Dysregulated mRNA Translation in Alzheimer’s Disease: Mechanisms, Therapeutic Opportunities, and Challenges for Clinical Translation

Abstract

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder and a growing public health concern. Its core pathological features include cerebral amyloid-β deposition and tau aggregation. mRNA translation is a tightly regulated process essential for gene expression and protein synthesis. It plays critical roles in neural development, synaptogenesis, and synaptic plasticity. In recent years, dysregulated translational control has been recognized as an important pathogenic mechanism in several neurological diseases, including AD. This review provides an integrated overview of the biological basis and mechanisms of dysregulated mRNA translation in AD, including translation initiation and elongation, local synaptic translation, abnormalities in RNA-binding proteins and stress granules, tRNA dysregulation, and the effects of AD-related genetic factors on translational control. Beyond the classic mechanisms of protein aggregation, the onset and progression of AD are also accompanied by abnormal translational regulation, involving initiation, elongation, and local synaptic translation. A major feature of these abnormalities is the coexistence of suppressed global protein synthesis and selectively enhanced translation of certain disease-related mRNAs, involving the integrated stress response, eIF2α/eIF4E signaling, eEF2K/eEF2 pathway, and RNA-binding protein dysfunction. Translational dysregulation provides a new perspective for understanding the molecular pathology of AD and has prompted the exploration of interventions targeting key nodes in translational control. Strategies targeting the integrated stress response, cap-dependent translation initiation, eEF2K, and tau–RNA-binding protein interactions have shown potential to improve synaptic function and cognitive phenotypes in animal models, although their safety and clinical translatability require further evaluation. This review provides an up-to-date reference for understanding translation-centered mechanisms in AD and evaluating emerging therapeutic strategies targeting translational control.

Introduction

Alzheimer’s disease (AD) is a multifactorial neurodegenerative disorder which is the most common cause of dementia in older adults worldwide.^1,2 It is estimated to account for approximately 60%–80% of all dementia cases, imposing a substantial burden on patients, families, caregivers, and healthcare systems.³ Currently, more than 50 million people are living with AD globally, which is expected to exceed 150 million by 2050, largely driven by population aging.^2,4

Clinically, AD is characterized by a gradual decline in cognition, daily functioning, and behavioral performance. Memory impairment is often the earliest and most prominent symptom, particularly difficulty in recalling recent events.⁵ Clinically, AD can be classified by age at onset into early-onset AD (EOAD) and late-onset AD (LOAD).⁶ Most patients develop symptoms at or after 65 years of age and are classified as having LOAD, which accounts for more than 90% of all cases. In contrast, EOAD is defined as onset before 65 years of age and is less common, representing approximately 5% to 10% of cases.⁶

The neuropathological basis of AD primarily involves extracellular deposition of amyloid-beta (Aβ), leading to amyloid plaque formation, and intracellular aggregation of hyperphosphorylated tau, resulting in neurofibrillary tangles. These pathological alterations usually begin many years before apparent memory symptoms. With continued accumulation of abnormal proteins, a series of downstream pathological events are triggered, including neuroinflammation, glial activation, synaptic injury, and neuronal loss. These interacting changes gradually drive the transition from subtle cognitive decline to overt dementia.^7–9 The pathogenesis of AD is complex. Studies have shown that pathogenic variants in PSEN1, PSEN2, and APP can cause familial AD (FAD), which follows an autosomal dominant inheritance pattern, whereas APOE is currently the most well-established risk gene associated with LOAD.¹⁰ Beyond the classic paradigm of abnormal protein aggregation, the onset and progression of AD involve multiple cellular pathways, including mitochondrial dysfunction, oxidative stress, neuroinflammation, and dysregulated mRNA translation.¹¹ In eukaryotic cells, mRNA translation is tightly regulated to control gene expression and maintain proteome homeostasis. This process supports cellular function and survival.^11,12

Recent studies suggest that RNA-level abnormalities are involved in the pathogenesis of AD, particularly through impaired mRNA translation and dysfunction of RNA-binding proteins (RBPs).^13,14 Translational dysregulation is present in both the preclinical and symptomatic stages of AD.¹⁵ Aβ42 can disrupt the synthesis of proteins related to synaptic function by affecting specific cap-dependent mRNA translation initiation pathways, thereby promoting AD pathogenesis.¹⁶ Studies have shown that multiple pathways exhibit disturbed protein synthesis during AD progression, including those involved in synaptic function, mitochondrial function, and lysosomal function.¹⁷ Defects in protein synthesis may precede the appearance of overt cognitive phenotypes during the disease course. Furthermore, interactions among tau, RNA, and the RBP TIA1 can drive tau phase separation at physiological concentrations and promote the formation of more toxic tau oligomers.¹⁸ Translation is also influenced by stress-responsive signaling pathways that integrate extracellular and intracellular cues. These include oxidative stress, endoplasmic reticulum stress mediated by the unfolded protein response, and metabolic deficits. These processes are associated with aging and neurodegeneration.¹¹ Despite recent findings on the relationship between AD and mRNA-related processes, the available evidence remains fragmented across different aspects of translational regulation, and an up-to-date comprehensive review is warranted.

To address this gap, we focus on the key mechanisms of mRNA translation dysregulation in AD. We introduce the basic mechanisms of translational regulation and summarize recent advances in the study of translational dysregulation in AD, including the underlying mechanisms and related therapeutic strategies. By bringing together recent findings on different stages and levels of translational control, we examine the links among these processes in AD and discuss the challenges associated with translating these findings into therapeutic approaches. This review provides an up-to-date reference on the role of dysregulated mRNA translation in AD and its potential therapeutic relevance.

Mechanisms of mRNA Translational Control

Protein translation is the final step of gene expression. It converts the nucleotide sequence encoded by mRNA into an amino acid sequence, thereby generating functional proteins. This process requires ribosomes to decode triplet codons in mature mRNAs and assemble the corresponding polypeptide chains. Translation is a complex and highly regulated process that is essential for normal cellular function and is subject to control at nearly every stage.¹⁹ In eukaryotic cells, translation proceeds through three major stages, including initiation, elongation, and termination (Figure 1). The major molecular events and regulatory mechanisms involved in these stages are summarized in Table 1.

Figure 1.

Overview of canonical eukaryotic mRNA translation. During initiation, the 40S ribosomal subunit associates with initiation factors and the eIF2–GTP–Met-tRNAi ternary complex to form the 43S PIC, which is recruited to the mRNA 5′ cap by the eIF4F complex. Following start-codon recognition and 60S subunit joining, the 80S ribosome enters elongation. eEF1A mediates aminoacyl-tRNA delivery, whereas eEF2 promotes ribosomal translocation; phosphorylation of eEF2 inhibits elongation. Translation terminates when eRF1 and eRF3 recognize a stop codon and promote peptide release, followed by ABCE1-mediated ribosome recycling. eIF2B promotes recycling of eIF2-GDP, whereas phosphorylation of eIF2α inhibits eIF2B and suppresses translation initiation. Adapted from¹¹. Numbers 1, 3, and 5 denote eIF1, eIF3, and eIF5, respectively. Arrows indicate activation or progression, inhibitory bars indicate repression, and p denotes phosphorylation.

Abbreviations: eIF, eukaryotic initiation factor; eEF, eukaryotic elongation factor; eRF, eukaryotic release factor; PIC, pre-initiation complex; 4E-BP, eIF4E-binding protein; ABCE1, ATP-binding cassette subfamily E member 1; GTP, guanosine triphosphate; GDP, guanosine diphosphate; 4A, 4E, and 4G denote eIF4A, eIF4E, and eIF4G, respectively; Met, methionine; Met-tRNAi, initiator methionyl-tRNA.

Table 1.

Stages and Regulatory Mechanisms of mRNA Translation

Stage	Molecular Events	Key Factors	Regulatory Mechanisms
Initiation	43S PIC recruitment; start codon recognition; 80S ribosome assembly	eIF2/eIF2B, eIF4F, 4E-BPs	eIF4F assembly; eIF2α phosphorylation; uORF-mediated selective translation
Elongation	Aminoacyl-tRNA delivery; peptide bond formation; ribosomal translocation	eEF1A, eEF2, eEF2K	eEF2 phosphorylation; regulation of eEF2K by Ca2+/calmodulin and AMPK
Termination and ribosome recycling	Stop codon recognition; peptide release; ribosome recycling	eRF1, eRF3, ABCE1	eRF1/eRF3-mediated termination; ABCE1-dependent ribosome recycling

Abbreviations: PIC, pre-initiation complex; eIF, eukaryotic initiation factor; eIF2B, eukaryotic initiation factor 2B; eIF4F, eukaryotic initiation factor 4F; 4E-BP, eIF4E-binding protein; uORF, upstream open reading frame; eEF1A, eukaryotic elongation factor 1A; eEF2, eukaryotic elongation factor 2; eEF2K, eukaryotic elongation factor 2 kinase; Ca²⁺, calcium ion; AMPK, AMP-activated protein kinase; eRF1, eukaryotic release factor 1; eRF3, eukaryotic release factor 3; ABCE1, ATP-binding cassette subfamily E member 1.

Initiation

Translation initiation is a major regulatory step in protein synthesis. It depends on accurate positioning of the ribosome and initiator tRNA at the start codon.²⁰ The 40S small ribosomal subunit associates with several initiation factors and the ternary complex, which consists of eIF2, GTP, and initiator methionyl-tRNA, to form the 43S pre-initiation complex (PIC).²¹ In canonical cap-dependent initiation, the 7-methylguanosine cap at the 5′ end of mRNA is recognized by the eIF4F complex, which includes eIF4E, eIF4A, and eIF4G. The eIF4F complex recruits the 43S PIC to the mRNA and supports scanning toward the AUG start codon. After start codon recognition, GTP hydrolysis on eIF2, release of initiation factors, and joining of the 60S subunit lead to formation of the 80S ribosome.^11,21

Initiation efficiency is tightly controlled by several mechanisms.²² eIF4E is a rate-limiting factor whose activity is controlled by phosphorylation and by competitive binding of 4E-binding proteins (4E-BPs), which inhibit eIF4F assembly and suppress cap-dependent initiation.²² Another important regulatory step is eIF2 cycling. After GTP hydrolysis, eIF2-GDP is recycled by eIF2B to support further rounds of initiation.¹⁹ Under cellular stress, phosphorylation of eIF2α at Ser51 inhibits eIF2B activity and suppresses global translation initiation. In addition, some mRNAs contain upstream open reading frames (uORFs) in their 5′ untranslated regions. These elements permit selective translation of specific stress-responsive transcripts even when global initiation is suppressed.²³ Thus, cells can downregulate global protein synthesis while preserving the production of proteins needed for stress adaptation.

Elongation

Following initiation, the ribosome proceeds into the elongation phase, during which it moves along the coding region and adds amino acids to the growing polypeptide chain. This process is primarily mediated by the coordinated functions of eEF1A and eEF2.²⁴ eEF1A delivers aminoacyl-tRNAs to the ribosomal A site, where correct codon–anticodon recognition allows peptide bond formation. eEF2 then promotes movement of the ribosome along the mRNA. With each cycle of aminoacyl-tRNA delivery, peptide bond formation, and translocation, the polypeptide chain is extended by one amino acid.

Elongation is also regulated in response to cellular conditions. eEF2 kinase (eEF2K), an atypical calcium/calmodulin-dependent kinase, controls protein synthesis by phosphorylating eEF2. Under stress conditions, including nutrient deprivation, hypoxia, and oxidative stress, eEF2K is activated and phosphorylates eEF2 at Thr56, reducing the affinity of eEF2 for the ribosome and slowing elongation.^25–27 eEF2K is regulated by Ca²⁺/calmodulin and AMPK signaling and is activated under energy limitation and other forms of cellular stress.²⁷ Through the eEF2K/eEF2 axis, cells can rapidly suppress elongation, reduce energy consumption, and limit protein synthesis under stress. In addition, neuronal expression of the tissue-restricted elongation factor isoform eEF1A2 further highlights the importance of elongation control for neuronal function and survival.²⁸

Termination and Ribosome Recycling

Translation termination begins when the ribosome encounters a stop codon. In eukaryotic cells, this step is mediated by eukaryotic release factor 1 (eRF1) together with the auxiliary GTP-binding factor eRF3.²⁹ Stop codon recognition in the A site promotes hydrolysis of the bond linking the nascent polypeptide to the tRNA in the P site, leading to release of the completed peptide.²⁹ After termination, ABCE1 promotes ribosome recycling by using ATP hydrolysis to separate the 80S ribosome into 40S and 60S subunits, allowing these subunits to be reused in subsequent rounds of translation.²⁹ These coordinated steps are essential for the fidelity and efficiency of protein synthesis.

Neuron- and Glia-Specific Translational Control

Because neurons have complex dendritic and axonal structures, protein synthesis in these cells shows distinct spatial and temporal features. In neurons, long-distance transport of mRNAs to dendrites and axons enables local translation in response to external cues. This process is required for proper neural circuit formation during development and for learning and memory.³⁰ Local mRNA transport and translation are therefore essential for neuronal function. In vivo, local translation contributes to axon survival, axon regeneration, and neural circuit assembly, and its dysregulation is associated with multiple neurological diseases.³¹ De novo protein synthesis within dendrites is required for long-term potentiation and the maintenance of long-term memory.³² Specific mRNAs contain cis-acting elements and bind trans-acting factors and RBPs to form messenger ribonucleoprotein granules. These granules are transported to specific subcellular regions by molecular motors along microtubules or microfilaments.^30,31 During transport, translation is usually repressed. After external stimulation, it is rapidly activated to produce proteins required at local sites.^30,31 Neuronal local translation is regulated by multiple RBPs and signaling pathways. FMRP is an RBP and translational repressor. FMRP deficiency causes fragile X syndrome and is associated with excessive protein synthesis and synaptic abnormalities.³³ Aβ42 can alter FMRP-regulated translation of proteins involved in synaptic function.¹⁶ Low concentrations of Aβ oligomers were shown to selectively alter the de novo proteome of primary cortical neurons without significantly changing total de novo protein synthesis.³⁴

mTOR is a major regulator of neuronal survival, differentiation, and development, and it also controls multiple metabolic and autophagy-related processes.³⁵ The mTOR pathway affects local translation by regulating factors such as 4E-BP and S6K. In AD, aberrant activation of mTOR signaling has been associated with impaired synaptic function and tau pathology.^35,36 Aβ- and tau-related pathology also interact at the synaptic level. High concentrations of Aβ oligomers can increase the synaptic recruitment of tau oligomers and exacerbate tau-associated impairment of synaptic potentiation.³⁷ Regional Aβ–tau interactions have also been associated with the onset and acceleration of tau spreading.³⁸

Neuron-specific translational regulation is also disrupted under disease conditions. Proteins implicated in neurodegenerative diseases, including TDP-43, FUS, and members of the hnRNP family, are involved in mRNA transport and local translational control. When these proteins are mutated or mislocalized, normal mRNA trafficking and local protein synthesis may be disrupted, contributing to synaptic dysfunction.^39,40 Therefore, clarifying neuron-specific mechanisms of translational regulation is important for understanding the pathogenesis of AD and related disorders. It may also help identify potential therapeutic strategies.

In addition, translational dysregulation in AD is not restricted to neurons but also shows cell-type- and subcellular compartment-specific features. In AD mouse models, the translatome of hippocampal perisynaptic astrocytic processes undergoes extensive remodeling, and these changes can occur before overt amyloid plaque deposition, indicating that altered local translation in PAPs may contribute to early synaptic dysfunction in AD.⁴¹ Astrocytic PERK–eIF2α signaling is also enhanced under AD-related pathological conditions and is accompanied by reduced protein synthesis, whereas astrocyte-specific deletion of PERK restores protein synthesis and attenuates AD-related pathology in mice.⁴² Following Aβ exposure, astrocytes can deliver the ribosomal protein Rps6 to neuronal axons through extracellular vesicles, thereby enhancing local protein synthesis in distal axons and improving synaptic integrity.⁴³ In addition, astrocytes and microglia exhibit distinct circadian translatomes that undergo different patterns of reprogramming in the presence of amyloid pathology.⁴⁴ These findings suggest that translational dysregulation in AD could be both cell-type- and subcellular compartment-specific and may also be modulated by neuron–glia interactions.

Translational Dysregulation in AD

Protein synthesis is disrupted at multiple regulatory levels in AD. These abnormalities include a global decline in mRNA translation, selective translation of specific transcripts, and impaired control of both translation initiation and elongation. Abnormal RBP and SG dynamics have also been increasingly associated with AD pathogenesis. In addition, altered tRNA function and activation of noncanonical translation pathways may further aggravate translational defects in AD (Figure 2).

Figure 2.

Mechanisms of mRNA translation dysregulation in AD. Global translational suppression is associated with aberrant ISR activation, tau-bound ribosomes, and altered abundance of translation factors in the hippocampal CA1 region. Selective mRNA translation is characterized by increased translation of BACE1 and APP mRNAs despite overall translational repression. Dysregulated synaptic translation initiation involves alterations in BDNF–TrkB, MAPK/ERK, Aβ42-associated signaling, and eIF4E-dependent cap initiation. Elongation impairment involves stress-dependent activation of the eEF2K–eEF2 pathway and tRF-associated ribosome stalling. RBP abnormalities include tau–TIA1- and Aβ42–PKR-associated SG formation, together with Musashi–tau-associated nuclear alterations involving Lamin B1. tRNA dysregulation includes reduced NSUN2- and ELP3-dependent tRNA modification and ANG-mediated tRNA cleavage, generating stress-induced tiRNAs that inhibit translation. Arrows indicate reported regulatory relationships, inhibitory bars indicate repression, ↑ and ↓ indicate increased and decreased abundance or activity, respectively, and p denotes phosphorylation.

Abbreviations: AD, Alzheimer’s disease; ISR, integrated stress response; BACE1, β-site APP-cleaving enzyme 1; APP, amyloid precursor protein; BDNF, brain-derived neurotrophic factor; TrkB, tropomyosin receptor kinase B; MAPK, mitogen-activated protein kinase; ERK, extracellular signal-regulated kinase; Aβ, amyloid-β; eIF4E, eukaryotic initiation factor 4E; eEF2K, eukaryotic elongation factor 2 kinase; eEF2, eukaryotic elongation factor 2; RBP, RNA-binding protein; TIA1, T-cell intracellular antigen 1; PKR, protein kinase R; SG, stress granule; MSI, Musashi; ANG, angiogenin; tRF, tRNA-derived fragment; tiRNA, stress-induced tRNA-derived RNA; NSUN2, NOP2/Sun RNA methyltransferase 2; ELP3, elongator acetyltransferase complex subunit 3.

Global Translational Suppression

Long-term memory consolidation and synaptic plasticity depend on neuronal protein synthesis.^8,45 The ISR is a conserved pathway that supports cell survival and homeostasis by regulating protein synthesis under intrinsic and extrinsic stress. One major intrinsic trigger of the ISR is ER stress, which occurs when protein-folding capacity is exceeded. In AD, accumulation of Aβ and pathological tau can disturb proteostasis and contribute to ER stress. Aβ oligomers can disturb intracellular Ca²⁺ homeostasis and promote Ca²⁺ release from ER stores.⁴⁶ Aβ oligomers have also been reported to aggravate ER stress through an ATF4–Parkin positive feedback loop.⁴⁷ In AD, aberrant ISR activation and the accompanying reduction in brain protein synthesis have been linked to memory impairment.⁸ Four ISR kinases, PERK, PKR, GCN2, and HRI, converge on phosphorylation of eIF2α at Ser51 in response to different stressors. Phosphorylation of eIF2α suppresses global protein synthesis. In AD, eIF2α phosphorylation is markedly increased, whereas the abundance of key eIF2B subunits is reduced. Phosphorylated eIF2α inhibits eIF2B and thereby broadly suppresses translation initiation.⁸ AD-related misfolded proteins can activate eIF2α kinases, particularly PERK and PKR, leading to eIF2α phosphorylation and translational repression.^42,48 More broadly, multiple forms of cellular stress can activate eIF2α kinases and induce eIF2α phosphorylation, which is a central mechanism of translational control in the ISR.^8,49 The duration of ISR activation appears to be an important determinant of its effects in AD. Acute ISR activation is generally transient and adaptive, with eIF2α phosphorylation reducing global protein synthesis and thereby relieving the protein-folding burden within the endoplasmic reticulum. This response facilitates the refolding or clearance of misfolded proteins and may provide short-term neuroprotection. In contrast, sustained ISR activation in AD maintains eIF2α phosphorylation and translational repression, which may contribute to proteostatic dysfunction, oxidative stress, mitochondrial dysfunction, and ultimately progressive neurodegeneration.⁴⁵ In addition to the ISR, tau-mediated ribotoxicity also contributes to translational defects in AD. Ribosomal dysfunction appears to occur early in the course of AD.⁵⁰ Pathological tau is able to interfere with key cellular processes required for protein synthesis. In AD, abnormal tau binds to ribosomes, compromises ribosomal activity, and disrupts RNA translation.^51,52 These changes may reduce the synthesis of PSD-95, an important synaptic protein, and further aggravate synaptic dysfunction.⁵² Altered expression of several translation-related factors, including eIF2α, eIF3η, eIF5, and eEF2, has also been observed in the hippocampal CA1 region of patients with AD.⁵⁰ Chronic ISR activation and tau-mediated ribotoxicity may contribute to broad translational repression in AD, resulting in impaired protein synthesis and synaptic dysfunction.

Selective mRNA Translation

Although AD is associated with global translational repression, some mRNAs are selectively translated more efficiently. Despite global translational repression, APP and BACE1 mRNAs can undergo selective translational upregulation under AD-related conditions. The BACE1 transcript contains a complex 5′ UTR with intrinsic inhibitory features that normally suppress its translation.^49,53 The BACE1 5′ UTR is long and highly structured and contains multiple upstream AUGs/uORFs that suppress basal translation. Under basal conditions, when eIF2α phosphorylation is low and the ternary complex is abundant, mRNAs with short 5′ UTRs are translated efficiently, whereas transcripts such as BACE1 with long, uORF-containing, and highly structured 5′ UTRs are translated less efficiently. Under cellular stress, eIF2α can be phosphorylated by PERK, which reduces ternary complex availability. As a result, translation of most mRNAs is suppressed, whereas BACE1 translation can increase despite the reduction in global protein synthesis. In this setting, scanning ribosomes are more likely to bypass inhibitory uORFs, continue scanning, and reinitiate at the main AUG codon of BACE1. This increases the translational efficiency of BACE1 mRNA under stress conditions. BACE1 is markedly elevated in AD and is enriched in dystrophic neurites surrounding plaques.⁵⁴ Because BACE1 is the initiating enzyme for Aβ production,⁵⁵ circulating BACE1 has been proposed as a candidate biomarker for AD diagnosis.⁵⁶ Selective translation driven by eIF2α phosphorylation may therefore increase BACE1 levels and promote Aβ production and amyloid accumulation. This process may also generate a positive feedback loop. Increased Aβ may sustain cellular stress, which further enhances BACE1 translation and leads to additional Aβ production. This mechanism may represent an important contributor to sporadic AD pathogenesis.⁴⁹

In AD, a paradox exists at the translational level: while global protein synthesis is suppressed, a subset of mRNAs escapes this repression and is translationally upregulated. This imbalance may contribute to pathology by disproportionately increasing the production of proteins involved in Aβ generation. APP mRNA is selectively upregulated at the translational level in AD through at least two distinct mechanisms. First, the APP 5′ UTR contains an iron-responsive element (IRE). Under conditions of iron overload—a feature associated with AD—binding of iron regulatory proteins (IRPs) to the IRE is reduced, thereby facilitating the recruitment of the eIF4F initiation complex and accelerating APP translation initiation.^57,58 Second, at synapses, neuronal activity can induce the local translation of APP mRNA. This process is regulated by FMRP, which dissociates from APP mRNA following mGluR activation, thereby increasing local APP synthesis and potentially providing additional substrate for Aβ production.⁵⁹ This selective translational upregulation of APP, together with increased translation of BACE1, may further promote Aβ production. Thus, translational regulation in AD is not simply reduced but altered. Selective upregulation of pro-pathogenic factors despite overall downregulation may contribute to disease progression.

Dysregulated Synaptic Translation Initiation

In neuronal dendrites, regulated mRNA translation provides an essential mechanism for synaptic plasticity and brain development.¹⁶ AD involves dysregulation at the level of cap-dependent initiation, particularly abnormal regulation of eIF4E. By supporting synaptic activity-dependent local protein synthesis, eIF4E contributes to synaptic plasticity and to the structural and functional adaptations underlying learning and memory.⁶⁰ Synaptic translation is tightly controlled by a complex composed of eIF4E, CYFIP, and FMRP. This complex regulates protein synthesis during synaptic activity and is critical for long-lasting synaptic plasticity.⁶¹ FMRP recruits CYFIP1 to bind eIF4E, thereby forming a complex that blocks assembly of eIF4F on specific mRNAs and keeps these transcripts translationally silent at synapses. BDNF is a key regulator of long-term potentiation (LTP) at hippocampal excitatory synapses and is involved in neuronal growth, survival, and synaptic plasticity. It is also an important mediator of learning and memory and can locally induce protein synthesis in the postsynaptic compartment.^60,62 The BDNF-TrkB pathway activates dendritic mRNA translation. TrkB signaling engages the PI3K–Akt and MAPK–ERK pathways in translational control.^60,62 BDNF-induced PI3K–mTOR signaling promotes 4E-BP1 phosphorylation and protein synthesis.⁶³ MAPK–ERK signaling activates Mnk1, leading to phosphorylation of eIF4E at Ser209. During dentate gyrus LTP, eIF4E phosphorylation promotes the release of translational repressors from eIF4E and facilitates translation initiation complex assembly.⁶⁴ mGluR signaling also regulates activity-dependent neuronal protein synthesis. Activation of group I mGluRs can engage both the PI3K–mTOR–4E-BP–eIF4E and ERK–Mnk1–eIF4E pathways, which converge on cap-dependent translation initiation.^65,66 In APOE4-exposed cortical neurons, stimulation of group I mGluRs restores protein synthesis in an mTORC1-dependent manner.⁶⁷ Studies have shown that Aβ42 induces Mnk1-dependent phosphorylation of eIF4E, promotes dissociation of eIF4E from CYFIP2, and ultimately leads to CYFIP2 ubiquitination and reduced expression.¹⁶ These changes are accompanied by increased translation of FMRP-regulated transcripts.¹⁶ Reduced CYFIP2 expression has been associated with Aβ deposition in the mouse thalamus, tau pathology in the entorhinal cortex and hippocampus, hippocampal gliosis and synapse loss, and impaired memory formation.¹⁶

Elongation Impairment

The maintenance of memory and synaptic plasticity relies on de novo protein synthesis, a process partly controlled by eEF2 and its kinase eEF2K. During mRNA translation, more than 95% of energy expenditure and amino acid use occurs during elongation.⁶⁸ Increased eEF2 phosphorylation and dysregulated mRNA translation have been implicated in AD.⁶⁸ In AD, eEF2 is excessively phosphorylated at Thr56 by eEF2K, which slows peptide elongation and suppresses global protein synthesis.⁶⁹ Studies have shown that genetic reduction of eEF2K in AD mouse models attenuates eEF2 hyperphosphorylation and improves impaired de novo protein synthesis, abnormal dendritic spine morphology, defective LTP, and memory deficits.⁷⁰ This intervention does not alter brain Aβ burden, indicating that the cognitive improvement mainly results from restoration of translational function rather than reduction of amyloid load.⁷⁰ Jan et al also reported that pharmacological or genetic inhibition of eEF2K protects neurons from the toxic effects of Aβ42 oligomers. This protective effect is partly associated with activation of the NRF2 antioxidant response pathway.⁷¹ These findings indicate that excessive eEF2K activity and abnormal eEF2 phosphorylation may contribute to elongation defects in AD. eEF2K overactivation in AD may be promoted by disease-related alterations in Ca²⁺ signaling and cellular energy homeostasis. Ca²⁺/calmodulin directly activates eEF2K, whereas AMPK promotes eEF2K activity under conditions of energy stress, providing two pathways through which cellular stress can inhibit translation elongation.^16,72,73 Impaired elongation can lead to ribosome stalling and collision on mRNAs, activating ribosome quality-control and stress-signaling pathways.^74,75 Small RNA fragments may also be involved.⁷⁶ Several tRNA-derived fragments (tRFs) are dysregulated in the hippocampus of patients with AD.⁷⁷ Certain tRFs can interact directly with ribosomes and induce ribosome stalling during translation.⁷⁶

RNA-Binding Proteins

RBP dysregulation and SG abnormalities are important features of AD pathology.^14,78 Under physiological conditions, RBPs promote assembly of ribonucleoprotein complexes and coordinate the RNA life cycle, including RNA processing, transport, translation, and degradation. In AD and other neurodegenerative diseases, many RBPs become mislocalized or aggregated, leading to disrupted RNA metabolism. RBPs may interact with tau, alter normal SG dynamics, and aggregate in tauopathies.^14,78 In these disorders, several RBPs accumulate directly in the cytoplasm and form toxic inclusions. Within these granules, some RBPs interact with tau and may contribute to neurodegenerative changes.^14,78 Cellular stress can further induce RBP aggregation and SG formation.⁷⁹ Aberrant interactions between tau and specific RBPs are therefore considered to contribute to toxic ribonucleoprotein granule formation in neurons.¹⁴

T cell intracellular antigen 1 (TIA1) is a core RBP required for SG formation and can sequester untranslated mRNAs during cellular stress. Given its central role in SG dynamics, TIA1 has been a focus in understanding AD-related SG pathology. Studies have shown that tau regulates the distribution of TIA1. Formation of tau–TIA1 complexes promotes excessive SG assembly and increases SG size, while the tau–TIA1 interaction facilitates tau misfolding and assembly at SG sites, generating toxic oligomers.^40,80,81 Knockdown or knockout of TIA1 suppresses tau misfolding and related toxicity in cultured hippocampal neurons, whereas overexpression of TIA1 promotes tau misfolding and neurodegeneration.⁴⁰ Pharmacological interventions that inhibit SG formation also suppress tau pathology.⁴⁰ In tau-transgenic mice, TIA1 haploinsufficiency markedly weakens the excessive SG response and reduces tau oligomer accumulation and neurodegeneration. Although TIA1 reduction increases neurofibrillary tangles, neuronal survival, behavioral outcomes, and lifespan are improved.⁸² Tau oligomers selectively bind RBPs, suggesting that these oligomers may act as the active tau species that drive SG formation and translational stress responses. In AD brain tissue, oligomeric tau strongly colocalizes with TIA1 and other RBP granule markers, particularly in vulnerable cortical and hippocampal regions.^82–84 In addition, Aβ42 can also promote SG formation through PKR activation.⁸⁵ The tau–TIA1 interaction promotes nuclear-to-cytoplasmic translocation of TIA1 and induces SG assembly. Over time, this process may generate persistent SGs and accelerate AD pathology. In cells treated with cycloheximide, which stabilizes polysomes and prevents stress granule formation, tau- and TIA1-containing granules are markedly reduced. In contrast, agents that promote polysome disassembly, such as puromycin, produce the opposite effect.⁸³ These findings suggest that targeting TIA1 may help reduce tau-related toxicity during disease progression.

The Musashi (MSI) family includes two homologous RBPs, Musashi1 and Musashi2. These proteins are highly expressed in neural stem and progenitor cells but are present at very low levels in mature neurons under normal conditions. In AD, however, they are abnormally re-expressed and co-aggregate with tau.¹⁴ Studies have shown that MSI levels are increased in brain tissue from patients with AD compared with age-matched nondemented individuals. Oligomeric MSI1 and MSI2 are present in cortical neurons in AD, and these high-molecular-weight complexes show strong colocalization with tau oligomers in the cytoplasm and around the nucleus.⁸⁶

In vitro studies indicate that tau oligomers regulate MSI through several mechanisms. They promote the nuclear localization and aggregation of both tau and MSI and also alter MSI nuclear transport.⁷⁸ As a result, the size and density of nuclear MSI-tau puncta progressively increase. Direct tau-MSI interaction also induces formation of characteristic intranuclear aggregates. These complexes can lead to nuclear instability and structural disruption, partly through suppression of Lamin B1 expression.⁷⁸ MSI1 and tau can enter the nucleus through Importin β, whereas MSI2 cannot. Partially overlapping round structures are observed between MSI1 and nuclear tau, whereas MSI2 and tau show completely overlapping puncta.⁷⁸ Aberrant interactions between MSI1 and nuclear tau may induce pathological inclusions.¹⁴ In MAPT P301L transgenic mice, targeted knockdown of Musashi1 markedly reduces oligomeric MSI1 and tau species. Multiple phosphorylated tau forms are decreased in the cortex, cerebellum, and especially the hippocampus, whereas synaptogenesis and cognitive function are improved.⁸⁷ These results indicate that MSI1 contributes to tau-mediated neurotoxicity and may be a potential therapeutic target in AD.

A common feature of the RBPs described above is their ability to undergo LLPS, which converts a homogeneous macromolecular system into dense liquid droplets. Many of these RBPs also form RNA granules, especially SGs, whose assembly largely depends on LLPS. Under stress, cap-dependent translation is generally suppressed, whereas selected transcripts may continue to be translated through alternative initiation mechanisms. This stress response is accompanied by cytoplasmic translocation of mRNA-binding proteins, which aggregate with transcripts to form SGs.⁷⁹ Under acute stress, SGs are generally transient structures that sequester untranslated mRNAs and associated proteins and contribute to cellular stress adaptation.⁸⁸ Under chronic or excessive stress, however, SGs may persist and thereby promote the pathological protein aggregation characteristic of age-related neurodegeneration.⁸⁹ In human AD brains and AD mouse models, prominent deposition of G3BP as well as TIA1 and TTP has been reported. Cytoplasmic granules containing TTP and G3BP are observed, although G3BP shows little colocalization with tau inclusions, whereas TTP colocalization appears mainly at late disease stages.⁷⁹ SG formation has been linked to several neurodegenerative diseases, including AD.⁷⁹ Studies indicate that G3BP2 may be particularly important in AD. Reduced neuronal G3BP2 expression has also been observed in sporadic AD brain.⁹⁰ Enhanced G3BP2-tau interaction occurs independently of NFTs, and knockdown of G3BP2 markedly aggravates tau pathology.⁹¹ G3BP2 can inhibit the interaction between tau and ribosomes, directly bind the MTBR region of tau, and prevent tau aggregation.⁹¹

Recent studies have identified additional RBPs involved in AD-related translational dysregulation. HNRNPA2B1 has also been shown to mediate the interaction between oligomeric tau and m6A-modified RNA, whereas HNRNPA2B1 knockdown attenuates oTau-induced translational suppression and associated neurodegeneration.⁹² PCBP2 is a KH-domain-containing RNA-binding protein. In AD, PCBP2 expression and condensate formation are increased, and PCBP2 condensates inhibit BACE1 mRNA degradation, thereby increasing BACE1 mRNA stability and BACE1 expression and indirectly promoting Aβ accumulation.⁹³ The small molecule CN-0928 reduces PCBP2 levels, suppresses condensate formation, and attenuates amyloid pathology and cognitive deficits in FAD mice.⁹³ STAU1 is a double-stranded RNA-binding protein whose expression is significantly increased in the brains of APP/PS1 mice. STAU1 promotes amyloidogenesis by stabilizing BACE1 mRNA and thereby increasing BACE1 expression.⁹⁴ These findings suggest that these RBPs may represent potential therapeutic targets, although their clinical relevance remains to be established. In addition, Aβ42 can trigger eIF2α phosphorylation and induce SG formation through activation of PKR.⁸⁵ Compared with fibrillar Aβ, monomeric and oligomeric Aβ42 induce more SG formation. In AD, both pathological tau and Aβ may aberrantly activate RBP- and SG-related pathways and thereby promote translational dysregulation and neurodegenerative changes.

tRNA Dysregulation in AD Pathogenesis

tRNAs are a class of functionally diverse non-coding RNAs that can be processed or cleaved to generate tRNA-derived small RNAs (tsRNAs), mainly including tRFs and tRNA-derived stress-induced RNAs (tiRNAs), which have emerged as important regulators in neurodegenerative diseases. In AD, tRF expression profiles are markedly altered in the hippocampus.^77,95 Studies indicate that both the modification status and cleavage patterns of tRNAs are altered in brain tissue from patients with AD, and the stress-associated ribonuclease ANG is significantly upregulated in the hippocampus. Under cellular stress, ANG cleaves mature tRNAs and promotes the generation of tiRNAs.⁷⁷ More recently, ANG was found to be dysregulated in a sex-dependent manner in AD, accompanied by altered levels of 5’-tiRNA^Gly-GCC, suggesting that ANG-mediated tRNA cleavage may contribute to AD pathology.⁹⁶ Because tRNA methylation can protect against cleavage, reduced tRNA methylation may increase the susceptibility of tRNAs to stress-induced cleavage.⁷⁷ Altered tRF profiles have also been detected in neuron-derived extracellular vesicles from patients with AD.⁹⁷

The functional effects of specific tsRNAs have begun to be defined. tRF5-ProAGG expression is increased in the AD hippocampus and rises with advancing neuropathological stage.⁷⁷ A recent study showed that tRF^Ala-AGC-3-M8 is reduced in AD mice and Aβ-treated cells. Increasing its expression in cell models suppresses the EphA7–ERK1/2–p70S6K signaling pathway and attenuates neuronal injury, tau hyperphosphorylation, and inflammatory responses.⁹⁸ tsRNAs can regulate protein translation at both the elongation and initiation stages. The tRNA^Pro 5′ half can bind directly to ribosomes, leading to peptidyl-tRNA retention and translation elongation arrest.⁷⁶ In addition, 5′-tiRNA^Cys is a guanine-rich 5′ stress-induced tRNA-derived fragment that can form G-quadruplex structures, displace eIF4F from capped mRNAs, suppress cap-dependent translation initiation, and promote SG formation.⁹⁹ PANDORA-seq can overcome sequencing biases caused by RNA modifications. Using this approach, a recent study identified alterations in mitochondrial tsRNAs associated with changes in lysosomal activity, which may contribute to AD progression.¹⁰⁰ Under chronic AD-related stress, abnormal tRNA cleavage may contribute to translational dysregulation by affecting translation elongation and cap-dependent initiation and by promoting SG formation.

tRNAs undergo diverse chemical modifications that are required for translational accuracy and efficiency.¹⁰¹ In recent years, RNA modification has increasingly been recognized as an important epitranscriptomic mechanism in both physiological and pathological contexts. Evidence suggests that small non-coding RNAs and their associated modifications may represent a new class of signaling molecules involved in AD onset and progression.¹⁰² Many enzymatic post-transcriptional modifications of tRNAs occur in the anticodon region. By supporting accurate codon recognition and maintenance of the reading frame, these modifications help ensure precise and efficient protein synthesis.¹⁰³ Significant hypermethylation at several mitochondrial tRNA position-9 sites has been observed in postmortem cerebellar tissue from patients with AD, suggesting altered mitochondrial tRNA modification in AD.¹⁰⁴ In the cortex of AD mice, both mitochondrial and cytosolic tRNAs exhibit m¹A hypomethylation, accompanied by reduced expression of m¹A-modifying enzymes. Knockdown of these enzymes exacerbates disease-related phenotypes in a Drosophila tau model, suggesting that reduced m¹A modification of tRNAs may contribute to AD pathogenesis.¹⁰⁵

Studies have shown that the tRNA epitranscriptome is remodeled in AD, and this remodeling may contribute to proteostasis imbalance.¹⁰⁵ ELP3, the catalytic subunit of the Elongator complex, is involved in the formation of modifications at the wobble U34 position of tRNAs, and its expression is reduced in brain tissue from patients with AD and in AD mouse models.¹⁰⁶ Reduced ELP3 expression results in loss of specific tRNA modifications and may lead to codon-specific translational defects and impaired proteostasis. In cells carrying familial AD mutations, ELP3 downregulation is accompanied by the accumulation of misfolded protein aggregates, whereas correction of the tRNA defects associated with ELP3 deficiency can restore proteostasis.¹⁰⁶ NSUN2 is an important methyltransferase that modifies non-coding RNAs, and neuronal NSUN2 deficiency can induce tRNA epitranscriptomic abnormalities and drive proteome remodeling, thereby affecting synaptic transmission and complex behaviors, including memory.¹⁰⁷ Kim et al reported that NSUN2 protein expression is reduced in the brains of patients with AD and that NSUN2 deficiency can promote tau hyperphosphorylation.¹⁰⁸ In neuron-specific Nsun2-deficient mice, reduced tRNA m⁵C modification is accompanied by selective depletion of tRNA^Gly isodecoders, resulting in glycine codon-specific translational defects and decreased expression of several Gly-rich proteins involved in synaptic function.¹⁰⁷ These findings suggest that disruption of tRNA-modifying enzymes may contribute to AD-related translational dysfunction and impaired neuronal proteostasis.

Genetic Determinants

Pathogenic variants in PSEN1, PSEN2, and APP are established genetic causes of autosomal dominant EOAD, whereas APOE4 is a major genetic risk factor for AD. Recent studies suggest that these genetic factors may influence mRNA translation at multiple regulatory levels. In the following sections, we summarize available evidence linking these genetic factors to translational abnormalities and discuss the possible underlying mechanisms (Figure 3).

Figure 3.

AD genetic factors and dysregulated mRNA translation. PSEN1 mutations impair adaptive UPR signaling and PERK–eIF2α-mediated translational attenuation, contributing to persistent endoplasmic reticulum stress and disturbed protein synthesis. APP mutations are associated with Aβ accumulation and activation of cellular stress pathways, including the ISR and UPR. APP mutation-associated Aβ42 activates PKR–eIF2α signaling and suppresses mRNA translation, whereas increased eIF4B phosphorylation is associated with altered translation-initiation signaling. APOE4-related signaling through NMDARs and L-VGCCs increases intracellular Ca²⁺ and engages the eEF2K–eEF2 pathway, resulting in translational repression. Together, these genetic factors converge on dysregulated mRNA translation through distinct regulatory mechanisms. Arrows indicate activation or pathway progression, inhibitory bars indicate repression, ↑ and ↓ indicate increased and decreased abundance or activity, respectively, and p denotes phosphorylation. Red crosses indicate impaired adaptive UPR responses associated with PSEN1 mutations.

Abbreviations: AD, Alzheimer’s disease; PSEN1, presenilin 1; ER, endoplasmic reticulum; UPR, unfolded protein response; PERK, protein kinase R-like ER kinase; eIF2α, eukaryotic initiation factor 2 alpha; APP, amyloid precursor protein; Aβ, amyloid-β; ISR, integrated stress response; PKR, protein kinase R; eIF4B, eukaryotic initiation factor 4B; APOE4, apolipoprotein E ε4 allele; NMDAR, N-methyl-D-aspartate receptor; L-VGCC, L-type voltage-gated calcium channel; eEF2K, eukaryotic elongation factor 2 kinase; eEF2, eukaryotic elongation factor 2.

PSEN1 and PSEN2

PSEN1 encodes a membrane-spanning protein with nine transmembrane domains. It is mainly localized to the ER, is an essential component of the γ-secretase complex, and plays a key role in APP processing and amyloid peptide generation. In addition to APP cleavage, PSEN1 is involved in several other biological processes, including Notch signaling, β-catenin processing, and Ca²⁺ regulation, and therefore has pleiotropic effects in the pathogenesis of AD and related disorders.¹⁰⁹ More than 300 PSEN1 mutations have been reported.¹¹⁰ Most reported PSEN1 variants are heterozygous and are inherited in an autosomal dominant manner. PSEN1 variants account for approximately 6% of EOAD.¹⁰⁹ Physiological protein folding, oligomerization, and post-translational modification occur in the ER and are essential for cellular function and survival.⁷ Under normal conditions, ER stress transiently suppresses global protein synthesis through PERK-mediated phosphorylation of eIF2α, whereas FAD mutations in PSEN1 weaken this adaptive UPR signaling, and under ER stress cells cannot adequately shut down protein synthesis, leading to excessive influx of nascent peptides into the ER and increased folding burden.⁷ In familial AD, PSEN1 mutations disrupt the adaptive UPR by impairing the activation of multiple UPR branches, including PERK, IRE1, and ATF6, causing neurons to fail to cope with ER stress and thereby promoting AD progression.⁷ These mutations also impair the PERK-eIF2α pathway, reducing PERK-dependent eIF2α phosphorylation and translational suppression under stress conditions.⁷ In addition, PSEN1 mutant neurons show markedly increased levels and accumulation of Aβ and tau proteins.¹¹¹ PSEN1 mutations also dysregulate mTORC1 signaling and weaken cellular responses to starvation. Even under starvation, the Akt–mTORC1–p70S6K pathway remains abnormally active, phosphorylation of downstream components remains elevated, and starvation-induced autophagy is suppressed.¹¹¹ In human neurons, PSEN1 mutations can disrupt Akt-mTORC1 signaling and promote abnormal tau accumulation.¹¹¹ Studies have also shown that the mTOR inhibitor Torin1 effectively reduces marked tau pathology in PSEN1 F105C neurons.¹¹¹ Presenilin loss has also been associated with increased mitochondrial Ca²⁺ and mTORC1 signaling, leading to disrupted redox and protein homeostasis and neuronal dysfunction.¹¹² PSEN1 L150P human iPSC-derived neurons also show impaired NMDAR-mediated Ca²⁺ signaling and abnormal activity-dependent eEF2 responses, indicating dysregulated neuronal translation.¹¹³ AD cases caused by PSEN2 mutations are relatively uncommon, and recent studies show that loss of PSEN2 accelerates AD-related pathology, including working memory deficits and impaired hippocampal LTP, as well as abnormalities in synaptic structure and endolysosomal function.¹¹⁴

APP

The APP gene, located on chromosome 21, consists of 18 exons and produces several isoforms through alternative splicing. APP encodes a type I transmembrane protein that is predominantly found at neuronal synapses. It is the precursor of Aβ and is trafficked through the secretory pathway before undergoing proteolytic processing in distinct cellular compartments.^115,116 APP is first cleaved by either α-secretase or β-secretase. The resulting C-terminal fragments are subsequently processed by γ-secretase within the transmembrane domain, leading to release of the APP intracellular domain together with p3 or Aβ peptides, respectively. APP amino acid substitutions are mainly located near the α-, β-, and γ-secretase cleavage sites. These substitutions can differentially affect Aβ production and aggregation. For example, the Swedish APP mutation increases total Aβ production by approximately 4- to 7-fold.^116,117 Aβ accumulation can activate cellular stress pathways, including the ISR and UPR.^7,48 In the brains of APP-mutant animal models, increased expression and phosphorylation of eIF4B have been observed. eIF4B is a translation initiation factor that facilitates cap-dependent translation initiation.¹¹⁸ In addition, studies have shown that brain tissue from transgenic AD mouse models carrying mutant APP exhibits increased eIF2α phosphorylation, PKR activation, and BACE1 expression.¹¹⁹ Although eIF2α phosphorylation suppresses global translation, translation of BACE1 mRNA is increased. The ISR also selectively upregulates ATF4, which can suppress CREB-dependent expression of memory-related genes and may contribute to AD progression.^48,120

Beyond translation initiation, APP mutations also affect local protein synthesis at synapses. Studies in APP-mutant AD model mice have shown dysregulated local translation in brain synaptosomes and impaired synaptic protein synthesis. After learning training, de novo synthesis of specific proteins is markedly enhanced in synaptosomes from control mice.¹²¹ Under pathological conditions driven by mutant APP, activity-dependent local protein synthesis at synapses is suppressed. APP itself is partly synthesized locally in synaptosomes and can undergo local translation at synaptic sites. These findings suggest that APP-related pathology may alter the local translational regulation of specific synaptic transcripts, including APP mRNA. In addition, oligomeric Aβ42 induces SG formation through the PACT/PKR pathway,⁸⁵ which may contribute to local translational dysregulation. Aβ oligomers can also activate stress-responsive kinase pathways, including CaMKK2-AMPK, which contributes to Aβ42 oligomer-induced synaptic dysfunction.¹²²

APOE

The APOE gene is located on chromosome 19 and has three major alleles, ε2, ε3, and ε4, which encode the three isoforms ApoE2, ApoE3, and ApoE4. Among them, APOE4 is the most common genetic risk factor for AD.^123–125 Almost all APOE4 homozygotes exhibit AD pathology.¹²⁶ ApoE is a lipid transport protein composed of 299 amino acids and has a molecular weight of approximately 34 kDa. Under normal physiological conditions, ApoE is mainly produced by astrocytes in the brain. However, neurons can also produce ApoE under conditions of stress, injury, or aging. When ApoE4 is expressed in neurons, it can induce Aβ and tau pathology, leading to neuroinflammation and neuronal injury and ultimately impairing learning and memory.¹²⁵ Primary microglia isolated from mice expressing human ApoE4 show altered morphology, reduced protein translation, and increased phosphorylation of eIF2α and its upstream kinases compared with ApoE3 microglia, consistent with increased cellular stress and ISR activation.¹²⁷ Acute exposure to ApoE4 causes a significant decrease in global protein synthesis in primary cortical neurons. ApoE4 can activate NMDARs and L-type voltage-gated calcium channels (L-VGCCs), resulting in a sustained rise in intracellular Ca²⁺ levels. Elevated Ca²⁺ then promotes persistent repression of translation through eEF2 phosphorylation and further interferes with NMDAR-mediated responses.¹²⁸ Through this disturbance of Ca²⁺ homeostasis, ApoE4 affects both basal neuronal protein synthesis and activity-dependent translational responses, which may contribute to ApoE4-related synaptic dysfunction.¹²⁸ The effect of ApoE4 on translation appears to occur independently of Aβ pathology, suggesting that ApoE4 may directly impair translational regulation. In E4FAD mouse models, treatment with the mTOR inhibitor rapamycin has been reported to restore brain function and reduce AD-related risk phenotypes in young asymptomatic mice carrying human APOE4.^129,130 These findings indicate that disturbed translational control may represent an early event in ApoE4-associated AD pathology.

Therapeutic Strategies Targeting Translational Dysregulation in AD

Translational dysregulation in AD involves multiple interacting pathways. Therapeutic strategies therefore need to focus on key regulatory nodes that influence protein synthesis and synaptic function. These approaches can be broadly classified into interventions that modulate specific translation-control pathways and those that act on upstream pathogenic proteins directly associated with translational stress (Figure 4).

Figure 4.

Therapeutic strategies targeting translational dysregulation in AD. (A) Targeting the ISR and translation initiation. GSK2606414 inhibits PERK-dependent signaling, whereas ISRIB activates eIF2B and partially restores translation initiation. Trazodone and dibenzoylmethane can relieve eIF2α-mediated translational repression. (B) Modulating cap-dependent translation initiation. The ERK–MNK1/2–eIF4E pathway contributes to physiological synaptic function and memory. eFT508 inhibits MNK1/2 and reduces eIF4E phosphorylation under conditions of dysregulated translation. 4EGI-1 disrupts the eIF4E–eIF4G interaction. Rapamycin inhibits mTORC1 signaling. Reduction of 4E-BP2 counteracts Aβ-induced translational repression. (C) Restoring translation elongation. Pharmacological inhibition of eEF2K with A-484954 or genetic reduction of eEF2K decreases eEF2 phosphorylation, restores translation elongation, and improves protein synthesis. (D) Targeting upstream drivers of translational stress. Reduction of TIA1 limits tau–TIA1-associated stress-granule responses and reduces tau aggregation and toxicity, whereas MAPTRx-mediated lowering of tau may indirectly reduce translational stress. Arrows indicate pathway relationships or downstream effects, inhibitory bars indicate inhibition, ↑ and ↓ indicate increased and decreased abundance or activity, respectively, and p denotes phosphorylation. Green arrows and inhibitory bars denote therapeutic interventions or beneficial downstream effects.

Abbreviations: AD, Alzheimer’s disease; ISR, integrated stress response; PERK, protein kinase R-like ER kinase; eIF2α, eukaryotic initiation factor 2 alpha; eIF2B, eukaryotic initiation factor 2B; MNK, MAPK-interacting kinase; eIF4E, eukaryotic initiation factor 4E; eIF4G, eukaryotic initiation factor 4G; mTORC1, mechanistic target of rapamycin complex 1; 4E-BP, eIF4E-binding protein; eEF2K, eukaryotic elongation factor 2 kinase; eEF2, eukaryotic elongation factor 2; TIA1, T-cell intracellular antigen 1; SG, stress granule; MAPTRx, MAPT-targeting antisense oligonucleotide; ERK, extracellular signal-regulated kinase; Aβ, amyloid-β; MAPT, microtubule-associated protein tau; ASO, antisense oligonucleotide.

Targeting the ISR and Initiation

Relieving the sustained inhibition of translation initiation caused by eIF2α phosphorylation has become an important therapeutic direction. Studies in AD models have shown that attenuating eIF2α-dependent ISR signaling can rescue hippocampal protein synthesis, synaptic plasticity, and memory.⁸ The PERK inhibitor GSK2606414 restores protein synthesis and alleviates neuropathology in prion disease models by blocking the eIF2α phosphorylation pathway. However, sustained PERK inhibition can cause pancreatic toxicity, limiting its potential for chronic treatment.¹³¹ In contrast, ISRIB, a small-molecule allosteric activator of eIF2B, can partially reverse translational repression without fully abolishing the protective ISR. Studies indicate that ISRIB can improve protein synthesis, synaptic plasticity, and memory in AD models.¹³² In aged mice, ISRIB attenuates age-related ISR activation and inflammatory changes and improves neuronal function and memory.¹³³ High-throughput screening has identified compounds such as trazodone and dibenzoylmethane that can cross the blood-brain barrier and reverse eIF2α-mediated translational repression in mouse models.¹³¹ In prion and tauopathy models, trazodone restores protein synthesis and exerts neuroprotective effects.^131,134 Some studies have not definitively demonstrated that trazodone slows cognitive decline in patients with dementia, and prospective studies are needed for validation.¹³⁵ These findings highlight the need to define a therapeutic window that attenuates chronic maladaptive ISR signaling without abolishing its physiological stress-protective functions.

Modulating the Cap-Dependent Initiation Machinery

Another strategy is to target the cap-dependent translation initiation pathway, particularly eIF4E-dependent translation.¹³⁶ eIF4E-dependent initiation is regulated by the ERK–MNK and mTORC1–4E-BP pathways.^65,136 MNK1/2 act downstream of the ERK cascade and phosphorylate eIF4E at Ser209, thereby modulating eIF4E-dependent translation.¹³⁶ Studies have found that the MNK inhibitor eFT508 can cross the blood-brain barrier, reduce cortical eIF4E phosphorylation, normalize abnormal neuronal maturation and synaptic activity, and rescue memory deficits in Fmr1-deficient mice.¹³⁷ Although these findings demonstrate CNS target engagement, the therapeutic efficacy of eFT508 in AD remains to be established.

mTORC1 controls translation by phosphorylating 4E-BP to relieve inhibition of eIF4E and by activating S6K and downstream ribosomal protein S6 signaling. Aberrant mTORC1 signaling has been implicated in AD and is associated with amyloid and tau pathology.³⁶ Inhibition of mTOR with rapamycin and related agents may help correct several core pathological features of AD, including abnormal protein aggregation and cognitive deficits.³⁶ However, recent clinical studies have highlighted uncertainties regarding CNS exposure and optimal dosing of rapamycin in AD.^138,139 Recent animal studies have also reported conflicting effects on amyloid pathology and behavioral outcomes.^140,141 These findings highlight the context-dependent effects and translational challenges of mTOR inhibition.

Genetic reduction of 4E-BP2 prevents Aβ oligomer-induced suppression of hippocampal protein synthesis and memory impairment and also improves memory deficits in aged APP/PS1 mice.¹⁴² In addition, studies in cancer models have reported that 4EGI-1 binds eIF4E, disrupts eIF4E-eIF4G binding, and inhibits cap-dependent translation without measurably affecting initiation factor-independent translation.¹⁴³ Because cap-dependent translation is also required for normal synaptic plasticity, excessive inhibition may interfere with physiological neuronal protein synthesis.

Restoring Translation Elongation

Impaired elongation represents another potential therapeutic target. Drugs that antagonize eEF2K, including A-484954, can rescue multiple pathophysiological abnormalities in AD model mice and may restore global elongation capacity, suggesting a feasible therapeutic strategy for AD and related dementia syndromes.⁶⁸ Genetic reduction of eEF2K similarly increases protein synthesis and improves synaptic and cognitive function.⁷⁰ Clinical translation of this strategy will require more selective, brain-penetrant eEF2K inhibitors together with confirmation of CNS target engagement and long-term tolerability.

Targeting Upstream Pathogenic Drivers Linked to Translation Stress

Given the tight link between tau pathology and SG-mediated translational arrest, strategies directly lowering TIA1 levels have shown promise in altering tau aggregation and toxicity in vivo.⁸² Antisense oligonucleotide (ASO) therapy targeting MAPT mRNA reduces tau expression and improves tau pathology, neuronal loss, and disease-related phenotypes in tauopathy models.¹⁴⁴ MAPTRx has shown dose-dependent tau reduction in a phase 1b clinical trial, with acceptable safety in patients with mild AD.¹⁴⁵ Further analyses showed reduced tau PET signal and exploratory trends toward slower clinical decline,¹⁴⁶ supporting the development of disease-modifying strategies that may indirectly reduce translational stress by lowering tau burden. However, its clinical efficacy and long-term safety remain uncertain, and repeated intrathecal administration may limit long-term use.^145,147

Conclusions and Future Directions

Important advances have been made in understanding AD pathogenesis. Increasing evidence suggests that dysregulated mRNA translation contributes to AD pathogenesis by impairing neuronal homeostasis and synaptic plasticity. Under physiological conditions, neurons rely on tightly controlled local protein synthesis in dendrites and at synapses to support synaptic plasticity and long-term memory.^148,149 This local control depends on RBPs, which coordinate mRNA transport, sequestration, translation, and degradation.⁴⁰ In AD, RBPs may become mislocalized or aggregated, and LLPS may be disrupted.^150,151 Translational dysregulation in AD does not simply reflect a uniform reduction in protein synthesis but encompasses global translational repression, selective enhancement of the translation of specific mRNAs, abnormalities in translation initiation and elongation, impaired activity-dependent local translation, and RBP/SG dysfunction and tRNA-related abnormalities. These alterations vary across regulatory pathways, cell types, and subcellular compartments and may disrupt proteostasis and synaptic function, thereby contributing to disease progression.

Precise regulation of RNA translation is essential for maintaining neuronal homeostasis. Increasing evidence suggests that translational dysregulation may be linked to selective vulnerability in the nervous system and may contribute to the onset and progression of several neurodegenerative diseases.^11,152 Key factors and regulatory nodes involved in different stages of translation have now been identified in AD. However, some gaps remain. In particular, relatively few studies have examined how neurons coordinate signaling pathways and the translatome to establish and maintain their highly specialized structure and function. Current mechanistic evidence is still derived mainly from animal models, whereas direct evidence from human AD brain tissue remains limited. In addition, the temporal sequence and mechanistic relationships among different forms of translational dysregulation during AD progression have not yet been fully defined. Although increasing attention has been paid to RBPs and phase separation-related events, the transcript classes, neuronal subcellular compartments, and cell populations involved at different stages of AD are still not well defined.

Despite encouraging preclinical findings, translation-targeted therapies still face important challenges in clinical practice. Translational control pathways are required for normal stress responses, synaptic plasticity, and proteostasis. The therapeutic effects may therefore depend on the extent and duration of pathway modulation, as well as the cellular context and disease stage. In addition, commonly used AD models do not fully reproduce the molecular heterogeneity of human disease.¹⁵³ Dose selection, CNS exposure and target engagement, long-term safety, and identification of the appropriate therapeutic context are important considerations for future clinical studies.

With advances in single-cell and spatial genomics, complex brain tissues can now be analyzed at substantially higher resolution. Single-cell translatomics enables isoform-specific profiling of translation at single-cell resolution and can reveal cell-type-specific differences in translational regulation.¹⁵⁴ Spatial and single-nucleus transcriptomic studies of AD brain have further shown that disease-associated molecular changes are cell-type- and spatial-context-specific and vary with pathological stage and genetic background.^155,156 Future studies integrating single-cell translatomics with spatial transcriptomics may help identify vulnerable neuronal and glial populations and determine where and at what stage translational dysregulation occurs during AD progression.^155–157 Restoring translational homeostasis may represent a therapeutic option for AD. Future studies should define cell-type- and pathway-specific targets, optimize delivery to relevant cell populations, and determine whether these interventions can achieve sustained functional benefits with acceptable safety.

Funding Statement

This work was supported by grants from the National Natural Science Foundation of China (82571350) and Zhejiang Clinovation Pride (Pre-dementia Stage of Alzheimer’s Disease).

Declaration of Generative AI and AI-Assisted Technologies in the Writing Process

During the preparation of this manuscript, the authors used ChatGPT (OpenAI; GPT-5.6 Sol) to improve the language, clarity, and readability of the manuscript. After using this tool, the authors reviewed and edited all AI-assisted revisions and confirmed the originality and accuracy of the manuscript. The authors take full responsibility for the integrity of the entire content, including the accuracy of all citations and references. The authors have reviewed the applicable terms of use for ChatGPT and confirm that its use for language editing is suitable for publication.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors report no conflicts of interest in this work.