Abstract
This narrative review aims to synthesize and critically evaluate the complex molecular mechanisms by which amyloid-β (Aβ) accumulation disrupts hippocampal synaptic plasticity, the cellular cornerstone of learning and memory in Alzheimer's disease (AD). AD is characterized by progressive hippocampus-dependent cognitive decline, strongly linked to impaired synaptic plasticity, the cellular basis of learning and memory. This review deciphers how Aβ accumulation orchestrates synaptic sabotage in the hippocampus. We detail the core molecular machinery of hippocampal synaptic plasticity, emphasizing glutamate receptor trafficking (NMDAR/AMPAR), Ca2+ signaling, and neurotrophin pathways (BDNF/TrkB). Central to AD pathogenesis, soluble Aβ oligomers initiate synaptic dysfunction by targeting receptors like PrPᶜ/mGluR5, triggering NMDAR overactivation (via Fyn/NR2B) and AMPAR endocytosis. Aβ further drives tau hyperphosphorylation (via GSK-3β/CDK5), leading to dendritic p-tau accumulation and destabilization of the postsynaptic density (PSD). Concurrently, Aβ activates microglia (via TLR4/TREM2) and astrocytes, promoting neuroinflammation (IL-1β, TNF-α, C1q) and complement-mediated synaptic phagocytosis. Aβ-induced oxidative stress (ROS/RNS, lipid peroxidation) and mitochondrial failure (mPTP opening, energy depletion) exacerbate Ca2+ dyshomeostasis and plasticity impairment. Critically, Aβ disrupts BDNF/TrkB signaling, promoting proBDNF/p75ᴺᵀᴿ-mediated spine loss and inhibiting CREB-dependent plasticity gene expression. These converging pathways—glutamate receptor dysregulation, p-tau toxicity, neuroinflammation, oxidative stress, mitochondrial dysfunction, and neurotrophic collapse—culminate in synaptic apoptosis, profound structural damage (spine loss, PSD dissolution), and functional deficits (long-term potentiation (LTP) blockade, enhanced long-term depression (LTD)). This molecular cascade directly underlies hippocampal circuit failure and cognitive decline in AD. Future research must address Aβ strain specificity, regional vulnerability, glial metabolic coupling, resilience mechanisms, and novel therapeutic strategies targeting these pathways.
Keywords
Introduction
Alzheimer's disease (AD) is a major global public health challenge and a progressive neurodegenerative disorder strongly linked to aging. Current estimates indicate over 55 million people live with dementia worldwide, a number expected to rise to 152 million by 2050, with AD accounting for 60–70% of cases. 1
The disease inflicts a devastating toll that extends far beyond the individual patient, placing an immense emotional, physical, and economic burden on families, caregivers, and healthcare systems. The total global cost of dementia, estimated to exceed one trillion US dollars annually, is a staggering financial burden that continues to rise, critically underscoring the urgent need for effective and cost-effective therapeutic strategies to alter the disease's relentless course.1,2 AD is marked by a progressive decline in cognitive function, most notably in episodic memory—the ability to recall personal experiences in detail. Early symptoms include forgetting recent events, misplacing items, missing appointments, and struggling to learn new information. Spatial disorientation, such as becoming lost in familiar places, is another key early sign, directly linked to damage in the hippocampus, a brain region essential for memory and navigation.3,4 The link between hippocampal integrity and memory function is one of the most robust correlations in clinical neuroscience. Neuroimaging studies consistently demonstrate significant volumetric loss, or atrophy, in the hippocampi of AD patients, and the degree of this shrinkage strongly correlates with the severity of memory impairment, firmly establishing the hippocampus as a critical site in the initial stages of the disease.3,5
The hippocampus does not store memories like a static repository; instead, it relies on a dynamic and malleable process known as synaptic plasticity to encode, consolidate, and retrieve information. Synaptic plasticity is the fundamental cellular mechanism by which experience modifies the strength and efficacy of communication between neurons. 6 It is widely accepted that the physical substrate of learning and memory resides in these activity-dependent changes in synaptic connections. The canonical experimental models for studying this phenomenon are LTP and LTD. The foundational concept of LTP as a cellular memory mechanism was first established decades ago, 7 and our understanding has since been profoundly refined, with comprehensive research elucidating the critical roles N-Methyl-D-aspartate receptors: NMDA Receptors (NMDARs) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) trafficking and signaling in these forms of synaptic plasticity.8,9 LTP is a persistent, activity-driven increase in synaptic strength, widely regarded as a primary cellular correlate for memory formation. Its counterpart, LTD, is a persistent decrease in synaptic strength, essential for memory refinement, the clearance of obsolete information, and maintaining overall synaptic homeostasis and stability. 10 The hippocampal circuit, particularly the classic trisynaptic pathway (perforant path → dentate gyrus → CA3 → CA1), has been the principal model system for elucidating the molecular mechanisms of LTP and LTD since their discovery. 11 The integrity of these plastic processes is paramount for cognitive health; when synaptic plasticity fails, the brain's capacity to learn, adapt, and form new memories disintegrates. 12 The central thesis of this narrative review is that the accumulation of amyloid-β (Aβ) peptides, a defining pathological hallmark of AD, orchestrates a multifaceted and convergent molecular sabotage of the exquisite machinery that underlies hippocampal synaptic plasticity. 13 This assault is not a singular event but a coordinated attack launched through several interconnected pathogenic streams that ultimately coalesce to disrupt synaptic function and trigger synaptic loss, laying the foundation for the cognitive collapse observed in AD. 14
The updated amyloid cascade hypothesis (ACH) proposes that the gradual buildup of Aβ, particularly Aβ42, and its conversion into soluble oligomers constitutes the initiating event in AD pathogenesis. 15 Although insoluble Aβ plaques are the most recognizable histopathological feature of the disease, current evidence shows that soluble Aβ oligomers—rather than plaque deposits—are the primary synaptotoxic species, consistent with and extending the synaptic failure hypothesis of AD.16,17 These oligomers are not inert bystanders but are pathologically active molecules that preferentially target and disrupt synaptic function long before the appearance of widespread plaques or overt neuronal death. 18 This review argues that Aβ oligomers impair hippocampal synaptic plasticity by disrupting glutamate receptor function through their interaction with cellular prion protein (PrPᶜ(and metabotropic glutamate receptor 5)mGluR5(. First, they directly induce AMPAR endocytosis, causing dysregulated calcium signaling and shifting the LTP/LTD balance toward synaptic depression, ultimately impairing memory formation. 19 Secondly, Aβ pathology initiates tau hyperphosphorylation. Normally, tau stabilizes axonal microtubules, which are critical for neuronal transport. In AD, hyperphosphorylation causes tau to misfolded and aggregate, contributing to neurodegeneration. 20
In AD, hyperphosphorylated tau detaches from microtubules and relocates to dendrites and spines. There, it disrupts the postsynaptic density by displacing scaffolding proteins such as postsynaptic density-95 (PSD-95), which in turn impairs the anchoring and trafficking of AMPARs- a process essential for synaptic strength and stability. 21 Third, Aβ triggers chronic neuroinflammation. It activates microglia via receptors like Toll-like receptor 4 (TLR4) and triggering receptor expressed on myeloid cells 2 (TREM2), leading to sustained release of pro-inflammatory cytokines (e.g., interleukin-1β (IL-1β) and tumor necrosis factor-alpha (TNF-α). This inflammatory response pathologically activates the complement cascade (e.g., C1q, C3), which tags synapses for elimination by microglial phagocytosis. This process, akin to aberrant adult synaptic pruning, results in widespread synapse loss. 22
Fourthly, the AD brain exists in a state of profound oxidative stress, a process comprehensively linked to Aβ accumulation and bioenergetics deficits that damages diverse cellular components. 23 Fourth, Aβ complexes with redox-active metals (copper, iron), catalyzing Fenton chemistry that generates reactive oxygen species (ROS) and reactive nitrogen species (RNS). This oxidative assault damages lipids (leading to the production of neurotoxic aldehydes like 4-hydroxynonenal, 4-HNE), proteins, and DNA, and impairs plasticity-related enzymes. This damage is closely linked to a fifth mechanism: mitochondrial failure. Notably, oxidative damage alone—even without amyloid pathology—is sufficient to cause learning and memory deficits, underscoring its central role in cognitive decline. 24
Fifth, mitochondrial dysfunction critically undermines synaptic energy supply as synapses require substantial adenosine triphosphate (ATP) for plasticity, primarily generated by local mitochondria. Aβ oligomers directly target mitochondria, disrupting the electron transport chain and impairing ATP production. This effect is worsened by altered communication at the mitochondria-associated endoplasmic reticulum (ER) membrane (MAM), further destabilizing neuronal energy balance and calcium signaling in AD. 25 These impairments lead to excessive ROS production and promote the opening of the mitochondrial permeability transition pore (mPTP)—an event that triggers a collapse of the mitochondrial membrane potential, catastrophic energy depletion, and the initiation of apoptotic signaling pathways, ultimately compromising the synaptic energy budget essential for plasticity. 26 Finally, Aβ disrupts neurotrophic support, particularly the brain-derived neurotrophic factor (BDNF)/its high-affinity receptor, Tropomyosin receptor kinase B (TrkB) signaling pathway. This system is crucial for synaptic plasticity, neuronal survival, and memory consolidation. Aβ suppresses both BDNF expression and TrkB signaling, effectively disabling this vital support mechanism. Given its central role, enhancing BDNF signaling is a promising therapeutic strategy for AD and related neurodegenerative conditions.27,28
This review addresses how Aβ peptides disrupt hippocampal synaptic plasticity through six interconnected pathways: 1) glutamate receptor dysregulation, 2) tau pathology and synaptic mislocalization, 3) neuroinflammatory synapse elimination, 4) oxidative stress and mitochondrial failure, 5) neurotrophic support collapse, and 6) activation of synaptic apoptotic pathways. By integrating molecular, cellular, and systems-level evidence, we link these mechanisms to circuit-level decay and cognitive symptoms. The synthesis also identifies key knowledge gaps—such as Aβ strain specificity, regional vulnerability, glial metabolic roles, and endogenous resilience—and discusses their therapeutic implications for preserving synapses and cognition in AD.
Literature search methodology
This narrative review synthesizes evidence from a broad body of literature to construct a coherent mechanistic framework. To ensure a comprehensive and up-to-date analysis, a systematic search strategy was employed. Primary literature searches were conducted in the PubMed/MEDLINE and Google Scholar databases from their inception through March 2024. Key search terms and their combinations included: “Alzheimer's disease,” “amyloid-beta,” “Aβ oligomers,” “hippocampus,” “synaptic plasticity,” “LTP,” “LTD,” “NMDA receptor,” “AMPA receptor,” “tau,” “neuroinflammation,” “microglia,” “astrocyte,” “complement,” “oxidative stress,” “mitochondrial dysfunction,” “BDNF,” “TrkB,” and “apoptosis.” The reference lists of identified key articles were also manually screened for additional relevant studies.
Given the narrative and integrative nature of this review, formal a priori inclusion/exclusion criteria were not defined in the manner of a systematic review. Instead, article selection was driven by the review's objective to decode converging molecular pathways. Priority was given to: (1) original research articles (in vitro, in vivo) elucidating mechanistic links between Aβ and synaptic dysfunction; (2) seminal reviews and hypothesis papers that have shaped the field's understanding; and (3) clinical studies correlating molecular findings with cognitive metrics or neuropathology. Literature not in English, purely descriptive, or unrelated to AD pathogenesis was excluded. The synthesis provides an interpretive analysis of the selected evidence to build a coherent mechanistic framework.
Hippocampal synaptic plasticity: core molecular machinery
The hippocampus, a seahorse-shaped structure nestled deep within the medial temporal lobe, stands as the neural epicenter for the formation of declarative memories, particularly episodic memories that define our personal experiences and spatial memories that enable navigation through our environment. 4 The capacity to form and retain memories depends on the dynamic adaptability of hippocampal synapses, a process known as synaptic plasticity. This refers to the activity-dependent strengthening or weakening of connections between neurons, which serves as the fundamental cellular basis for learning and memory. The two most widely studied forms are LTP, a persistent increase in synaptic efficacy after high-frequency stimulation, and LTD, a persistent decrease following low-frequency stimulation. 7 These opposing processes allow for the refinement of neural circuits, the encoding of new information, and the updating of stored memories.29,30 Understanding the intricate molecular machinery underpinning hippocampal synaptic plasticity is not merely an academic exercise; it is an absolute prerequisite for deciphering how this exquisitely tuned system is dismantled in AD by Aβ accumulation.29,30 This section delves into the essential molecular pathways and components that orchestrate the induction, expression, and maintenance of synaptic plasticity within hippocampal circuits, focusing on the glutamatergic synapses of the trisynaptic pathway (entorhinal cortex → dentate gyrus → CA3 → CA1 → subiculum), which are particularly vulnerable in AD.31,32
Glutamate, the primary excitatory neurotransmitter in the mammalian hippocampus acts upon two major classes of ionotropic receptors at hippocampal synapses: NMDARs and AMPARs. Their precise regulation and trafficking constitute the cornerstone of synaptic plasticity. NMDARs function as coincidence detectors, pivotal for the induction of both LTP and LTD. Their activation requires glutamate binding, a co-agonist (glycine/D-serine), and sufficient postsynaptic depolarization to remove the Mg2+ block from the channel pore. 33 This unique property allows NMDARs to sense correlated pre- and postsynaptic activity, triggering Ca2+ influx specifically when the postsynaptic neuron is sufficiently depolarized, as occurs during high-frequency firing patterns associated with learning. The subunit composition of NMDARs significantly influences their functional properties. While typically heterotetrameric assemblies of two obligatory GluN1 subunits and two regulatory subunits (GluN2A-D or GluN3A-B), the GluN2 subunits, particularly GluN2A and GluN2B, dictate channel kinetics, Mg2+ sensitivity, and intracellular signaling complex formation. 34 GluN2B-containing receptors exhibit slower decay kinetics, higher Ca2+ permeability, and stronger interactions with scaffolding proteins linking them to downstream signaling cascades compared to GluN2A-containing receptors. The developmental switch from predominantly GluN2B to GluN2A subunits and the dynamic regulation of subunit composition in adulthood are crucial for shaping plasticity thresholds and metaplasticity—the ability of synapses to modify their future plasticity based on prior activity. 35 In contrast, AMPARs (composed of GluA1-GluA4 subunits) mediate fast excitatory transmission and are central to expressing changes in synaptic strength. Notably, GluA2-lacking AMPARs are calcium-permeable. 36 The trafficking of AMPARs into the synapse is a dynamic process essential for plasticity. During LTP, GluA1-containing AMPARs are rapidly inserted into the postsynaptic membrane, increasing synaptic strength. 37 This is facilitated by phosphorylation of GluA1 (e.g., at Ser831 by CaMKII/PKC and Ser845 by PKA), which promote receptor insertion and stabilization. 38
Conversely, LTD expression primarily involves the clathrin-dependent endocytosis of AMPARs, particularly GluA2/GluA3-containing receptors, mediated by dephosphorylation events and interactions with specific endocytic proteins. 39 The constant cycling of AMPARs under basal conditions and their activity-regulated insertion or removal ensure that synaptic strength can be rapidly and persistently modified in response to experience.
The influx of Ca2+ through NMDARs and, in specific contexts, voltage-gated calcium channels (VGCCs) or Ca2+-permeable AMPARs, serves as the critical second messenger triggering the biochemical cascades underlying synaptic plasticity. The amplitude, duration, and spatial localization of the Ca2+ signal is exquisitely decoded by a network of Ca2+-binding proteins and effector enzymes, determining whether LTP or LTD ensues. 9 Large, rapid calcium surges during LTP induction strongly activate calcium/calmodulin-dependent protein kinase II (CaMKII). In the hippocampus, CaMKII exists as a multi-subunit holoenzyme (mainly α and β isoforms). When bound by Ca2+/calmodulin, it autophosphorylates at Thr286 (α-isoform), allowing it to remain active independently of calcium.40,41 Autophosphorylated CaMKII translocates to the PSD where it phosphorylates key substrates, including the AMPAR subunit GluA1 at Ser831 (enhancing channel conductance) and the scaffolding protein SAP97 (promoting GluA1 insertion), thereby stabilizing its active state in the PSD nanodomain. This persistent CaMKII activity is considered a molecular memory trace essential for the early phase of LTP and the stabilization of synaptic potentiation. 42 Conversely, moderate or prolonged Ca2+ elevations, often associated with LTD-inducing protocols, preferentially activate the Ca2+/calmodulin-dependent phosphatase calcineurin. Activated calcineurin dephosphorylates inhibitor-1 (I-1), relieving its inhibition of protein phosphatase 1 (PP1). 43 Active PP1 then dephosphorylates key substrates, including GluA1 at Ser845 (promoting AMPAR endocytosis) and the transcription factor cAMP response element-binding protein (CREB), contributing to synaptic weakening and gene expression changes associated with LTD. The balance between CaMKII (kinase) and calcineurin/PP1 (phosphatase) activities is thus a fundamental determinant of the direction of synaptic plasticity, acting as a molecular switch sensitive to the precise characteristics of the Ca2+ transient. 44
Beyond the core CaMKII-calcineurin axis, a constellation of plasticity-related kinase signaling pathways integrates synaptic activity with long-term changes in gene expression and protein synthesis, crucial for the late phases of LTP and LTD consolidation. The extracellular signal-regulated kinase/mitogen-activated protein kinase (ERK/MAPK) pathway is a major conduit linking NMDAR and Ca2+ signals to nuclear events. Ca2+ influx can activate Ras small GTPases (e.g., Ras, Rap1/2) through RasGRF (Ras Guanine nucleotide Releasing Factor), leading to the sequential phosphorylation and activation of Raf, MEK (MAPK/ERK kinase), and finally ERK1/2. 45 Active ERK translocates to the nucleus where it phosphorylates transcription factors like Elk-1, which associates with serum response factor (SRF) to regulate immediate early genes (IEGs) such as c-Fos and Zif268, and CREB (at Ser133), a master regulator of activity-dependent gene expression. 46 Phosphorylated CREB (pCREB) recruits the coactivator CBP (CREB-binding protein) to stimulate the transcription of plasticity-related genes (PRGs) including BDNF, Arc, and cpg15, which encode proteins essential for synaptic growth, strengthening, and stabilization during late-LTP. 47 The phosphoinositide 3-kinase (PI3 K)/Akt (Protein Kinase B) pathway represents another critical signaling cascade activated by synaptic activity and neurotrophins. PI3 K is recruited to the membrane by receptor tyrosine kinases (e.g., TrkB) or adhesion molecules (e.g., integrins), generating phosphatidylinositol (34,5)-trisphosphate (PIP3). PIP3 then recruits Akt to the membrane where it is phosphorylated and activated by PDK1 and mTORC2. 48 Active Akt promotes neuronal survival, growth, and plasticity by phosphorylating and inhibiting pro-apoptotic factors (e.g., Bad, GSK-3β), activating mTORC1 (mammalian Target Of Rapamycin Complex 1), and regulating glucose metabolism. mTORC1 activation stimulates cap-dependent protein synthesis by phosphorylating key regulators like 4E-BP (Eukaryotic translation initiation factor 4E-Binding Protein) and S6K1 (S6 Kinase 1), leading to the local translation of specific mRNAs at synapses, a process indispensable for the maintenance of LTP and synaptic structural changes. 49 These kinase pathways (ERK/MAPK, PI3 K/Akt/mTOR) thus translate transient synaptic signals into sustained functional and structural alterations, bridging the gap between synaptic activation and nuclear gene transcription or local protein synthesis.
The molecular complexity and precision required for synaptic plasticity necessitate a highly organized structural framework. This is provided by a dense network of scaffolding, adaptor, and cytoskeletal proteins within the PSD, a specialized electron-dense structure opposed to the presynaptic active zone. PSD-95, a member of the membrane-associated guanylate kinase (MAGUK) family, is a central organizer of the PSD. It contains three PDZ domains that bind directly to the C-termini of NMDARs (GluN2 subunits) and certain potassium channels, as well as to other scaffolding proteins. 50
Through its guanylate kinase (GK) and SH3 domains, PSD-95 interacts with a multitude of signaling molecules, including SynGAP (Synaptic GTPase Activating Protein), neuronal nitric oxide synthase (nNOS), and adhesion molecules like neuroligins, which in turn bind presynaptic neurexins, contributing to trans-synaptic adhesion and alignment. PSD-95 also clusters AMPARs indirectly via transmembrane AMPA receptor regulatory proteins (TARPs) or directly via auxiliary subunits like CKAMP44, and it links receptors to the actin cytoskeleton via proteins like SPAR and IRSp53. 51 Shank proteins (Shank1-3) are large scaffolding proteins that occupy a deeper layer within the PSD. They act as master organizers, binding directly to Homer proteins via their proline-rich domain and to GKAP (Guanylate Kinase-Associated Protein, which binds PSD-95) via their PDZ domain, thereby forming a critical PSD-95/GKAP/Shank/Homer superstructure. 50 The Homer proteins (Homer1-3), particularly the long isoforms, are adaptor proteins that bind to Shank and also to group 1 metabotropic glutamate receptors (mGluR1/5) via an EVH1 domain. This physical linkage between ionotropic (NMDARs/AMPARs via PSD-95/Shank) and metabotropic (mGluR5 via Homer) glutamate receptors allows for the coordinated signaling necessary for complex forms of plasticity. Furthermore, Shank proteins provide a platform for actin regulatory proteins (e.g., cortactin, Abp1, α-fodrin) and link to the ER via Homer and inositol trisphosphate receptors (IP3Rs), facilitating Ca2+-induced Ca2+ release (CICR) and localized Ca2+ signaling microdomains crucial for plasticity induction.52,53 This intricate PSD nano-architecture ensures that receptors, signaling enzymes, cytoskeletal elements, and organelles are precisely positioned for rapid, efficient, and localized signal transduction in response to synaptic activity.
Finally, neurotrophic factors, particularly BDNF, play a pivotal role in modulating hippocampal synaptic plasticity, influencing both functional changes and structural rearrangements. BDNF signals primarily through its high-affinity receptor tyrosine kinase, TrkB. Activity-dependent release of BDNF, often triggered by Ca2+ influx through NMDARs or VGCCs, leads to TrkB dimerization and autophosphorylation on tyrosine residues within its intracellular domain. 48 This recruits adaptor proteins (Shc, Grb2) and enzymes (PI3 K, Phospholipase C gamma (PLCγ)) to specific phospho-tyrosine sites, activating the major signaling pathways mentioned earlier: PI3 K/Akt, Ras/ERK/ MAPK, and PLCγ. PLCγ activation is particularly important for acute synaptic effects. PLCγ hydrolyzes phosphatidylinositol 4,5-bisphosphate (PIP2) to generate inositol trisphosphate (IP3) and diacylglycerol (DAG). IP3 binds to IP3 receptors on the ER, releasing intracellular Ca2+ stores, while DAG activates specific isoforms of protein kinase C (PKC). 54 This PLCγ-PKC pathway contributes to neurotransmitter release enhancement, NMDAR potentiation, and activation of the translation regulator eEF2 kinase. Critically, BDNF-TrkB signaling strongly activates the CREB transcription factor, primarily via the ERK/ MAPK and CaMKIV pathways. pCREB drives the transcription of a wide array of PRGs, including its own ligand Bdnf, creating a positive feedback loop that sustains plasticity-related changes. 55 Furthermore, BDNF-TrkB signaling potently stimulates the mTORC1 pathway, primarily via PI3 K/Akt, leading to the phosphorylation of 4E-BP and S6 K, thereby enhancing cap-dependent translation initiation and protein synthesis necessary for late-LTP consolidation and synaptic growth. 56 BDNF also promotes actin cytoskeleton remodeling within dendritic spines through Rho GTPase regulation (e.g., Rac1 activation), facilitating spine enlargement and stabilization associated with LTP. 57 The interplay between BDNF signaling and the core plasticity machinery (glutamate receptors, Ca2+ signaling, kinases, PSD scaffolds) underscores its role as a potent modulator and permissive factor for enduring synaptic change. Deficits in BDNF signaling, as occur in AD, therefore represent a critical point of failure in the plasticity apparatus. 58
In summary, hippocampal synaptic plasticity is governed by a highly sophisticated and interconnected molecular machinery. Glutamate receptor trafficking, particularly the subunit-specific regulation and dynamic movement of NMDARs and AMPARs, dictates synaptic strength and the threshold for plasticity induction. The ensuing Ca2+ influx is decoded by sensor proteins like CaMKII and calcineurin, acting as a molecular switch between LTP and LTD pathways.59,60 Downstream kinase cascades (ERK/MAPK, PI3 K/Akt/mTOR) translate synaptic signals into long-term changes via nuclear gene transcription and local protein synthesis. This complex signaling occurs within a meticulously organized structural framework provided by PSD scaffolding proteins (PSD-95, Shank, Homer) that cluster receptors, anchor signaling complexes, link to the actin cytoskeleton, and integrate with intracellular organelles. Finally, neurotrophin signaling, particularly through BDNF and TrkB, powerfully modulates these processes, enhancing synaptic efficacy, promoting structural changes, and driving the gene expression and protein synthesis essential for lasting plasticity.60,61 This intricate symphony of molecules and pathways forms the resilient yet vulnerable foundation upon which learning and memory are built – a foundation systematically sabotaged by the pathological processes initiated by Aβ accumulation in AD (Table 1).
Core hippocampal synaptic plasticity machinery.
Aβ accumulation: initiating synaptic pathogenesis
AD is fundamentally characterized by the insidious accumulation of Aβ peptides, a pathological process that instigates a cascade of molecular events culminating in synaptic dysfunction within the hippocampus. This hippocampal vulnerability is paramount, as this brain region orchestrates learning, memory consolidation, and spatial navigation—functions severely impaired in AD. The journey from soluble Aβ monomers to pathological aggregates, particularly soluble oligomers, represents the critical initiating event in synaptic pathogenesis. 17 These oligomers exhibit a sinister tropism for synapses, binding with high affinity to specific neuronal surface receptors and triggering a domino effect of intracellular disruptions. This section delves into the molecular intricacies of four primary Aβ-driven insults—oligomerization, tau hyperphosphorylation, neuroinflammation, and oxidative stress—that collectively sabotage hippocampal synaptic plasticity, the cellular foundation of cognition.
The genesis of synaptic toxicity lies in the misfolding and aggregation of Aβ, primarily the amino Aβ42, which displays a heightened propensity for oligomerization compared to its Aβ40 counterpart. Aβ monomers, derived from the sequential proteolytic cleavage of the amyloid-β protein precursor (AβPP) by β- and γ-secretases, undergo conformational changes that favor the formation of soluble, non-fibrillar oligomers. These Aβ oligomers, ranging from dimers to dodecamers such as the pathogenic Aβ*56 species, are now recognized as the primary neurotoxic species, exerting their deleterious effects long before the appearance of insoluble plaques.17,18,64 Their potency stems from their ability to diffuse freely within the extracellular space and bind with exquisite specificity to synaptic membranes
The interaction of Aβ oligomers with synaptic receptors turns the synapse into a site of pathological signaling, disrupting the balance necessary for plasticity. Excitatory synapses, particularly those enriched in NMDARs and AMPARs, are especially vulnerable, explaining early deficits in hippocampal circuits involved in episodic and spatial memory. Aberrant signaling leads to AMPAR internalization through an LTD-like endocytic pathway, triggered by Ca2+ overload and phosphatase activation, resulting in reduced synaptic strength and impaired LTP, undermining the cellular foundation of memory. 70
Concurrently, Aβ oligomers induce the pathological transformation of tau, a microtubule-associated protein critical for axonal transport and neuronal stability. Under normal conditions, tau stabilizes microtubules, enabling efficient trafficking of organelles, vesicles, and receptors along neuronal processes. However, Aβ oligomers trigger a cascade that hyperphosphorylates tau, altering its function and localization, primarily via dysregulation of key kinases, including glycogen synthase kinase-3β (GSK-3β) and cyclin-dependent kinase 5 (CDK5).71,72
Aβ oligomers activate GSK-3β through mechanisms including Wnt pathway disruption and NMDA receptor–mediated Ca2+ influx, which cleaves the GSK-3β inhibitor p35 into the more stable activator p25, subsequently hyperactivating CDK5. 72 Hyperactive GSK-3β and CDK5 phosphorylate tau at pathological sites such as serine 202 and threonine 205 (AT8 antibody), reducing its microtubule affinity and causing detachment and microtubule destabilization. 73 Hyperphosphorylated tau undergoes missorting, aberrantly accumulating in the somatodendritic compartment and dendritic spines, 74 displacing scaffolding proteins like PSD-95 and impairing AMPAR trafficking by retaining GluA1 subunits in the Golgi apparatus, leading to synaptic dysfunction.21,75 Furthermore, specific phosphorylation events, such as at serine 396, render tau susceptible to caspase cleavage, generating toxic fragments that further destabilize the actin cytoskeleton within spines, contributing to spine loss. 76 Thus, Aβ initiates a tauopathy that directly dismantles the synaptic machinery.
The synaptic toxicity induced by Aβ is amplified by activation of microglia, leading to chronic neuroinflammation. Aβ oligomers act as damage-associated molecular patterns (DAMPs) engaging Pattern Recognition Receptors (PRRs) such as TLR4 and TREM2.77,78 Binding of Aβ oligomers to TLR4 initiates a pro-inflammatory signaling cascade via myeloid differentiation primary response 88 (MyD88), culminating in the activation of nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB). This transcription factor drives the expression of potent pro-inflammatory cytokines like IL-1β and TNF-α. 68 While TREM2 can enhance Aβ clearance, sustained engagement by Aβ aggregates, particularly with AD risk variants like R47H, promotes maladaptive inflammation and impaired phagocytosis. 78 Activated microglia undergo morphological changes (becoming amoeboid) and release a barrage of inflammatory mediators. IL-1β and TNF-α are particularly detrimental to synapses. IL-1β signaling through its receptor (IL-1R) on astrocytes can downregulate the expression of the glutamate transporter EAAT2, leading to impaired glutamate reuptake, synaptic glutamate spillover, and excitotoxic NMDAR overactivation. 79 TNF-α, signaling primarily through TNF receptor 1 (TNFR1), activates pathways that enhance the activity of STEP. STEP dephosphorylates key tyrosine residues on the GluA2 subunit of AMPARs and the NR2B subunit of NMDARs, promoting receptor endocytosis and synaptic weakening. 80 Furthermore, activated microglia release complement proteins, notably C1q, which initiates the classical complement cascade. C1q binds to synapses, tagging them for elimination via opsonization. This leads to the cleavage of C3 to C3b, which binds to complement receptor 3 (CR3, CD11b/CD18) on microglia, signaling them to phagocytose the tagged synapses—a process termed synaptic pruning gone awry. This complement-mediated synaptic loss is particularly prominent in the hippocampus during early AD stages. 81
A final, pervasive mechanism by which Aβ accumulation initiates synaptic pathogenesis is through the generation of oxidative stress. Aβ peptides, particularly oligomeric Aβ-42, possess a high affinity for transition metals like copper (Cu) and iron (Fe). These metals bind to the histidine residues within Aβ, forming Aβ-metal complexes. 82 These complexes are redox-active, meaning they can cycle between different oxidation states (e.g., Cu+ ↔ Cu2+, Fe2+ ↔ Fe3+) and catalyze the production of ROS through Fenton and Haber-Weiss reactions. In the Fenton reaction, hydrogen peroxide (H2O2), which can be generated by mitochondrial leakage or enzymatic activity, reacts with reduced metals (Fe2+ or Cu+) bound to Aβ, yielding highly reactive hydroxyl radicals (•OH). These hydroxyl radicals are indiscriminate in their attack, causing extensive damage to cellular macromolecules. A primary target is lipids within neuronal and synaptic membranes. Lipid peroxidation occurs when hydroxyl radicals abstract hydrogen atoms from polyunsaturated fatty acids (PUFAs), initiating a chain reaction that propagates membrane damage.83,84 A key end-product of this peroxidation is 4-HNE, a highly reactive aldehyde. 4-HNE readily forms covalent adducts with proteins, including critical synaptic proteins like glutamate transporters (EAATs), ionotropic glutamate receptors (NMDARs, AMPARs), and enzymes involved in energy metabolism and antioxidant defense. 85 Furthermore, lipid peroxidation and 4-HNE adduct formation can impair the function of glucose transporters in neurons, contributing to a bioenergetic deficit that exacerbates synaptic failure. 86 These modifications impair protein function, leading to synaptic dysfunction. For instance, 4-HNE modification of NMDARs can alter their gating properties and Ca2+ permeability, exacerbating Ca2+ dyshomeostasis. Oxidative stress also targets proteins directly through carbonylation, an irreversible modification where reactive carbonyl groups (e.g., aldehydes, ketones) are introduced into amino acid side chains (lysine, arginine, proline, threonine), often rendering the protein dysfunctional. 87 Synaptic proteins involved in plasticity, such as CaMKII, are highly susceptible to oxidative inactivation, directly impairing LTP. Furthermore, ROS damage mitochondrial DNA and proteins, impairing the electron transport chain and reducing ATP production. 88 This energy crisis compromises ATP-dependent processes crucial for synaptic function, including neurotransmitter vesicle loading, receptor trafficking, and actin cytoskeleton dynamics underlying spine morphology. The resulting mitochondrial dysfunction creates a vicious cycle, generating more ROS and further amplifying oxidative stress.
In summary, Aβ accumulation, particularly in the form of soluble oligomers, is the seminal event that launches a multi-pronged attack on hippocampal synaptic integrity. The oligomers themselves directly target synaptic receptors like PrPᶜ and mGluR5, initiating aberrant signaling that dysregulates Ca2+ and promotes glutamate receptor internalization. Simultaneously, Aβ sets tau pathology in motion by hyperactivating kinases GSK-3β and CDK5, leading to tau hyperphosphorylation and its toxic mislocalization to dendrites, where it disrupts synaptic scaffolding and receptor trafficking. Aβ also functions as a potent DAMP, activating microglia via TLR4 and TREM2, resulting in a chronic neuroinflammatory milieu characterized by the release of cytokines (IL-1β, TNF-α) and complement proteins (C1q).68,81 These inflammatory mediators induce excitotoxicity, promote synaptic receptor endocytosis via phosphatases like STEP, and tag synapses for phagocytic elimination. Finally, Aβ-bound redox-active metals catalyze the production of devastating ROS via Fenton chemistry, driving widespread oxidative damage through lipid peroxidation (yielding 4-HNE) and protein carbonylation. This oxidative assault directly damages synaptic proteins, impairs mitochondrial function, and depletes energy reserves. These four interconnected pathways—oligomer-mediated receptor binding, tauopathy, neuroinflammation, and oxidative stress—act synergistically and sequentially to dismantle the molecular machinery of hippocampal synaptic plasticity long before overt neuronal death occurs. This intricate molecular sabotage initiated by Aβ accumulation fundamentally undermines the hippocampus's ability to encode, store, and retrieve memories, marking the onset of the cognitive decline that defines AD (Table 2).
Aβ oligomerization and synaptic targeting.
Molecular mechanisms of plasticity disruption
Aβ oligomer-mediated glutamate receptor dysfunction
Building upon the foundational insults instigated by Aβ accumulation within the hippocampus, as delineated in the preceding section, the direct and profound sabotage of synaptic plasticity mechanisms emerges as a critical nexus in AD pathogenesis. Central to this disruption is the specific targeting and dysregulation of the glutamatergic system—the primary excitatory neurotransmitter system underpinning synaptic plasticity, learning, and memory. Soluble Aβ oligomers, recognized as the principal neurotoxic species, 18 execute a sophisticated molecular attack on both NMDARs and AMPARs, the key mediators of activity-dependent synaptic strengthening (LTP) and weakening (LTD). 70 This section dissects the intricate cascade of events triggered by Aβ oligomers at the synapse. We focus on three core mechanisms: 1) NMDAR overactivation leading to calcium toxicity, 2) AMPAR internalization driven by aberrant phosphatase signaling, and 3) pathological facilitation of LTD through NMDAR trafficking alterations. Collectively, these processes dismantle the hippocampus's capacity for adaptive change (Table 3).
Aβ oligomer-mediated glutamate receptor dysfunction.
The assault commences with the highly specific synaptic targeting of Aβ oligomers, primarily mediated through their high-affinity binding to the PrPᶜ. This interaction is not merely passive anchoring; it initiates a conformational change in PrPᶜ that serves as a platform for recruiting and activating key intracellular signaling molecules. Crucially, Aβ oligomer binding to the PrPᶜ complex, which often involves co-receptors like mGluR5, leads to the robust activation of the Src-family tyrosine kinase Fyn. 19 Fyn activation is a pivotal event in Aβ-mediated synaptic toxicity. Once activated, Fyn phosphorylates specific tyrosine residues on the cytoplasmic tail of the NR2B subunit of the NMDAR. 102 This Fyn-mediated tyrosine phosphorylation potentiates NMDAR function by increasing channel open probability, prolonging channel open time, and reducing magnesium block sensitivity at resting membrane potentials led to enhance NMDAR currents and Ca2+ influx.90,102,103
This aberrant potentiation transforms physiological NMDAR signaling into a pathological state. Under normal conditions, NMDAR activation, particularly during high-frequency stimulation typical of LTP induction, allows a controlled influx of calcium ions (Ca2+), which acts as a critical second messenger activating downstream plasticity kinases like CaMKII and initiating gene transcription programs. However, Aβ oligomer-induced Fyn-mediated NR2B hyperphosphorylation results in sustained and excessive Ca2+ influx through NMDARs, even during basal synaptic activity or sub-threshold stimulation that would normally be insufficient to trigger significant Ca2+ entry. 90 This chronic NMDAR overactivation and the consequent Ca2+ overload create a state of excitotoxicity within the synapse. The excessive intracellular Ca2+ overwhelms the neuron's buffering systems (e.g., ER uptake, mitochondria, calcium-binding proteins) and activates a plethora of Ca2+-dependent enzymes with detrimental consequences. Furthermore, Aβ oligomers induce neuronal oxidative stress through NMDAR-dependent mechanisms that exacerbate mitochondrial dysfunction and contribute to synaptic failure. 99 These include proteases like calpains, which degrade critical synaptic proteins such as spectrin and PSD-95; phosphatases like calcineurin, which counter synaptic strengthening pathways; and nitric oxide synthases, generating RNS contributing to oxidative stress. Furthermore, sustained high intracellular Ca2+ can trigger mPTP opening, leading to mitochondrial swelling, rupture, and the release of pro-apoptotic factors, initiating a cascade towards synaptic failure and neuronal death. 100 Thus, the Aβ-PrPᶜ-Fyn-NR2B axis effectively hijacks the NMDAR, converting a key mediator of synaptic plasticity into a potent engine of synaptic degeneration and neuronal dysfunction through uncontrolled Ca2+ influx.
Concurrently, Aβ oligomers exploit another major glutamate receptor pathway – the mGluR5 – to instigate the removal of AMPARs from the PSD, thereby weakening synaptic strength and impairing LTP. Aβ oligomers directly or indirectly engage mGluR5 signaling. Binding to the PrPᶜ-mGluR5 complex or potentially interacting with mGluR5 independently, Aβ oligomers trigger sustained activation of Gq/11 proteins coupled to mGluR5. 67 This activation leads to the stimulation of phospholipase Cβ (PLCβ), which hydrolyzes PIP2 into IP3 and DAG. While IP3 contributes to intracellular Ca2+ release from the ER, amplifying the Ca2+ dyshomeostasis initiated by NMDAR overactivation, the DAG pathway activates PKC. Crucially, persistent mGluR5 signaling downstream of Aβ oligomers strongly activates the STEP.104,105 STEP is a brain-specific phosphatase that antagonizes tyrosine kinase signaling and is a potent negative regulator of synaptic strength. Aβ-mediated mGluR5 activation leads to STEP upregulation at both the transcriptional and post-translational levels. Increased STEP activity specifically targets AMPARs. STEP dephosphorylates the GluA2 subunit of AMPARs at Tyr876, a site critical for stabilizing AMPARs at the synapse by regulating interactions with scaffolding proteins. More significantly in the context of Aβ-induced plasticity disruption, STEP dephosphorylates the GluA1 subunit at Ser845.38,98 Phosphorylation of GluA1 at Ser845, primarily by PKA, is essential for promoting AMPAR exocytosis, anchoring AMPARs at the synapse, and facilitating LTP by increasing AMPAR single-channel conductance and synaptic incorporation. Dephosphorylation of GluA1 Ser845 by activated STEP disrupts these processes. It promotes the endocytosis of GluA1-containing AMPARs by facilitating the binding of adaptor proteins like AP2 to the receptor's cytoplasmic tail, leading to clathrin-coated pit formation and internalization. 93 This internalization is accompanied by broader disruption of synaptic architecture, as critical scaffolding proteins like PSD-95 undergo significant reorganization, leading to disintegration of receptor signaling complexes and further compromising synaptic integrity. 101 The internalized AMPARs are then targeted for degradation or recycling, effectively reducing the number of functional AMPARs available at the synapse. This STEP-mediated AMPAR endocytosis results in a net weakening of synaptic transmission. It directly impairs the expression and maintenance of LTP, which relies heavily on the synaptic insertion and stabilization of GluA1-containing AMPARs phosphorylated at Ser845.38,106 Consequently, Aβ oligomers, via mGluR5-driven STEP activation, dismantle the molecular machinery responsible for strengthening synapses in response to learning-relevant activity, contributing significantly to the early cognitive deficits observed in AD models and patients.
The deleterious effects of Aβ oligomers on glutamate receptors extend beyond the potentiation of NMDARs and the internalization of AMPARs; they also actively promote synaptic weakening by facilitating LTD, a process physiologically involved in forgetting and synaptic refinement, but pathologically exaggerated in AD. Aβ oligomers create a synaptic environment highly permissive to LTD induction and maintenance through several converging mechanisms. Firstly, the chronic NMDAR overactivation and Ca2+ overload discussed earlier preferentially activate protein phosphatase 2B)PP2B(and PP1. 95 Calcineurin, activated by sustained moderate elevations in intracellular Ca2+, dephosphorylates inhibitor-1 (I-1), relieving its inhibition of PP1. Active PP1 then dephosphorylates key synaptic proteins, including GluA1 at Ser845 (acting in concert with STEP) and other substrates necessary for maintaining synaptic strength. 43 This phosphatase dominance directly facilitates the molecular cascade required for LTD, which involves AMPAR endocytosis. Secondly, Aβ oligomers induce the internalization of NMDARs themselves, particularly those containing the NR2B subunit. 96 While Aβ initially potentiates NR2B-containing NMDARs via Fyn phosphorylation, prolonged exposure or specific signaling pathways downstream of Aβ oligomers (potentially involving STEP or calcineurin) trigger clathrin-dependent endocytosis of NMDARs. This internalization is often subunit-specific, favoring NR2B-containing receptors. The loss of synaptic NMDARs, particularly NR2B-NMDARs which are crucial for certain forms of LTP and synaptic integration, further tilts the balance towards synaptic weakening. However, the facilitation of LTD by Aβ is more complex than mere receptor loss. Aβ oligomers appear to lower the threshold for LTD induction and stabilize the depressed state. The chronic low-level activation of mGluR5 by Aβ mimics the agonist (DHPG)-induced chemical LTD pathway. 96 Furthermore, the Aβ-induced rise in basal intracellular Ca2+ primes calcineurin/PP1 activity, meaning that even weak synaptic stimulation that would normally be subthreshold can now activate the LTD machinery. This results in an aberrant, protein synthesis-dependent form of LTD that is resistant to depotentiation (reversal). 107 Importantly, this Aβ-facilitated LTD shares mechanistic similarities with mGluR-LTD and NMDAR-dependent LTD but occurs under conditions where normal LTP should prevail. The net effect is a profound shift in the synaptic plasticity metaplasticity, the ability of synapses to undergo future plasticity, favoring persistent depression over potentiation. Synapses become “locked” in a depressed state, unable to respond effectively to learning signals, thereby directly impairing the hippocampus's ability to encode new memories and consolidate existing ones. 108
In summary, Aβ oligomers orchestrate a multi-pronged attack on hippocampal glutamatergic synaptic plasticity through receptor-specific dysregulation. By binding to the PrPᶜ/mGluR5 complex, they initiate Fyn kinase activation, leading to NR2B hyperphosphorylation, NMDAR overactivation, and catastrophic Ca2+ overload, triggering excitotoxic cascades. Simultaneously, Aβ hijacks mGluR5 signaling to upregulate and activate STEP, which dephosphorylates GluA1 at Ser845, promoting AMPAR endocytosis and crippling LTP. Furthermore, Aβ exposure lowers the threshold for LTD induction, chronically activates LTD-related phosphatases (calcineurin, PP1), and promotes NMDAR internalization, creating a synaptic environment pathologically biased towards persistent depression. These interlinked mechanisms, NMDAR-driven excitotoxicity, STEP-mediated AMPAR removal, and facilitated LTD, act synergistically to dismantle the molecular foundation of synaptic plasticity long before overt neuronal loss occurs. 22 The cumulative effect is the functional silencing of critical hippocampal synapses, directly underlying the early and profound deficits in learning and memory that characterize AD. This intricate sabotage of glutamate receptor function represents a core molecular pathway through which Aβ accumulation translates into the cognitive devastation of AD (Table 3).
Phosphorylated tau-driven synaptic impairment
While Aβ oligomers initiate the assault on hippocampal synapses, the subsequent emergence and dendritic accumulation of phosphorylated tau (p-tau) represent a critical downstream effector pathway that directly dismantles synaptic architecture and function, acting as a potent molecular saboteur of plasticity. Although historically associated with neurofibrillary tangles and late-stage neurodegeneration, compelling evidence now positions pathological p-tau as an early and active mediator of synaptic failure, occurring independently of overt tangle formation and neuronal death. 100 This p-tau-driven impairment operates through distinct but interconnected mechanisms: the aberrant mislocalization of p-tau to dendritic spines where it disrupts the PSD nano-organization; the destabilization of microtubule networks essential for receptor trafficking; and the activation of proteolytic cascades that cleave critical synaptic scaffolding proteins. Together, these processes orchestrate a profound structural and functional decimation of hippocampal synapses, directly contributing to the plasticity deficits underlying cognitive decline in AD.
The pathological journey of tau in AD begins with its transformation from a predominantly axonal, microtubule-stabilizing protein into a hyperphosphorylated state that mislocalizes to the somatodendritic compartment, including dendritic spines. This mislocalization is not a passive consequence of disease but an active process instigated by Aβ oligomers, primarily through the dysregulation of kinases like GSK-3β and CDK5, as detailed in the preceding section. While GSK-3β plays a central role in tau hyperphosphorylation, other kinases including cyclin- CDK5 contribute significantly to pathological tau phosphorylation through proline-directed phosphorylation mechanisms. 109 Conversely, reduced activity of protein phosphatases such as PP2A disrupts the delicate balance of tau phosphorylation-dephosphorylation dynamics, further promoting tau accumulation. 110
Hyperphosphorylation at specific epitopes, such as serine 202 and threonine 205 (recognized by the AT8 antibody), serine 396, serine 404, and threonine 231, drastically reduces tau's affinity for microtubules. 73 Detached from microtubules, p-tau accumulates in the soma and invades dendritic shafts and spines, compartments where it is virtually absent under physiological conditions. 74 This dendritic accumulation is particularly toxic. Within spines, p-tau interacts aberrantly with key components of the PSD, the dense protein network that anchors and organizes glutamate receptors and signaling molecules. A primary target is postsynaptic density protein-95 (PSD-95), a major scaffolding protein essential for clustering and stabilizing NMDARs and AMPARs at the synapse. 51 p-tau, through mechanisms involving direct binding or steric hindrance, displaces PSD-95 from its normal location within the PSD. 21 This displacement is catastrophic for synaptic integrity. PSD-95 acts as a central organizer, forming nano-domains that precisely localize receptors, kinases, and adhesion molecules to ensure efficient signal transduction during plasticity events like LTP. 111 Displacement of PSD-95 by p-tau leads to the disassembly of these critical nano-domains. Consequently, AMPARs, which rely heavily on PSD-95 for their synaptic retention and positioning within signaling complexes, become destabilized and diffuse away from the synapse. 75 This nanoscale disorganization results in a significant reduction in the number of functional AMPARs at the synapse, directly impairing synaptic strength and the ability to sustain LTP, the cellular correlate of learning and memory. Importantly, this p-tau-induced displacement of PSD-95 and AMPAR disassembly occurs early in AD pathogenesis, often preceding overt synapse loss, and correlates strongly with initial cognitive impairments in animal models and humans. 75 The mislocalized p-tau essentially acts as a molecular wrecking ball within the dendritic spine, dismantling the meticulously organized PSD architecture necessary for effective glutamatergic transmission and plasticity.
Beyond disrupting the synaptic scaffold, p-tau exerts a devastating effect on the intracellular transport highways, the microtubules, fundamentally crippling the delivery of essential synaptic components, particularly AMPARs, to the synapse. In healthy neurons, microtubules serve as tracks for motor proteins (kinesins and dyneins) that transport cargo, including neurotransmitter receptors packaged in vesicles, from the cell body (soma) and Golgi apparatus to distal dendrites and spines. Physiological tau binds to and stabilizes these microtubules, promoting efficient transport. 112 However, hyperphosphorylated tau loses this stabilizing function. Instead, it acts in a dominant-negative manner. Detached p-tau sequesters normal tau and other microtubule-associated proteins (MAPs), preventing them from stabilizing microtubules. 113 Furthermore, p-tau can directly promote microtubule disassembly. The result is profound microtubule destabilization and fragmentation, particularly within the intricate dendritic arbor where long-distance transport is crucial. 114 This collapse of the microtubule network has dire consequences for synaptic maintenance and plasticity. AMPARs, synthesized in the soma, are assembled and processed in the Golgi apparatus before being packaged into vesicles for transport to the dendritic membrane. Microtubule destabilization caused by p-tau severely disrupts this anterograde transport. Vesicles containing AMPAR subunits, particularly GluA1, the subunit critical for activity-dependent synaptic delivery during LTP, fail to be efficiently transported along the destabilized microtubules. 115 Imaging studies reveal that GluA1-containing vesicles accumulate in the soma and proximal dendrites, trapped in the Golgi apparatus or failing to embark on the dendritic journey due to impaired kinesin-mediated motility. 116 This “Golgi retention” of GluA1 prevents the replenishment of synaptic AMPAR pools. Synapses are dynamic structures; AMPARs undergo constant turnover, being internalized, recycled, degraded, and replaced. The p-tau-induced blockade of GluA1 trafficking means that synapses cannot replace internalized or degraded receptors, nor can they deliver new receptors required for synaptic strengthening during learning. Consequently, synaptic AMPAR content progressively declines. This trafficking deficit is not limited to GluA1; other synaptic proteins necessary for plasticity, including other AMPAR subunits, NMDARs, and signaling molecules, are similarly affected by the breakdown in microtubule-dependent transport. 117 This mechanism provides a crucial link between somatic tau pathology and remote synaptic dysfunction, explaining how pathological changes in the neuronal cell body can incapacitate synapses located microns away in the dendritic tree. The destabilization of microtubules by p-tau thus strangles the supply lines that sustain synaptic strength and adaptability.
Adding a third layer of destruction, p-tau, particularly when phosphorylated at specific residues like serine 396, acts as a potent trigger for localized caspase activation within synapses, leading to the irreversible cleavage of essential structural proteins. Caspase-3, a key executioner protease typically associated with apoptotic cell death, is increasingly recognized for its non-apoptotic roles in synaptic function and dysfunction. Pathological p-tau, especially species phosphorylated at Ser396, exhibits a heightened propensity to activate caspase-3. 76 This activation can occur through several interconnected pathways. One mechanism involves the direct or indirect activation of upstream caspases (e.g., caspase-9 via the intrinsic mitochondrial pathway, potentially triggered by Aβ-induced stress or p-tau itself) that then cleave and activate caspase-3. 118 Alternatively, p-tau can promote caspase-3 activation through calcium dysregulation or by facilitating the aggregation of proteins that form platforms for caspase activation. Caspase-3 not only cleaves key synaptic proteins but also directly processes tau at Asp421, generating aggregation-prone fragments that accelerate neurofibrillary tangle formation. 119 This caspase-mediated tau cleavage creates a vicious cycle where truncated tau species exhibit enhanced toxicity and further promote caspase activation. Once activated, caspase-3 cleaves specific synaptic substrates with devastating consequences. Two primary targets are αII-spectrin and PSD-95. 120 αII-spectrin is a major component of the membrane-associated periodic skeleton that underlies the plasma membrane and provides structural integrity to axons, dendrites, and spines. Cleavage of αII-spectrin by caspase-3 fragments the cytoskeleton, leading to spine destabilization, loss of spine head volume, and ultimately, spine collapse. 121 This structural disintegration directly undermines the physical basis for synaptic contact and plasticity. Simultaneously, caspase-3 cleaves PSD-95, the very scaffolding protein already displaced by p-tau. 122 Caspase-mediated cleavage of PSD-95 generates a stable C-terminal fragment that lacks the ability to cluster receptors and organize the PSD, effectively dismantling the synaptic signaling hub. 75 This cleavage event is irreversible and represents a point of no return for the synapse. The loss of intact PSD-95 further accelerates AMPAR dispersal and internalization. Moreover, caspase-3 activation creates a vicious cycle: cleaved tau fragments generated by caspases (including fragments resulting from cleavage at Asp421) are themselves more prone to aggregation and exhibit enhanced toxicity, potentially further stimulating caspase activity and synaptic damage. 123 Importantly, this caspase-3 activation and synaptic protein cleavage can occur locally within dendrites and spines, in a compartmentalized manner, without immediately triggering full-blown neuronal apoptosis. 124 This phenomenon, termed “synaptosis” or synaptic apoptosis, represents a selective pruning mechanism where synapses degenerate due to localized caspase activity, contributing significantly to the early, synapse-specific pathology observed in AD. The p-tau-induced activation of caspase-3 and the subsequent proteolysis of spectrin and PSD-95 thus deliver a terminal blow to synaptic integrity, severing structural supports and dismantling the molecular organization essential for plasticity.125,126 In conclusion, p-tau is far more than a passive marker of neurodegeneration; it is an active and potent executor of synaptic sabotage within the hippocampus. Its Aβ-induced mislocalization to dendrites allows it to directly invade spines, where it displaces PSD-95, dismantling AMPAR nanodomains and destabilizing the synaptic scaffold. Concurrently, p-tau destabilizes microtubules, crippling the anterograde transport system and trapping critical plasticity proteins like GluA1 in the Golgi, starving synapses of essential components. 75 Finally, p-tau, especially when phosphorylated at sites like Ser396, ignites localized caspase-3 activation, leading to the irreversible cleavage of structural pillars like spectrin and PSD-95, culminating in spine collapse and functional elimination. These three core mechanisms—PSD disorganization, trafficking failure, and synaptic proteolysis—operate synergistically and often sequentially. The mislocalized p-tau first disrupts the existing synaptic structure, then prevents its repair and replenishment by blocking trafficking, and finally activates proteases that dismantle it irreversibly.20,127 This p-tau-driven triad of synaptic impairment occurs early in AD pathogenesis, downstream of Aβ but often amplifying Aβ's toxic effects, and is sufficient to cause significant deficits in hippocampal synaptic plasticity, including impaired LTP and enhanced LTD, well before widespread neuronal death. Targeting the molecular pathways underlying p-tau's synaptic toxicity—preventing its dendritic mislocalization, stabilizing microtubules, or inhibiting localized caspase activation—holds significant promise for preserving synaptic function and cognitive resilience in AD. Understanding and countering this p-tau driven synaptic sabotage is paramount for developing effective interventions aimed at halting the progression of cognitive decline20,127 (Table 4).
The tau-induced synaptic dysfunction cascade in AD.
Neuroinflammation and synaptic elimination
The insidious progression of AD is characterized not only by the accumulation of Aβ plaques but by a profound dismantling of synaptic networks, particularly within the hippocampus—a structure indispensable for learning and memory. Central to this synaptic sabotage is neuroinflammation, wherein Aβ oligomers activate glial cells, triggering cascades that culminate in the elimination of functional synapses. This section delineates the molecular triad—complement-mediated phagocytosis, astrocyte-driven excitotoxicity, and cytokine-induced AMPAR internalization—through which neuroinflammation orchestrates synaptic degeneration 22 (key pathways are summarized in Table 5.
Neuroinflammatory pathways.
Microglia, the resident immune sentinels of the brain, undergo pathological activation upon encountering soluble Aβ42 oligomers. These oligomers engage microglial pattern recognition receptors, including TLR4 and TREM2, initiating a pro-inflammatory signaling cascade.132,133 Activation of these receptors can lead to NLRP3 inflammasome formation and enhanced production of IL-1β and IL-18, which contribute to AD pathology progression. 68 Activated microglia release an array of cytotoxic factors, among which the complement protein C1q serves as the linchpin for synaptic elimination. C1q, the initiator of the classical complement cascade, binds to phosphatidylserine residues exposed on synaptic membranes—a canonical “eat-me” signal amplified by Aβ-induced oxidative stress and neuronal damage. 134 This opsonization event recruits complement component C3, which is cleaved into C3b by convertase enzymes. C3b then covalently attaches to synaptic surfaces, marking them for destruction. The microglial complement receptor 3 (CR3; CD11b/CD18 integrin) recognizes bound C3b, activating phagocytic machinery that engulfs and degrades “tagged” synapses. Pioneering work by Hong et al. demonstrated that Aβ deposits colocalize with C1q and C3 in AD mouse models, while CR3 deficiency rescues synaptic loss and cognitive deficits, underscoring this pathway's lethality. 22 Critically, Aβ oligomers exacerbate this process by upregulating neuronal C1q expression via transforming growth factor-beta (TGF-β) signaling, creating a feedforward loop of synaptic elimination. 139 The hippocampus, with its high density of complement receptors and susceptibility to Aβ deposition, becomes a hotspot for this aberrant pruning, eroding dendritic spines essential for LTP.
Simultaneously, astrocytes—the other key glial players—transition into a reactive state in response to Aβ. These reactive astrocytes, influenced by activated microglia, acquire toxic properties and disrupt synaptic function. 136 Activated astrocytes release IL-1β, a potent pro-inflammatory cytokine that binds to neuronal IL-1R. This engagement activates the c-Jun N-terminal kinase (JNK) pathway, wherein JNK phosphorylates the transcription factor c-Jun, driving its nuclear translocation and binding to the AP-1 promoter site. This suppresses the expression of the excitatory amino acid transporter 2 (EAAT2; GLT-1 in rodents), the primary glutamate transporter responsible for >90% of synaptic glutamate clearance. 135 EAAT2 downregulation leads to extracellular glutamate accumulation, resulting in spillover and excessive activation of extrasynaptic NMDARs. The resultant calcium influx triggers mitochondrial ROS production, calpain activation, and dendritic atrophy. Rothstein et al. first linked EAAT2 dysfunction to excitotoxicity in neurodegenerative contexts, but in AD, hippocampal EAAT2 loss correlates directly with Aβ burden and cognitive decline. 140 Notably, IL-1β amplifies its own release via NF-κB signaling, creating a self-sustaining inflammatory loop. 79 Pharmacological inhibition of JNK (e.g., SP600125) restores EAAT2 expression and rescues LTP in Aβ-exposed hippocampal slices, as demonstrated by Wang et al., highlighting the therapeutic potential of targeting this axis. This excitotoxic milieu not only impairs synaptic transmission but also primes synapses for complement-mediated elimination by exposing phosphatidylserine—a convergence point amplifying synaptic loss. 137
TNF-α, secreted by both Aβ-activated microglia and astrocytes, constitutes the third arm of neuroinflammatory synaptic sabotage. TNF-α binds to neuronal TNF receptor 1 (TNF-R1), activating NADPH oxidase (NOX) and generating superoxide radicals. These radicals oxidize and activate STEP, a key regulator of synaptic stability. Activated STEP dephosphorylates the GluA1 subunit of AMPARs at serine residue 845 (Ser845)—a critical site phosphorylated by PKA to stabilize surface AMPARs and facilitate synaptic insertion. Dephosphorylation promotes GluA1 endocytosis via clathrin-dependent pathways, effectively decoupling synapses from plasticity mechanisms. 80 Zhang et al. established that Aβ oligomers induce STEP accumulation in hippocampal neurons, while genetic STEP knockout in AD mouse models prevents AMPAR internalization and LTP impairment. 104 TNF-α further disrupts plasticity by suppressing activity-regulated cytoskeleton-associated protein (Arc) synthesis—a protein essential for AMPAR trafficking—via MAPK. 141 Crucially, TNF-α and IL-1β synergize: IL-1β enhances STEP transcription, while TNF-α post-translationally activates it, creating a cytokine convergence that maximizes synaptic weakening.
The interplay of these pathways dismantles hippocampal synapses through complementary mechanisms: complement tagging eliminates spines marked by excitotoxic damage or phosphatidylserine exposure; glutamate spillover from EAAT2 downregulation fuels calcium overload, attracting microglial phagocytosis; and the TNF-α/STEP axis removes AMPARs, erasing synaptic efficacy. The hippocampus's vulnerability arises from its high Aβ uptake (mediated by low-density lipoprotein receptor-related protein 1 [LRP1] on CA1 neurons), dense microglial populations, and reliance on EAAT2 for glutamate clearance. Human postmortem studies confirm elevated C1q, IL-1β, and TNF-α in the AD hippocampus, correlating with synapse loss quantified by PSD-95 reduction and cognitive decline.22,142
Therapeutic strategies targeting these pathways are emerging. Complement inhibition via anti-C1q antibodies (e.g., ANX005) reduces synaptic loss in preclinical AD models. 22 EAAT2 enhancers like ceftriaxone restore glutamate homeostasis, reversing excitotoxicity. 143 TNF-α blockers (e.g., etanercept) and STEP antisense oligonucleotides rescue LTP and memory in AD mice, though clinical translation requires optimization.104,144 In summary, Aβ hijacks neuroinflammatory machinery to execute synaptic sabotage: microglia eliminate synapses via complement opsonization, astrocytes induce excitotoxicity through IL-1β/JNK/EAAT2 suppression, and TNF-α activates STEP to internalize AMPARs. This triad converts adaptive synaptic remodeling into pathological elimination, crippling hippocampal plasticity and accelerating cognitive decline (Table 5).
Oxidative stress and mitochondrial failure
The relentless accumulation of Aβ oligomers within the hippocampal formation initiates a catastrophic cascade of oxidative damage and bioenergetic collapse that fundamentally dismantles synaptic plasticity machinery. 15 This section delineates the molecular choreography through which Aβ-induced ROS production, lipid peroxidation, and mitochondrial permeability transition converge to disrupt calcium homeostasis, deplete cellular energy reserves, and dismantle the actin cytoskeleton essential for dendritic spine integrity. The hippocampus, with its high metabolic demand and dense synaptic architecture, becomes disproportionately vulnerable to these insults, converting a hub of learning and memory into an epicenter of pathological decay.
Aβ oligomers, particularly soluble Aβ42 species, exhibit a high affinity for transition metals such as copper (Cu2+) and iron (Fe2+) via coordination bonds involving histidine residues at positions 6, 13, and 14. 145 This metal binding facilitates redox cycling wherein Aβ-bound Cu2+ reduces molecular oxygen (O2) to superoxide (O2•−), subsequently dissimulated to hydrogen peroxide (H2O2) by superoxide dismutase. Through Fenton chemistry, Aβ-Cu/Fe complexes catalyze the conversion of H2O2 into hydroxyl radicals (•OH), among the most reactive and destructive oxidants in biological systems. Opazo et al. (2002) demonstrated that Aβ42-Cu complexes generate ROS at rates 6-fold higher than metal-free Aβ, a phenomenon quantified using electron paramagnetic resonance spectroscopy. 145 The hippocampus, characterized by elevated baseline oxidative stress due to high oxygen consumption and relatively low antioxidant defenses (e.g., glutathione peroxidase), suffers disproportionate protein carbonylation and nucleic acid oxidation. 146 Lipid peroxidation, a hallmark of oxidative damage in AD, generates 4-HNE which forms covalent adducts with synaptic proteins, leading to widespread membrane damage and protein dysfunction. 147 Crucially, this oxidative milieu directly targets CaMKII, a master regulator of LTP. Beyond the Fenton chemistry driven by Aβ-metal complexes, the overproduction of reactive RNS, particularly nitric oxide (NO) via nNOS activation, constitutes a major component of the oxidative assault in AD. Inhibition of nNOS has been shown to significantly attenuate brain tissue oxidative damage and its associated functional deficits, highlighting a key pathway intersecting excitotoxicity, oxidative stress, and synaptic failure 148 Oxidation of methionine residues 281 and 282 within CaMKII's autoinhibitory domain induces conformational changes that impede autophosphorylation at Thr286—an event essential for the kinase's calcium-independent activity during LTP maintenance. 149 Jiang et al. (2017) documented a 60% reduction in CaMKII activity in hippocampal slices exposed to Aβ oligomers, while overexpression of methionine sulfoxide reductase A, which repairs oxidized methionine residues, restored LTP magnitude in APP/PS1 transgenic mice. 150 This oxidation further destabilizes CaMKII's binding to the NMDAR-PSD-95 complex, uncoupling synaptic activity from downstream plasticity effectors like CREB phosphorylation.
The deluge of ROS propagates membrane damage through lipid peroxidation, wherein •OH radicals abstract hydrogen atoms from PUFAs in synaptic membranes, generating lipid alkyl radicals (L•) that react with oxygen to form lipid peroxyl radicals (LOO•). These reactive intermediates decompose into α,β-unsaturated aldehydes, most notably 4-HNE, which diffuse far beyond their site of origin due to their lipophilicity. 150 4-HNE forms covalent Michael adducts with nucleophilic residues (cysteine, histidine, lysine) on critical synaptic proteins. NMDAR subunits GluN2A and GluN2B are particularly vulnerable, with 4-HNE adduct formation at Cys399 of GluN2A increasing channel open probability by 40% and prolonging mean open time by 2.3-fold, as demonstrated by single-channel patch-clamp recordings in hippocampal neurons. 151 This pathological “gain-of-function” drives excessive calcium influx, overwhelming endogenous buffers (e.g., calbindin) and saturating mitochondrial uptake capacity. Concurrently, 4-HNE inhibits plasma membrane calcium ATPase (PMCA) and sarco/ER Ca2+-ATPase (SERCA) by adduct formation at key cysteine residues in their catalytic domains. 152 The oxidative stress metabolite 4-HNE promotes Alzheimer protofibril formation, creating a vicious cycle that further exacerbates Aβ pathology and synaptic dysfunction 153 The resulting calcium dyshomeostasis activates calpain proteases, which truncate spectrin, tau, and PSD-95, precipitating dendritic spine retraction. he pivotal role of BDNF in memory processes is further substantiated by interventional studies demonstrating that pharmacological enhancement of BDNF signaling, such as through angiotensin-converting enzyme inhibition in a scopolamine-induced model, is sufficient to rescue learning and memory deficit. 154
Critically, 4-HNE also impairs glutamate uptake by forming adducts with cysteine residues of the astrocytic glutamate transporter EAAT2 (GLT-1), exacerbating excitotoxic stress and creating a vicious cycle of ROS generation and calcium overload.155,156
Mitochondria emerge as both sources and victims of Aβ-induced oxidative damage. Aβ oligomers bind to mitochondrial Aβ-binding alcohol dehydrogenase (ABAD), inhibiting complex IV and disrupting electron transport chain function, leading to excessive ROS production and energy crisis. 157 Aβ oligomers translocate to mitochondria via the translocase of the outer membrane (TOM) complex, binding to ABAD in the mitochondrial matrix. This interaction inhibits complex IV (cytochrome c (Cyt c) oxidase), reducing electron transport chain efficiency and elevating mitochondrial ROS (mtROS) production by 300% in hippocampal neurons, as measured by MitoSOX fluorescence. 157 mtROS oxidize cardiolipin, a phospholipid critical for inner mitochondrial membrane integrity, prompting CCyt c detachment and oligomerization of the permeability transition pore (mPTP). The mPTP—a pathological mega channel composed of cyclophilin D (CypD), adenine nucleotide translocator, and voltage-dependent anion channel—opens irreversibly under Aβ stress, dissipating the mitochondrial membrane potential (ΔΨm). Du et al. demonstrated using patch-clamp mitoplast recordings that Aβ oligomers decrease the threshold for calcium-induced mPTP opening by 65% in hippocampal mitochondria. 26 Persistent mPTP opening uncouples oxidative phosphorylation, collapsing ATP synthesis by >50% within hippocampal synapses. This energy crisis cripple's actin-regulatory pathways: ATP-dependent kinases like LIM kinase (LIMK) fail to phosphorylate cofilin, while ATP depletion impairs slingshot phosphatase (SSH1) inactivation. Consequently, dephosphorylated cofilin undergoes a conformational shift exposing its actin-binding domain, severing F-actin filaments at a 1:1 stoichiometry. Bamburg et al. (2010) observed cofilin-actin rods in 70% of hippocampal dendrites in Aβ-infused rats, physically obstructing synaptic vesicle transport and collapsing spine heads. 156 Mitochondrial failure amplifies this catastrophe through CCyt c release, which activates caspase-3 to cleave actin-stabilizing proteins like spectrin and gelsolin. Furthermore, mtROS activate p38 MAPK, which phosphorylates tau at Ser422, promoting its translocation to dendritic spines where it sequesters actin-regulatory proteins and accelerates cofilin activation. 158
The interplay between oxidative stress and mitochondrial failure extends beyond structural collapse to impair synaptic plasticity at functional levels. 159 ATP depletion suppresses NMDA receptor exocytosis and AMPAR trafficking, 160 while ROS-oxidized syntaxin and SNAP-25 disrupt SNARE complex assembly, inhibiting vesicular release. The combined assault on CaMKII, actin dynamics, and vesicle cycling manifests as LTP impairment in the Schaffer collateral-CA1 pathway—reducing potentiation by 60–80% in Aβ-exposed hippocampal slices. Therapeutic strategies targeting this axis include metal chelators like PBT2, which reduced Aβ-Cu ROS production by 80% in a phase II trial, and mitochondrial protectants such as SS-31 (elamipretide), which binds to cardiolipin, preventing mPTP opening and restoring ATP levels by 40% in AD models. 161 Inhibition of lipid peroxidation with ferrostatin-1 or modulation of cofilin phosphorylation with the S3 peptide offers further promise for synaptic rescue. 162 Critically, these pathways intersect with broader AD pathogenesis: mtROS activate BACE1, accelerating Aβ production, while 4-HNE adducts impair insulin signaling, exacerbating tau hyperphosphorylation.163,164 The hippocampal vulnerability to this “triple threat” of CaMKII oxidation sabotaging LTP induction, 4-HNE hijacking calcium homeostasis, and mPTP-mediated ATP depletion collapsing structural plasticity underscores its role as a nexus of molecular sabotage in AD (Table 6).
Oxidative stress and mitochondrial failure.
BDNF/TrkB signaling attenuation: molecular mechanisms
The progressive erosion of hippocampal synaptic plasticity in AD is profoundly influenced by the systematic impairment of BDNF and its high-affinity receptor, TrkB. BDNF-TrkB signaling serves as a crucial regulator of neuronal survival, dendritic complexity, spine maturation, and LTP, the cellular basis of learning and memory. 48 Within the hippocampus, a region highly dependent on neurotrophic support for plasticity, Aβ oligomers execute a multi-faceted attack on this essential pathway. 166 This section details the molecular mechanisms through which Aβ accumulation disrupts BDNF synthesis, alters BDNF processing toward an inactive form, and impairs TrkB signal transduction, leading to the collapse of synaptic resilience and plasticity necessary for cognitive function. 28
Aβ oligomers, particularly soluble Aβ42 species enriched at hippocampal synapses, initiate BDNF/TrkB signaling attenuation through transcriptional repression of the BDNF gene. A key mechanism involves Aβ-induced upregulation of specific microRNAs (miRNAs) that silence gene expression by binding to complementary sequences in target mRNAs. Notably, Aβ exposure elevates levels of microRNA-206 (miR-206) in hippocampal neurons. Lee et al. demonstrated that miR-206 directly targets the 3’ untranslated region (3'-UTR) of BDNF mRNA, promoting its degradation and reducing BDNF protein synthesis. 167 This effect is particularly pronounced in the hippocampus, where miR-206 expression inversely correlates with BDNF levels in AD patients and transgenic models. The induction of miR-206 by Aβ involves dysregulation of transcription factors like REST (RE1-Silencing Transcription Factor), which is aberrantly activated in AD and promotes expression of plasticity-repressing miRNAs, including miR-206. 168 Furthermore, Aβ-triggered oxidative stress and neuroinflammation exacerbate this transcriptional repression by activating stress-responsive kinases like p38 MAPK, creating a feedforward loop that sustains BDNF deficiency. 169
Beyond suppressing BDNF synthesis, Aβ accumulation pathologically alters processing of the BDNF precursor protein, proBDNF. BDNF is initially synthesized as proBDNF, which is cleaved intracellularly by furin or extracellularly by plasmin/tissue plasminogen activator (tPA) to yield mature BDNF (mBDNF). mBDNF preferentially binds TrkB, promoting neuronal survival, synaptic strengthening, and LTP, whereas proBDNF exhibits high affinity for the p75 neurotrophin receptor (p75NTR), typically activating pathways leading to apoptosis, synaptic weakening, and LTD. 170 Aβ oligomers profoundly disrupt this balance. Oxidative stress, a key consequence of Aβ accumulation, impairs enzymatic activity of proteases involved in proBDNF conversion, particularly furin and plasmin. Pang et al. provided early evidence linking oxidative stress to reduced plasmin activity and proBDNF accumulation. 171 More recently, Yang et al. showed that Aβ directly inhibits the plasminogen activation system in hippocampal neurons, causing significant proBDNF accumulation. 172 Consequently, the hippocampal environment in AD shifts toward excess proBDNF relative to mBDNF. This proBDNF binds p75NTR with high affinity, triggering activation of RhoA, a small GTPase. Activated RhoA stimulates Rho-associated kinase (ROCK), which phosphorylates and inactivates the actin-depolymerizing factor cofilin. Paradoxically, sustained ROCK activation leads to increased cofilin activity through downstream mechanisms. Activated cofilin severs actin filaments, the primary structural components of dendritic spines, causing rapid spine collapse and retraction, thereby eliminating synaptic connections essential for plasticity. 173
The third major mechanism of Aβ-induced BDNF/TrkB impairment targets the TrkB receptor complex and its downstream signaling cascades, rendering neurons unresponsive to available mBDNF. TrkB is a receptor tyrosine kinase that undergoes dimerization and trans-autophosphorylation at specific tyrosine residues upon mBDNF binding. 174
These phospho-tyrosines serve as docking sites for adaptor proteins, initiating key signaling pathways essential for synaptic plasticity. 175 One critical pathway involves the recruitment of PLCγ to phospho-Tyr785. 54 PLCγ catalyzes the hydrolysis of PIP2 into IP3 and DAG. The functional indispensability of this specific TrkB-PLCγ pathway for hippocampal plasticity and memory has been firmly established through genetic studies. 176 IP3 mobilizes calcium from intracellular stores, while DAG activates PKC, particularly the PKCε in neurons. Activated PKCε phosphorylates numerous substrates, including the transcription factor cAMP response CREB at Ser133. PCREB translocates to the nucleus, binds cAMP response elements in promoters of plasticity-related genes such as Arc and BDNF itself, and enhances their transcription, creating a positive feedback loop vital for sustaining synaptic strength and LTP. 54 Aβ oligomers severely damage this TrkB-PLCγ-PKCε-CREB axis. First, Aβ disrupts the membrane microenvironment, altering lipid raft composition and potentially impairing TrkB dimerization or autophosphorylation efficiency. 65 More significantly, Aβ activates several phosphatases that directly antagonize TrkB signaling. Aβ-induced activation of STEP also targets TrkB. STEP dephosphorylates critical activation tyrosines on TrkB (e.g., Tyr816), terminating kinase activity and downstream signaling. 104 Additionally, Aβ-triggered calcium dyshomeostasis activates calcineurin (protein phosphatase 2B, PP2B). Calcineurin dephosphorylates CREB at Ser133, counteracting effects of PKCε and other CREB kinases. 43 Zhang et al. demonstrated that Aβ oligomers significantly reduce CREB phosphorylation in hippocampal neurons, an effect blocked by calcineurin inhibitors. 94 Consequently, impaired TrkB-PLCγ-PKCε signaling and CREB dephosphorylation dramatically reduce transcription of synaptic plasticity genes. Arc mRNA and protein levels are significantly diminished in the AD hippocampus, severely compromising synaptic adaptability. 177 Crucially, reduced BDNF transcription downstream of pCREB exacerbates initial BDNF deficiency caused by miR-206, creating a devastating feedforward cycle of neurotrophic support failure.
These three primary mechanisms—miRNA-mediated BDNF mRNA degradation, oxidative stress-induced proBDNF accumulation/p75NTR activation, and TrkB-PLCγ-PKCε-CREB signaling impairment—operate synergistically within the AD hippocampus, amplifying synaptic toxicity. For instance, oxidative stress impairing proBDNF conversion also directly damages mitochondria, reducing ATP levels required for efficient TrkB trafficking and signaling complex assembly. 27 Pro-inflammatory cytokines like TNF-α, elevated due to Aβ-driven neuroinflammation, further suppress BDNF transcription and potentiate p75NTR signaling. 178 Conversely, loss of BDNF/TrkB trophic support increases neuronal vulnerability to Aβ-induced excitotoxicity, oxidative damage, and inflammatory insults. 28 The spatial and temporal convergence of these insults within hippocampal circuitry is particularly devastating. The CA1 region, essential for associative memory and highly dependent on BDNF for LTP maintenance, exhibits some of the earliest and most severe reductions in BDNF levels and TrkB signaling in AD. 179 Resulting synaptic pathology includes simplified dendritic arbors, loss of thin plastic spines, impaired LTP induction and maintenance, and enhanced susceptibility to LTD, directly contributing to failure of hippocampal-dependent memory processes.
The critical role of BDNF/TrkB attenuation in AD pathogenesis highlights its therapeutic potential. Strategies to boost BDNF levels include intranasal BDNF delivery, TrkB agonists like 7,8-dihydroxyflavone (7,8-DHF), and approaches to inhibit miR-206 or enhance proBDNF conversion (e.g., tPA activators). Promoting TrkB signaling resilience through phosphatase inhibition (targeting STEP or calcineurin) or directly enhancing CREB activity are also promising avenues. Nagahara et al. demonstrated that restoring BDNF levels via viral vector delivery in aged rats reverses synaptic loss and improves memory, providing proof-of-concept. 180 However, challenges remain, including targeted hippocampal delivery, avoiding proBDNF/p75NTR side effects, and intervening before irreversible synaptic loss occurs. Understanding these molecular sabotage mechanisms provides essential insights for developing refined therapies to rescue neurotrophic support and restore hippocampal synaptic plasticity in AD (see Table 7 for summary).
BDNF/TrkB signaling attenuation.
Apoptotic pathway convergence
The insidious dismantling of hippocampal synaptic plasticity in AD reaches its catastrophic zenith with the activation and convergence of apoptotic pathways. While neuronal death is the ultimate endpoint, the apoptotic machinery is co-opted much earlier in the disease process, specifically targeting the molecular and structural integrity of synapses themselves. 118 Aβ oligomers, acting as the central instigator, synergize with the various stressors they induce—oxidative stress, neuroinflammation, excitotoxicity, mitochondrial failure, and trophic withdrawal—to trigger the programmed demolition of synaptic components. 183 This section meticulously dissects the molecular mechanisms by which Aβ-driven stressors activate the intrinsic (mitochondrial), extrinsic (death receptor), and ER stress-mediated apoptotic pathways, culminating in the proteolytic cleavage of critical synaptic scaffolding proteins, the dissolution of dendritic spines, and the irreversible collapse of plasticity mechanisms essential for hippocampal function. 184 This synaptic apoptosis represents a point of no return in the molecular sabotage orchestrated by Aβ. 185
The intrinsic apoptotic pathway serves as a primary executioner in Aβ-induced synaptic degeneration. 186 Its activation is fundamentally driven by profound oxidative stress and mitochondrial dysfunction. 187 ROS, generated abundantly through Aβ-bound metal catalysis and particularly from damaged mtROSBAV, act as potent activators of pro-apoptotic Bcl-2 family proteins, notably Bcl-2-associated X protein (Bax) and Bcl-2 antagonist/killer (Bak). Under physiological conditions, these proteins are restrained by anti-apoptotic members like Bcl-2 and Bcl-xL. However, sustained Aβ exposure disrupts this balance. 188 ROS directly oxidize and inactivate anti-apoptotic Bcl-2 family members while promoting the conformational activation and oligomerization of Bax and Bak. 189
Aβ oligomers specifically induce the mitochondrial translocation of Bax in hippocampal neurons, a process amplified by concurrent mtROS production. 101 Once activated, Bax and Bak oligomerize and integrate into the outer mitochondrial membrane, forming large pores that lead to the catastrophic release of apoptogenic factors. 190 The most critical of these for synaptic demolition is CCyt c, which acts as a crucial signaling molecule within the synaptic compartment. 183 Cytosolic CCyt c binds to Apoptotic protease-activating factor 1 (Apaf-1), triggering formation of the apoptosome complex and activation of caspase-9. 191 Active caspase-9 then cleaves and activates executioner caspases, primarily caspase-3. 118 Mattson et al. (1998) provided seminal evidence linking Aβ exposure to CCyt c release and caspase activation in neurons.183,192 Activated caspase-3 plays a pivotal role in synaptic dismantling before engaging in nuclear apoptosis 118 Within dendrites and spines, caspase-3 targets key structural proteins essential for synaptic integrity. 193 It cleaves spectrin, destabilizing synaptic architecture, and crucially cleaves PSD-95, the master scaffolding protein of the excitatory postsynaptic density. 118 Cleavage of PSD-95 at D295 dissociates it from NMDARs and other signaling complexes, leading to disassembly of the postsynaptic density and internalization of glutamate receptors. 125 D'Amelio et al. (2011) showed that caspase-3 activation within hippocampal dendrites occurs early in AD models and correlates strongly with PSD-95 degradation and synapse loss, preceding overt neuronal death. 118 Thus, the intrinsic pathway utilizes caspase-3 as a molecular scalpel to surgically dismantle synapses from within. 186
Simultaneously, Aβ accumulation potently activates the extrinsic apoptotic pathway through induction of chronic neuroinflammation. 192 Activated microglia and astrocytes release pro-inflammatory cytokines, with TNF-α being a major driver. 184 TNF-α binds to TNFR1 on hippocampal neurons, inducing trimerization and recruitment of adaptor proteins TNF receptor-associated death domain and Fas-associated protein with death domain (FADD), forming the death-inducing signaling complex (DISC). 192 Within the DISC, procaspase-8 undergoes autocatalytic activation through proximity-induced dimerization and cleavage. Doll et al. provided compelling evidence using single-cell RNA sequencing that hippocampal neurons in AD models exhibit upregulated TNFR1 and FADD expression, correlating with increased caspase-8 activity. 192 Active caspase-8, the initiator caspase of the extrinsic pathway, then directly cleaves and activates executioner caspases like caspase-3 and caspase-7. However, in many neuronal contexts, including hippocampal neurons exposed to Aβ, the direct activation of executioner caspases by caspase-8 is often insufficient to trigger full apoptosis. Instead, caspase-8 cleaves the BH3-only protein Bid, generating a truncated, potent form called truncated Bid (tBid). tBid translocates to the mitochondria where it potently activates Bax and Bak, thereby directly linking the extrinsic pathway to the intrinsic mitochondrial pathway and amplifying CCyt c release. 190 This mitochondrial amplification loop is critical for robust caspase-3 activation and synaptic damage in AD. 184 The convergence occurs at the level of caspase-3 activation within the synapse, where it cleaves the same suite of synaptic targets (PSD-95, spectrin) as triggered by the intrinsic pathway. 118 Therefore, neuroinflammation-driven TNF-α signaling not only directly contributes to synaptic dysfunction (e.g., via STEP activation) but also converges onto the mitochondrial pathway to execute synaptic demolition via caspase-3. 186
Aβ oligomers also inflict severe stress on the ER, a key organelle for protein folding, calcium storage, and lipid synthesis. The accumulation of misfolded Aβ peptides, either extracellularly or potentially intracellularly, disrupts ER homeostasis, leading to ER stress and the activation of the unfolded protein response (UPR). 194 Aβ can directly interact with ER membranes or indirectly cause ER stress through calcium dysregulation (release from ER stores via IP3R/RyR channels overloaded by Aβ-induced signals) and oxidative stress damaging ER chaperones. 195 The UPR is mediated by three main ER-resident sensors: IRE1 (Inositol-requiring enzyme 1), ATF6 (Activating transcription factor 6), and PERK (PKR-like ER kinase). While initially adaptive, chronic or severe ER stress, as occurs in AD, shifts the UPR towards pro-apoptotic signaling, predominantly through the PERK-eukaryotic initiation factor 2 (eIF2α) (PERK- eIF2α)pathway. 196 Aβ oligomers have been shown to directly activate PERK or prolong its activation stat. 197 Upon activation, PERK dimerizes and autophosphorylates, leading to the phosphorylation of the alpha subunit of eIF2α at Ser51. 198 p-eIF2α attenuates general protein translation, a protective measure to reduce the load of new proteins entering the stressed ER. However, paradoxically, p-eIF2α specifically enhances the translation of select mRNAs, including that of ATF4 (Activating transcription factor 4). 199 ATF4 translocates to the nucleus and induces the expression of genes involved in amino acid metabolism, redox homeostasis, and crucially, the transcription factor CHOP (C/EBP homologous protein, also known as GADD153). 200 CHOP is a major mediator of ER stress-induced apoptosis 201 Smith et al. (2023) demonstrated that Aβ oligomers induce a sustained PERK/eIF2α/ATF4/CHOP axis in hippocampal neurons, with CHOP levels significantly elevated in AD brains. 202 CHOP transcriptionally represses anti-apoptotic Bcl-2 while inducing the expression of pro-apoptotic BH3-only proteins, particularly BIM (Bcl-2 interacting mediator of cell death) and PUMA (p53 upregulated modulator of apoptosis). 202 BIM and PUMA then directly activate Bax and Bak, leading to mitochondrial outer membrane permeabilization, CCyt c release, apoptosome formation, and caspase-9/-3 activation – thus converging onto the same execution pathway as the intrinsic and extrinsic routes. 194 The activation of BIM and PUMA provides a direct molecular link from Aβ-induced ER stress to mitochondrial apoptosis and subsequent synaptic caspase-3 activation. 202 Furthermore, CHOP can also contribute to oxidative stress by downregulating antioxidant enzymes and promoting ROS production, creating a vicious cycle that further exacerbates all apoptotic pathways.203,204 The convergence of these three major apoptotic pathways—intrinsic, extrinsic, and ER stress-mediated—onto the activation of executioner caspases, primarily caspase-3, within the synaptic compartment represents the final common pathway for synaptic elimination in AD. 118 Caspase-3 acts as the central executioner, cleaving critical synaptic substrates like PSD-95 and spectrin, thereby dissolving the postsynaptic density and collapsing the dendritic spine cytoskeleton.118,205 This proteolytic dismantling is not merely a consequence of dying neurons but an active process targeting synapses specifically, representing “synaptic apoptosis” or “synaptosis”. 186 The vulnerability of the hippocampus to this convergence of pathological mechanisms is heightened by several intrinsic features: its high metabolic rate, reliance on precise calcium signaling, dense innervation (which makes it a hotspot for Aβ accumulation and neuroinflammation), and the critical dependence of its synaptic plasticity on scaffolding proteins such as PSD-95. The activation of these damaging pathways does not occur in isolation; rather, significant cross-talk among them synergistically amplifies neuronal and synaptic injury. For instance, ROS (intrinsic trigger) can exacerbate ER stress and potentiate TNF-α signaling. 206 TNF-α (extrinsic trigger) generates ROS and can induce ER stress. 207 ER stress (via CHOP) induces expression of death receptors like DR5, sensitizing neurons to extrinsic ligands. The PERK/eIF2α/ATF4/CHOP signaling branch of the unfolded protein response mediates neuron death induced by soluble oligomeric Aβ peptide. The BH3-only proteins (Bid, BIM, PUMA) serve as critical integrators, funneling signals from all three pathways towards Bax/Bak activation and mitochondrial permeabilization. 208 Furthermore, caspases can feedback to amplify upstream events; caspase-3 can cleave and activate other procaspases and cleave proteins involved in calcium homeostasis, worsening ER stress.
This intricate network of interactions ensures that once initiated by Aβ, the apoptotic machinery efficiently dismantles synapses, leading to the characteristic synaptic loss and plasticity failure in the hippocampus. 186
Targeting this apoptotic convergence holds significant therapeutic potential, although challenges remain due to the essential physiological roles of caspases in development and normal synaptic pruning. 209 Strategies include inhibiting caspase activation (e.g., small molecule caspase inhibitors like Minocycline, Emricasan), blocking upstream triggers (e.g., antioxidants for ROS, TNF-α inhibitors like Etanercept), modulating the UPR (e.g., PERK inhibitors like GSK2606414 or ISRIB to reverse eIF2α phosphorylation), and stabilizing mitochondria (e.g., CypD inhibitors, mPTP blockers). 188 Louneva et al. (2008) showed elevated caspase-3 activity and caspase-cleaved substrates like PSD-95 fragments in the AD hippocampus, providing a clear rationale for intervention. 125 Preclinical studies in AD models demonstrate that caspase inhibition can reduce synaptic loss and improve cognitive function. 189 However, the timing and specificity of intervention are critical to avoid disrupting essential cellular processes. 196 Understanding the precise molecular mechanisms of synaptic apoptosis, the points of convergence, and the key integrators like caspase-3, provides crucial insights for developing neuroprotective strategies aimed at preserving hippocampal synaptic plasticity and halting cognitive decline in AD by preventing this final stage of molecular sabotage 3 (Table 8).
Apoptotic pathway convergence in AD hippocampus.
Structural and functional consequences
The relentless molecular sabotage orchestrated by Aβ oligomers within the hippocampus, as detailed in the preceding sections, inevitably manifests as profound structural and functional alterations at the synaptic, cellular, and circuit levels. These changes represent the direct physical and physiological underpinnings of the cognitive collapse characteristic of AD. 16 The hippocampus, a structure exquisitely designed for learning, memory encoding, and spatial navigation, undergoes a devastating transformation, losing its fundamental plasticity and computational capabilities. This section comprehensively details the specific structural derangements, the visible erosion of synaptic architecture, and the consequent functional deficits, the crippling of synaptic communication and network dynamics, that arise directly from the Aβ-induced molecular pathologies. 213 Understanding these consequences is paramount, as they form the crucial bridge between the molecular mechanisms of synaptic disruption and the clinical expression of dementia, ultimately explaining the failure of hippocampal-dependent memory processes. 16
At the most fundamental level, the assault on hippocampal synapses results in significant synaptic loss. This is not merely a quantitative reduction but a qualitative dismantling of the intricate structural machinery essential for plasticity. Perhaps the most visually striking and functionally consequential structural change is the dramatic reduction in dendritic spines, the primary sites of excitatory synaptic transmission. Within the CA1 region, a critical hub for associative memory, postmortem studies and in vivo imaging in animal models consistently reveal a loss exceeding 50% of dendritic spines, with a particular vulnerability observed in the thin, highly plastic spines. 214 These thin spines are the substrates for new learning and memory formation; their plasticity allows for rapid changes in synaptic strength. Scheff et al. (2007) provided seminal quantitative evidence, demonstrating a 45-55% loss of synapses in the CA1 stratum radiatum in mild cognitive impairment (MCI) and AD compared to age-matched controls, correlating strongly with cognitive decline indices. 215 This spine loss is not uniform; it reflects the specific targeting of synapses by Aβ oligomers and the downstream pathways they activate. For instance, spines bearing GluN2B-containing NMDARs, crucial for plasticity induction but also susceptible to Aβ-induced excitotoxicity and internalization, are disproportionately affected. 97 Furthermore, the complement-mediated synaptic pruning actively eliminates synapses deemed dysfunctional, contributing significantly to this spine loss, particularly in the early stages of synaptic pathology within the dentate gyrus and CA regions. 22 The loss is further amplified by the cytoskeletal collapse driven by cofilin activation and RhoA signaling downstream of proBDNF/p75ᴺᵀᴿ, which actively dismantles the actin backbone of spines, causing their retraction and disappearance. 216
Concomitant with spine loss is the dissolution of the PSD nano-cluster organization. The PSD is a dense, highly organized protein matrix beneath the postsynaptic membrane, acting as a scaffold for neurotransmitter receptors, signaling molecules, and structural proteins. It orchestrates the precise localization and dynamic trafficking of AMPA and NMDA receptors, essential for synaptic transmission and plasticity. Aβ oligomers, through multiple converging pathways, disrupt this nano-architecture. p-tau mislocalized to dendritic spines physically displaces key scaffolding proteins like PSD-95 from the PSD. 75 PSD-95 is crucial for anchoring AMPARs and coupling NMDARs to downstream signaling pathways; its displacement leads to the disassembly of AMPAR nanodomains, scattering receptors and uncoupling them from signaling complexes. 217 Caspase-3 activation, triggered directly by Aβ-induced mitochondrial failure and ER stress or indirectly by p-tau, cleaves core PSD components such as PSD-95 and spectrin (αII-spectrin/fodrin), further destabilizing the entire postsynaptic structure. 118 Oxidative stress, through lipid peroxidation products like 4-HNE, can form adducts on PSD proteins, altering their function and interactions. 218 The STEP phosphatase, upregulated by Aβ via TNF-α signaling, also targets PSD proteins, including STEP61 which resides within the PSD and can dephosphorylate key components. 94 The net result is a fragmented, disorganized PSD incapable of efficiently clustering receptors or supporting the molecular machinery necessary for plasticity, effectively silencing the synapse even before its physical disappearance. 63
The presynaptic compartment is equally ravaged, leading to profound presynaptic vesicle depletion. This is often quantified by the significant reduction in levels of presynaptic marker proteins, most notably synaptophysin. Synaptophysin is a major integral membrane glycoprotein of small synaptic vesicles, and its levels are widely regarded as a reliable indicator of synaptic density and presynaptic integrity. Numerous studies have documented substantial decreases (30–50% or more) in synaptophysin immunoreactivity in the AD hippocampus, particularly in the CA1 and subiculum, correlating strongly with cognitive impairment severity and often preceding significant neuronal loss. 219 This depletion stems from multiple Aβ-driven mechanisms. Mitochondrial failure and ATP depletion (ripple the energy-intensive processes of vesicle cycling, docking, and neurotransmitter loading. 220 Excitotoxicity resulting from glutamate spillover due to impaired EAAT2 function can damage presynaptic terminals directly. 221 Furthermore, the collapse of the postsynaptic density and dendritic spine retraction eliminates the target for presynaptic vesicle release, leading to a functional and ultimately structural disconnection. 222 The loss of neurotrophic support, particularly BDNF further compromises presynaptic function, as BDNF-TrkB signaling is critical for regulating synaptic vesicle protein expression and release probability. 168 The depletion of synaptic vesicles directly impairs the reliability and fidelity of neurotransmitter release, contributing to synaptic transmission failure. 223
These structural insults collectively translate into devastating functional deficits in synaptic plasticity and network activity. The most extensively studied and arguably most critical impairment is the failure of LTP at hippocampal synapses, particularly the Schaffer collateral-CA1 pathway, which is the canonical model for associative learning. Aβ oligomers potently suppress LTP induction and maintenance through a multitude of the molecular mechanisms previously described. NMDAR dysfunction, including overactivation leading to calcium overload and subsequent internalization or modification by 4-HNE altering channel kinetics directly impairs the primary trigger for LTP. 224 The oxidation and inhibition of CaMKII, a master kinase essential for LTP expression through AMPAR phosphorylation and insertion, severs a key signaling pathway downstream of NMDAR activation. 225 Impaired TrkB signaling deprives the synapse of the trophic support necessary for LTP consolidation and stabilization. 54 The structural disintegration of the PSD (loss of PSD-95, cleavage of spectrin) physically removes the scaffold needed for AMPAR incorporation during LTP. 226 Consequently, high-frequency stimulation that robustly induces persistent LTP in healthy tissue fails to do so, or elicits only a transient and unstable potentiation, in Aβ-rich environments. Shankar et al. (2008) demonstrated that soluble Aβ oligomers extracted from AD brain tissue potently inhibit LTP in hippocampal slices and impair memory consolidation in vivo, providing a direct link to human pathology. 18 This LTP blockade fundamentally undermines the hippocampus's ability to strengthen connections in response to experience, preventing the encoding of new memories. 227
Paradoxically, while LTP is suppressed, LTD is often enhanced or pathologically facilitated in the AD hippocampus. Aβ oligomers lower the threshold for inducing LTD and can enhance its magnitude and duration. This pathological tilt towards synaptic weakening is driven by several mechanisms. Aβ binding to mGluR5 activates signaling pathways involving STEP, which dephosphorylates GluA1 at Ser845, promoting AMPAR endocytosis, a hallmark of LTD. 93 Aβ-induced NMDAR internalization may preferentially affect GluN2A-containing receptors, potentially shifting NMDAR composition towards GluN2B-dominated synapses that are more prone to calcium-dependent phosphatase activation (calcineurin/PP1) characteristic of LTD. 228 The accumulation of proBDNF and activation of p75ᴺᵀᴿ signaling actively promote synaptic weakening and retraction, mimicking and potentiating LTD mechanisms. 229 The depletion of synaptic AMPARs due to STEP activation, PSD destabilization, and potentially complement-mediated removal further lowers basal synaptic strength, creating a state primed for further depression. 230 This imbalance, favoring synaptic weakening over strengthening, creates a neural environment where experiences fail to leave lasting positive traces (impaired LTP), while negative or inactive connections are excessively eroded (enhanced LTD), actively degrading existing memory representations and preventing new stable ones from forming. 231
Beyond these classic forms of Hebbian plasticity, Aβ accumulation severely disrupts more complex forms of synaptic computation and network dynamics essential for information processing. Gamma oscillations (γ, 30–100 Hz) are fast rhythmic patterns of neuronal firing synchronized across populations of neurons, crucial for binding features of an object or event, attention, and memory retrieval. The hippocampus generates prominent gamma oscillations, particularly during exploration and memory tasks. Aβ oligomers potently disrupt gamma oscillations in vitro and in vivo. This disruption arises from impaired synaptic inhibition, as Aβ damages parvalbumin-positive (PV+) fast-spiking interneurons, which are critical pacemakers for gamma rhythms. 232 Aβ-induced oxidative stress and mitochondrial dysfunction compromise the energy-demanding high-frequency firing of these interneurons. 233 Furthermore, impaired glutamatergic transmission onto these interneurons, due to presynaptic vesicle depletion and postsynaptic AMPAR dysfunction, reduces their excitatory drive. 234 The dysregulation of NMDAR function also plays a role, as NMDAR hypofunction on interneurons can reduce network synchrony. 235 Knobloch et al. (2007) showed that Aβ oligomers selectively impair gamma oscillations but not slower rhythms in hippocampal slices, and similar disruptions are observed in AD mouse models and human patients. 236 The functional consequence is a failure in the precise temporal coordination of neuronal assemblies necessary for encoding and retrieving coherent memories. 237
Another critical form of plasticity disrupted is spike-timing-dependent plasticity (STDP). STDP is a Hebbian learning rule where the precise timing of pre- and postsynaptic action potentials determines whether a synapse is potentiated or depressed: if the presynaptic spike precedes the postsynaptic spike within a narrow time window (tens of milliseconds), LTP occurs; if the order is reversed, LTD results. STDP is believed to be fundamental for learning sequences, temporal coding, and refining neural circuits. Aβ oligomers severely impair STDP in hippocampal neurons. The mechanisms likely involve the same pathways that disrupt LTP/LTD, but with an added layer of sensitivity to timing precision. NMDAR dysfunction, particularly alterations in kinetics or subunit composition (e.g., reduced GluN2A, increased GluN2B), can disrupt the precise calcium influx dynamics required for STDP. 238 Impaired AMPAR trafficking affects the immediate postsynaptic depolarization needed for effective backpropagating action potentials, altering the temporal window for plasticity. 239 The overall reduction in synaptic strength and reliability makes it harder to achieve the precise spike timing necessary. 240 Consequently, the hippocampus loses its ability to adaptively modify synaptic weights based on the precise temporal correlations of neuronal activity, impairing the learning of sequences and associations that depend on timing. 241
The structural and functional consequences outlined—the loss of synaptic contacts, the dissolution of synaptic nano-architecture, the failure to potentiate connections, the pathological weakening of synapses, and the breakdown of rhythmic coordination and precise timing-dependent learning—collectively represent the catastrophic failure of hippocampal synaptic plasticity. This is not merely the loss of individual synapses but the collapse of the intricate network dynamics that transform transient experiences into lasting memories and enable spatial navigation. The CA1 region, the output node of the hippocampal trisynaptic circuit, bears the brunt of spine loss and LTP impairment, directly explaining the profound deficits in episodic and associative memory. 242 The dentate gyrus, vital for pattern separation (distinguishing similar experiences), suffers from complement-mediated pruning and disrupted gamma oscillations, leading to confusion and memory interference. 243 The subiculum, another major output region, shows significant synaptic loss, contributing to disorientation. This widespread synaptic and network pathology, directly instigated and amplified by Aβ oligomers through the molecular mechanisms described throughout this review, transforms the dynamic, plastic hippocampus into a structure incapable of its core cognitive functions, paving the path towards dementia (Table 9). 244
Structural and functional consequences.
Bridging mechanisms to cognitive decline
The intricate molecular cascades triggered by Aβ accumulation ultimately converge to erode hippocampal synaptic integrity. This erosion directly links cellular pathology to the defining clinical hallmark of AD: progressive cognitive decline. 16 This translation from synaptic sabotage to measurable deficits in learning, memory, and executive function represents a critical nexus in AD pathogenesis. Understanding these bridging mechanisms is paramount, as they elucidate how the molecular chaos orchestrated by Aβ oligomers, hyperphosphorylated tau, neuroinflammatory onslaught, oxidative stress, mitochondrial collapse, and neurotrophic failure culminates in the failure of hippocampal-dependent cognition. 246 Each mechanism disrupts distinct facets of hippocampal information processing, collectively dismantling the neural architecture essential for navigating the complexities of human experience.
A primary conduit linking Aβ pathology to cognitive impairment is the blockade of LTP, the electrophysiological cornerstone of synaptic strengthening underlying learning. Aβ oligomers, particularly soluble Aβ42 assemblies, potently inhibit LTP in the hippocampal Schaffer collateral-CA1 pathway, a phenomenon extensively documented in rodent models. This inhibition stems from Aβ's targeted assault on glutamatergic signaling. As detailed previously, Aβ binding to cellular PrPᶜ or mGluR5 activates Fyn kinase, leading to phosphorylation of the NR2B subunit of NMDARs. This aberrant phosphorylation promotes excessive NMDAR opening, resulting in pathological calcium influx. 19 Concurrently, Aβ-triggered signaling cascades activate STEP, which dephosphorylates GluA1 subunits of AMPARs at Ser845, promoting their endocytosis. 104 The combined effect is a catastrophic disruption of the calcium signaling microdomains essential for LTP induction and a net reduction in synaptic AMPARs, the primary mediators of fast excitatory transmission. The functional consequence of this LTP blockade is starkly evident in spatial memory deficits. In vivo electrophysiology studies in Aβ-infused or transgenic AD mouse models consistently reveal impaired LTP magnitude and duration in the CA1 region, directly correlating with profound errors in spatial navigation tasks. The radial arm maze, a gold standard for assessing hippocampal-dependent spatial working and reference memory, provides compelling evidence. Aβ-overexpressing mice (e.g., APP/PS1, Tg2576) exhibit significantly increased working memory errors (revisiting baited arms within a trial) and reference memory errors (entering unbaited arms), reflecting an inability to form or utilize a stable spatial map.247,248 Critically, pharmacological interventions that rescue NMDAR function (e.g., memantine, partial NMDAR antagonists) or enhance AMPAR trafficking not only restore LTP magnitude in these models but also significantly ameliorate radial arm maze performance.249,250 Similarly, genetic reduction of STEP expression prevents Aβ-induced GluA1 internalization and rescues both LTP deficits and spatial memory impairments. This causal chain—Aβ oligomers → NMDAR/AMPAR dysregulation → LTP blockade → spatial disorientation—establishes a direct molecular pathway from Aβ accumulation to a core cognitive deficit in early AD: the loss of spatial orientation and navigation skills, often manifesting clinically as getting lost in familiar environments. 104
Complementing the functional impairment of synaptic plasticity is the devastating structural loss of synapses, primarily through Aβ and tau-driven dendritic spine elimination. Dendritic spines, the postsynaptic sites harboring glutamate receptors, are dynamic structures whose density and morphology are fundamental to synaptic strength and network connectivity. 214 Aβ oligomers initiate spine loss through multiple converging pathways: promoting calcineurin-dependent cofilin activation leading to actin depolymerization, inducing caspase-3-mediated cleavage of synaptic scaffolds like PSD-95, and triggering tau hyperphosphorylation and missorting. Crucially, hyperphosphorylated tau (p-tau), particularly at sites like Ser202/Thr205 (AT8 epitope) and Ser396, actively contributes to spine pathology. Dendritic p-tau displaces PSD-95 from the postsynaptic density, destabilizing the AMPAR nanodomain and impairing receptor trafficking. 21 Furthermore, p-tau disrupts microtubule-dependent transport, leading to the accumulation of vesicles and organelles in the soma and impaired delivery of AMPARs and other critical synaptic components to spines. 128 The consequence is a progressive, often dramatic, reduction in spine density, particularly affecting thin, plastic spines crucial for learning new information, observed in both AD mouse models and postmortem human hippocampal tissue. This structural synaptic loss is inextricably linked to the decline of episodic memory—the ability to recall specific autobiographical events—a hallmark of AD. Robust evidence for this bridge comes from human biomarker studies correlating cerebrospinal fluid (CSF) measures with cognitive performance. Longitudinal cohort studies consistently demonstrate that lower CSF levels of Aβ42 (reflecting Aβ sequestration into plaques) and higher levels of phosphorylated tau (p-tau, e.g., p-tau181, p-tau217) are strongly associated with accelerated rates of episodic memory decline, measured by tests like the Free and Cued Selective Reminding Test or the Rey Auditory Verbal Learning Test.251,252 Importantly, advanced neuroimaging techniques like high-resolution structural magnetic resonance imaging (MRI) and positron emission tomography (PET) with synaptic vesicle glycoprotein 2A (SV2A) ligands (e.g., [11C]UCB-J) reveal that the rate of hippocampal atrophy and synaptic density loss in the medial temporal lobe, including the hippocampus proper and entorhinal cortex, mediates the relationship between CSF Aβ42/p-tau levels and episodic memory performance.253,254 Individuals with both pathological CSF Aβ42 and elevated p-tau exhibit the most rapid synaptic loss and the steepest decline in episodic memory function. This convergence of biochemical, structural, and cognitive data firmly establishes dendritic spine loss, driven by the synergistic toxicity of Aβ and p-tau, as a primary structural correlate of the erosion of personal, event-based memory that defines the lived experience of AD.
Beyond the direct synaptic targeting by Aβ and tau, dysregulated hippocampal innate immunity, specifically aberrant complement-mediated synaptic pruning, forms a third critical bridge to cognitive decline, with particular impact on pattern separation. In the AD brain, the complement cascade, a key innate immune component, is maladaptively co-opted. Microglia, activated via Aβ oligomer binding to receptors such as TLR4 and TREM2, release the initiator protein C1q. C1q binds to and tags synapses for elimination. This triggers the cleavage of C3 to C3b, which opsonizes the tagged synapses. Microglial complement receptor 3 (CR3, integrin αMβ2) then binds C3b, initiating phagocytic engulfment and synaptic removal, a process termed synaptic pruning.22,255 While developmental pruning is essential for refining neural circuits, chronic, Aβ-driven microglial activation leads to excessive and inappropriate complement-mediated pruning in the adult hippocampus. The dentate gyrus (DG) subfield of the hippocampus is exquisitely vulnerable to this process. The DG is the primary entry point for cortical information into the hippocampus and is critically responsible for pattern separation, the computational process of transforming similar input patterns into highly distinct, non-overlapping output representations in the downstream CA3 region. This function is vital for disambiguating similar experiences and storing them as distinct memories without interference (e.g., remembering where you parked your car today versus yesterday). 22 Complement proteins, particularly C1q and C3, are markedly upregulated in the DG in AD models and human AD brains. Genetic ablation of C1q, C3, or CR3 in AD mouse models (e.g., APP/PS1; C3-/-, APP/PS1; CR3-/-) significantly reduces microglial engulfment of synapses, preserves synaptic markers like synaptophysin and PSD-95 specifically in the DG, and rescues deficits in pattern separation.256,257 Behaviorally, pattern separation is assessed using tasks like the spontaneous location recognition (SLR) task with similar locations, or the continuous spatial alternation task in a T-maze with varying degrees of spatial similarity. Aβ-overexpressing mice exhibit profound impairments in distinguishing highly similar spatial contexts, a deficit rescued by complement pathway inhibition. 258 In humans, functional MRI (fMRI) studies reveal impaired DG/CA3 activation patterns during tasks requiring discrimination of highly similar visual scenes or objects in individuals with MCI and early AD, correlating with CSF Aβ levels. 259 Moreover, PET imaging of translocator protein (TSPO), a marker of microglial activation, shows heightened signal in the DG/CA3 region in MCI/AD patients, correlating with both CSF complement component levels (e.g., C3) and pattern separation performance deficits. 260 Thus, Aβ-induced activation of the complement cascade → excessive microglial pruning of DG synapses → failure of pattern separation → confusion between similar experiences represents a distinct pathway contributing to the cognitive disorientation and memory interference characteristic of AD.
Finally, the pervasive mitochondrial dysfunction instigated by Aβ accumulation provides a fourth bridge to cognitive decline, particularly impacting cognitive flexibility and executive functions reliant on hippocampal-prefrontal circuits. Aβ oligomers, especially those localized to mitochondria (e.g., via binding to the mitochondrial matrix protein ABAD), directly impair mitochondrial function. They inhibit key respiratory chain complexes (notably complex IV, CCyt c oxidase), increase ROS production, and promote the opening of the mPTP. 255 This leads to a cascade of detrimental events: collapse of mitochondrial membrane potential (ΔΨm), impaired ATP synthesis, calcium buffering failure, and release of pro-apoptotic factors. Synaptically, the energy deficit and oxidative stress impair critical processes: ATP-dependent actin dynamics necessary for spine motility and stability are compromised, ROS oxidatively modify key plasticity enzymes like CaMKII (inactivating it), and calcium dysregulation disrupts signaling cascades. 26 While this impacts all hippocampal subfields, the functional consequence extends beyond pure memory formation to impair cognitive flexibility—the ability to adapt behavior in response to changing rules or contingencies, a higher-order executive function dependent on intact hippocampal-prefrontal cortex interactions. The attentional set-shifting task (ASST), adapted from human neuropsychology (Wisconsin Card Sorting Test), effectively probes this in rodents. It requires learning to discriminate between stimuli based on different perceptual dimensions (e.g., odor versus texture) and then flexibly shifting attention to a previously irrelevant dimension when the reward contingency changes. Aβ-overexpressing mice exhibit specific deficits in the extra-dimensional shift stage of the ASST, where they perseverate on the previously relevant dimension despite negative feedback, indicating impaired cognitive flexibility. 261 Critically, this deficit aligns with mitochondrial dysfunction markers—such as reduced CCyt c oxidase activity and elevated ROS, measured in the hippocampus and prefrontal cortex. Pharmacological interventions targeting mitochondrial health demonstrate causality: mitochondrial antioxidants (e.g., MitoQ, SS-31) and agents that enhance mitochondrial biogenesis (e.g., peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α) activators) restore mitochondrial function by reducing ROS and increasing ATP, improve hippocampal synaptic integrity, and specifically rescue the extra-dimensional shift deficit in AD animal models.161,262 Similarly, agents blocking mPTP opening (e.g., cyclosporine A analogs) confer protection. In humans, deficits in cognitive flexibility, assessed by tasks like Trail-Making tests or set-shifting paradigms, are prominent early features of AD. Magnetic resonance spectroscopy studies reveal reduced levels of high-energy phosphates (e.g., phosphocreatine) and increased markers of oxidative stress in the hippocampi of MCI/AD patients, which correlate with the severity of executive function impairments, including set-shifting deficits. 263 Therefore, the pathway Aβ → mitochondrial dysfunction (ROS, energy failure, Ca2+ dysregulation) → impaired synaptic plasticity and network coordination → reduced cognitive flexibility underscores how Aβ-induced bioenergetic collapse disrupts the dynamic neural computations necessary for adaptive behavior, contributing significantly to the rigidity and impaired problem-solving observed in AD. 264
In summation, the progression from Aβ accumulation to cognitive decline in AD spans multiple intersecting molecular pathways within the hippocampus. The convergence of oxidative stress, trophic factor loss, and synaptic dysfunction produces learning and memory impairments, while receptor dysfunction directly disrupts the synaptic substrates required for spatial navigation. The combined actions of Aβ oligomers and dendritic p-tau on synaptic structure—manifested as spine loss and reflected in CSF biomarkers—progressively erode the neural foundation of episodic memory. 265 The dysregulation of the complement system, hijacked by Aβ-driven neuroinflammation, leads to excessive synaptic pruning in the dentate gyrus, crippling the essential computational process of pattern separation and leading to confusion. 22 Finally, the pervasive mitochondrial failure induced by Aβ disrupts the bioenergetic and signaling foundation of synaptic plasticity, impairing not only memory formation but also the cognitive flexibility required for adaptive behavior. 264 These mechanisms are not isolated; Aβ-induced oxidative stress exacerbates tau pathology and neuroinflammation, 266 complement activation can be amplified by synaptic debris resulting from mitochondrial failure, 267 and the energy crisis impairs all aspects of synaptic repair and plasticity. 214 This intricate web of molecular sabotage ultimately unravels the complex tapestry of hippocampal function, transforming molecular lesions within synapses into the profound cognitive disarray that defines AD. 16 Targeting these specific bridging mechanisms—rescuing LTP, preventing spine loss, inhibiting aberrant complement pruning, and restoring mitochondrial health—represents a promising multi-pronged therapeutic strategy aimed not just at removing pathological hallmarks, but at preserving and restoring the cognitive essence of the individual 268 (Table 10).
Bridging mechanisms to cognitive decline.
Gender and racial differences in hippocampal synaptic vulnerability in AD
The molecular mechanisms of Aβ-induced synaptic sabotage detailed in this review represent a core pathological framework. However, the significant clinical heterogeneity in AD susceptibility, progression, and symptomatology underscores the potential influence of demographic factors such as biological sex and genetic ancestry. Emerging evidence suggests these factors may modulate key pathways, including hippocampal glutamate receptor function and neuroinflammatory synapse elimination, thereby contributing to differential vulnerability.275,276 An integration of this perspective is crucial for a comprehensive understanding of AD pathogenesis and the development of equitable therapeutics.
Epidemiological data consistently show that women have a higher age-adjusted prevalence of AD and often experience more rapid cognitive decline than men. 277 This disparity may be rooted in sex-specific vulnerabilities within hippocampal synaptic plasticity circuits. Preclinical studies indicate that female rodent models of AD often exhibit more severe deficits in hippocampal LTP and greater synaptic loss compared to males 278 A key mechanistic candidate is the neuroprotective role of sex hormones, particularly 17β-estradiol. Estradiol enhances synaptic resilience by potentiating NMDAR function through increased synaptic localization and phosphorylation of the GluN2B subunit—a primary target of the Aβ/Fyn kinase signaling axis described. 279 Furthermore, estradiol robustly activates BDNF/ TrkB signaling, a pathway whose collapse is central to Aβ toxicity. 280 The perimenopausal decline in estrogen may thus remove a critical buffer against Aβ-induced excitotoxicity and trophic withdrawal, rendering the female hippocampus more susceptible. Sex differences also extend to neuroimmune responses. Microglia, the primary mediators of complement-dependent synaptic pruning, exhibit sexually dimorphic phenotypes, with evidence suggesting female microglia may adopt a more reactive state in some AD models, potentially accelerating inflammatory synapse loss. 281
Racial and ethnic disparities in AD risk are well-documented, with African American and Hispanic populations exhibiting higher incidence and prevalence compared to non-Hispanic White populations. 282 While socio-economic and healthcare access factors are major contributors, biological differences rooted in population genetics may also influence synaptic resilience. Genome-wide association studies have identified AD risk variants with differing allele frequencies across ancestral groups. Importantly, several of these variants are in genes encoding synaptic proteins or regulators of neuroinflammation. For instance, polymorphisms in the GRIN2B gene, which encodes the GluN2B NMDAR subunit, have been associated with AD risk and cognitive performance in a population-specific manner. 283 Given the pivotal role of GluN2B in Aβ-induced NMDAR hyperactivation and calcium dyshomeostasis, genetic variation affecting its function could differentially modulate synaptic vulnerability. Similarly, variants in genes regulating microglial activity and the complement cascade, such as TREM2, CR1, and CLU, show varying frequencies across ancestries.284,285 As the C1q-complement pathway is a potent mediator of Aβ-triggered synaptic elimination, such genetic differences could lead to divergent neuroinflammatory responses to Aβ pathology, altering the trajectory of hippocampal synaptic erosion. However, direct mechanistic studies linking population-specific genetic variants to defined alterations in glutamate receptor trafficking, LTP/LTD balance, or synaptic density in AD are profoundly lacking, as most pathophysiological research utilizes model systems that do not reflect human genetic diversity.
Addressing these knowledge gaps demands coordinated, multi-disciplinary work. Future efforts should include preclinical models incorporating genetic diversity and controlled hormonal environments. Human postmortem studies from diverse, well-characterized brain banks are needed to relate synaptic proteins (e.g., GluA1, PSD-95), receptor phosphorylation, and glial markers to sex, genetic ancestry, and cognitive status. In vivo, combining advanced synaptic PET imaging (e.g., SV2A tracers) and fluid biomarkers with demographic and genetic data in large diverse cohorts will be essential for monitoring synaptic integrity across populations. Clinical trials of synaptic-resilience–targeted therapies, such as TrkB agonists, STEP inhibitors, and complement modulators, should be powered for predefined analyses to detect differential responses across sex and genetic backgrounds. Overall, although Aβ-driven synaptic failure forms the central AD pathway, its extent is modulated by gender and racial factors through hormonal influences, genetic variation in synaptic and immune genes, and socio-environmental determinants of brain health. Defining these interactions is crucial for precision medicine and equitable AD prevention and treatment.
Geographical considerations and global consistency of evidence
The research synthesized in this review stems from a broad international effort, with studies from North America, Europe, Asia, and beyond, enabling evaluation of the universality of proposed molecular pathways. Notably, core mechanisms of Aβ-induced synaptic dysfunction—such as oligomer binding to PrPᶜ/mGluR5, Fyn-dependent NMDAR dysregulation, tau hyperphosphorylation and mislocalization, complement-driven pruning, and mitochondrial deficits—have been consistently reproduced across diverse regions and experimental contexts, highlighting their conserved role in AD pathogenesis.
At the same time, global literature reveals important contextual nuances. Population-specific studies have clarified the effects of genetic variants (e.g., TREM2 R47H) on neuroinflammatory signaling, while regional epidemiological differences (e.g., metabolic health, co-pathology profiles) may influence the prominence or manifestation of synaptic failure pathways. Thus, while Aβ-driven synaptic toxicity forms a robust central framework, population-level heterogeneity adds an essential layer of complexity.
Overall, the international concordance in mechanistic evidence strongly supports the core model presented here, while emphasizing a key future priority: therapeutic development and clinical trials must incorporate geographical and genetic diversity to ensure broad applicability of strategies targeting convergent pathways, such as synaptic resilience enhancement or modulation of neuroinflammation. The existing cross-national reproducibility provides a solid foundation for such translational progress
Conclusion and future directions
The intricate molecular narrative woven throughout this review unequivocally positions Aβ oligomers as the primary instigators of a devastating cascade that sabotages hippocampal synaptic plasticity, the fundamental cellular process underpinning learning and memory, ultimately leading to the cognitive collapse characteristic of AD. 286 Rather than a linear cascade, AD involves a synergistic convergence of interacting pathological axes that collectively dismantle hippocampal synaptic structure and function, rather than a simple linear cascade. The emerging synthesis is that soluble Aβ oligomers, especially toxic Aβ42 assemblies, initiate synaptic pathology by engaging membrane receptors such as PrPᶜ and mGluR5, activating Fyn kinase and inducing aberrant phosphorylation of NR2B on NMDARs. 19 This drives pathological NMDAR hyperactivation and postsynaptic Ca2+ overload, disrupting critical Ca2+ microdomains.19,250 Parallel Aβ-induced signaling elevates STEP activity, causing GluA1 dephosphorylation, AMPAR endocytosis, and weakened excitatory transmission necessary for LTP. 94 These glutamatergic disruptions are amplified by Aβ-triggered tau pathology. Aβ activates kinases such as GSK-3β and CDK5, promoting tau hyperphosphorylation at sites including Ser202/Thr205 and Ser396.21,128 p-tau then mislocalizes to dendrites, where it disrupts synaptic structure by displacing PSD-95 from the postsynaptic density and dismantling AMPAR-supporting nanodomains. 21 Additionally, p-tau destabilizes microtubules and impairs trafficking of key synaptic cargos, including AMPARs, thereby depriving spines of essential components. 128
Aβ oligomers act as the primary molecular trigger, initiating a toxic cascade that directly disrupts synaptic plasticity in the hippocampus, the cellular foundation of learning and memory. This synaptic failure is the critical link between initial protein pathology and the progressive cognitive collapse observed in AD.15,286 The core synthesis, emerging from decades of relentless research, can be distilled as follows: Soluble Aβ oligomers, particularly neurotoxic assemblies of Aβ42, initiate synaptic pathogenesis by hijacking specific membrane receptors such as PrPᶜ and mGluR5. 287
Aβ oligomers activate Fyn kinase, causing aberrant phosphorylation of the NR2B subunit of NMDARs, leading to pathological receptor activity. Synaptic dysfunction is further exacerbated by Aβ-induced neuroinflammation, as engagement of microglial receptors such as TLR4 and TREM2 triggers the release of pro-inflammatory cytokines (IL-1β, TNF-α) and complement proteins, including C1q.22,255 Microglial C1q deposition on synapses tags them for elimination via the complement cascade (C1q → C3 → C3b binding to microglial CR3), leading to phagocytic engulfment and destruction of functional synapses, a maladaptive recapitulation of developmental pruning gone awry in the adult brain.22,255,257 Astrocytes, activated in response to microglial signals and directly by Aβ, contribute by downregulating glutamate transporters (e.g., EAAT2 via IL-1β/JNK signaling), leading to glutamate spillover, excitotoxicity, and further synaptic damage. 135 Adding to this multifaceted assault is the Aβ-induced bioenergetic crisis centered on mitochondrial failure. Aβ oligomers, particularly those interacting with mitochondrial proteins like ABAD, directly inhibit the electron transport chain (especially complex IV), increase ROS production, and promote the opening of the mPTP.26,157 This results in collapse of the mitochondrial membrane potential, severe ATP depletion, failure of mitochondrial Ca2+ buffering, and release of pro-apoptotic factors. Synaptically, the energy deficit cripples ATP-dependent processes like actin dynamics (essential for spine stability and motility), while mitochondrial ROS oxidatively damage key plasticity enzymes such as CaMKII, rendering them dysfunctional.26,97 The catastrophic convergence of glutamate receptor dysregulation, p-tau–induced PSD destabilization, inflammatory synaptic elimination, and mitochondrial energy failure generates a destructive cascade that surpasses endogenous neuroprotection. A major consequence is disruption of the BDNF/TrkB axis: Aβ suppresses BDNF via mechanisms such as miR-206 upregulation and impairs TrkB signaling, while oxidative stress increases proBDNF, which activates p75ᴺᵀᴿ–RhoA pathways to drive spine collapse.167,229 This results in a profound loss of trophic support essential for synaptic stability and plasticity.
These insults culminate in synaptic-targeted apoptotic signaling. Intrinsic apoptosis arises from Aβ-induced ROS and Ca2+ imbalance, promoting Bax/Bak oligomerization, CCyt c release, and caspase-9/-3 activation.26,118 Extrinsic apoptosis is driven by TNF-α and related cytokines activating death receptors, 135 while Aβ-induced ER stress triggers the PERK/eIF2α/CHOP pathway, upregulating BIM and PUMA. 288 These routes converge on caspase-mediated cleavage of key synaptic proteins (e.g., PSD-95, spectrin), producing irreversible “synaptic apoptosis”. 118 The structural consequence is a profound loss of synapses, particularly thin, plastic spines in the hippocampus, dissolution of PSD nano-clusters, and presynaptic vesicle depletion. Functionally, this manifests as a severe impairment in LTP, facilitation of LTD, disruption of network oscillations (e.g., gamma rhythms), and failure of STDP, the electrophysiological signatures of learning and memory mechanisms gone awry. 108 Specific synaptic failures directly cause AD defining cognitive deficits: Aβ-blocked LTP impairs spatial memory, Aβ/p-tau spine loss disrupts episodic memory, excessive complement pruning in the dentate gyrus hampers pattern separation, and mitochondrial failure reduces cognitive flexibility. This progression from Aβ oligomers to dementia involves coordinated molecular sabotage, including receptor dysfunction, tau pathology, inflammatory phagocytosis, and bioenergetic collapse, which together degrade synaptic support, trigger synaptic loss, and cause ultimate hippocampal circuit failure. 286
Despite major progress, substantial gaps still impede the development of effective therapies. A key unresolved issue is the specificity of Aβ strains and assemblies. Although soluble oligomers are recognized as the primary synaptotoxic species, the exact structural features of the most pathogenic assemblies (e.g., dodecamers, trimers, paranuclei) and their distinct effects on different synaptic plasticity mechanisms remain insufficiently defined.289,290 Furthermore, how the dynamic interconversion between Aβ oligomers, protofibrils, and fibrils, and the balance among them affects synaptic toxicity at different disease stages remains unclear. Key unanswered questions include whether fibril formation sequesters toxic oligomers or instead catalyzes further oligomer generation, and how distinct Aβ strains engage specific receptors (e.g., PrPᶜ, mGluR5) to trigger unique pathological pathways. Resolving this complexity requires advanced techniques such as cryo-electron microscopy, single-molecule fluorescence, and conformation-specific antibodies to identify the most promising therapeutic targets.290–292 Linked to this is the enigma of subfield vulnerability within the hippocampus. Why are certain regions, like the DG and CA1, seemingly more susceptible to Aβ and tau toxicity in early stages, while CA3 might exhibit different resilience or vulnerability patterns? 293 The selective vulnerability of hippocampal regions in AD may result from differences in receptor expression, glial function, or intrinsic neuronal properties. 243 Clarifying this molecular basis could uncover protective mechanisms for therapy, guide targeted interventions, and explain specific cognitive deficits like impaired pattern separation. A key unresolved issue is how Aβ disrupts metabolic coupling between astrocytes and neurons, particularly the astrocyte-neuron lactate shuttle, potentially by impairing astrocyte glucose metabolism or lactate export.294,295 Understanding whether enhancing this coupling with ketone bodies can bypass Aβ-induced mitochondrial defects is critical for developing metabolic therapies. Investigating why some synapses resist degeneration despite pathology could reveal new therapeutic targets. A key resilience factor is the use of ketone bodies as an alternative energy source, which bypasses glycolysis to fuel mitochondria. Evidence indicates elevating ketone levels improves mitochondrial function, reduces oxidative stress, and enhances cognition in models and patients296,297
Exploring endogenous mechanisms of synaptic resilience presents a promising therapeutic direction. Key questions include why certain synapses withstand degeneration despite comparable Aβ and tau burdens. Defining the molecular signatures of resilient synapses may reveal novel targets. Among proposed resilience factors, alternative energy substrates, especially ketones, are notable. Ketone bodies (β-hydroxybutyrate, acetoacetate) offer efficient mitochondrial fuel by bypassing glycolysis and entering the tricarboxylic acid cycle. Evidence indicates that elevating ketone levels via ketogenic diets, MCTs, or exogenous ketone esters/salts can improve mitochondrial function, reduce oxidative stress, enhance synaptic markers, and improve cognition in AD models and preliminarily in humans.293,298 Critical mechanistic questions remain: Do ketones enhance synaptic ATP production, support neurotransmitter synthesis (e.g., glutamate, GABA), reduce mitochondrial ROS, modulate inflammation, or alter epigenetic regulation of neuroprotective genes? Clarifying these pathways could optimize ketone-based interventions. Additional resilience factors also warrant investigation, including protein chaperone upregulation, enhanced synaptic autophagy flux, and activation of neurotrophic or antioxidant pathways.
Ultimately, translating these mechanistic insights into therapeutics requires combinatorial, multi-target strategies, given the repeated failures of monotherapies, particularly Aβ-centric approaches:
Aβ oligomer-targeted strategies and microglial modulation: Moving beyond bulk Aβ removal, future efforts need precision targeting of specific synaptotoxic oligomers using conformation-specific antibodies, small molecule inhibitors of oligomer formation or receptor binding (e.g., blocking PrPᶜ or mGluR5), or enhancing clearance mechanisms.19,290,298 Critically, modulating microglial function is paramount. Instead of broad suppression, strategies should aim to promote a protective, phagocytic phenotype (e.g., via TREM2 agonism) that clears Aβ and debris while dampening excessive inflammation and complement-mediated synaptic pruning.22,255,299 Inhibiting specific components of the complement cascade (e.g., C1q, C3, C3aR, C5aR) also holds significant promise for preventing aberrant synaptic loss.22,255,257 TrkB signaling agonists: Addressing the BDNF/TrkB signaling collapse is crucial. Direct TrkB agonists, positive allosteric modulators of TrkB, or strategies to boost endogenous BDNF levels (e.g., inhibiting miR-206, promoting BDNF gene expression, facilitating proBDNF cleavage) offer potential routes to enhance synaptic resilience, promote plasticity, and support neuronal survival collapse.167,229,300 Tau immunotherapy and kinase modulation: Targeting pathogenic tau, particularly dendritic p-tau, is essential. Tau immunotherapies designed to clear pathological tau aggregates or block its spread are in active development.21,301 Equally important is inhibiting the kinases responsible for pathogenic tau phosphorylation (e.g., GSK-3β, CDK5) or activating phosphatases that can dephosphorylate tau.21,128 Caspase inhibitors: Given the central role of caspase activation in synaptic apoptosis and the cleavage of synaptic proteins, targeted caspase inhibition, particularly of caspase-3 and −6, represents a promising strategy to halt the final stages of synaptic destruction.118,302 Developing brain-penetrant, specific caspase inhibitors with favorable safety profiles is a key challenge. Mitochondrial protectants and antioxidants: Protecting mitochondria from Aβ-induced damage is vital. Strategies include mitochondrial-targeted antioxidants (e.g., MitoQ, SS-31), inhibitors of mPTP opening (e.g., cyclosporine A analogs), compounds enhancing mitochondrial biogenesis (e.g., PGC-1α activators), and providing alternative fuels like ketones to support energy production26,157,293,303 Combination therapies: Given the interconnected nature of the pathways, rationally designed combination therapies targeting Aβ/tau pathology, neuroinflammation, mitochondrial dysfunction, and synaptic support simultaneously or in sequence will likely be necessary for robust efficacy. For instance, clearing Aβ oligomers while simultaneously boosting TrkB signaling and inhibiting caspases might protect synapses more effectively than any single approach (as summarized in Figure 1).
Converging molecular pathways of aβ-induced synaptic dysfunction in AD. Schematic overview illustrating how soluble amyloid-β (Aβ) oligomers act as a primary trigger, disrupting hippocampal synaptic plasticity through six interlinked pathogenic pathways:
In conclusion, Aβ-induced synaptic dysfunction in the hippocampus involves a complex, multi-pathway cascade initiated by soluble oligomers. This process disrupts glutamate signaling, promotes tau pathology, triggers inflammatory synapse elimination, impairs mitochondrial function, and ultimately leads to trophic support collapse and synaptic loss. Decoding this molecular cascade not only clarifies the pathogenesis of AD but also offers a roadmap for therapy. Future progress depends on addressing key unresolved questions, such as Aβ strain specificity, regional vulnerability, glial metabolic roles, and endogenous resilience mechanisms. Developing integrated, multi-targeted interventions informed by this molecular logic is essential to preserve synaptic integrity and cognitive function, moving closer to preventing or mitigating the impact of AD.
Footnotes
Acknowledgements
We utilized artificial intelligence (AI) tools for language polishing and editing; however, the scientific content, accuracy, and interpretations remain the sole responsibility of the authors.
Author contribution(s)
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.