Local Protein Synthesis at Synapses: A Driver for Synapse Diversification

ABSTRACT

Local protein synthesis within neuronal processes seems to be crucial for the rapid and dynamic remodeling of the proteome at synaptic compartments. Indeed, this capability enables neurons to swiftly adapt their synaptic functions in response to activity. In this review, we first explore the diverse mechanisms that allow the targeted transport of mRNAs into both dendrites and axons. Then, we report evidence that local mRNAs are actively recruited for protein synthesis during plasticity. Finally, we highlight how this molecular complexity contributes to the establishment and stabilization of memory traces, or engrams, within neural circuits. We propose that presynaptic protein synthesis is a pivotal factor driving the diversification of presynaptic terminals, a process we foresee as essential for the durable consolidation and specificity of engrams.

Local protein synthesis is abundant in neuronal processes. This localized translation of mRNA is believed to allow neurons to rapidly modify synaptic strength in response to activity. We review mechanisms guiding mRNA transport into dendrites and axons, emphasizing their selective recruitment for translation during synaptic plasticity. Such spatially restricted protein synthesis provides the molecular flexibility required for long‐term information storage. We propose that presynaptic translation in particular plays a central role in shaping terminal diversity, thereby supporting the consolidation and specificity of memory engrams across neural circuits.

Abbreviations

4E‐BP

eIF4E‐binding protein

AD

Alzheimer's disease

ALS

amyotrophic lateral sclerosis

AMPA

α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazolepropionic acid receptor

Arc

activity‐regulated cytoskeleton‐associated protein

BDNF

brain‐derived neurotrophic factor

BORC

lysosome‐kinesin adaptor BLOC‐one‐related complex

CA3

Cornu Ammonis region 3

CAMKII

calcium/calmodulin kinase 2

CPEB

cytoplasmic polyadenylation element binding protein

CREB

cAMP response element‐binding protein

ER

endoplasmic reticulum

ErbB4

Erb‐B2 receptor tyrosine kinase 4

FMRP

fragile X messenger ribonucleoprotein

FTD

frontotemporal dementia

FTO

fat mass and obesity‐associated protein

GFP

green fluorescence protein

IDR

intrinsically disordered region

IEG

immediate early gene

KH

K‐homology domain

KIBRA

kidney and brain expressed adaptor protein

LLPS

liquid–liquid phase separation

LTD

long‐term depression

LTP

long‐term potentiation

m6A

N6‐methyladenosine

miRNA

microRNA

mRNA

messenger RNA

mTOR

mammalian target of rapamycin

NMDA

N‐methyl‐D‐aspartate

p‐bodies

processing bodies

PIN1

peptidyl‐prolyl cis‐trans isomerase NIMA‐interacting 1

PKA

protein kinase A

PKMζ

protein kinase M zeta

PTBP

polytrimidine tract‐binding protein

PTM

posttranslational modification

PV+

parvalbumin‐positive

RBP

RNA‐binding protein

RGG

arginine‐glycine–glycine

RISC

RNA‐induced silencing complex

RNP

ribonucleoprotein

S6K

P70 S6 kinase

SG

stress granule

SST+

somatostatin‐positive

TDP‐43

TAR DNA‐binding protein 43

UTR

untranslated region

ZBP1

zipcode binding protein 1

1. Introduction

Local protein synthesis in neuronal dendrites and axons has emerged as a critical mechanism enabling synaptic autonomy and rapid proteome remodeling far from the cell body. Early ultrastructural studies revealed polyribosomes beneath dendritic spines (Bourne and Harris 2012), and metabolic labeling within isolated dendrite compartments demonstrated de novo translation following brain‐derived neurotrophic factor (BDNF) or electrical stimulation, linking local synthesis to synaptic potentiation (Kang and Schuman 1996). This was further supported by reporter mRNAs (e.g., myc‐ or GFP‐tagged CAMKIIα constructs), which, when delivered to transected dendrites, yielded protein locally upon synaptic stimulation (Aakalu et al. 2001). In fact, high‐resolution and sequencing‐based approaches have since cataloged thousands of mRNAs and ribosomes within dendrites and axons, revealing compartment‐specific transcriptomes and translatomes that differ markedly from the soma (e.g., Shigeoka et al. 2016; Hafner et al. 2019; Tushev et al. 2018; Glock et al. 2021). In axons, super‐resolution imaging and puromycin‐incorporation assays demonstrated that a majority of mature presynaptic boutons harbor mRNAs, ribosomes, and ~40% display ongoing translation after only 5 min metabolic labeling under basal conditions (Hafner et al. 2019). Together, these multi‐modal data—from electron microscopy, metabolic labeling, reporter assays, sequencing, and fluorescence microscopy—demonstrate that both dendrites and axons can regulate their local proteome during synaptic plasticity, development, and homeostasis.

Nonetheless, the extent of the contribution of local protein synthesis in the induction and maintenance of synaptic plasticity on the pre‐ and postsynaptic sides of the synapse remains to be fully understood. For instance, the kinetics of the translational response in various synaptic compartments have not yet been elucidated. Additionally, the nature of proteins undergoing chronic versus acute synthesis under different plasticity contexts remains unknown. Finally, the contribution of local protein synthesis to higher cognitive processes is rarely discussed.

In this review, we start by discussing the latest studies investigating the mechanisms allowing mRNA localization and local translational regulation. We go on with reviewing evidence of mRNA translation regulation by plasticity. Ultimately, we discuss how local protein synthesis particularly in presynaptic boutons could be a key mechanism enabling synapse diversification, thereby allowing the saliency of memory engram in neuronal networks.

2. mRNA Mechanisms of Transport

The highly polarized architecture of neurons presents a logistical challenge for the spatial regulation of gene expression (Sun and Schuman 2022; Hafner and Triesch 2023). In fact, as much as 90% of the neuronal cytoplasm is in its processes (i.e., dendritic arbor and axon) (Sun and Schuman 2022). Understanding the mechanisms that govern mRNA transport and localization is key to elucidating how neurons achieve spatial and temporal precision to create highly specialized and semi‐autonomous cell compartments such as synapses. Additionally, elucidating how mRNA is trafficked in neurons to distal subcellular compartments, including both the molecular state and type of macromolecular complexes it is packaged in, is crucial to better understand how local protein synthesis can be induced during plasticity.

2.1. Formation of RNA Granules

Rather than being transported as single naked strands, mRNA molecules are packaged into highly organized ribonucleoprotein (RNP) complexes. RNP granules are distinct, membrane‐free intracellular RNA‐protein complexes and play crucial roles in various cellular processes including RNA localization, processing, and translation (Fritzsche et al. 2013). The interactions found inside RNPs are not only limited to RNA‐protein, but they can also involve RNA–RNA and protein–protein interactions (An et al. 2021; Ripin and Parker 2023). Cumulative evidence suggests that most RNP granules, if not all, are formed via a physicochemical phenomenon called liquid–liquid phase separation (LLPS), during which the macromolecules that make up the RNP condense together into a dense, liquid‐like phase. It is quite likely that the forces involved in LLPS can influence the accessibility of RNA‐to‐protein binding domains found inside the granule to domains outside the granule. Thus, any regulation affecting phase separation could regulate for instance granule composition, transport, or disassembly.

One subclass of RNP complexes, consisting of RNA molecules and RNA‐binding proteins (RBPs), is called the RNA granule. The individual contribution of RNA molecules and RBPs during the LLPS process and the formation of an RNA granule is still under debate in the field. On the one hand, RNA has been found to influence intracellular phase transitions and is capable of promoting LLPS on its own (Shin and Brangwynne 2017). For instance, the presence of RNA can induce phase separation of some RBPs such as polytrimidine tract‐binding protein (PTBP) which contains four RNA recognition motifs, or the N‐terminal fragment of hnRNPA1 that contains only two RNA recognition motifs (Li et al. 2012). Interestingly, those two RBPs play important roles in RNA metabolism including mRNA stability, 3′ end formation, and splicing (Castelo‐Branco et al. 2004; Kim et al. 2005). Moreover, the length of an RNA molecule has also been identified to be an important factor for LLPS as this can affect the viscosity of the phase‐separated condensates. Short polyadenylated RNAs can reduce LAF‐1 protein‐containing condensate viscosity, thereby enhancing internal molecular mobility and accessibility of RNA and protein components (Elbaum‐Garfinkle et al. 2015). In contrast, longer RNAs increase the viscosity of Whi3 condensates, leading to slower molecular exchange (Zhang et al. 2015). Thus, changes in transcript length and viscosity are believed to tune whether a granule behaves as a highly dynamic hub to regulate RNA transport or as a more stable compartment that protects its RNA cargo. On the other hand, RBPs can also drive phase separation through intrinsic multivalency conferred by features such as RNA‐binding domains, dynamic posttranslational modifications (PTMs), or intrinsically disordered regions (IDRs) (Wiedner and Giudice 2021). It has been shown that different RNA‐binding domains, such as arginine‐glycine–glycine (RGG) motifs, K‐homology domains (KHs), and zinc‐finger domains, facilitate RBP‐RNA interactions and modulate phase behavior (Lunde et al. 2007; Thandapani et al. 2013). Importantly, PTMs such as arginine methylation and phosphorylation showed fine‐tuning properties in condensate formation by altering their interaction strengths (Qamar et al. 2018; Wang et al. 2018). In conclusion, both RNA molecules as well as proteins seem to be able to drive phase separation and the formation of an RNP granule.

The diverse mechanisms of phase separation contribute to the formation and regulation of various RNA granules observed in neurons, including processing bodies (p‐bodies), RNA‐induced silencing complex (RISC), and stress granules (SGs) mostly localized in the soma and neurites, transport granules predominantly in dendrites, and activity‐dependent translation granules in the synapses (Bauer et al. 2023; Cui et al. 2024). Despite their distinct identities, these granules frequently share a core of overlapping factors and can dynamically exchange components. For instance, Argonaute protein 2, the key component of RISC, can also be present in other RNP complexes (Ehses et al. 2022).

2.2. Regulation and Role of RNA Granules

Emerging evidence has highlighted the importance of post‐transcriptional RNA modifications, specifically N6‐methyladenosine (m6A), as regulators of mRNA dynamics during synaptic plasticity. The m6A modification has been implicated in several neuronal processes, including mRNA stability, splicing, translation, localization, and interactions with RNA‐binding proteins (Leighton et al. 2018). mRNAs harboring multiple m6A residues can preferentially bind the cytosolic m6A readers YTHDF1‐3, which enhances their incorporation into phase‐separated granules (Ries et al. 2019). Although the YTHDF proteins bind largely overlapping mRNA pools, they exert different regulatory effects, introducing an additional layer of complexity in m6A‐mediated RNA dynamics. Notably, in a study by Walters et al., the RNA demethylase fat mass and obesity‐associated protein (FTO) has been observed to localize in synapses (Walters et al. 2017). Interestingly, in two separate studies, it was observed that both a knockdown of FTO in the mouse dorsal hippocampus as well as a complete knockout of FTO in the mouse medial prefrontal cortex can enhance the formation and consolidation of fear memory, suggesting that RNA methylation dynamics in the synapse may indeed play a role during plasticity and memory formation (Widagdo et al. 2016; Walters et al. 2017).

Regardless of how RNA granules are formed, the role of RBPs in mRNA localization and translation is undeniably important. RBPs with their target mRNAs can act as cytoskeletal adaptors and/or translational silencers to transport their cargo to subcellular locations. Once on site, RBPs can either act as translational repressors or activators of their mRNA targets, thus providing a way to control translation spatially and temporally (Besse and Ephrussi 2008). With the help of unbiased proteomic studies, different catalogs of RBPs were described in axons and dendrites, depending on their mission. Among these RBPs, the Fragile X Messenger RibonucleoProtein (FMRP) plays a key role in regulating the local translation of numerous neuronal mRNAs. FMRP is encoded by the Fmr1 gene, which is silenced in Fragile X syndrome due to an abnormal expansion of a CGG repeat sequence that prevents the production of the FMRP protein (Fu et al. 1991). In hippocampal neurons, axonal FMRP has been shown to regulate axon guidance (Li et al. 2009), growth cone motility (Antar et al. 2006), and synapse formation (Hanson and Madison 2007) in part through translational control of Map1b. Similarly, TAR DNA‐binding protein 43 (TDP‐43), implicated in amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD), shuttles between nucleus and cytoplasm to regulate RNA metabolism and axonal transport. In the motor neurons, TDP‐43 contributes to axon outgrowth (Fallini et al. 2012) and branching by regulating Futsch mRNA, the drosophila homolog of MAP1B (Ymlahi‐Ouazzani et al. 2010). Zipcode binding protein 1 (ZBP1) was the first RBP found to regulate axon guidance, and its local regulation of β‐actin mRNA in response to guidance cues is conserved in several species (Zhang et al. 2001; Leung et al. 2006; Welshhans and Bassell 2011). TDP‐43 also interacts with Staufen1 and FMRP to transport Rac1‐containing RNPs in the mouse dendrites in both anterograde and retrograde directions (Chu et al. 2019).

Understanding the dynamic properties and functional relationships among neuronal granules is important as their dysregulation is highly involved in neurodegenerative disorders. Through recent years, the predominant role of SGs in neurodegenerative diseases was uncovered. Both ALS and FTD show a strong relation to SGs and TDP‐43, encoded by Tardbp the disease‐causing gene in both disorders (Chen and Cohen 2019). TDP‐43 protein co‐localizes with SG proteins, including G3BP1 and TIA1 in oxidative stress conditions (Colombrita et al. 2009). Similar co‐localization patterns were later identified for other ALS‐associated proteins, including SOD1 and FUS (Baron et al. 2013; Gal et al. 2016; Lee et al. 2020). These findings were further confirmed in following studies in animal models showing comparable co‐localization of TIA1 and eIF3η with FUS in FUS‐R521C knock‐in mice (Zhang et al. 2020). Beyond SGs, the recruitment of RNA and RBPs into pathological inclusions represents a common theme in neurological disorders (Rai et al. 2021). The microtubule‐binding protein tau, whose mutations and oligomerization are linked to Alzheimer's disease (AD) and tauopathies, not only associates with RBPs (Gunawardana et al. 2015) but also accumulates in pathological aggregates rich in RNAs (Lester et al. 2021). Significantly, tau has been shown to assemble into dynamic condensates both in vitro and in cortical mouse neurons (Wegmann et al. 2018). These in vitro condensates can also mature from a reversible liquid‐like state into an irreversibly aggregated state (Lu et al. 2022). Moreover, Apicco et al. demonstrated that tau directly interacts with the SG‐nucleating RBP TIA1, and reducing TIA1 levels in a tauopathy mouse model mitigates tau‐mediated neurodegeneration and synaptic loss (Apicco et al. 2018). This study established that aberrant tau–RBP interactions not only promote toxic aggregation but also perturb RNA metabolism, local translation, and synaptic maintenance (Apicco et al. 2018). Together, these findings suggest a mechanistic framework in which neuronal RNP granules, acting as local concentrators of interaction‐prone molecules, may transition from dynamic to more static assemblies over time (Alberti and Hyman 2016, 2021). Such static assemblies could hinder mRNA accessibility and translation, ultimately leading to deficits in local protein synthesis and synapse formation or retention—features commonly observed in both neurodevelopmental and neurodegenerative disorders. This process is often exacerbated by disease‐causing mutations and risk factors such as aging or chronic stress (Alberti and Hyman 2021; Hipp et al. 2019).

2.3. Thumbs Up for mRNA

The journey of localized mRNA in different neuronal compartments starts with the transcription of nascent mRNA in a nucleus. Once in the cytosol, certain mRNAs are packed into RNPs and then travel to their destination. However, a single neuron can harbor thousands of pre‐ and postsynapses, thus making it impossible to provide each of those at all times mRNA coding for all locally translated synaptic proteins. Thus, in 2011, Doyle and Kiebler hypothesized a continuous bidirectional flow of RNPs, with their words, “like a conveyor belt in a Japanese sushi restaurant serving all potential customers,” and their employment in activated synapses (Doyle and Kiebler 2011). Such bidirectional transport of RNPs in the dendrites, specifically near synapses, was in fact observed in many studies (e.g., Knowles et al. 1996; Köhrmann et al. 1999; Dynes and Steward 2007).

Different mRNAs interacts with various combinations of RBPs which influences their mode of transportation. Most RNPs move inside the cytoplasm using molecular motors: dynein and kinesin superfamilies using microtubules and myosin using actin filaments. For instance, the KIF5 family of kinesin is known to cargo RNPs in axons and dendrites. One of its most renowned interactors is FMRP, already mentioned above for its role in fragile X syndrome. When isolated from Fmr1 knockout mice, KIF5 showed a significantly decreased interaction with Dlgap4 mRNA (Dictenberg et al. 2008), encoding SAP90/PSD95‐associated protein 4, an important excitatory postsynaptic scaffolding complex member (Wang, Bai, et al. 2023). In contrast, dynein motor proteins have been found to be involved in dendritic mRNA localization in Staufen1‐bound mRNAs in synapses (Gershoni‐Emek et al. 2016). Another example are TDP‐43‐FMRP‐Staufen1 complexes transporting Rac1 mRNA in dendrite which can recruit both kinesin‐1 and dynein motor proteins (Chu et al. 2019). To conclude, the composition of an RNA granule can dictate the complement of intracellular motors that will be used for its localization.

Additionally, alternative secondary modes of transport have been uncovered for some mRNAs such as hitchhiking strategy using moving organelles. ANXA11, an ALS‐associated protein, possesses an N‐terminal domain that drives phase separation and a C‐terminal domain that interacts with lysosomes to operate the cotransport (Liao et al. 2019). Additionally, De Pace and colleagues also demonstrated that lysosomes contribute to mRNA transport by using BORC (one of the components for lysosome‐kinesin‐1/3 coupling) knock out mice and observed a significant decrease in mRNAs coding for mitochondrial and ribosomal proteins in axons (De Pace et al. 2024). Interestingly, these losses were associated with defects that cause axonal swelling, a typical phenotype observed during neurodegeneration. Additionally, Cioni and colleagues studied RNA granule transport with endosomes in Xenopus retinal ganglion cells (Cioni et al. 2019). By labeling RNA granules with Cy3 and endosome markers Rab5 (early endosome) and Rab7 (late endosome) with GFP, they were able to investigate such cotransport in axons. Although detections were mostly static or oscillatory, they still observed both retrograde and anterograde transport of RNA granules via endosomes.

Beyond transport, different organelles also organize local mRNA translation within neuronal compartments. Koppers et al. showed that axonal ER tubules bind to ribosomes and act as local sites of protein synthesis (Koppers et al. 2024). The ER protein P180/RRBP1 anchors ribosomes and promotes translation of mRNAs enriched in axons. Disrupting ER‐ribosome contacts reduces local translation and alters axon morphology, indicating that specialized ER domains regulate translation in axons.

Similarly, Biever et al. proposed that presynaptic regions may use transient ER‐like membranes for local synthesis of synaptic proteins. Although classical rough ER is absent from axons, co‐translational targeting via the SRP pathway could support translation and membrane insertion of localized synaptic proteins (Biever et al. 2019). In addition, in a review by Nguyen et al., the authors discuss how ER membranes interact with biomolecular condensates and RNA granules, suggesting that ER–condensate contact sites dynamically regulate mRNA localization and translation (Nguyen et al. 2024). Together, these findings indicate that the neuronal ER forms specialized, adaptable platforms that coordinate local protein synthesis in both axonal and presynaptic regions.

2.4. mRNA Targeting Motifs and Alternative Transport Mechanisms

Several lines of evidence suggest that cis‐acting elements contain the answer to “How does mRNA know where to go?”. Those elements are specific sequences mostly found in 3′ UTRs, also called “zipcodes” allowing the RBP to recognize the mRNA for transport, without affecting its coding sequence (Anderson and Kedersha 2006; Singh et al. 2015). One of the most studied zipcode binding proteins ZBP1 is associated with β‐actin mRNA transport and translation in dendrites. It has the ability to bind a 54‐nucleotide sequence within 3′ UTRs with its KH3 and KH4 domains (Nicastro et al. 2017; Zhang et al. 2001). The role of ZBP1‐ β‐actin RNP in axon growth regulation is well established (Zhang et al. 2001). In dendrites, this RNP complex has been shown to modulate dendritic filopodia density and promote the formation of filopodial synapses (Eom et al. 2003). This provides a clear example of how cis‐acting elements and their binding proteins regulate local translation to shape neuronal development. The 3′ UTR sequences of several dendritically localized mRNAs, such as Camk2a, Arc3, and Mapt have been defined as cis‐acting elements (Rook et al. 2000; Dynes and Steward 2007; Huang et al. 2003). However, many cis‐acting elements remain to be identified. Recently, the development of a neuronal zipcode identification protocol (N‐zip) enabled the systematic mapping of candidate zipcodes across tiled 3′ UTR fragments in mouse cortical neurons (Mendonsa et al. 2023). This high‐throughput approach identified AU‐rich motifs in 3′ UTRs and let‐7 microRNA binding sites as functional cis‐elements that mediate mRNA localization in neurites (Figure 1).

FIGURE 1.

Schematic representation of mRNA recruitment in pre‐ and post‐synapses during plasticity. Different sets of RNA‐binding proteins (RBPs) are involved in axonal and dendritic mRNA targeting. Created in https://BioRender.com.

Beyond serving as recognition sites for RBPs, 3′ UTRs also play an important role in regulating mRNA stability and half‐life (Tian and Manley 2017). Moreover, the untranslated regions of mRNAs seem to play a critical role in regulating mRNA transport during LTP and plasticity. Notably, alternative splicing of the 3′ UTR has been observed to alter the subcellular distribution of specific mRNA species (Ciolli Mattioli et al. 2019). Furthermore, neuronal activity can dynamically alter the composition of different 3′ UTR isoforms found in distinct cellular compartments (Tushev et al. 2018). These activity‐driven changes could be induced either by de novo transcription of specific isoforms or through local remodeling of existing 3′ UTR isoforms. Globally, Tushev et al. showed a correlation indicating that longer 3′ UTR mRNA isoforms tend to have longer half‐lives and to localize further away from the somatic compartment (Tushev et al. 2018). Strikingly, computational work from Fonkeu and colleagues suggested that, for mRNAs which have the capacity to use microtubule‐mediated transport, the main regulator of mRNA accumulation in neurites is mRNA half‐life. Indeed, within physiological ranges parameters like non‐directed mobility or synthesis rates have only marginal effects (Fonkeu et al. 2019).

While the active transport of mRNA within RNP complexes is the predominant mechanism for synaptic localization, emerging evidence suggests the existence of alternative or complementary routes for RNA delivery to synapses. One exciting alternative route is the virus‐like capsids of Arc protein, one of the most important regulators of synaptic plasticity (Bramham et al. 2010). Endogenous activity‐regulated cytoskeleton‐associated protein (Arc) creates virus‐like capsids via self‐assembling into oligomers and shows Arc mRNA‐binding properties similar to retroviral Gag proteins. Additionally, it can be released from rat neurons in extracellular vesicles, and it transfers its own mRNA to other neurons via endocytosis. As naked mRNA cannot efficiently cross membranes on its own, its extracellular transfer generally requires the aid of membrane fusion proteins. Arc capsids may therefore interact with endosomal membranes to deliver their mRNA cargo while the protein is disassembled (Pastuzyn et al. 2018). Similarly, in the fly neuromuscular junction, Drosophila Arc1 protein transports its mRNA from neurons to muscle cells. dArc1 binds to the 3′ UTRs sequence of its own mRNA to form capsid‐like structures (Ashley et al. 2018). Importantly, blocking dArc1 mRNA transfer impairs synaptic plasticity (Ashley et al. 2018). Moreover, Müller et al. discovered that ribosomes are transported from glial cells to sciatic nerve axons (Müller et al. 2018). Such a mechanism could enable coordinated changes across neuronal networks, possibly contributing to the stabilization and strengthening of neural circuits and/or memory engrams.

We have explored how mRNAs make their journey to distant synaptic destinations. In the following paragraphs, we will discuss how mRNA translation is locally regulated and the role of local protein synthesis in synaptic plasticity and its implications for learning and memory.

3. Role of Local Translation During Plasticity

The idea that synapses can locally produce their own proteins was first proposed by Steward and Levy in the early 1980s after observing in electron micrographs polyribosomes in dendritic spines, challenging the traditional view that all proteins are synthesized in the soma of a neuron and transported to distal subcellular compartments (Steward and Levy 1982). Local mRNA translation at the synapse is believed to be one of the mechanisms essential for dynamic remodeling of the synaptic proteome. Translating mRNA locally in the synapse offers several advantages over expressing them in the soma and transporting them to distal synapses. First, local synthesis of synaptic proteins is much more energy efficient as one mRNA copy can produce several copies of a protein. Secondly, it enables faster responses to synaptic activity by bypassing somatic signaling. Additionally, local protein synthesis permits neurons to express proteins key to synaptic plasticity with higher spatial precision, as they can be expressed in an activity‐dependent manner directly at or near activated synapses.

3.1. Influence of Synaptic Plasticity on mRNA Mobility

How do neurons ensure that mRNA required for synaptic remodeling can mobilize to activated synapses when they undergo synaptic plasticity? As we will see below, in some cases, the mRNAs encoding for important plasticity proteins are already present at the synapse but remain translationally repressed. However, synapses are made of thousands of different types of proteins; thus, it is not feasible for neurons to maintain all the mRNAs needed for synaptic function and plasticity inside all synapses at all times (Sheng and Kim 2011). Therefore, when synapses undergo plasticity, neurons have evolved several mechanisms that allow them to transport and localize specific mRNAs in response to synaptic activity.

In the past few decades, a lot of effort has been put into the identification of mRNA transcripts whose transport is enhanced by neuronal activity. A well‐studied example is the immediate early gene (IEG) Arc, also known as Arg 3.1, a protein that has been shown to play an important role in various forms of synaptic plasticity and during memory formation (Guzowski et al. 2000; Plath et al. 2006; Waung et al. 2008). Arc mRNA is transcribed within minutes of neuronal activation and rapidly translocates from the nucleus to the cell soma and dendrites where it can be locally translated (Link et al. 1995; Lyford et al. 1995; Guzowski et al. 1999). Arc mRNA accumulates particularly near active synapses, which have been shown to require activation of NMDA receptors (Steward et al. 1998; Steward and Worley 2001). Nonetheless, the precise mechanism through which Arc mRNA is released is not yet fully understood. RNA granules positive for both CaMKIIα and Arc have been associated with 42 different proteins in dendrites, including the translational repressor FMRP (Kanai et al. 2004). Earlier studies have shown that synaptic stimulation via metabotropic or AMPA‐type glutamatergic receptors significantly reduces FMRP levels without affecting its mRNA (Antar et al. 2004; Narayanan et al. 2007). Loss of FMRP following stimulation could alleviate the repressive effect it normally has on mRNA translation, allowing FMRP‐bound mRNAs to be actively translated in an activity‐dependent manner. Knockout of Fmr1 in mice has indeed shown that FMRP is a repressor of translation of several dendritic mRNAs, including Arc (Zalfa et al. 2003).

Next to FMRP, there are other proteins that have been associated with the anchoring and release of mRNAs in an activity‐dependent manner. For instance, the localization of β‐actin mRNA inside ZBP1‐containing complexes toward dendrites and the axonal growth cone is inducible by KCl depolarization and BDNF respectively (Tiruchinapalli et al. 2003; Sasaki et al. 2010). Furthermore, prior to synaptic stimulation, β‐actin mRNA is abundantly present in dendrites but remains in a translationally repressed state, also known as a masked state (Buxbaum et al. 2014). Chemical induction of long‐term plasticity (LTP) leads to brief unmasking of the β‐actin mRNA, allowing it to be locally translated in response to stimulation. The increase in detectable β‐actin mRNA following stimulation withdrew to baseline within 30 min, which suggests that after a rapid local synthesis response of β‐actin, it returns to its masked state again, preparing it to respond to the next round of stimulation (Buxbaum et al. 2014).

3.2. Postsynaptic Regulation of mRNA Translation

Following synaptic activity, local protein synthesis is rapidly promoted inside the stimulated synapse, relying on both mRNA transcripts that were already residing in the synapse, as well as mRNA transcripts that are quickly being transported towards the stimulated synapse in an activity‐dependent manner. The upregulation in local protein synthesis can allow for the remodeling of the synaptic proteome and supplies the synapse with the building blocks needed to increase its size and strength. An interesting question is how neurons selectively regulate which mRNA transcripts are translated in response to synaptic activity, and how they contribute to plasticity and memory encoding and retrieval. In dendrites and spines, several independent mechanisms have been identified that regulate the translation of mRNA following synaptic activity.

First, some proteins that have been identified as regulators of local protein synthesis in an activity‐dependent manner. One well‐studied example is CaMKII which plays a major role during LTP, learning, and memory formation (Lisman et al. 2012; Yasuda et al. 2022; Bayer and Giese 2025). Following synaptic stimulation, the influx of intracellular calcium can activate CaMKII via Ca²⁺/calmodulin binding (Schulman and Greengard 1978; Saitoh and Schwartz 1985). Activated CaMKII plays a highly versatile role during LTP by interacting with many synaptic proteins that together regulate a broad range of functions important to LTP (Lisman et al. 2012; Rostas and Skelding 2023). Importantly, one of the functions activated CaMKII regulates during LTP is local protein synthesis. This is achieved by phosphorylating proteins such as the cytoplasmic polyadenylation element binding protein (CPEB) which in turn initiates the translation of mRNA containing the cytoplasmic polyadenylation element (Atkins et al. 2004).

Another example of a protein that can regulate mRNA translation in dendrites during plasticity is the mammalian target of rapamycin (mTOR) pathway. Both BDNF and stimulation of NMDA receptors can enhance mTORC1 activity, which can subsequently phosphorylate its downstream targets eIF4E‐binding protein (4E‐BP) and P70 S6 Kinase (S6K) to regulate local translation and promote LTP (Hoeffer and Klann 2010; Switon et al. 2017; Khamsing et al. 2021). Taken together, these examples show that following synaptic stimulation, key plasticity proteins such as CaMKII and mTORC1 can activate signaling cascades that upregulate local protein synthesis to promote plasticity and LTP.

It is not surprising that mRNA translation induced by synaptic stimulation can persist well beyond the transient period of stimulation, as sustained synaptic changes and reprogramming of the local translatome are crucial for plasticity and memory formation. Recently, a study by Tsokas et al., showed that the continuous interaction between two proteins, the kidney and brain expressed adaptor protein (KIBRA) and protein kinase M zeta (PKMζ), is essential for the maintenance of LTP and long‐term memory (Tsokas et al. 2024). Under basal conditions, peptidyl‐prolyl cis‐trans isomerase NIMA‐interacting 1 (PIN1) is constitutively active and represses the translation of several kinases, including PKMζ, in dendritic spines (Westmark et al. 2010). Following glutaminergic stimulation, PIN1 activity is downregulated, alleviating the inhibitory effect it has on its target mRNA, thus allowing these to be locally expressed. Downregulation of PIN1 allows for the synthesis of PKMζ, which can then phosphorylate serine 16 of PIN1, thereby further inhibiting PIN1 forming a positive‐feedback loop that maintains the translation of PKMζ mRNA over time (Westmark et al. 2010; Sacktor 2012). This mechanism highlights how brief synaptic stimulations can initiate long‐lasting changes in the local translational landscape of synapses. Persistent synthesis of plasticity‐related proteins supports the consolidation and maintenance of synaptic changes underlying LTP. It is very likely that additional positive translational feedback loops, that remain to be identified pre‐ and postsynaptically, are switched on during LTP and are involved in memory maintenance.

Interestingly, there is evidence that microRNA (miRNA) species can regulate protein synthesis in dendritic spines during synapse homeostasis and plasticity. miRNAs are small non‐coding RNA molecules of typically 19–25 nucleotides in length, which are capable of binding to their complementary target sequence thereby regulating gene expression (reviewed in Gulyaeva and Kushlinskiy 2016; O'Brien et al. 2018). One example of a miRNA that regulates gene expression in the synapse is miR‐134, first described by Schratt and colleagues (Schratt et al. 2006). miR‐134 has several target genes in the synapse. By interacting with the 3′ UTR of those dendritic mRNAs, it plays a role in dendritic spine morphogenesis, regulating the expression of LIMK1 and Pum2 (Schratt et al. 2006; Fiore et al. 2014). Additionally, miR‐134 is capable of inhibiting synaptic plasticity by downregulating the expression of the transcription factor cAMP Response Element‐Binding Protein (CREB) (Zhu et al. 2015). Besides miR‐134, other dendritically localized miRNA species have been implicated in LTP such as miR‐26a, miR‐384‐5p, and miR‐335‐5p (Gu et al. 2015; Capitano et al. 2017). Collectively, these studies provide evidence that local translation in the dendrites and spines during plasticity and LTP can be regulated by miRNA species.

3.3. Presynaptic Function and Local Protein Synthesis

Local protein synthesis does not occur exclusively in dendrites and dendritic spines. In fact, axons and their presynaptic boutons also contain mRNA and all the ribosomal machineries needed to locally produce their own proteins (Holt and Schuman 2013; Shigeoka et al. 2016; Arey et al. 2019; Hafner et al. 2019). A critical component of local translation in axons is the axonal endoplasmic reticulum (ER), which if disrupted can impair the localization of ribosomes along the axon, leading to a reduction in local translation that affects neuronal development (Koppers et al. 2024).

The RNA‐binding protein FMRP, known for its role in regulating postsynaptic translation, also plays a significant part in presynaptic protein synthesis. Parvin et al., demonstrated that FMRP regulates the local translation of active zone protein Munc18‐1, which plays an important role during neurotransmitter release (Parvin et al. 2019). Consistent with these findings, Monday et al. reported that in mossy fiber‐CA3 synapses, FMRP regulates local protein synthesis during plasticity through its involvement in presynaptic structural remodeling and neurotransmitter release (Monday et al. 2022). The mechanism through which FMRP regulates protein synthesis in the presynapse in the context of plasticity is not fully understood, though it is believed that the general mechanisms of mRNA binding, release and protein interactions are shared between FMRP in axons and dendrites (Wang, Sela‐Donenfeld, and Wang 2023).

Building on the topic of proteins that are involved in local translation in both pre‐ and postsynaptic plasticity, the mTOR pathway has also been implicated in both sides of the synapse. Presynaptic regulation of local translation through mTOR requires retrograde signaling of endocannabinoids and the activation of presynaptic cannabinoid type 1 receptors, which have been observed to be essential for the long‐lasting reduction of GABA release in hippocampal inhibitory synapses of rodents (Younts et al. 2016).

By integrating translating ribosome affinity purification (TRAP) with RNAseq, Ostroff et al., characterized the translatome of cortical axons projecting to the amygdala in rats and identified over 1200 mRNAs that were differentially expressed during the formation and consolidation of associative memory (Ostroff et al. 2019). Furthermore, the highly conserved RNA‐binding protein Imp is vital for the trafficking of mRNA to Mushroom Body γ axons in Drosophila (de Queiroz et al. 2025). Disruption of Imp‐dependent mRNA transport is detrimental for the consolidation of long‐term memories, but not the formation of short‐term memories.

Local axonal mRNA translation also plays an essential role during presynaptic plasticity in regulating neurotransmission. For example, local synthesis of VGLUT2, among 271 other transcripts, is modulated by the RNA‐binding protein CPEB2 in mice (Lu et al. 2024). Using a conditional Cpeb2 knockout model, it was shown that CPEB2‐mediated local synthesis of VGLUT2 is linked to presynaptic translation‐dependent LTP. Additionally, local synthesis in the axon was shown to be required for burst neurotransmission between neocortical layer 5 pyramidal synapses (Wong et al. 2024). This effect was driven by presynaptic NMDAR and mTOR signaling which triggers axonal mRNA translation during high‐frequency activity.

3.4. Diversity of Transcriptomes in Neuronal Cell Types Suggests Differential Regulations

The brain is composed of many different cell types, each characterized by their own transcriptomic and proteomic signature. Recent single‐neuron transcriptomic studies analyzing millions of neurons have identified up to 34 neuron classes and thousands of neuronal clusters based on their transcriptomic profile (Siletti et al. 2023; Yao et al. 2023). Also on the proteomic level, many different neuronal subclasses and synapse types can be identified (Aburaya et al. 2020; van Oostrum et al. 2023; Unterauer et al. 2024). Moreover, the expression of mRNA isoforms can vary across distinct cell types and brain regions (Joglekar et al. 2021; Patowary et al. 2024; Wolfe et al. 2024). It is expected that most mechanisms underlying synaptic plasticity are shared between different cell types, such as the induction of immediate early genes like Arc, c‐fos and NPAS4, calcium signaling cascades, the activation of central plasticity‐related proteins such as CaMKII and PKA. However, given the large number of different neuronal subclasses, each with a different transcriptome, important cell type‐specific differences are expected to exist.

A study by Bernard et al. demonstrated that the regulation of protein synthesis extends beyond cell type specificity and can occur even in a synapse type‐specific manner (Bernard et al. 2022). The authors investigated the formation of two specific synaptic connections, parvalbumin‐positive (PV+) and somatostatin‐positive (SST+) interneurons. Under basal conditions, mTORC1 activity is suppressed by the tuberous sclerosis proteins Tsc1 and Tsc2, and deletion of Tsc2 leads to hyperactivity of mTORC1. Interestingly, in a Tsc2 knockout genotype, they observe that this only affects the formation of excitatory synapses onto PV+ interneurons, but not those onto SST+ interneurons. This synapse‐specific effect can be attributed to the Erb‐B2 receptor tyrosine kinase 4 (ErbB4) protein, which is selectively expressed in PV+ interneurons. Activation of ErbB4 in PV+ cells leads to inhibition of Tsc2 activity, which induces mTORC1‐mediated local protein synthesis. Taken together, these findings highlight a mechanism through which local protein synthesis can be regulated in both a cell type‐ and synapse type‐specific manner.

So far, a direct examination of how local translation differs across distinct neuronal cell or synapse types in the context of synaptic plasticity and memory formation is still lacking. Investigating the local proteomic and transcriptomic changes before and after the induction of plasticity in a synapse‐specific manner would provide critical insights into how local protein synthesis supports plasticity and memory encoding.

4. Discussion

As described above, transporting mRNAs to the right compartment at the right time is crucial for neuronal function. The correct transport into dendrites and axons is believed to be important to ensure that distal compartments can rapidly adjust their protein composition in response to local demands. Importantly, this process is not uniform but highly specific. Indeed, neurons can target mRNAs to distinct synaptic or subcellular sites. As highlighted above this specificity seems to rely on several complex mechanisms: (1) the assembly of mRNAs and their RBPs into phase‐separated granules, (2) the recognition of cis‐acting targeting motifs and/or epitranscriptomic modifications such as m6A within 3′ UTRs, (3) the recruitment of motor proteins or organelle‐based “hitchhiking” systems for long‐range trafficking, and (4) capture mechanisms at delivery sites at or near synapses.

Cis‐acting elements within 3′ UTRs provide one layer of this specificity. An example is the sequence‐specific RNP complex between β‐actin and ZBP1 (Zhang et al. 2001). On the one hand, high‐throughput approaches have identified AU‐rich regions within 3′ UTRs as neurite‐localized targeting motifs (Mendonsa et al. 2023). On the other hand, studies of 3′ UTR isoform regulation have shown that neuronal activity can remodel their transcript isoforms, thereby altering mRNA localization patterns (Tushev et al. 2018; Ciolli Mattioli et al. 2019). Nevertheless, sequence elements alone do not explain the diversity of localization behaviors across compartments, suggesting that additional mechanistic layers are likely required.

RBPs contribute to this regulation by interpreting UTR motifs, packaging transcripts into mRNA granules, and linking cargos to kinesin and dynein motors (Antar et al. 2006; Li et al. 2009; Chu et al. 2019). They also can act as repressors, keeping mRNAs translationally silent during transport, and releasing them at their synaptic destinations (Besse and Ephrussi 2008). Post‐transcriptional factors and intrinsically disordered regions further modulate the properties of RNA granules formed by phase separation (Wiedner and Giudice 2021). Despite these advances, we still lack an understanding of how RBPs coordinate in different neuronal architectures. Neurons with relatively simple structures like cerebellar granule cells with their single branching point axons, may not require the same transport selectivity as hippocampal CA1/3 pyramidal neurons, which must distribute mRNAs into axons with extensive collateral branching. How these architectural differences shape the decision rules for mRNA routing remains an open question.

Epitranscriptomic modifications add another dimension in the mRNA transport system. For instance, m6A has been shown to regulate mRNA stability and translation (Leighton et al. 2018), but recent work also points to a role in transport (Martinez De La Cruz et al. 2025). In sympathetic neurons, NGF stimulation promotes m6A deposition and enhances axonal delivery of transcripts such as Trp53inp2, required for axonal growth and survival. This finding suggests that methylation can bias mRNAs towards long‐range trafficking. Whether similar mechanisms operate in cortical and hippocampal neurons to ensure selective supply of distal synapses is unknown, but such routing could be particularly important in axonal systems with elaborate axonal branching. Integrating m6A into models of transport selectivity will therefore be essential for understanding how different architectures meet their local protein demands.

Synaptic activity further modifies mRNA dynamics. Arc mRNA is rapidly induced and targeted to synapses in an NMDA receptor‐dependent manner (Guzowski et al. 1999; Steward and Worley 2001). β‐actin mRNA, often stored in a masked state within ZBP1 complexes, can be transiently unmasked during LTP to allow its rapid local protein synthesis (Buxbaum et al. 2014; Tiruchinapalli et al. 2003). FMRP degradation following stimulation relieves repression of its bound transcripts, among which is Arc mRNA (Zalfa et al. 2003; Narayanan et al. 2007). These findings illustrate how the same molecular factors can contribute to transport granule formation and respond dynamically to plasticity signals, ensuring selective translation where needed.

To date, most of our knowledge on local translation of localized mRNA in mature neurons comes from studies focusing on the postsynaptic compartment. That is because the very presence of mRNA in presynapses was long debated, before several independent research groups brought evidence of mRNA localization and translation in inhibitory and excitatory presynapses (Younts et al. 2016; Shigeoka et al. 2016; Hafner et al. 2019; Monday et al. 2022). Notably, while the induction of LTP recruits locally in postsynapses polyribosomes (e.g., Ostroff et al. 2002), a similar phenomenon for presynapses has not yet been described. This could be because it is technically challenging to visualize ribosomes using EM in presynaptic compartments due to the high density of lipids, or because mRNA translation in presynapses might primarily rely on monosome translation (one ribosome per mRNA). Of note, unless synaptic mRNA translation is studied using super‐resolution approaches and/or on isolated synaptic compartments (e.g., synaptosomes), it is not possible to distinguish pre‐ and postsynaptic signals.

If mRNA translation is differentially used and/or regulated in pre‐ and postsynaptic compartments, it raises the question of whether it has similar functions in both compartments. Fonkeu and colleagues build a computational model dissecting the contribution of mRNA and protein motion in dendritic protein distribution using CamKIIα as a case study. Using realistic dynamic parameters extracted from experimental data, the authors showed that about one half of the CamKIIα proteins are synthesized in dendrites and the other half in the soma (Fonkeu et al. 2019). Importantly, local synthesis of CamKIIα in the dendrite was essential to provide sufficient amounts of CamKIIα proteins to synapses. This work suggests that local protein synthesis in dendrites is not only important for plasticity but also used as a key mechanism to supply proteins to synapses. Is this also the case in presynapses? Or is presynaptic local protein synthesis mostly a vector of synapse diversification? These questions remain to be answered.

Presynaptic plasticity is commonly defined as a modulation in release probability (Pr) of synaptic vesicles following an action potential, a parameter that directly correlates to the strength of an individual synapse (Dürst et al. 2022). Two independent studies from 1993 have both shown that hippocampal excitatory synapses can be divided into two populations based on their Pr. The majority of the synapses exhibit a low Pr (< 0.09), whereas the high Pr population displays a sixfold increase in Pr (Hessler et al. 1993; Rosenmund et al. 1993). In a study investigating mossy fibers to CA3 neurons synapses, Monday and colleagues reported that LTP at this synapse induced enlargement of mossy fibers presynaptic boutons in a local protein synthesis‐dependent manner (Monday et al. 2022). Interestingly, this form of presynaptic plasticity appears to be under the control of FMRP, as deletion of presynaptic FMRP occluded structural and functional mossy fibers LTP (Monday et al. 2022). Moreover, inhibitory presynapses from hippocampal interneurons have also been shown to exhibit protein synthesis‐dependent long‐term depression (LTD) (Monday et al. 2020; Younts et al. 2016). Multiple factors have been identified in the past that control the Pr of synaptic vesicles, including structural remodeling of the active zone (Weyhersmüller et al. 2011), the modulation of the number of readily releasable vesicle pool (Dobrunz and Stevens 1997), local intracellular calcium dynamics (Schneggenburger and Neher 2000), and the availability of presynaptic proteins such as the SNARE machinery (Finley et al. 2002). Importantly, many of these factors could be upregulated in a synapse‐specific manner through local protein synthesis to minimally affect neighboring presynapses during memory formation (Figure 2).

FIGURE 2.

Schematic representation of memory engrams in brain neuronal networks. We hypothesized that presynaptic translation modifies vesicle release probability by increasing the number of release sites and intracellular calcium as well as by modifying the lipid composition of the active zone, thus, facilitating fusion. Created in https://BioRender.com.

The importance of the presynapse during memory formation on the system level should not be overlooked. The presynaptic compartment may offer unique advantages for memory formation and engram stability. For example, when a neuron that is part of an engram fires during memory recall, it is desirable that synaptic release occurs selectively in presynaptic terminals projecting to other engram neurons, while presynapses projecting onto non‐engram neurons remain unaffected. Through selective upregulation of the release probability in presynaptic terminals connected to other engram neurons, which is potentially supported through local protein synthesis, a neural network can strengthen engram‐specific communication, thereby aiding in the selective reactivation of a memory trace.

Moreover, the idea that presynapses contribute significantly to memory formation aligns well with the currently existing models on systems consolidation. The hippocampus plays a central role in the initial encoding, stabilization and consolidation of memory. During memory consolidation, engram networks migrate away from the hippocampus toward neocortical regions of the brain (Wiltgen et al. 2004; Takashima et al. 2009). Many studies have shown that over time after a memory has been consolidated, these cortical memory networks can be retrieved independently of the hippocampus (Wiltgen and Silva 2007; Winocur et al. 2007). The migration of an engram is an active process that involves the reorganization of the engram population, where new neurons are recruited into the memory trace and others gradually lose their relevance (Dudai 2004; Frankland and Bontempi 2005). Seemingly, hippocampal projections that initially were critical for the consolidation of long‐term memory become dispensable, meaning that the potentiated spines they are connected to no longer receive the proper stimulation and become irrelevant. Despite this, memories can still be preserved, suggesting that presynaptic mechanisms may play a significant role in the stabilization of the memory. Provided that a subset of the engram population remains stable during the reorganization of the engram, the memory can be maintained. Strikingly, presynaptic boutons have also been shown to be much more stable than their postsynapse counterparts, with roughly 80% of axonal boutons persisting over 12 months in the mouse barrel cortex (Qiao et al. 2016). Differences in synapse potentiation decays between the hippocampus (fast decay) and the cortex (slow decay) could explain such a decoupling.

High neuronal excitability is known to be a key factor in determining which neurons are preferentially recruited into memory engrams. However, given the vast number of neuronal subtypes, each with their own transcriptional profile, it is plausible that the transcriptomic identity of a neuron may also play a role during engram allocation. Differences in IEG expression or certain combinations of IEG expression might bias specific neurons for their allocation into memory engrams. Recently, a preprint by Arai and colleagues investigated the co‐expression of three popular IEGs that are often used as a marker for engram cells: c‐fos, Arc and NPAS4, in different brain regions in response to different experiences (Arai et al. 2025). They found that in response to different experiences (e.g., adverse vs. rewarding), different brain regions express different combinations of IEGs. This observation could explain why different brain regions are often specialized in the encoding of specific types and aspects of memory (Thorup 2024). Furthermore, overexpression of the transcription factor CREB in a subset of amygdala neurons in mice increased their excitability and was preferentially allocated to the fear memory engram (Han et al. 2007; Zhou et al. 2009). It remains unclear whether higher neuronal excitability results in transcriptional changes and IEG expression, or whether intrinsic transcriptional diversity dictates neuronal excitability, thereby influencing the allocation into memory engrams.

Taking this idea a step further, presynapse diversification may also play a significant role during neuron allocation into memory engrams. For example, presynapses with distinct transcriptomic profiles could potentially exhibit a higher release probability, thereby increasing the likelihood of recruiting its postsynaptic partner into the engram. This could be achieved through the differential expression of genes related to for example synaptic vesicle recycling, calcium dynamics, or other proteins that regulate presynaptic function. In conclusion, molecular heterogeneity in the presynapse may offer the potential to enhance the flexibility and adaptability of engram formation. Future studies integrating super‐resolution microscopy with spatial omics may be able to test and validate this model.

Author Contributions

Ezgi Daskin: conceptualization, visualization, writing – original draft, writing – review and editing. Stanley Van: conceptualization, visualization, writing – original draft, writing – review and editing. Anne‐Sophie Hafner: conceptualization, funding acquisition, project administration, supervision, visualization, writing – original draft, writing – review and editing.

Funding

This work was supported by H2020 European Research Council (101076961).

Conflicts of Interest

The authors declare no conflicts of interest.

Daskin, E. , Van S., and Hafner A.‐S.. 2025. “Local Protein Synthesis at Synapses: A Driver for Synapse Diversification.” Journal of Neurochemistry 169, no. 11: e70308. 10.1111/jnc.70308.

Data Availability Statement

Data sharing is not applicable to this review because no new data were created or analyzed.