, followed by rabbit Alexa Fluor 488 secondary antibody for visualization. For F-actin staining, acti-stain 670 was incubated for 1.5 h, washed and imaged under confocal microscope. Immunostaining for 6E10, synaptophysin, and homer followed a similar protocol. Permeabilization was performed using 0.3% Triton X-100, followed by a 1-hour incubation with the primary antibody at room temperature, and then incubation with the respective secondary antibody for visualization.

Image acquisition and analysis

Confocal images were acquired using a Carl Zeiss LSM780 laser scanning system with an oil-immersion objective (63x/1.40 NA). The images were captured at a resolution of 512 × 512 with a 16-bit depth, a zoom factor of 3, a pinhole set at 1 airy unit, and a step size interval of 0.45 μm for z-stack acquisition. The pinhole set at 1 airy unit (AU) ensures that the pinhole diameter matches the Airy radius, which captures most of the light intensity, thereby minimizing diffraction and aberration and resulting in the highest possible resolution. Increasing the pinhole diameter above 1 AU allows more extra focal light to reach the detector, which can enhance intensity but reduces resolution. Conversely, reducing the pinhole diameter below 1 AU improves resolution but reduces signal intensity, due to the restricted amount of light entering the detector. MetaMorph software (version 7.8.0.0, 2013; Molecular Devices) [34] was used to quantify F-actin levels as phalloidin intensity and measure GluA2 and GluN1 intensity as described previously [7]. Phalloidin-stained primary cortical neuronal cultures were used to analyze and reconstruct dendritic spines using Neurolucida 360, as described in previous studies [7, 35]. The measurement of colocalization between PSD-95 with GluA2 and GluN1 was conducted in the form of Pearson’s correlation coefficients using the JACoP (Just Another Co-localization Plugin) plugin in ImageJ, as described [36, 37].

Plasmids and generation of LV-PSD-95–mClover3

pMDLg/pRRE (Addgene plasmid No. 12251; [38]; RRID: Addgene_12251), pRSV-Rev (Addgene plasmid No. 12253; [39]; RRID: Addgene_12253), pMD2.G (Addgene plasmid No. 12259; [40]; RRID: Addgene_12259) were a gift from Didier Trono. FH95pUp95GW (B4) was a gift from Robert Malenka and Oliver Schluter and Weifeng Xu (Addgene plasmid No. 74013; [41]; RRID: Addgene_74013). pLVX-mCherry-Actin (Cat. No. 631078) was purchased from Clontech Laboratories, Inc. pKanCMV-mClover3-18aa-Tubulin was a gift from Michael Lin (Addgene plasmid No. 74253; [42]; RRID: Addgene_74253). and a kind gift from Dr. Narendrakumar Ramanan, Indian Institute of Science, Bangalore. To achieve this, the PSD-95 sequence was amplified from the FH95pUp95GW (B4) plasmid using specific primers: PSD95-XbaI-For (5’-GACGTCTCTAGACCACCATGGACTGTCTCTGTATAGTG-3’) and PSD95-AgeI-Rev (5’-ATTAATACCGGTGAGTCTCTCTCGGGCTGGGACCCAG-3’). Simultaneously, the mClover3 gene was amplified from the pKanCMV-mClover3-18aa-Tubulin plasmid using the following primers: mClover3-AgeI-For (5’-ATTAATACCGGTATGGTGAGCAAGGGCGAGGAG-3’) and mClover3-SalI-Rev (5’-CGGATCGTCGACTTACTTGTACAGCTCGTCCAT-3’) utilizing the Q5 High-Fidelity PCR Kit. pRRLsinPPTeGFP vector and inserts were enzymatically digested using the corresponding restriction enzymes and then ligated together employing T4 DNA ligase at 16 °C for overnight. The ligation mixture was then introduced into DH5-α cells and screened for positive clones, ensuring the successful integration of the desired DNA sequences.

Lentiviral particles were prepared for PSD-95-mClover3 according to Ritter B et al., protocol [43]. Briefly, HEK293T cells were plated in DMEM high glucose medium at a density of 15 × 10⁶ cells/plate on day 1. Calcium phosphate–mediated transfection method was used to deliver PSD-95-mClover3 plasmid DNA into the cells. DNA for the packaging mix along with expression plasmid was incubated with CaCl₂ and mix with tube containing HBS by continuously bubbling. Then, allow DNA/calcium-phosphate precipitate to form for 25 min at room temperature. Carefully added the transfection mixture to the cell culture plate and remove the medium from the plate post 8-hours of transfection. Virus-producing cells started releasing the virus into the medium 24 h post-transfection. Supernatant was collected at 24, 36 and 48 h post transfection and centrifuged at 55,000×g for 2 h at 4⁰C to obtain the viral particle expressing PSD-95-mClover3. The active particles in resuspended pellet were determined by titration in HEK293T cells.

pLV-actin-mcherry transfection to primary cortical neurons

Primary cortical neurons cultured for 7 days undergo transfection using lipofectamine 2000 reagent. Neurobasal medium supplemented with 200 mM L-glutamine was used as the incubation medium [44]. Next, 1.5 µg pLV-actin-mCherry plasmid and 3.3 µl lipofectamine were diluted in 100 µl neurobasal medium per 35 mm glass bottom dish, mixed with the solution, and incubated for 30 min at room temperature. The conditioned medium of the neurons was then transferred to a new 35 mm dish and placed in a 37 °C incubator. Following this, 500 µl of incubation medium and 200 µl of the DNA plus lipofectamine mixture were added to the neurons and incubated at 37 °C for 40 min. Finally, the transfection mixture was replaced with the conditioned medium.

Fluorescence recovery after photobleaching

Fluorescence recovery after photobleaching (FRAP) experiments were conducted on primary cortical neuronal cultures to investigate the dynamics of actin when expressed alone and when co-expressed with PSD-95. For co-expression, neurons were transfected with pLV-actin-mCherry on day 7 in vitro (DIV7) and transduced with LV-PSD-95-mClover3 on the following day. FRAP experiments were conducted three days after transfection by placing the samples in a temperature-controlled chamber with 5% CO₂ using a live cell imaging-compatible solution (HEPES buffer, pH 7.5, containing 10 mM Glucose, 120 mM NaCl, 3 mM KCl, 1 mM MgCl₂, 2 mM CaCl₂, and 10 mM HEPES). The images were captured using a 63x oil-immersion objective, with a resolution of 512 × 512 pixels and a zoom factor of 2. Region of interest (ROI) was bleached by a single iteration of 100% laser power from 561 nm wavelengths. After capturing five pre-bleach images, the ROI underwent bleaching, followed by the acquisition of 45 post-bleach images. The rate of image acquisition was set at 1 s per frame. The images were acquired every second to 1 min to monitor the recovery process. Extended recovery time beyond 1 min was not pursued due to significant photobleaching effects observed under the experimental conditions. MetaMorph software was used for image acquisition and data collection. Normalized fluorescence intensity was calculated as described previously [45]. The recovery curve was fitted using GraphPad Prism 9 software with the formula, Y = Y0 + (Plateau − Y0) × (1 − exp(− Kx). Y0 = Y value when X = 0, Plateau = Y value at infinite X, K = rate constant (1/x), t½ = ln (2)/K. The mobile fraction was calculated from normalized recovery curve using the equation, M_f = (F_∞ - F₀) / (F_i - F₀) × 100 where, F_∞ = intensity at time t post-bleach, F₀ = intensity immediately after the bleach pulse, F_i = intensity at time t pre-bleach and the half time of recovery was extracted from the fitted curve of respective analysis. FRAP experiments were conducted to assess the dynamics of PSD-95 in both WT and APP/PS1 cortical cultures at DIV21 using identical parameters. ROI was bleached by a single iteration of 100% laser power from 488 nm wavelengths.

Aβ₁₋₄₂ enzyme linked immunosorbent assay (ELISA)

Aβ₁₋₄₂ concentrations were measured with the SensoLyte Anti-Mouse/Rat β-Amyloid (1–42) quantitative ELISA (Cat. No. AS-55554, AnaSpec, Inc., Fremont, CA, USA) following the manufacturer’s protocol. Cortex tissue was homogenized in a buffer containing 5 M Guanidine HCl, 50 mM Tris HCl (pH = 8.0), and protease inhibitors, then incubated at room temperature for 4 h while mixing. Samples were diluted 1:40 with the provided buffer, and 100 µL of the diluted sample was added to each well in duplicates. For conditioned media, 100 µL was added to each well without dilution. Aβ levels (pg/µL) were determined by comparing the optical density at 450 nm to a standard curve. Protein concentrations in tissue samples were measured using the BCA assay from the homogenates. Aβ₁₋₄₂ levels were normalized to total protein content and expressed as picograms of Aβ₁₋₄₂ per milligram of total protein (pg/mg) for tissue lysate and picograms of Aβ₁₋₄₂ per mL for conditioned media.

Mass spectrometry proteomics data

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium [46] via the PRoteomics IDEntifications (PRIDE) partner repository with the dataset identifier < PXD050802>. The workflow for mass spectrometry analysis followed the schematic representation depicted in Graphical Abstract (Fig. 1a). This workflow included sample preparation and mass spectrometry analysis as described in detail in the supplementary information. Proteins that consistently appear in all three biological replicates were selected for subsequent analysis from both WT and APP/PS1 mice. The abundance obtained from mass spectrometry analysis was normalized to an internal control (β-actin) within each sample [47].

Bioinformatics

ClueGO plugin within Cytoscape 3.10.1 [48] was used to analyze functional enrichment in Gene Ontology processes and pathways, queried against the Mus musculus database and incorporated GO, KEGG, and REACTOME annotations. Functional clusters were visualized as nodes with a kappa score cutoff of 0.4, displaying only the most significant term in each node. Overlapping nodes representing functionally related items appeared as multicoloured nodes. The web app, VolcaNoseR [49] was employed to generate volcano plots illustrating significantly differentially expressed actin-interacting proteins in APP/PS1 [50]. A fold change of > 1 indicated increased expression, and a fold change of <-1 indicated decreased expression, corresponding to a p-value of < 0.05 in an unpaired, two-tailed Student t-test. The Morpheus Heatmap Generator [51] was utilized to create a heatmap depicting the relative abundance of significantly differentially expressed proteins in WT and APP/PS1.

Statistical analysis

Statistical analysis was performed using GraphPad Prism software (version 9.01, GraphPad Software Inc) [52]. The statistical comparisons between the WT and APP/PS1 groups for assessing glutamatergic receptor levels and interaction studies were analyzed using the two-tailed, unpaired, Mann–Whitney “U” test. Experiments involving jasplakinolide injection or treatment involving vehicle and treated groups from WT and APP/PS1 were analyzed with two-way ANOVA, Tukey’s multiple comparisons test, with a difference of at least P < 0.05 considered statistically significant. Studies involved in determining PSD-95 actin association in human subject with NCI, MCI and AD were analyzed with one-way ANOVA on ranks (Kruskal-Wallis test), followed by Dunn’s multiple comparisons test. Values were expressed as mean ± S.E.M. FRAP experiments were statistically analyzed and tested with one-way Welch’s t-test and expressed as mean ± SD. Detailed descriptions of statistical analysis are provided in Table 4, Additional file 5.

Results

Synaptic actin interactome from APP/PS1 mice reveals proteins critical for synaptic functions

Actin performs critical functions within the synapse to ensure its structural integrity and dynamic adaptability. Previous studies from our laboratory on the APP/PS1 mouse model revealed that the depolymerization of synaptosomal F-actin starts as early as one month of age [7]. We also observed a concurrent decrease in the reduced form of F-actin and an increase in glutathionylated synaptosomal actin [53]. In this study, we examined the synaptic actin interactome in APP/PS1 mice using mass spectrometry. We utilized immunoprecipitation and quantitative mass spectrometry to identify and characterize the set of actin-associated proteins in synaptosomes (Fig. 1a). Proteins that consistently appeared in all three biological replicates were selected for subsequent analysis from both WT and APP/PS1 mice (Table 1, Additional file 2). A total of 176 and 177 proteins were identified as interacting with actin in WT and APP/PS1 mice, respectively (Fig. 1b). Notably, 122 proteins were identified as common entities between the two groups (Fig. 1b). To identify biological processes associated with these 122 proteins, Gene Ontology (GO) enrichment analysis was performed. Gene Ontology analysis of these actin–interacting proteins revealed enrichment in functional categories related to post synapse organization, structural constituent of post synapse, actin filament depolymerization and maintenance of postsynaptic specialization structure etc. (Fig. 1c).

Among the proteins common to both the WT and APP/PS1 mice, 24 exhibited significant differences, with 22 proteins downregulated and 2 proteins upregulated in APP/PS1 mice (Fig. 2, a-d) (Table 2, Additional file 3). The GO enrichment analysis of significantly altered proteins revealed pathways implicated in the structural components of the synapse, post-synapse and cytoskeleton, among others (Fig. 2, a and b). For instance, the downregulated proteins in the structural constituents of post synapse include synapse-associated protein-102 (SAP-102), post synaptic density-95 (PSD-95), synaptic Ras GTPase-activating protein 1 (SynGAP1), growth associated protein-43 (GAP-43) and calcium/calmodulin dependent protein kinase II beta (Camk2b). Among the upregulated proteins are the neurofilament medium polypeptide and the Cytochrome b-c1 complex subunit Rieske (Fig. 2, c and d). This finding substantiates the diverse roles of actin in regulating both the structure and function of synapses.

Fig. 2

Differential expression of actin interacting proteins identified from WT and APP/PS1 mice. (A) GO analysis depicting functionally enriched pathways of significantly differentially expressed proteins (DEPs) in both WT and APP/PS1. (B) Volcano plot illustrating significantly differentially expressed proteins in both the groups. Proteins with a fold change > 1 are depicted in red, indicating increased expression, while proteins with a fold change <-1 are shown in green, indicating decreased expression, all corresponding to a significance level of p < 0.05. A total of 24 proteins were identified, with 22 proteins downregulated and 2 proteins upregulated. (C) Visualization of Gene Ontology (GO) term enrichment using the ClueGO / CluePedia plugin within Cytoscape. The network displays the connectivity of significant GO terms associated with differentially expressed proteins (DEPs) in both groups, with functional nodes and edges representing shared relationships based on a kappa score of 0.4. Only GO terms with a significance level of p ≤ 0.05 are included in the enrichment analysis. Differentially expressed proteins (DEPs) specific to each group are highlighted in red font. (D) Heatmap displaying the expression levels of significantly differentially expressed proteins from each sample in both groups

Disruption of the PSD-95–actin interaction in middle-aged APP/PS1 mice

We initially identified a disruption in the PSD-95–actin interaction at six months of age in APP/PS1 mice using our mass spectrometry (Fig. 2, a and b). To further validate and explore the dynamics of this interaction, we aimed to conduct immunoprecipitation assays in synaptosomes isolated from one- (adolescent) and nine-month-old (middle-aged) WT and APP/PS1 mice. We observed that PSD-95 associates with actin in both the WT and APP/PS1 mice at one-month of age, but no significant differences were detected between the two groups (Fig. 3, a and b) However, we detected a significant disruption in the PSD-95–actin interaction in middle-aged APP/PS1 mice (Fig. 3, c and d), as evidenced by mass spectrometry data. Furthermore, we conducted immunoprecipitation assays on F-actin fractions isolated from the synaptosomes of nine-month-old WT and APP/PS1 mice. Our analysis revealed that the association of PSD-95 with actin was similar in both WT and APP/PS1 mice, with no significant difference observed between the two groups (Fig. 3, e and f). Our analysis indicated that the depolymerization of F-actin is a contributing factor to the disrupted interaction between PSD-95 and actin rather than the decrease in PSD-95 levels observed at nine-months-of age in the APP/PS1 mice (Fig. 1a, Additional file 1).

Fig. 3

Memory recall and PSD-95 actin association restored by jasplakinolide in APP/PS1 mice. Representative immunoblot showing the association between PSD-95 and actin in WT and APP/PS1 mice at one-month (adolescent-ADL) (a, b), nine-month of age (middle-aged-MA) (c, d), and in F-actin fraction of synaptosomes in middle-aged APP/PS1 mice (e, f) (n = 8). G) Schematic of contextual fear conditioning memory paradigm. Jasplakinolide injection (h) rescued fear memory in middle-aged APP/PS1 mice (n = 8 (WT vehicle, APP/PS1 vehicle, APP/PS1 jas), n = 6 (WT jas), restored (i-j) synaptic F-actin levels (n = 8 (WT vehicle, APP/PS1 vehicle, APP/PS1 jas), n = 6 (WT jas), and (k-l) PSD-95 actin association (n = 8 (WT vehicle, APP/PS1 vehicle, APP/PS1 jas), n = 6 (WT jas). Statistical comparison between groups was conducted using either an unpaired two-tailed t-test (a-f) or a two-way ANOVA followed by Tukey’s post test. (h-l). Data are presented as mean ± SEM. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Memory recall and the PSD-95–actin association is restored by jasplakinolide in APP/PS1 mice

To investigate the effects of actin stabilization on memory recall and the PSD-95–actin association, we administered jasplakinolide to WT and APP/PS1 mice and assessed their recall memory. We found that APP/PS1 mice had impaired memory recall compared to that of WT mice, as indicated by their significantly reduced freezing response (Fig. 3, g and h). After jasplakinolide infusion, the reduced freezing response in the APP/PS1 mice was restored compared to that in the vehicle-treated mice, but no comparable difference was observed in WT mice (Fig. 3h). Following jasplakinolide injection and memory recall, cerebral cortex samples from these mice were analyzed to assess synaptic F-actin levels and the association between PSD-95 and actin. Synaptic F-actin levels were restored in APP/PS1 mice treated with jasplakinolide (Fig. 3, I and j). The disruption of association between PSD-95 and actin in nine-month-old APP/PS1 mice was reversed upon treatment with jasplakinolide compared to vehicle (Fig. 3, k and l). Together, our results provide evidence that the depolymerization of actin underlies the observed deficits as shown by the rescued memory recall, the restoration of synaptic F-actin levels and PSD-95 actin association by actin stabilization in APP/PS1 mice.

Altered AMPA and NMDA receptor levels and interactions with PSD-95 in APP/PS1 mice

PSD-95 orchestrates the clustering and anchoring of AMPA and NMDA receptors at the excitatory synapses, ensuring their stable localization for synaptic neurotransmission, signaling underlying cognitive processes and plasticity [54]. It interacts directly with NMDA receptors through the PDZ domain, whereas the interaction with AMPA receptors occurs through an auxiliary subunit, namely transmembrane AMPA receptor regulatory proteins (TARPs) [54]. We therefore aimed to determine the AMPA and NMDA receptor subunit levels and their associations with PSD-95 in one- and nine-month-old WT and APP/PS1 mice. The levels of AMPA receptor subunits, GluA1 and GluA2 were decreased at nine-month of age in APP/PS1 mice compared to that of WT, with no comparable difference between the groups at one-month of age (Fig. 4, a–d). Interestingly, the alteration in the levels of NMDA receptor subunits occurs as early as one-month of age in APP/PS1 mice. Specifically, the levels of the NMDA receptor subunits, GluN1, GluN2A and GluN2B were significantly decreased at one-month of age (Fig. 4e, f and g). Among these subunits, GluN1 and GluN2B exhibited a sustained decrease in expression at three (Fig. 2, a and c, Additional file 1) and nine months in APP/PS1 mice compared with WT mice (Fig. 4, h and j). However, no significant difference was detected in the expression levels of the GluN2A subunit at later ages (Fig. 4i) (Fig. 2b, Additional file 1).

Fig. 4

Changes in AMPA and NMDA receptor levels in APP/PS1 mice. (a) GluA1 levels in one-month and (b) nine-month-old WT and APP/PS1 mice (n = 8). (c) GluA2 levels in one-month and (d) nine-month-old WT and APP/PS1 mice (n = 8). (e) GluN1 levels in one-month and (f) nine-month-old WT and APP/PS1 mice (n = 8). (g) GluN2A levels in one-month and (h) nine-month-old WT and APP/PS1 mice (n = 8). (i) GluN2B levels in one-month and (j) nine-month-old WT and APP/PS1 mice (n = 8). Statistical comparison between groups was conducted using an unpaired two-tailed t-test. Data are presented as mean ± SEM. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

The observed alterations in AMPA and NMDA receptor subunit levels prompted an investigation into their association with PSD-95 in APP/PS1 mice compared to WT controls. The associations of GluA1 and GluA2 with PSD-95 were disrupted at nine-months of age in APP/PS1 mice (Fig. 5, a–d). However, the association between GluN1 and GluN2B with PSD-95 was enhanced at one-month of age in APP/PS1 mice, despite the significantly reduced receptor levels observed in synaptosomes at a similar age. Nonetheless, at nine-months of age, there was no significant difference in the association between the WT and APP/PS1 mice. Additionally, the association of GluN2A with PSD-95 is not affected at one- and nine-months of age between WT and APP/PS1 mice (Fig. 5, e–p).

Fig. 5

Altered PSD-95 association with AMPA and NMDA receptors in APP/PS1 mice. PSD-95 association with GluA1 (a-b) and GluA2 (c-d) in middle-aged APP/PS1 mice (n = 8). PSD-95 association with GluN1 in one-month- (e-f) and in nine-month-old APP/PS1 mice (g-h) (n = 8). PSD-95 association with GluN2A in one-month-(i-j) and in nine-month-old APP/PS1 mice (k-l) (n = 8). PSD-95 association with GluN2B in one-month- (m-n) and in nine-month-old APP/PS1 mice (o-p) (n = 8). Statistical comparison between groups was conducted using an unpaired two-tailed t-test. Data are presented as mean ± SEM. Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

F-actin levels and dendritic spine density are restored by jasplakinolide treatment in APP/PS1 neurites

Synaptic dysfunction in AD manifests as a gradual loss of synapses, which strongly correlates with cognitive decline [2, 3, 55, 56]. This synaptic deterioration becomes apparent early in AD progression, as dendritic spines undergo progressive changes in both number and morphology, indicating compromised synaptic integrity. Building upon prior investigations from our laboratory that demonstrated a reduction in dendritic spine density and F-actin levels within the tertiary neurites of APP/PS1 mice [7], our objective was to assess whether treatment with an F-actin stabilizing agent could effectively restore F-actin levels and dendritic spine density. Jasplakinolide, a membrane permeable natural product, binds to the pointed end of F-actin in a 1:1 ratio, thereby impeding the dissociation of monomers from the filaments. The minimum concentration necessary for F-actin stabilization is > 6 nM, while at a concentration of 20 nM, jasplakinolide induces actin assembly rather than merely stabilizing the filaments [32, 57]. We observed a significant increase in F-actin levels (Fig. 6, a and b) and dendritic spine density (Fig. 6, a and c) in jasplakinolide treated APP/PS1 neurites compared to those in vehicle treated neurite. In contrast, a comparable difference was not observed between the WT groups.

Fig. 6

Jasplakinolide treatment restores F-actin levels and dendritic spine density in APP/PS1 neurites. (a) Representative images of WT and APP/PS1 primary cortical neuronal cultures treated with jasplakinolide and stained with phalloidin (F-actin), along with 3D reconstructions. (b) Quantification of F-actin levels from tertiary neurites. (c) Quantification of total dendritic spine density normalized to WT vehicle. (d) Quantification of number of spine/µm and (e) total number of spines observed. Statistical comparison between groups were conducted using two-way ANOVA followed by Tukey’s post test. Data are presented as mean ± SEM from three independent experiment (20 neurites from each independent experiment). Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar = 5 μm

Next, we utilized Pearson correlation coefficient (PCC) analysis to assess the degree of colocalization between F-actin and PSD-95 in tertiary neurites from WT and APP/PS1 mice. We observed a significant decrease in the colocalization of F-actin with PSD-95 in APP/PS1 neurites compared to that in WT neurites. However, the treatment of jasplakinolide increased the colocalization in APP/PS1 neurites compared to those treated with vehicle. There was no significant difference in the degree of colocalization among the WT groups (Fig. 7, a and b).

Fig. 7

Jasplakinolide enhances PSD-95 colocalization with actin in APP/PS1 neurites. (a) Representative images of jasplakinolide-treated WT and APP/PS1 primary cortical neurons stained with phalloidin and anti-PSD-95 antibody. (b) Colocalization of PSD-95 and actin quantified using pearson’s correlation coefficient. Statistical comparison between groups were conducted using two-way ANOVA followed by Tukey’s post test. Data are presented as mean ± SEM (n = 3, 11–12 neurites from each independent experiment). Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar = 5 μm

Jasplakinolide treatment restores AMPA and NMDA receptor levels in APP/PS1 neurites

Our investigations, utilizing synaptosomes isolated from APP/PS1 mice, revealed a decrease in AMPA and NMDA receptor subunits across different ages. Further, the colocalization of F-actin with PSD-95 exhibited a reduction in APP/PS1 neurites. PSD-95 plays a crucial role in tethering glutamatergic receptors to the synaptic membrane, facilitating synaptic neurotransmission. Consequently, our aim was to determine the synaptic levels of the receptors GluA2 and GluN1, in primary cortical neurons derived from both WT and APP/PS1 mice. We observed a significant decrease in the levels of GluA2 (Fig. 8, a and b) and GluN1 (Fig. 3. a and b, Additional file 1) receptors in neurites from APP/PS1 mice at DIV21. To determine the impact of F-actin stabilization and its association with PSD-95 on the synaptic localization of AMPA and NMDA receptors, primary cortical neurons were treated with jasplakinolide, and the receptor levels were subsequently assessed. Levels of GluA2 (Fig. 8, a and b) and GluN1 (Fig. 3. a and b, Additional file 1) in neurites of APP/PS1 mice were restored following treatment with jasplakinolide compared with vehicle.

Fig. 8

Synaptic levels of GluA2 are increased in jasplakinolide-treated APP/PS1 neurites. (a) Representative images of jasplakinolide-treated WT and APP/PS1 primary cortical neurons stained with phalloidin, anti-PSD-95 antibody and anti-GluA2 antibody. (b) Quantification of GluA2 levels. (c) Colocalization of PSD-95 and GluA2 quantified using pearson’s correlation coefficient. Statistical comparison between groups were conducted using two-way ANOVA followed by Tukey’s post test. Data are presented as mean ± SEM (n = 3, 15 neurites from each independent experiment). Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar = 5 μm

We then conducted a PCC analysis to assess the colocalization of PSD-95 with GluA2 and GluN1 in both WT and APP/PS1 neurites. We observed a reduced colocalization of PSD-95 with GluA2 (Fig. 8, a and c) and GluN1 (Fig. 3. a and c, Additional file 1) in APP/PS1 neurites than in WT neurites. Upon F-actin stabilization by jasplakinolide, PSD-95 colocalization with the receptors (GluA2; Fig. 8a and c and GluN1; Fig. 3. a and c, Additional file 1) increased significantly in APP/PS1 neurites compared to vehicle-treated, whereas in WT groups, no significant difference in colocalization was observed.

Fluorescence recovery after photobleaching (FRAP) analysis shows a decreased actin recovery rate in primary cortical neurons co-expressing with PSD-95

We conducted Fluorescence recovery after photobleaching (FRAP) experiments to investigate the dynamics of actin-mCherry alone and actin-mCherry coexpressed with PSD-95-mClover3 in primary cortical neuronal cultures. Our analysis aimed to elucidate how the interaction between actin and PSD-95 influences the recovery rates of fluorescence intensity post photobleaching. In cells expressing actin-mCherry alone, we observed a rapid recovery of fluorescence intensity following photobleaching, which is indicative of dynamic actin turnover within the cellular environment (Fig. 9, a and b) (Fig. 4, a and b, Additional file 1). Intriguingly, in cells coexpressing actin-mCherry with PSD-95-mClover3, we observed a significant reduction in the rate of FRAP recovery compared to that in cells expressing actin-mCherry alone (Fig. 9, a and b) (Fig. 4, a and b, Additional file 1), along with a corresponding decrease in the mobile fraction (Fig. 9c) (Fig. 4c, Additional file 1) and an increase in the half time to recovery of actin-mCherry (Fig. 9d) (Fig. 4d, Additional file 1). The prolonged half-life indicates a slower turnover rate of actin molecules, likely influenced by the interaction with PSD-95. The reduced FRAP recovery rate observed in the presence of the interaction between actin and PSD-95 suggests a potential mechanism for PSD-95 localization and dynamics within the synaptic environment. It is plausible that the interaction between actin and PSD-95 facilitates PSD-95 localization to the synaptic membrane, thereby increasing its availability within the postsynaptic density.

Fig. 9

Frap experiments shows decreased actin recovery rate in primary cortical neuronal cultures co-expressed with PSD-95. (a) Representative images showing pre-bleach, bleach, and post-bleach stages of actin fluorescence when expressed alone and when co-expressed with PSD-95 in WT neurites. (b) The plot illustrates the recovery of fluorescence intensity over time. (c) Mobile fraction and (d) halftime of recovery extracted from the fitted curve. Statistical comparison between groups were conducted using an unpaired, two-tailed t-test with Welch’s correction. Data are presented as mean ± SD (n = 16–18 cumulative from three independent experiments). Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar = 5 μm

We further conducted FRAP experiments by overexpressing PSD-95-mClover3 in primary cortical neuronal cultures from WT and APP/PS1 mice to investigate how Aβ-mediated synaptic dysfunctions affect PSD-95 dynamics within the cell. We observed that PSD-95 in primary cortical neuronal cultures from APP/PS1 mice exhibited a faster recovery of fluorescence intensity following photobleaching compared to neurons from WT mice (Fig. 10, a, b and e, f). This is accompanied by a higher mobile fraction (Fig. 9, c and g) and a shortened half-time to recovery (Fig. 9, d and h), suggesting the potential influence of Aβ-mediated synaptic dysfunctions on PSD-95 dynamics, possibly through the loss of interactions and affecting its localization within the cell.

Fig. 10

Differential dynamics of PSD-95 in WT and APP/PS1 primary neuronal cultures. Representative images of neurites (a) and dendritic spines (e) showing pre-bleach, bleach, and post-bleach stages of PSD-95 fluorescence in WT and APP/PS1 neurite. (b, f) The plot illustrates the recovery of fluorescence intensity over time. (c, g) Mobile fraction and (d, h) halftime of recovery extracted from the fitted curve. Statistical comparison between groups were conducted using an unpaired, two-tailed t-test with Welch’s correction. Data are presented as mean ± SD (n = 16–18 cumulative from three independent experiments). Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Scale bar = 5 μm

PSD-95–actin association is disrupted in the cortex of MCI and AD postmortem brains

To determine whether the disruption observed in the PSD-95–actin association in APP/PS1 mice extends to human subjects with AD, we conducted immunoprecipitation studies using synaptosomes isolated from postmortem dorsolateral prefrontal cortex tissue samples of individuals with no cognitive impairment (NCI), mild cognitive impairment (MCI), and AD. A significant reduction in the association was observed in MCI and AD subjects (Fig. 11, a and b), indicating a perturbation in this critical synaptic interaction during disease progression. We found a significant correlation between the disruption of PSD-95 actin association and cognitive performance, as well as an inverse correlation with pathological scores such as β–amyloid and Braak staging (Fig. 11, c–e). However, we did not detect a significant correlation between the disruption of the PSD-95–actin association and tangle density (Fig. 11, f).

Fig. 11

Disrupted PSD-95 actin association in human AD patients. (a, b) Representative immunoblot and quantification showing decreased PSD-95 actin association in MCI and AD patients compared with NCI. Statistical comparison between groups were conducted using one-way ANOVA on ranks followed by Dunn’s multiple comparison test. Data are presented as mean ± SEM (n = 10). Correlation analysis of PSD-95 actin association in human subjects with respect to cognitive performance (c), Braak score (d), amyloid-β deposition (e) tau tangle pathology (f) (n = 30, 10 from each group). Statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Discussion

Our study aimed to elucidate the synaptic actin interactome in an AD mouse model, APP/PS1. Our findings highlighted several actin-binding proteins crucial for synaptic structure and function, particularly PSD-95. We observed a significant decrease in the PSD-95-actin association in the synaptosomes of middle-aged APP/PS1 mice compared to those of WT. Deficits in memory recall, PSD-95-actin association, and synaptic F-actin levels were restored by actin stabilization with jasplakinolide in APP/PS1 mice. In primary cortical neurons from APP/PS1 mice, actin stabilization restored the decreased synaptic levels of AMPA and NMDA receptors; reversed the decreased colocalization of PSD-95 with actin, AMPA and NMDA receptors; and increased F-actin levels and dendritic spine density. FRAP experiments further supported the actin-PSD-95 association as evidenced by the reduced recovery of actin when coexpressed with PSD-95. Analysis of postmortem human tissue from NCI, MCI, and AD subjects revealed disrupted PSD-95–actin interactions, emphasizing the clinical relevance of our findings. Thus, our study demonstrated the altered PSD-95 actin association in primary cortical neurons from APP/PS1 mice, in synaptosomes of APP/PS1 mice and in human cortical tissues from NCI, MCI and AD.

Synapse loss and structural alterations characterize synaptic dysfunction, which has emerged as a prominent correlate of cognitive decline in AD [55, 58,59,60]. Synaptic aberrations manifest early in the disease continuum, preceding the onset of behavioral symptoms. Studies, conducted in our laboratory as well as in other laboratories, have demonstrated the loss of synapses and F-actin at an early-stage in an AD mouse model [61]. F-actin cytoskeleton plays a crucial role in dendritic spine morphology and synaptic plasticity. F-actin influences dendritic spine size, which subsequently correlates with PSD size and AMPAR content, thereby impacting synaptic strength. Mass spectrometry-based quantitative proteomics and network analysis revealed protein hubs in AD including proteins linked to the organization and regulation of the actin cytoskeleton, which are negatively correlated with the AD phenotype and reflect impaired actin dynamics in AD [62]. The phosphoproteomics analysis of the AD brain also revealed changes in the phosphorylation status of proteins associated with the cytoskeletal network, including actin, spectrin, and neurofilament proteins [63]. Here, we characterized the synaptic actin interactome in APP/PS1 mice and identified significant dysregulation of actin binding proteins, suggesting a critical interplay between actin dynamics and synaptic functions. The differential expression of proteins, including synapse-associated protein-102 (SAP-102), calcium/calmodulin-dependent protein kinase type II β (CamK2b), post synaptic density-95 (PSD-95), tropomodulin-2 (Tmod2), neurofilament medium chain (Nefm) and growth associated protein-43 (GAP-43) in APP/PS1 mice, among others, suggested specific molecular perturbations associated with AD pathology. Members of the disk-large family, SAP-102 and PSD-95 [64], directly interact with GluN2A or GluN2B subunits and are involved in the synaptic clustering of NMDA receptors. The interaction of SAP-102 with GluN2B is critical for the diffusion of NMDA receptor to synaptic sites [65]. These proteins are also important for synaptogenesis and AMPA receptor trafficking [66]. PSD-95 deficiency is associated with reduced AMPA and NMDA receptor expression and affects prefrontal cortex associated functions such as cognition and working memory [67]. The appropriate localization of PSD-95 at the postsynaptic density is critical for its functions, such as strengthening excitatory synapses, surface localization of receptors, and synaptic plasticity. Despite their critical roles in synaptic function, the AD inferior temporal cortex exhibited significantly reduced PSD-95 and SAP-102 mRNA and protein levels that varied in accordance with disease severity [68]. CamK2b, abundant in excitatory synapses, plays a central role in synaptic plasticity, learning, and memory, with its F-actin bundling activity crucial for maintaining dendritic spine structure [69]. The binding of F-actin to CamK2b in cerebellar Purkinje cells regulates bidirectional synaptic plasticity [70]. Tmod2 plays a role in regulating dendritic arborization as an actin slow-growing end capping protein [71]. GAP-43 plays a crucial role in neuronal development and synaptogenesis, while also being involved in the regulation of axonal outgrowth and synaptic plasticity, particularly in learning and memory processes [72]. Recent research has suggested that GAP-43 may serve as an early biomarker of synaptic dysfunction during AD progression [73]. The dysregulation of these proteins observed in APP/PS1 mice has profound implications for synaptic function, delineating essential components involved in AD pathology.

Elucidating our understanding of the synaptic actin interactome, we showed that the association between PSD-95 and actin is disrupted in synaptosomes from middle-aged APP/PS1 mice. Furthermore, our examination of the F-actin fraction in synaptosomes revealed no significant difference among the groups, suggesting that the disruption of PSD-95 actin interaction in middle-aged APP/PS1 mice may be primarily driven by alterations in F-actin dynamics rather than changes in PSD-95 levels per se. At one month of age, although there is evidence of F-actin loss, the interaction between PSD-95 and actin remains intact. This suggests that the synaptic environment, in terms of protein interaction, is still relatively preserved. The lack of significant changes in PSD-95 levels or its interaction with actin might be attributed to the early stage of synaptic pathology, where synaptic loss has not yet advanced significantly. As F-actin loss progresses, it could disrupt the structural integrity and functional dynamics of the synapse. Consequently, the disruption of the PSD-95-actin interaction may be a downstream effect of synaptic loss and dysregulation in AD. These disruptions are further exacerbated by pathological changes at the synapse associated with plaque deposition. Therefore, the absence of early alterations in this interaction at one month suggests that such disruptions may emerge as a consequence rather than an initial indicator of synaptic dysfunction. Remarkably, synaptosomes prepared from human subjects with MCI and AD also exhibited disrupted interactions compared to NCI. In our previous study, we observed a gradation in the disruption of F-actin filaments between MCI and AD [7]. However, we did not observe a corresponding gradation in the disrupted interaction between these groups. This discrepancy may be attributed to the pattern of PSD-95 levels in these tissues, which did not differ significantly between MCI and AD, thus reflecting the disrupted interaction. Additionally, variability in tissue samples could contribute to these findings, as differences in sample quality, disease progression, and individual patient factors may obscure subtle gradations in protein interactions. These decreases in interaction in AD correlated significantly with impaired cognitive performance and inversely with pathological scores such as amyloid, and Braak staging. Importantly, similar disruptions in PSD-95 actin interaction have been observed in both AD mouse models and human subjects with MCI and AD. Brain tissues from AD mouse models and individuals with AD exhibit a decrease in PSD-95 levels (Fig. 1b, Additional file 1) [74, 75]. PSD-95, characterized by multiple protein structural domains, posttranslational modifications, and protein–protein interactions, mediates the synaptic localization and function of PSD-95 at excitatory synapses. Palmitoylation of PSD-95 at Cys3 and Cys5 near its N-terminus is critical for its accumulation at postsynaptic sites, facilitating the postsynaptic targeting of AMPA receptors [76,77,78]. Ephrin-B3 has emerged as a potential postsynaptic anchoring protein for PSD-95, as mutations in the C terminus of ephrin-B3 disrupt PSD-95 binding and affect both PSD-95 and ephrin-B3 postsynaptic localization [79, 80]. Additionally, α-actinin, an F-actin cross-linking protein, acts as a critical PSD-95 anchor regulating postsynaptic AMPA receptors, thus forming an α-actinin/PSD-95/AMPAR axis that docks to F-actin [80]. Therefore, the depolymerization of F-actin, which affects its interaction with PSD-95, may compromise the stability and localization of the PSD-95 scaffold, leading to synaptic dysfunction in AD.

Studies have demonstrated that actin and its regulatory proteins play a crucial role in memory formation and consolidation, as shown by impaired consolidation of contextual fear memory in mice infused with an F-actin depolymerizing agent [81, 82]. Our research demonstrated that F-actin stabilization with jasplakinolide reversed the memory recall deficit observed in middle-aged APP/PS1 mice, and also restored the PSD-95–actin association and synaptic F-actin levels. The restored interaction between PSD-95 and actin suggests that actin depolymerization is a contributing factor to the observed deficits, highlighting the importance of actin dynamics in learning and memory processes. These findings suggest that preserving actin dynamics and synaptic structure could mitigate synaptic dysfunction and ameliorate cognitive deficits associated with AD.

PSD-95 plays a pivotal role in synaptic plasticity by mediating the regulated delivery of AMPA-type glutamate receptors to synapses, which is essential for processes such as LTP and experience-driven synaptic strengthening [78]. Through PDZ domain interactions and downstream signaling modulation, PSD-95 controls activity-dependent AMPAR incorporation, influencing synaptic efficacy and plasticity in response to neural activity and environmental stimuli [83, 84]. PSD-95 plays a dual role in regulating synaptic NMDA receptors by stabilizing their surface expression through direct binding to GluN2B and promoting synaptic exclusion and degradation of the negative regulator STEP61, which reduces NMDAR surface expression by dephosphorylating GluN2B [85]. Conversely, Aβ triggers immediate Ca²⁺ influx through GluN2B-containing NMDARs, leading to impaired NMDAR-dependent LTP and reduced surface expression of NMDARs, ultimately impacting synaptic transmission and plasticity [86,87,88]. NMDAR subunit levels were reduced in the cortex and hippocampus of postmortem AD brains [89, 90]. A recent study suggested that PSD-95 overexpression actively protects synapses from Aβ toxicity and improves its interaction with the NMDAR C-terminal domain [91]. Our study revealed age-dependent alterations in AMPA and NMDA receptor subunits in APP/PS1 mice, with significant decreases in GluA1 and GluA2 levels at nine-months compared to those in WT mice. This reduction in AMPA receptor subunits coincided with disrupted association with PSD-95, suggesting impaired synaptic clustering and anchoring mechanisms in middle aged APP/PS1 mice. A decrease in these receptor levels and disrupted PSD-95 association may lead to synaptic destabilization, impaired neurotransmission, and synaptic plasticity deficits, contributing to cognitive impairment in AD. Conversely, changes in NMDA receptor subunits, particularly GluN1 and GluN2B, were evident as early as one month of age, indicating early disruption of the NMDA receptor composition in APP/PS1 mice. However, although the association between GluN1 and GluN2B with PSD-95 was initially enhanced at one month, this effect was not sustained at later ages, highlighting complex regulatory mechanisms underlying NMDA receptor interactions and synaptic stability in AD-related pathology. Furthermore, the preserved interactions and levels of GluN2A subunits observed in APP/PS1 mice may be ascertained to their involvement in protective pathways against Aβ-induced synaptic dysfunction, whereas pathways associated with GluN2B subunits may exacerbate neuronal susceptibility to such dysfunction [92]. Studies have shown that Aβ oligomers bind to synapses, leading to a reduction in the surface expression of GluN2B but not GluN2A [93]. Our results align with the findings that the synaptic modification threshold is shifted to favor LTP over LTD in young APP/PS1 mice, whereas adult APP/PS1 mice show reduced LTP and increased LTD compared to WT mice [94]. This could be attributed to the differences in AMPA levels and the impaired association with PSD-95 observed in middle-aged APP/PS1 mice.

F-actin stabilization in primary cortical neurons from APP/PS1 mice underscores the crucial role of actin in synaptic function. This F-actin stabilization restored the reduced levels of synaptic AMPA and NMDA receptors; reversed the decreased colocalization of PSD-95 with actin, AMPA, and NMDA receptors; and increased F-actin levels and dendritic spine density, emphasizing the significance of F-actin stabilization in promoting actin-PSD-95 colocalization and synaptic integrity. Furthermore, it led to increased levels of synaptic AMPA and NMDA receptors, likely enhancing synaptic transmission and plasticity.

In our FRAP experiments, we observed a slower recovery rate and reduced mobile fraction of actin when coexpressed with PSD-95, indicating an altered dynamics or stability of actin in the presence of PSD-95. Electron microscopy studies revealed that actin filaments can extend into the postsynaptic density to interact with proteins present in the PSD [15, 95, 96], providing a plausible explanation for the observed interaction with PSD-95. Furthermore, PSD-95 exhibited an increased recovery rate and mobile fraction with a shortened halftime of recovery in primary neurons from APP/PS1 mice compared to that of WT, suggesting altered dynamics in AD pathology. The increased mobile fraction could be attributed to amyloid beta-mediated F-actin loss or other interacting proteins at the synapse, reflecting synaptic dysfunction in AD. This finding is consistent with studies demonstrating a much slower recovery of fluorescence associated with scaffolding proteins, typically spanning tens of minutes [97, 98], as observed in the WT. Inferentially, these results imply that F-actin may play a role in stabilizing PSD-95 at the synapse, contributing to synaptic organization and function.

It is crucial to consider the regulation of actin polymerization in dendritic spines, specifically the roles of cofilin and its upstream regulator, LIMK1, for therapeutic strategies aimed at mitigating synaptic dysfunction and cognitive impairment in AD. Aberrant regulation of cofilin has been implicated in AD [99, 100], including the phosphorylation [101] and subsequent inactivation of cofilin-1 [102]. Studies have shown that cofilin-actin rods increase in postmortem AD brains, mouse models, and APP/PS1 neurons, leading to enhanced synaptic dysfunction and neurodegeneration [100, 103]. An increased phosphorylation of LIMK1 has been observed in AD pathology-rich regions of the cortex in human AD brains, suggesting that enhanced LIMK1 activity may promote cofilin phosphorylation and potentially promote actin polymerization [104]. Overexpression of LIMK1 in hippocampal excitatory neurons has been shown to rescue impaired long-term potentiation and improve social behavior in APP/PS1 mice [105], while LIMK1-deficient mice exhibit abnormal spine morphology and impaired learning and memory [106]. Conversely, other studies have demonstrated that the ROCK2-LIMK1 pathway mediates the detrimental effects of Aβ42 oligomers on dendritic spine degeneration and synaptotoxicity. Aβ-mediated dendritic spine degeneration in hippocampal neurons can be rescued by treatment with a LIMK1 inhibitor in hAPP mice. Pharmacologic inhibition of LIMK1 has been shown to confer dendritic spine resilience against amyloid-beta (Aβ) toxicity, indicating a protective role against Aβ-induced synaptic loss [107]. Moreover, cofilin phosphorylation has been identified in the postsynaptic compartment of excitatory synapses in primary neuronal cultures exposed to Aβ oligomers, with this phosphorylation alone being sufficient to induce Aβ oligomer-induced synaptic dysfunction, even before actin-cofilin rod formation [101]. Despite these controversies, recent studies suggest that enhancing the LIMK1-cofilin axis can improve memory in wild-type old mice, highlighting the importance of actin polymerization for synaptic structural plasticity and cognition [108]. Overall, the increase in cofilin phosphorylation observed in AD may initially represent a compensatory mechanism to restore disrupted actin polymerization. However, persistent dysregulation of the LIMK1-cofilin axis over time can exacerbate synaptic impairment and contribute to further cognitive decline. Considering these findings, a better understanding and careful consideration of the complex interplay between LIMK1, cofilin phosphorylation, and actin polymerization are required for therapeutic strategies aimed at promoting actin stabilization while preventing aberrant actin structures to optimize synaptic resilience and cognitive function in AD.

Conclusion

Our investigation of the synaptic actin interactome in AD provides valuable insights into the molecular mechanisms underlying synaptic dysfunction and cognitive decline in the disease. The dysregulation of actin dynamics and synaptic protein interactions represents a critical nexus in AD pathophysiology. Stabilization of F-actin using compounds such as jasplakinolide restores synaptic integrity and ameliorates cognitive deficits in AD models, highlighting the therapeutic potential of modulating cytoskeletal dynamics. Overall, our research underscores the critical importance of the PSD-95–actin interaction in regulating synaptic function, emphasizing its potential as a promising therapeutic target for alleviating cognitive deficits associated with Alzheimer’s disease.