Cytoskeletal regulation guides neuronal trafficking to effectively supply the synapse

Abstract

The development and proper function of the brain requires the formation of highly complex neuronal circuitry. These circuits are shaped from synaptic connections between neurons and must be maintained over a lifetime. The formation and continued maintenance of synapses requires accurate trafficking of pre- and postsynaptic components along the axon and dendrite, respectively, necessitating deliberate and specialized delivery strategies to replenish essential synaptic components. Maintenance of synaptic transmission also requires readily accessible energy stores, produced in part by localized mitochondria, that are tightly regulated with activity level. In this review, we focus on recent developments in our understanding of the cytoskeletal environment of axons and dendrites, examining how local regulation of cytoskeletal dynamics and organelle trafficking promotes synapse-specific delivery and plasticity. These new insights shed light on the complex and coordinated role cytoskeletal elements play in establishing and maintaining neuronal circuitry.

Introduction

The brain hosts an incredible assembly of neurons intricately connected to one another, a complex array of synapses that serve as the physical substrate encoding the human condition. With estimates of ~85 billion neurons in the human brain that each are capable of making thousands of synapses, there are hundreds of trillions of synapses that must be maintained over the lifespan of an individual^1,2. To support the ongoing function of this extraordinarily complex network, neurons must ensure a continual supply of new material to replenish synaptic components and replace aged or damaged proteins. But how are these essential cargos trafficked to the appropriate cellular locations? There has been considerable progress on this question over recent years. Here, we discuss these developments, focusing on novel insights into the molecular and cellular mechanisms regulating long-distance trafficking and synaptic delivery in neurons.

Neurons connect to one another by forming synapses between long cellular projections, called axons and dendrites. Typically, presynapses in axons oppose postsynapses in dendrites of adjacent neurons. These synaptic connections allow neurons to rapidly send chemical or electrical signals through the neuronal circuit. Synapse formation requires the precise delivery of pre- and postsynaptic components to contact sites between axons and dendrites^3,4. Importantly, once formed, synapses rely on continued re-supply of synaptic components for their ongoing function. At presynaptic regions, dense core vesicles (DCV) and synaptic vesicle precursors (SVPs) must be appropriately transported along the axon to replenish the active zone and locally generated synaptic vesicles (SVs)⁵. Postsynaptically, DCVs and neurotransmitters must navigate the more complex cytoskeletal organization of dendrites in order to correctly localize to postsynaptic densities. Additionally, both pre- and post-synapses require a steady supply of energy provided primarily by local mitochondria⁶. These mitochondria must be trafficked, docked, and maintained at synaptic sites, where they not only provide energy but also act as scaffolds for activity-dependent local translation of synaptic components^7,8. Appropriate regulation of sorting, trafficking, and delivery of these cargos and organelles is essential for maintenance of synaptic function.

Given the importance of the continued delivery of synaptic components, understanding the molecular and cellular mechanisms dictating long-distance trafficking in axons and dendrites is paramount. Synaptic components, like many other neuronal cargos, are primarily transported along microtubules: cytoskeletal polymers that provide structural integrity to axons and dendrites and serve as the ‘tracks’ upon which motor proteins traffic cargo. Microtubules are composed of tubulin heterodimers that assemble to form hollow, dynamic tubes with intrinsic polarity. Neurons exhibit distinct patterns of microtubule orientation depending on intracellular location. This orientation regulates the direction of transport by microtubule motor proteins, and therefore directs cargo delivery. Microtubule tracks can further be differentiated by decoration with microtubule-associated proteins (MAPs), and by post-translational modifications (PTMs) to either the inside or outside surfaces of the hollow microtubule polymer.

The motors that carry cargos along microtubules are also subject to multiple layers of regulation that influence motor activation, cargo selection, motility, and delivery location. There are two types of motor families involved in long-distance neuronal trafficking, kinesins and dynein. During neuronal transport, kinesin motors carry components toward the more dynamic microtubule plus end while dynein transports cargos toward the microtubule minus end. Distinctions within the axonal and dendritic environments, in concert with cellular cues at pre- and postsynaptic sites, dictate which motors carry what cargos where. However, our understanding of the intricate and dynamic layers of regulation directing neuronal trafficking is still developing. In this review, we will discuss cytoskeletal mechanisms directing synaptic component delivery with particular attention to axonal vs. dendritic microtubule environment, regulation of synapse-destined motors, the actin-myosin cytoskeleton in synaptic trafficking, and recently identified cargo-specific delivery mechanisms.

Microtubule environment regulates long-distance neuronal trafficking

The distinctive and polarized geometry of neurons necessitates a unique reliance on intracellular transportation mechanisms. Within a single neuron in the central nervous system, there are thousands of postsynaptic boutons distributed over millimeters of dendritic arbor and hundreds of thousands of en passant presynapses in an axon. Thus, the need for efficient and precise targeting of synaptic components is evident. While both dendrites and axons are polarized projections extending from the soma, their internal environments diverge greatly to provide distinct subcellular compartments. There are fundamental differences in microtubule organization, the MAP/PTM environment, and the motors/cargos that are found in each compartment (Figure 1). In the following section, we will review our current understanding of the unique axonal and dendritic microtubule environments.

Figure 1:

Multiple levels of regulation control long-distance microtubule trafficking in dendrites and axons. Neurons contain distinct subcellular compartments: axon (in black) hosts presynapses that pass on propagating signals to neighboring dendrites (in blue) that host postsynapses. A) Microtubules are formed from the polymerization of α/β-tubulin heterodimers. GTP-tubulin is added to the growing plus end, and after lattice incorporation the GTP-tubulin is hydrolyzed to GDP-tubulin. B, C) Microtubules are distinctly organized depending on which neuronal process they inhabit. Axonal microtubules are uniformly organized with GTP-rich dynamic plus-ends facing the distal axon. In contrast, microtubules are organized with mixed polarity in mammalian dendrites. D, E) Microtubule motor traffic is distinct in axons and dendrites. In axons, kinesin-1, −2, and −3 family motors move anterogradely toward the microtubule plus-end, while dynein/dynactin motors are responsible for retrograde trafficking toward the microtubule minus-end. In dendrites, kinesin-2, −3, and −4 family motors and dynein/dynactin motors can move cargo in either the anterograde or retrograde direction, depending on the orientation of their individual microtubule tracks. F, G) Microtubule post-translational modifications (PTMs) are differentially distributed across neuronal compartments. Axons contain acetylated, polyglutamylated, polyaminated, detyrosinated, and tyrosinated microtubules, with detyrosination enriched in the axon shaft and tyrosination enriched in the distal axon. Dendrites contain polyglutamylated, polyaminated, acetylated, and tyrosinated microtubules, with acetylation enriched on plus-end-out microtubule arrays and tyrosination enriched on plus-end-in microtubule arrays. H, I) Microtubule-associated proteins (MAPs) preferentially localize to microtubules in distinct subcellular compartments and regulate microtubule behavior and spatially distinguish motor accessibility. MAP7, MAP9, and Tau can decorate both axonal and dendritic microtubules, while DCX, DCLK1, MAP2, and SEPT9 are specifically enriched on dendritic microtubules.

Microtubule Fundamentals:

Microtubules are intrinsically polarized filaments formed from the polymerization of tubulin heterodimers; each heterodimer is an obligate dimer of one α-tubulin protein and one β-tubulin protein (Figure 1A). Heterodimers assemble head-to-tail to form longitudinal chains called protofilaments, and lateral interactions between neighboring protofilaments form the cylindrical microtubule lattice. A typical neuronal microtubule contains 13 protofilaments. The intrinsic polarity of the microtubule arises from the position of α- and β-tubulin along the lattice. The dynamic plus end of the microtubule terminates with an open β-tubulin, which houses an exchangeable GTP binding site. The hydrolysis of GTP to GDP in β-tubulin as heterodimers are incorporated into the lattice is responsible for the dynamic instability property of microtubules. Dynamic instability encompasses four stages: 1) polymerization, or microtubule growth, 2) catastrophe, the transition from growing to shrinking, 3) depolymerization, or microtubule shrinkage, and 4) rescue, the transition from shrinking to growing. The human genome contains numerous tubulin isotypes that vary modestly by amino acid sequence but exhibit differential expression across cell types and developmental stages. In neurons, the most prominent α-tubulin gene is TUBA1A, which accounts for over 95% of α-tubulin in the developing brain^9–11. Neuronal β-tubulin is less dominated by a single isotype, with the neuron-specific TUBB3 sharing the β-tubulin landscape with TUBB5, TUBB2B, TUBB2C, and TUBB4^12,13. Distinct isotypes have been shown in purified systems to influence microtubule protofilament number¹⁴, dynamic properties^14,15, and motor activity¹⁶. Whether neuronal microtubules within cells are formed from distinct pools of tubulin isotypes or a mixed isotype population is yet to be determined, but there is evidence that the loss of specific isotypes can be compensated for functionally^12,17.

Microtubule organization:

One of the most striking differences in the cytoskeletal environment of the axon versus that of the dendrite is the organization of microtubules. Microtubules in the axon are organized in a tiled array¹⁸ of overlapping microtubules that are uniformly oriented with their plus end directed outward toward the axonal terminal¹⁹ (~95% plus end out; Figure 1C). In contrast, mammalian dendrites contain microtubules of mixed polarity with some plus ends facing out and others facing in toward the soma¹⁹ (~60% plus end out; Figure 1B). This divergence in organization of axonal and dendritic microtubules greatly influences regulation of long-range trafficking, including which microtubule motors/cargos can enter each compartment and where the cargos ultimately get deposited.

Microtubule polarity directly influences the directionality of transport by motor proteins. There are two major classes of microtubule-based motor proteins that drive the long-range transport of organelle and vesicular cargos in neurons. Both classes hydrolyze ATP to produce the force necessary to ‘walk’ along microtubule tracks, but the direction of motion along microtubules is distinct. During neuronal trafficking, kinesin motors drive movement toward the microtubule plus end. Humans express 45 kinesin genes that are classified into distinct families and can perform numerous, diverse roles^20,21. Kinesin motors belonging to kinesin-1, kinesin-2, kinesin-3, and kinesin-4 families have been implicated in long-range neuronal transport. For additional information on specific kinesin families, the following reviews are recommended: kinesin-1²², kinesin-2²³, kinesin-3²⁴, kinesin family comparisons^20,25,26.

For movement toward the microtubule minus end, cells rely on a single motor, cytoplasmic dynein-1, that works in conjunction with a required activator, dynactin. Dynein drives the transport of a wide range of cargos, with specificity of cargo binding regulated by a growing family of activating adaptors, discussed below. The following reviews are suggested for further insight into the function and mechanics of dynein motors^27,28.

Due to the intrinsic polarity of the axonal microtubule network (Figure 1B, C), cargo transport toward the axon terminal (anterograde transport) is performed by kinesin motors and transport back toward the soma (retrograde transport) is performed by dynein (Figure 1I). In dendrites, where microtubules of mixed polarity reside, regulation of directional transport is more complex (Figure 1H). Motors, their cargos, and specific regulatory mechanisms will be explored later in this review.

Post-translational modifications (PTMs):

Microtubules are formed from the polymerization of tubulin heterodimers; each α- and β-tubulin subunit has a C-terminal tail domain that extends from the lattice surface. These tubulin tails can be modified post-translationally, either when the tubulin heterodimer is free in solution or alternatively when the dimers are assembled into a microtubule lattice. While many PTMs were identified decades ago, only recently have we begun to understand the spatial regulation of PTMs within neurons and how they regulate long-range transport. PTMs can influence numerous aspects of microtubule behavior, including assembly dynamics, polymer stability, MAP binding, and motor transport (Table 1). For example, kinesin-1 prefers binding to and walking along stable, acetylated^29,30 and/or polyglutamylated³¹ microtubules, kinesin-2 prefers detyrosinated microtubules¹⁶, kinesin-3 dissociates preferentially from GTP-rich dynamic microtubule plus ends³², and the dynein-dynactin complex binds preferentially to dynamic plus-ends enriched in tyrosinated tubulin^33,34.

Table 1:

Known neuronal microtubule PTM subcellular compartment localization and implications on MAPs and microtubules.

PTM	Specialized Localization	Microtubule implications	MAP implications
Detyrosination	Axon Shaft35,36	Longer-lived microtubules158	Kinesin-2 binding16
Tyrosination	Neurites35 Distal axon33,159 Dendrite plus-end out arrays29	Newly assembled microtubules158,159	Dynein-dynactin binding via the p150Glued CAP-Gly-domain 33,34
Polyglutamylation	Potentially ubiquitous		Tunes spastin activity160,161 Regulates MAP1A162, MAP1B162, MAP2163, Tau163, and kinesin-1 binding31 Promotes KIF1A pausing164
Acetylation	Axon35,165 Dendrite plus-end in arrays29	Longer-lived microtubules166,167 Promotes lattice flexibility168,169	Kinesin-1 binding29,30 Dynein binding170
Polyamination	Potentially ubiquitous	Microtubule cold stabilization171,172

PTMs are distributed throughout the neuron in both axons and dendrites, with some clearly enriched in specific subcellular compartments such as detyrosination of microtubules along the axon shaft^33,35,36 (Figure 1D, E, Table 1). However, our knowledge of specific PTM localization has been limited in the past by available labeling and imaging technology. Recent innovative optical nanoscopy strategies have helped reveal relationships between dendritic microtubule orientation, stability, and PTMs²⁹. Tas et al. used motor-PAINT to demonstrate that in dendrites, bundles of stable, acetylated microtubules are mostly oriented plus-end in (toward the soma) while dynamic, tyrosinated microtubule bundles are oriented plus-end out²⁹ (Figure 1D). This organization specifies cellular ‘highways’ biasing motor traffic direction and restricting motor access to the dendrite, allowing tyrosination-preferring kinesin-3 to enter dendrites on the plus-end out arrays while restricting acetylation-biased kinesin-1. Future studies are needed to provide a more comprehensive view of which PTMs exist on which neuronal microtubules. However, the model put forward by Tas et al. suggests that instead of a single microtubule populated by a medley of diverse PTMs, specific microtubules are populated by an individual type of PTM. How these distinct microtubule populations arise within the same cellular region must be elucidated in future studies.

Microtubule-associated proteins (MAPs):

Specific combinations of MAPs present on axonal and dendritic microtubules can spatially distinguish motor binding and/or motility. Recent studies have helped refine our understanding of MAP decoration across different neuronal regions and how these MAPs influence long-range transport. In axons (Figure 1G, Table 2), where kinesin-1 and kinesin-3 motors perform the majority of long-range cargo transport, abundant MAP7 promotes binding of the primarily axonal kinesin-1^37–40, while Tau slows transport of kinesin-1^37,38,41,42 and kinesin-3^37,38 cargos. In dendrites (Figure 1F, Table 2), kinesin-3 processivity is enhanced by the presence of DCX^38,43, DCLK1^38,43,44, MAP9³⁸, and SEPT9⁴⁵ while the same MAPs deter axonal-specific kinesin-1 from motility along dendritic microtubules^38,44. How established and newly generated axonal and dendritic microtubules become decorated by MAPs is still uncertain. Are all axonal/dendritic microtubules decorated by the same combination of MAPs (as displayed in Figure 1F, G)? Or are microtubules sharing the same subcellular space decorated by distinct MAPs? These and other questions remain to be answered, and discoveries will likely be aided by advances in super-resolution microscopy.

Table 2:

Known neuronal structural MAPs: subcellular compartment localization and implications for microtubules and motors.

MAP	Reported enriched localization		Motor implications	Reported microtubule implications
MAP1A		Dendrite (adult)173		Stabilizes microtubules - decreases dynamicity174
MAP1B	Axon (development)175 Axon44	Dendrite44	Interferes with LIS1-dynein interaction176	Stabilizes microtubules - reduces depolymerization rate177
MAP1S		Dendrite178		Stabilizes and bundles microtubules178
MAP7	Axon37,38	Dendrite37,38	Kinesin-137–40 Kinesin-337,38	Stabilizes microtubules179
Tau	Axon37,38,180	Dendrite37,38,181	Kinesin-137,38,41,42 Kinesin-337,38	Stabilizes microtubules - promotes assembly182, increases microtubule rigidity183, bundles microtubules184
MAP9	Axon38	Dendrite38	Kinesin-138 Kinesin-338 Dynein38
MAP2	Axon initial segment185	Dendrite44,180	Kinesin-138,185 Kinesin-338	Stabilizes microtubules - reduces depolymerization rate177, increases microtubule rigidity183, bundles microtubules184
DCX	Migrating neurons186 Axon growth cone187	Dendrite38	Kinesin-138 Kinesin-338,43	Stabilizes and bundles microtubules188, promotes formation of 13 protofilament microtubules189, promotes microtubule straightness190
DCLK1	Neurite growth cone191	Dendrite38,44	Kinesin-138 Kinesin-338,43,44

Intriguingly, the velocity of specific kinesins can vary depending on whether the cargo is trafficked through the dendrite or the axon⁴⁶. This suggests that the differential regulation and/or restrictions of a subcellular compartment, potentially via the microtubule organization, PTMs, and MAPs discussed above, influences the way a motor moves through the neuronal environment. Motor regulation extends beyond the microtubule track that it traverses, however. In the following section, we will discuss the motors involved in long-distance neuronal trafficking, including how/where they move and what they carry.

Regulation of neuronal microtubule motors influences synaptic component trafficking:

Kinesin and dynein motors are responsible for carrying diverse cargos (Table 3), many of which are essential for synaptic function. Not all cargos are carried at the same rate throughout the neuron, however. Many synaptic components are membrane-bound cargos that undergo fast axonal transport (200-400 mm/day⁴⁷). Synaptic vesicle proteins, membrane-associated receptors, neurotransmitters, neuropeptides, and mitochondria all fall within this transport class. Other synaptic components, notably clathrin⁴⁸ and synapsin⁴⁹, are cytosolic/soluble proteins that undergo slow axonal transport (2-8 mm/day⁴⁷). As the same motor types can carry cargos from both transport classes, the difference in transport rate is not due to intrinsic differences in motor velocity. Rather, the difference stems from how often the component is in motion. Motor-driven transport follows a saltatory, or ‘stop and go’, mode of transport. The slow components spend more time in the ‘stop’ phase which decreases the overall transport rate. While the slow axonal transport of cytosolic cargoes and their synaptic targeting is not fully understood, the numerous, diverse proteins traveling at these speeds are hypothesized to form macromolecular complexes traveling with microfilaments⁵⁰ or through short-lived interactions with fast-moving vesicles⁵¹.

Table 3: Known motor-cargo interactions and preferred subcellular compartment.

Green checkmarks denote kinesin detected in the indicated subcellular compartment. Red x’s denote a lack of consensus on the localization preferences of the indicated kinesins.

Kinesin family	Known neuronal cargos	Gene	Reported Preference
Kinesin-1	Presynaptic membrane - Syntaxin 1192, SNAP25193 APP-containing vesicles194 APOER2 and Reelin-containing vesicles195 TrkB vesicles196 Mitochondria197 Lysosomes197 Neurofilaments198 Retrograde trafficking proteins dynein and LIS180,199 mRNA complexes200 AMPA111 and GABA112 receptors	KIF5A	Axon52
		KIF5B	Axon46,52
		KIF5C	Axon46,52
Kinesin-2	Membranous organelles62 Spectrin-containing plasma membrane precursors63	KIF3A/B	Axon52
		KIF3A/C	Axon52	Dendrite52
	NMDA receptors64,65	KIF17	Axon52	Dendrite65
Kinesin-3	Synaptic vesicle precursors201 - Synaptophysin201, synaptotagmin-1201, VAMP2, SV2A, Rab3A, neurexins Dense core vesicles202 - BDNF202, NPY202, synaptotagmin-IV203, VMAT2204 AMPA receptors113 Late endosomes and lysosomes*205 Peroxisomes44 Tropomyosin Receptor Kinase A (TrkA)206 β-Secretase 1 (BACE1)207 Alphaherpesvirus particles208	KIF1A	Axon46,52,209	Dendrit46,52,209
	Mitochondria210 SCG10 / Stathmin-2211 PSD-95212	KIF1Bα	Axon211,212	Dendrite212
	Synaptic vesicle precursors213 Late endosomes and lysosomes*205 mRNPs214	KIF1Bβ	Axon46,213	Dendrite46,214
	Dense core vesicles215 - Rab6215, BICDR-1215, NPY215, BDNF215, Sema3a215 14-3-3 family members*216 Peroxisomes44	KIF1C	Axon44	Dendrite44
	Early endosomes*205 Peroxisomes44 Serotonin type 1A receptor217	KIF13A	Axon52	Dendrit4652
	Early endosomes*205 PIP3 vesicles218	KIF13B/G AKIN	Axon46,52,218	Dendrite5246,218
	Endosomes114 AMPA and NGF receptors114	KIF16B		Dendrite114
Kinesin-4	Peroxisomes44 K+-dependent Na+/Ca2+ exchanger NCKX2219	KIF21A	Axon44,52,75	Dendrite44,75
	Peroxisomes44 TrkB receptors76	KIF21B	Axon44,52 75	Dendrite44,52,75,76
Cytoplasmic dynein	Mitochondria in axon91 and dendrite137 Early endosomes in axon220 and dendrite221 Late endosomes/lysosomes in axons222,223 Signaling endosomes in axon81,220,224 and dendrite122 Synaptic vesicle precursors in axons32 Dense core vesicles in axons225 Autophagosomes86	DYNC1H1	Axon	Dendrite

Asterisks indicates a presumed neuronal cargo that has not been directly confirmed in neurons.

Determining where each motor is permitted within a neuron has been essential in linking specific motors to their diverse cargos. In combination with studies into axonal vs. dendritic PTM/MAP regulation of motor proteins, which are frequently performed using reconstituted systems with purified components, researchers have sought to discover where each motor/cargo is trafficked within the cell. New visualization techniques continue to improve our understanding of motor/cargo localization. For example, expression of fluorescently tagged, constitutively active motor domains from kinesin superfamily members in neuronal systems⁵² provided insight into the compartment specificity (axon and/or dendrite) of different types of kinesin. However, visualizing cargo-bound kinesins by labeling the vesicle-binding tail region⁴⁶ revealed a more complicated picture, as the axonal or dendritic compartment preference sometimes differs between the constitutively active motor domain and either the full-length or cargo-binding regions of the motor (Table 3). Additionally, binding of distinct adaptor proteins can also impact motor localization. In this section we will introduce the kinesin families responsible for long-range neuronal trafficking, including their known cargos, their subcellular compartment preference, and mechanisms for activating and regulating motility.

Kinesin-1:

Intracellular motors belonging to the kinesin-1 family were the first identified^53,54, and are often referred to as ‘conventional kinesin’. Expression of the kinesin-1 genes KIF5A and KIF5C is neuron-specific while KIF5B is ubiquitously expressed⁵⁵; collectively these kinesin-1 motors are responsible for trafficking many diverse cargos primarily into axons (Table 3). In neurons, functional kinesin-1 motors are generally hetero-tetramers comprised of two KIF5 kinesin heavy chain (KHC) homodimers and two kinesin light chain (KLC) homodimers (KLC1 or KLC2⁵⁶). KLCs regulate cargo binding, yet how they are capable of discriminating among structurally unrelated adaptor proteins to differentially regulate transport of numerous cargos remains largely unclear. Kinesin-1 folds to form an auto-inhibited state in the cytosol, with motility activated through cargo binding. Recent work based on crystallography proposes a two-step cargo activation model⁵⁷, in which the KLCs gate access to KHCs⁵⁸. First, cargo/KLC binding induces structural changes in the KHC dimer leading to unfolding, freeing the motor domains from autoinhibition by interaction with KHC tail domains. Then, cargo binding to the KHC tail is required to activate motility. Once activated, kinesin-1 motors drive processive, plus-end directed movement at velocities of 0.1-1 μm/s.

Kinesin-2:

Kinesin-2 family members assemble into more diverse oligomeric states. KIF3A, KIF3B, and KIF3C can hetero-oligomerize with an additional associated protein, KAP, to form the functional heterotrimeric kinesin-2 complexes KIF3A/B/KAP3 or KIF3A/C/KAP3^59–61. While both heterotrimers transport membranous organelles and spectrin-containing plasma membrane precursors, KIF3A/B/KAP3 functions primarily in axons while KIF3A/C/KAP3 is found in both axonal and dendritic compartments^52,62,63 (Table 3). In contrast, the kinesin-2 motor KIF17 homodimerizes to form a functional motor unit⁶⁴. KIF17 has been found associated with vesicles containing N-methyl-D- aspartate (NMDA)-type glutamate receptors within dendrites^64,65, although constitutively active versions of KIF17 localize to both axons and dendrites⁵². Recently, it was demonstrated that KIF17 cargos are restricted from the axonal compartment, further supporting the dendritic localization of KIF17⁶⁶. Kinesin-2 motors remain poorly studied in neurons and additional investigation is required to provide a more comprehensive view of their roles and regulation.

Kinesin-3:

Neuronally expressed kinesin-3 family members responsible for long-distance trafficking encompass three subfamilies and six distinct genes: KIF1A, KIF1B (expressed as two alternative isoforms KIF1Bα and KIF1Bβ⁶⁷), KIF1C, KIF13A, KIF13B, and KIF16B (Table 3). Numerous structural differences set kinesin-3 motors apart from other kinesins. Family-specific residue changes in the motor domain including a stretch of positively charged lysine residues in loop 12 (K-loop) promote binding to and processivity on microtubules, gaining them the moniker ‘marathon runners’⁶⁸. Motors in the kinesin-3 family also contain a number of other unique domains that influence their cargo/regulator binding, inhibited state, and activation^69–72. For example, the two isoforms of KIF1B diverge in both their motor⁷³ and cargo binding domains⁷⁴, providing distinct splice variants with unique motor characteristics that transport distinct cargos (Table 3). Unlike many other kinesins, kinesin-3 motors are found as inactive monomers in the cell that can dimerize and activate upon cargo binding⁶⁸. Work from Soppina and colleagues demonstrates the importance of the neck coil (NC) and coiled-coil 1 (CC1) domains in cargo-mediated activation of kinesin-3 motors, proposing that cargo binding induces release of intramolecular interactions between the NC and CC1 regions and increases the effective concentration of the motor to favor dimerization and super-processive motility⁶⁸. While kinesin-3 proteins are found in both axons and dendrites, KIF13 family members may preferentially traffic into axons (Table 3). The kinesin-3 motor family encompasses numerous neuronally expressed, super-processive kinesin proteins capable of rapid (2-5 μm/s), long-distance trafficking of diverse cargos in the axon and dendrite, many of which are essential for synaptic function.

Kinesin-4:

Kinesin-4 family members KIF21A and KIF21B are neuronally enriched motors that share significant sequence similarity with each other but only limited similarity within their motor domains to other kinesins⁷⁵. KIF21A and KIF21B can transport disparate cargos and function in distinct subcellular compartments (Table 3). While KIF21A is primarily axonal, KIF21B functions in both axons and dendrites^75,76. KIF21B motors also regulate microtubule dynamics, with neuronal activity tuning the balance between cargo transport and cytoskeletal remodeling⁷⁶. Missense mutations or loss of KIF21B cause neurodevelopmental abnormalities in humans and knockout mice^77,78, consistent with the importance of known cellular roles of this motor in the trafficking of BDNF/TrkB and the surface expression of synaptic receptors^76,77.

Cytoplasmic Dynein:

As the sole minus-end-directed motor responsible for long-distance trafficking in neurons, dynein motors carry diverse cargos along microtubules in both dendrites (anterograde and retrograde) and axons (retrograde). Because of its minus-end motility, active dynein motors are excluded from walking into the axon and thus selectively drive cargo into the mixed-polarity microtubule environment of the dendrite⁷⁹. Dynein relies on kinesin-1 motors for localization out to axon terminals⁸⁰. Dynein is a larger and more complex motor than the kinesins discussed above. The core of the dynein motor is a dimer of dynein heavy chains, large (~500 kDa) polypeptides with homology to AAA-family proteins (ATPases associated with diverse cellular activities) that include the catalytic ATPase sites and microtubule-binding domains. Additional non-catalytic subunits include two intermediate chains, two light intermediate chains, and additional light chains²⁸. Dynein is tightly regulated by the binding of additional co-factors. These include the multi-subunit dynactin complex and cargo-specific activating adaptors such as the BICD, HOOK, LIS1 and HAP/TRAK families of proteins^81,82. The stable tripartite dynein/dynactin/activating adaptor complex drives processive motility toward the microtubule minus end at velocities of ~1 μm/sec. Recent reviews provide more in-depth discussions of dynein activation and adaptor regulation⁸³ and dynein mechanics²⁷.

Motor-cargo organization:

Most vesicular cargos moving within neurons have multiple motor types bound simultaneously. Opposing kinesin and dynein motors can drive the motility of cargos such as lysosomes or mitochondria alternately toward either the plus or minus end of the microtubule^84,85. Even cargos that move over long distances in a single direction, such as autophagosomes, copurify with both kinesin and dynein motors⁸⁶. One model for regulation of these opposing motors posits a ‘tug-of-war’ situation where net cargo transport is directed by which motor team ‘wins’. Some neuronal cargos such as lysosomes move bidirectionally with frequent pausing and may fit this model⁸⁴. An alternative model posits key roles for motor-associated scaffolding proteins in regulating opposing motors bound to the same organelle⁸⁷. Scaffolding proteins such as JIP1 and JIP3 bind to both kinesin and dynein and can directly regulate motor activity on cargos such as autophagosomes and lysosomes^88–90. This coordinate regulation ensures that only a single motor type will be active at a time (avoiding an active motor ‘tug-of-war’) and allows for more processive unidirectional activity, such as that exhibited by axonal autophagosomes⁸⁸. Mitochondria also move bidirectionally due to associated kinesin-1 and dynein motors, with evidence for coordinate regulation of the opposing motors⁹¹. Further work is required to more fully define the mechanisms regulating motor coordination on mitochondria and other cellular organelles.

Multiple types of kinesins, such as kinesin-1 and kinesin-3 motors, sometimes coordinate to transport the same cargos. Recent work with kinesin complexes moving on immobilized microtubules in vitro⁹² confirms previous reports that the slower-moving kinesin-1 dictates transport speeds⁹³. This result is ascribed to kinesin-1’s ability to resist force-dependent detachment compared to kinesin-3⁹⁴ and supports the model that detachment rate is the driving motor characteristic determining team motility. Thus, the presence of motor teams on neuronal cargos may be especially important for kinesin-2 and -3 motors, which readily detach from the microtubule under load⁹⁴. In cells, when a two-motor complex containing one kinesin-1 and one kinesin-3 is present, which motor is active depends on the microtubule state⁹³. While the slower, more-stably-bound kinesin-1 typically dominates two-motor motility events, faster kinesin-3 ‘wins’ under conditions where more dynamic microtubules are present⁹³. Thus, cargo transport distance as well as transport velocity are likely to be determined by the biophysical properties of the specific motors involved.

Cargos require multiple motors working simultaneously for processive motility. Recently, a study using a novel in vitro 2D supported lipid bilayer system predicted the number of kinesin-1 motors necessary to drive processive movement of spherical vesicles, with 35 kinesin-1 motors proposed to move a 100 nm vesicle traveling 10 μm, and 6 motors required for a 30 nm vesicle traveling the same distance⁹⁵. These in vitro predictions can be compared to biological measurements that suggest 1-2 kinesin-1 and 6-12 dynein motors co-purify with 100 nm brain lysosomes⁸⁴. Teamwork between motors with faster detachment/reattachment kinetics (such as kinesin-2 and kinesin-3) may enhance transport of large cargos without increasing motor copy number.

Motors are capable of carrying numerous cargos throughout the neuron. But how are motors linked to their specified cargos? The answer to this seemingly basic question is complex, and continued research in this area highlights the numerous, diverse mechanisms at play to specify precise motor-cargo interactions. Depending on the motor, cargo interactions can be achieved through the following binding modes: motor-scaffold, motor-signaling protein, motor-Rab GTPase, motor-receptor, and/or motor-lipid binding. For example, kinesin-3 motors KIF1A and KIF1B are capable of direct interaction with phospholipid membranes through their pleckstrin homology (PH) domains; this interaction is important for synaptic vesicle precursor transport in C. elegans⁷⁰. For additional insight into motor-cargo interactions important for synaptic component trafficking, we refer the reader to the following review⁹⁶. Continued research into specific regulations for each motor, and for each of the motors’ cargos, will further define the complexities of neuronal motor trafficking. We will discuss additional mechanisms dictating synaptic cargo deposition in the following section.

Cytoskeletal regulation at the synapse directs cargo delivery

Cytoskeletal regulation of the axonal and dendritic environment clearly influences long-distance trafficking. However, once activated motors direct their cargos into the correct cellular compartment, trekking along microtubules modified by various MAPs and PTMs, how are these cargos properly distributed to synaptic regions? In this section, we will cover recent studies that shed light on local regulation of the cytoskeleton and motor/cargo interactions at synaptic sites to promote the delivery of essential pre- and postsynaptic components (Figure 2 and 3, respectively).

Figure 2:

Local regulation of cytoskeletal and motor dynamics dictates precise delivery of presynaptic components. A) Synaptic vesicle precursors are trafficked along axonal microtubules by KIF1A, which preferentially detaches from GTP-tubulin-rich microtubule ends at en passant presynapses³². B) Synaptic vesicle exchange is facilitated by actin elongation¹⁰², myosin-V-driven movement along actin filaments¹⁰⁴, and activity-dependent transport on augmin/γ-tubulin-nucleated microtubules¹⁰³. C) Dense core vesicles (DCVs) can be axonally trafficked by KIF1C, once autoinhibition has been relieved by Hook3 or PTPN21¹⁰⁶, or by KIF1A bound to Synaptotagmin-4 (Syt4)¹⁰⁵. At active presynapses, phosphorylated JNK phosphorylates Syt4, which destabilizes KIF1A-Syt4 binding to promote capture of released DCVs by actin¹⁰⁵. DCVs also undergo long-distance axonal circulation via kinesin (anterograde) and dynein (retrograde)¹⁰⁸.

Figure 3:

Cytoskeletal regulation of postsynaptic components. A) For dendritic trafficking of DCVs, calcium stimulates DCV loading onto KIF1A motors by promoting KIF1A-calmodulin (CaM) binding. At dendritic spines, postsynaptic density proteins TANC2 and liprin-α directly interact with KIF1A to capture DCVs¹¹⁰. B) NMDA receptors are transported along dendritic microtubules by KIF17, which interacts with the Mint1-containing scaffolding complex to bind NMDA receptor subunit NR2B⁶⁵. Upon arrival at postsynaptic regions, CaMKII binds to the tail of KIF17 to trigger the release of cargo⁶⁵. Additionally, Neurobeachin (NBEA)-containing tubular structures extending from Rab4-positive recycling endosomes transiently enter dendritic spines to promote recycling of NMDA receptors¹¹⁶. KIF21B and dynein help regulate receptor surface expression¹¹⁶. C) NBEA, which localizes to the ERGIC and Golgi complex, also regulates targeting of AMPA and GABA neurotransmitter receptors to synapses¹¹⁷. The axonally enriched kinesin-1, KIF5, is directed to dendrites by association with adaptor proteins GRIP1 and/or HAP1 to traffic AMPA or GABA receptors, respectively^111,112,120.

Synaptic vesicle precursor transport vesicles (SVPs):

Continued delivery of SVPs to presynapses along the axon shaft and at the axon terminal is necessary to replenish active zone components and SV proteins. SVPs and SVs by nature share similar components, making differentiation by fluorescent labeling challenging, but they are functionally and morphologically distinct. Unlike SVs, which are small, electron-lucent vesicles that are locally generated and clustered at presynapses⁵, SVPs are membranous organelles that are transported via fast axonal transport with an anterograde bias and preferential retention at presynaptic regions^32,97. Recently, it was discovered that en passant presynaptic regions along the axon shaft are enriched for dynamic, GTP-tubulin-rich microtubule ends³². This microtubule regulation directs the delivery of SVPs transported by KIF1A, which preferentially detaches from GTP-tubulin³² (Figure 2A). KIF1A disease-associated mutations that either alter this detachment preference³² or hyperactivate the motor⁹⁸ perturb SVP delivery and weaken presynapses. The upstream regulation that promotes enrichment of microtubule ends in presynaptic regions remains largely unexplored; future studies are needed to further elucidate how the microtubule environment at en passant presynapses is locally differentiated. Interestingly, kinesin-1 motors appear to have an opposing response to GTP-tubulin-rich regions and bind more tightly to microtubule ends⁹⁹, which raises the possibility that kinesin-1-carried cargos may ‘ride’ along the growing microtubule as it extends beyond the presynapse, bypassing presynaptic regions and reaching targets further along the axon. How this may influence the trafficking of synaptic-destined kinesin-1 cargos such as mitochondria and presynaptic membrane remains unexplored.

Synaptic vesicles (SVs):

SVs cluster at presynapses where they are exocytosed to release neurotransmitter into the synaptic cleft to stimulate the postsynapse. Upon neuronal stimulation, SVs undergo bidirectional interbouton translocation important for appropriate neurotransmitter release^100–103. SV clusters therefore can be shared amongst multiple presynaptic boutons, primarily via transport initiated by actin filament elongation independent of microtubule transport¹⁰² (Figure 2B). Nanometer-resolution tracking of individual SVs also implicates actin/myosin-V trafficking in activity-induced, long-distance inter-bouton SV exchange¹⁰⁴ (Figure 2B). Controversially, a new study implicates microtubule transport in activity-dependent SV sharing between en passant boutons. This study reveals a role for gamma-tubulin/augmin-mediated de novo microtubule nucleation at excitatory presynaptic boutons in SV interbouton transport and neurotransmitter release¹⁰³. These newly nucleated microtubules help span the distance between presynapses, facilitating both anterograde and retrograde interbouton SV motility and ultimately exocytosis upon neuronal stimulation¹⁰³ (Figure 2B). Additional insight is needed to understand what mechanisms regulate the on-loading and off-loading of SV cargo from microtubules and actin at presynapses and thus locally control synaptic strength and replenishment.

Dense core vesicles (DCVs):

Many DCVs contain active zone components; when axonally trafficked DCVs reach presynapses these cargos are used to assemble or maintain the active zone. It has been recently shown that DCVs carrying presynaptic component Synaptotagmin-4 (Syt4) are trafficked by the kinesin-3 KIF1A¹⁰⁵. The Syt4-KIF1A interaction is regulated by phosphorylation: the kinase JNK phosphorylates Syt4 at active presynapses to destabilize Syt4-KIF1A binding, leading to activity-induced capture of DCVs by actin¹⁰⁵ (Figure 2C). This work demonstrates how subtle, local changes to the cargo/motor interaction can have potent consequences on long-distance transport. Axonal DCVs can also be trafficked to the presynapse by the kinesin-3 KIF1C¹⁰⁶. New data reveal that KIF1C dimer autoinhibition is released by regulator binding to the KIF1C shaft region, either by protein tyrosine phosphatase N21 (PTPN21) or the cargo adapter Hook3. This motor activation stimulates the transport of KIF1C cargos including DCVs¹⁰⁶ (Figure 2C). As Hook3 activates both dynein/dynactin¹⁰⁷ and KIF1C¹⁰⁶, this mechanism could provide integration of bidirectional DCV transport¹⁰⁶. The mechanism directing KIF1C unloading at presynapses remains to be determined, but it is possible that KIF1C shares a similar mechanism as fellow kinesin-3 family member KIF1A³², leading to preferential detachment at presynaptic, GTP-tubulin-rich zones. In addition to anterograde trafficking by kinesin motors, retrograde DCV trafficking by dynein may also help supply en passant synapses. In Drosophila motoneurons, 90% of DCVs moving anterogradely bypass synapses to reach the axon tip, switch direction, and travel back along the axon carried by dynein¹⁰⁸. Of these retrogradely trafficked DCVs, 10% are captured as they pass a bouton, similar to the rate of capture from the anterograde-moving population. Interestingly, uncaptured DCV do not continue moving to the soma, but instead switch direction again and begin their anterograde march back out to the axon tip. This suggests a ‘conveyor belt’ mode of synaptic replenishment, with DCVs circulating back and forth along the axon with sporadic capture at en passant boutons^108,109. Whether this long range DCV circulation mechanism is at play in mammalian systems is yet to be determined.

While many DCVs are axonally trafficked to the presynapse, some DCVs containing neuropeptides such as BDNF are shuttled along the dendrite to postsynaptic regions. Ca²⁺-bound calmodulin (CaM) directly binds KIF1A’s stalk domain to promote loading of DCVs and enhance KIF1A motility¹¹⁰. Once these KIF1A-driven DCVs reach dendritic spines, static postsynaptic density proteins liprin-α and TANC2 capture them¹¹⁰ (Figure 3A). Interestingly, this Ca²⁺/CaM mechanism dictates loading of synaptic-destined KIF1A cargos, which is distinct from previously observed Ca²⁺-dependent unloading mechanisms, such as Ca²⁺/CaMKII-mediated disruption of KIF17 cargos or Ca²⁺/MIRO stalling of mitochondria (discussed in following sections). Whether KIF1A is released from DCVs upon liprin-α and TANC2 capture is yet to be determined, but as Ca²⁺ concentrations are generally higher at postsynaptic spines, it’s Ca²⁺/CaM-mediated cargo interaction may be maintained.

Neurotransmitter receptors:

Neurotransmission and synaptic strength depend on appropriate targeting and expression of receptors at postsynaptic densities. Kinesin-1^111,112, kinesin-2^64,65, and kinesin-3^113,114 family members have all been implicated in receptor trafficking (Table 3). NMDA receptor trafficking by kinesin-2 KIF17 is regulated by the Mint1-containing scaffolding complex, which binds NMDA receptor subunit NR2B to link the NMDA-containing vesicle to KIF17⁶⁵ (Figure 3B). Cargo binding to KIF17 relieves motor autoinhibition and promotes motility into the axon. However, upon entry into the axon KIF17 vesicles are stalled and dynein promotes re-direction to dendrites⁶⁶. At postsynaptic regions, the KIF17-Mint1 interaction is disrupted by local phosphorylation of the KIF17 tail by CaMKII⁶⁵. CaMKII is enriched at postsynaptic spines where it is activated by Ca²⁺ influx through NMDA receptors¹¹⁵. Thus, KIF17’s NMDAR-containing cargos are released specifically at dendritic spines⁶⁵ (Figure 3B). Once in the dendritic spine, NMDA receptors are regulated by Neurobeachin (NBEA)-containing tubular structures extending from Rab4-positive recycling endosomes, which transiently enter dendritic spines to promote recycling of NMDA receptors. KIF21B and dynein also help regulate receptor surface expression¹¹⁶ (Figure 3B).

Beyond its function at the dendritic spine, NBEA also localizes to the endoplasmic-reticulum-Golgi intermediate compartment (ERGIC) and Golgi complex where it regulates sorting of kainate, AMPA, NMDA, and GABA_A neurotransmitter receptors to expedite their postsynaptic delivery^116–119 (Figure 3C). Surprisingly, the axonally enriched kinesin-1, KIF5, is directed to dendrites by association with adaptor proteins GRIP1¹¹¹ and HAP1¹¹² to traffic AMPA or GABA receptors, respectively¹²⁰ (Figure 3C). Once at the postsynapse, AMPA receptors are recruited in part due to the novel cytoskeletal element CARMIL3 shown to regulate postsynaptic actin capping¹²¹.

These studies demonstrate that while some kinesins may generally operate in the axon, interactions with adaptors can drive them into the dendrite to deliver important postsynaptic cargos. These observations are consistent with induced recruitment assays showing that axon-preferring motors such as kinesin-1 can also drive long-distance cargo motility in dendrites, albeit less effectively than dendritic motors such as KIF21B and dynein¹²². More work is required to understand the relative roles of microtubule organization, MAP decoration, PTMs, adaptors, and cargo interactions in regulating axonal vs. dendritic targeting and transport. Further, much of the work investigating pre- and postsynaptic cargo delivery largely ignores how the presence of motor teams may influence cargo release mechanisms. The many layers of regulation likely allow for the specific delivery of cargos to sites of need in response to dynamic changes in neuronal activity.

Mitochondria trafficking to and docking at the synapse:

The brain consumes up to 20% of energy produced in the body, with synaptic transmission the most energetically expensive process¹²³. While there is debate over the sources and use of neuronal energy, it is thought that glycolysis/GAPDH-mediated ATP generation and mitochondrial oxidative phosphorylation primarily meet demands. There is compelling evidence for glycolysis providing energy for axonal vesicle transport^124,125 and maintaining high ATP levels during sustained synaptic transmission¹²⁶, but mitochondria are thought to be imperative for synaptic tasks such as release and recycling of neurotransmitter-containing vesicles at the presynapse and generation of postsynaptic currents at postsynaptic boutons⁶. New evidence suggests that local synaptic mitochondria are also important for activity-dependent generation of synaptic components by supplying energy as well as acting as scaffolds for local translation^7,8. In this section, we will briefly address current models of mitochondrial trafficking, synaptic docking, and localized function at the synapse. For more in-depth information, we direct the reader to the following recent reviews focused on synaptic mitochondria^6,127,128.

Synaptic transmission necessitates enormous local energy supplies which are in part fulfilled by mitochondria residing at pre- and postsynaptic regions. How are mitochondria trafficked and retained at these synapses? Recent work in the Aplysia sea slug demonstrates that initial synapse formation prompts enhanced mitochondrial trafficking, dependent on increased cAMP signaling, transcription, and protein synthesis in the presynaptic neuron¹²⁹. While mitochondrial trafficking has been attributed to both kinesin-1 (KIF5) and kinesin-3 (KIF1Bα), the role of kinesin-1 in mitochondrial motility is the most well-defined. All kinesin-1 family members, KIF5A, KIF5B, and KIF5C, are capable of transporting mitochondria^130,131. Kinesin-1 associates with mitochondria through direct interactions with the motor adaptors Mitochondrial Rho GTPases (MIROs)¹³¹ and Trafficking kinesin-binding proteins (TRAK1 and TRAK2 in vertebrates¹³²; Milton in Drosophila¹³³). MIRO localizes to the mitochondrial outer membrane protein and contains Ca²⁺-binding EF hands¹³⁴, which have been proposed to respond to local elevations in Ca²⁺ by promoting mitochondrial detachment from microtubules at presynapses^{131,135–137}. There are two neuronally expressed MIRO genes, RHOT1 and RHOT2, encoding the proteins MIRO1 and MIRO2, respectively; MIRO1 is the main regulator of mitochondrial trafficking¹³⁸. MIROs interact with TRAKs. TRAK1, which can bind both kinesin-1 and dynein, is required for bidirectional axonal mitochondrial trafficking while TRAK2 predominantly mediates dendritic mitochondrial transport through dynein binding^137,139. In addition to Ca²⁺-dependent mitochondria stalling, local glucose can also regulate mitochondrial positioning at synapses. TRAK proteins can form a complex with O-GlcNAc transferase 110 kDa subunit (OGT), a catalytic enzyme whose activity depends on glucose availability¹⁴⁰. Therefore, local increase in extracellular glucose levels leads to the O-GlcNAcylated state of TRAK1, which in turn promotes mitochondrial stalling¹⁴⁰.

Recent evidence suggest that once mitochondria reach individual presynapses, their presence is correlated to increased stability of those boutons¹⁴¹. Mitochondria are positioned more closely to en passant boutons than terminal boutons. Furthermore, not all presynapses (en passant or terminal) are associated with stable mitochondria, but the presence of localized synaptic mitochondria may influence synaptic longevity¹⁴¹. What directs and retains mitochondria to particular synaptic sites remains poorly understood.

Only 10% of axonal mitochondria are mobile in mature axons, the rest remain tethered in place¹²⁸. Many of these static mitochondria reside at synaptic sites. What keeps them there? Axon-targeted syntaphilin (SNPH) acts as a docking receptor specific for axonal mitochondria through a unique interaction with the microtubule-based cytoskeleton at presynaptic sites¹⁴². SNPH-bound mitochondria lose the ability to bidirectionally migrate; this immobilization is dependent on SNPH’s microtubule binding domain, which is necessary and sufficient to stall axonal mitochondria¹⁴². The actin cytoskeleton has also recently been implicated in immobilizing mitochondria at presynaptic sites¹⁴³. Dendritic regulation of mitochondrial docking remains less characterized. However, recent data suggest that both the actin and microtubule cytoskeletons are required for dendritic mitochondria tethering⁸. Future studies are required to determine how dendritic dynein-and/or KIF1Bα-trafficked mitochondria are unloaded and anchored to postsynaptic sites, and to decipher the mechanisms leading to the activity-dependent regulation of mitochondrial transport and function.

Actin-myosin cytoskeleton in synaptic trafficking:

While the bulk of long-distance neuronal trafficking is mediated by the microtubule cytoskeleton, actin and myosin also play important roles in synaptic function, including organelle biogenesis, organelle transport driven either by actin polymerization or myosin motors, localized cargo anchoring, and physical barriers to halt cargo motility (discussed more thoroughly in the following recent reviews^144–146). Of note, polarized actin filaments help define the dendritic cargo-restricting AIS¹⁴⁷. Additionally, actin along the axon shaft is organized in periodic rings interconnected by spectrin¹⁴⁸. These actin structures, along with neurofilament and microtubule cytoskeletal elements, dictate axon diameter, and actomyosin-II has even been recently shown to facilitate axon expansion and contraction^149,150. As neuronal motors/cargos must navigate the narrow axon shaft to reach their synaptic targets, this dynamic radial expansion aids efficient axonal trafficking of large cargos such as mitochondria, autophagosomes, and large endosomes¹⁵⁰. At presynapses, in addition to the previously highlighted role for actin-mediated interbouton SV exchange, actin plays an essential role in Ca²⁺-mediated presynaptic SV release and recycling (recently reviewed by Chanaday et al.¹⁵¹).

In the dendrite, actin plays a pivotal role in forming and maintaining postsynaptic spines. A periodic actin structure (similar to axonal actin rings) defines spine neck regions¹⁵², while branched actin filament networks are present in the head of the spine. As previously mentioned, actin at the postsynapse not only dictates spine shape but can also act to anchor cargos and help transport them into the spine head. Recently, actin remodeling in the base of postsynaptic spines has been demonstrated to promote microtubule entry¹⁵³. This remodeling is dependent on NMDAR-mediated Ca²⁺ influx, providing a link between synaptic activity and microtubule extension into postsynaptic spines. Together with evidence for targeted KIF1A/synaptotagmin-IV trafficking into hippocampal dendritic spines¹⁵⁴, these studies provide a snapshot into a complex postsynaptic delivery system involving activity-dependent actin remodeling, dynamic microtubule targeting, and specific kinesin/cargo motility regulation. They also raise numerous questions including how microtubules are steered into spines, how specific motors/cargo complexes are permitted entry into the spine via the targeted microtubule, and how the temporal dynamics of this multi-layered regulation unfolds.

Given the importance of local actin in synapse function, how is actin delivered along axons and dendrites? A new study challenges previously held assumptions that actin polymers are transported along microtubules, instead providing support for a model in which biased actin polymerization generates slow axonal transport of actin independent of microtubule-based motor trafficking¹⁵⁵. Whether this mechanism is conserved in the dendritic compartment is yet to be determined. For additional information on neuronal and synaptic actin-myosin regulation, we direct the reader to the following reviews^144,156,157.

Conclusions

Neurons utilize numerous cytoskeletal strategies to spatially regulate subcellular compartments and local synaptic regions to appropriately deliver necessary synaptic components. Both the organization and the dynamics of microtubules provide critical spatial cues that locally direct traffic. Differences in dendritic vs. axonal microtubule polarity have long been acknowledged to be important for the directed sorting of compartment-specific cargos. Recent progress has added detail to this cellular map by indicating how distinct distributions of PTMs and MAPs either enhance or restrict motor accessibility locally, although much remains to be discovered. For example, with new discovery of microtubule “highways” decorated with specific PTMs in the dendrite²⁹, questions emerge regarding how microtubule modifications are set-up and maintained and whether a similar system is utilized by the axon. Further, it is yet to be determined how MAPs are organized along dendritic and axonal microtubules. Do they also adhere to a segregated organization along microtubules, with a single MAP or specific collection of MAPs decorating a specific microtubule, or are they more ubiquitously dispersed? If they are segregated to distinct microtubule populations, how is this organization established and maintained? How might this system of PTM/MAP microtubule regulation influence cargo movement driven by motor teams comprised of distinct motor types? Beyond the microtubule track, numerous outstanding questions exist regarding how specific motor proteins associate with their varied cargos, and how long-range motility is regulated to span vast distances and ultimately deliver cargos at appropriate target sites. As highlighted in this review, the answers to these questions will likely vary depending on the motor and cargo of interest. Understanding these diverse mechanisms will be especially significant for determining synaptic network regulation.

Recent work highlights the critical role for microtubule dynamics in directing traffic at synapses, with growing evidence that local induction of microtubule assembly or modulation of polymer stability can directly influence synaptic strength. Additionally, new studies highlight the complexity of actin/microtubule interactions at both pre- and postsynaptic sites, and how this local regulation further directs the deposition of specific cargos. Continued research is needed to fully explore how neurons selectively regulate synaptic regions, and even individual synapses, to specify delivery and docking of cargos, such as mitochondria and receptors, which influence synapse strength and longevity. In summary, the progress made over the past few years has revelated much of the integrated levels of control that are required to effectively supply the synapse. These mechanisms support both the development and the life-long plasticity of the neural networks that power our ability to think, learn, and move.