molecular motors and mechanisms of directional transport in neurons

© 2005 Nature Publishing Group

R E V I E W S

NATURE REVIEWS | NEUROSCIENCE VOLUME 6 | MARCH 2005 | 201

A neuron has a highly polarized structure.A typical neu-ron comprises a cell body, several short, thick, taperingdendrites and one long, thin axon. Most of the proteinsthat are needed in the axon and synaptic terminals aresynthesized in the cell body and transported along theaxon in membranous organelles or protein complexes1.Most dendritic proteins are also transported from the cellbody, but several specific mRNAs are transported intodendrites to support local protein synthesis2 (BOX 1).

In the axon and dendrites, microtubules run in alongitudinal orientation3,4, and serve as rails alongwhich membranous organelles and macromolecularcomplexes can be transported5. A microtubule is a long,hollow cylinder that is made of a polymer of α- and β-tubulins and has a diameter of 25 nm. It has intrinsicpolarity, with a fast-growing ‘plus end’ and an opposite,slow-growing ‘minus end’6. Microtubules in axons anddistal dendrites are unipolar, with the plus end pointingaway from the cell body7,8. However, the microtubulesin proximal dendrites are of mixed polarity8. The orga-nization of microtubules also differs between axons anddendrites (BOX 2).

MOLECULAR MOTORS of the kinesin and dynein super-families move along microtubules. Many kinesinsuperfamily proteins (KIFs) move towards the plusend of microtubules (‘plus-end-directed motors’) and

participate in ANTEROGRADE TRANSPORT, selectively trans-porting molecules from the cell body to axons anddendrites. By contrast, RETROGRADE TRANSPORT, from theaxonal or dendritic terminals to the cell body, is car-ried out mostly by cytoplasmic dyneins, which areminus-end-directed motors5,9–12.

Selective transport to axons and dendrites has beenstudied from several viewpoints, including whichsequences of selectively transported proteins function asselective targeting signals, and whether the basic mecha-nism is one of selective transport or selective retention(whereby cargoes would be transported to both axonsand dendrites, and selectively eliminated by endocytosisfrom the inappropriate destination). However, manyseemingly unrelated sequences have been identified astargeting signals, and the identification of the targetingsequences of specific proteins has not always clarified theunderlying sorting mechanisms. Both selective transportand selective retention seem to occur, depending on thecargoes involved, but it is not clear how some cargoes aretransported selectively, whereas others are transportednonselectively. Understanding the mechanisms of sort-ing, selective transport and recognition is an importantendeavour. This review focuses on recent developmentsthat relate to the mechanisms of selective transport, withparticular emphasis on the role of KIFs.

MOLECULAR MOTORS ANDMECHANISMS OF DIRECTIONALTRANSPORT IN NEURONSNobutaka Hirokawa* and Reiko Takemura‡

Abstract | Intracellular transport is fundamental for neuronal morphogenesis, function and survival.Many proteins are selectively transported to either axons or dendrites. In addition, some specificmRNAs are transported to dendrites for local translation. Proteins of the kinesin superfamilyparticipate in selective transport by using adaptor or scaffolding proteins to recognize and bindcargoes. The molecular components of RNA-transporting granules have been identified, and it isbecoming clear how cargoes are directed to axons and dendrites by kinesin superfamily proteins.Here we discuss the molecular mechanisms of directional axonal and dendritic transport withspecific emphasis on the role of motor proteins and their mechanisms of cargo recognition.

MOLECULAR MOTOR

SUPERFAMILIES

Kinesin and dynein superfamilyproteins move alongmicrotubules, and myosinsuperfamily proteins movealong actin filaments by ATPhydrolysis.

*Department of Cell Biologyand Anatomy, GraduateSchool of Medicine,University of Tokyo,Hongo 7-3-1, Bunkyo-ku,Tokyo 113-0033, Japan.‡Okinaka MemorialInstitute for MedicalResearch, Toranomon 2-2-2,Minato-ku, Tokyo 105-8470,Japan.Correspondence to N.H.e-mail: [email protected]:10.1038/nrn1624Published online 15 February 2005


ANTEROGRADE AND

RETROGRADE TRANSPORT

The direction of anterogradeaxonal transport is from the cellbody to the synapses, whereasthat of retrograde axonaltransport is from the synapses tothe cell body.

FAST AXONAL TRANSPORT

Membranous organelles aretransported by fast axonaltransport at ~400 mm day–1,whereas cytosolic andcytoskeletal proteins aretransported by slow axonaltransport at ~0.2–2.5 mm day–1.

202 | MARCH 2005 | VOLUME 6 www.nature.com/reviews/neuro

R E V I E W S

motor, middle motor and carboxy (C)-terminalmotor types (referred to as N-kinesins, M-kinesinsand C-kinesins, respectively)5,14,15 (FIG. 2). There are 39N-kinesin genes, three M-kinesins (Kif2a, Kif2b andKif2c) and three C-kinesins (KifC1, KifC2 and KifC3)14.

Recently, a group of researchers in the field establisheda new standardized nomenclature for the classificationof kinesins — based on 14 large families, Kinesin 1 to14 — to facilitate understanding of the evolutionaryrelatedness of genes that have been identified withdifferent names in various phylogenies19. To minimizeconfusion, they decided that the names of individualmotors should remain unchanged, but that the standard-ized family name should be included in any publicationto clarify the evolutionary relatedness of individualmotors. The families and their members are listed inTABLE 1. For reviews of the properties and functions of the main KIF genes in mice and humans, see REFS 5,15,16,20–23; see also BOX 3.

All KIFs have a globular motor domain that showshigh degrees of homology and contains a microtubule-binding sequence and an ATP-binding sequence, but,outside the motor domain, each KIF has a uniquesequence. The diversity of these cargo-binding domainsexplains how KIFs can transport numerous differentcargoes. Many KIFs are expressed primarily in thenervous system, but KIFs are also expressed in othertissues and participate in various types of intracellulartransport.

In vitro, KIFs show microtubule-dependent motoractivity, and each KIF has a characteristic velocity. Thevelocities, which vary from 0.2 µm s–1 to 1.5 µm s–1,are consistent with the speed of fast axonal transportin vivo5,15. N-kinesins and M-kinesins generally movetowards the plus end of microtubules, whereas C-kinesinsseem to move towards the minus end. M-kinesins canalso depolymerize microtubules.

KIF5 monomers have an N-terminal globularmotor domain and neck, stalk and tail domains, andthey form homodimers. At the end of the stalkdomain, two kinesin light chains (KLCs) form fanlikeends24,25 (FIG. 1). However, KIFs have varied molecularshapes. Although many KIFs form homodimers,KIF1A and KIF1B (including the alteratively splicedforms KIF1Bα and KIF1Bβ) are monomeric26,27. Bycontrast, KIF3B and KIF3C form heterodimers withKIF3A. The resulting KIF3A–KIF3B or KIF3A–KIF3Cheterodimers assemble with kinesin superfamily-associated protein 3 (KAP3) to form a heterotrimericcomplex called the KIF3 complex or kinesin II.

Axonal cargoes and motorsIn the axon, many membranous organelles are trans-ported from the cell body to the synaptic terminals.Their contents include the components of synapticvesicles and plasma membrane at synaptic term-inals26,31–36, ion channels37–40, adhesion molecules33,41–43

and molecules that are abundant in growth cones34,44.Mitochondria are also transported27,45. Several KIFs are involved in the transport of these cargoes(FIG. 3a,b; TABLE 2).

Kinesin superfamily proteinsThe kinesin superfamily is a large gene family ofmicrotubule-dependent motors with 45 members inmice and humans5,13–16. The ‘conventional kinesin’(originally called kinesin) was the first member to bediscovered. It was identified biochemically as a candi-date microtubule-dependent motor for anterogradeFAST AXONAL TRANSPORT17,18. Subsequently, molecularcloning identified KIFs 1–5 — murine KIFs that arehomologous to kinesin at their motor domains13 (FIG. 1).Kinesin corresponds to KIF5. A combination of mole-cular biological approaches with a protein search ofgenome databases using a basic local alignment searchtool (BLAST) has identified 45 KIF genes14.

The 45 murine and human KIF genes have beenclassified into three types on the basis of the positionsof their motor domains: the amino (N)-terminal

Box 1 | Morphological characteristic of axons and dendrites

Axons and dendrites differ in terms of their shape, cytoskeletal content andorganization, and membrane protein composition134. When dendrites originate fromthe cell body, they tend to have a large diameter and taper off gradually. The transitionfrom the cell body to proximal dendrites is gradual, and the cytoplasmic architectures ofproximal dendrites and the cell body are similar. In particular, the endoplasmicreticulum and ribosomes are almost as abundant in the proximal dendrites as in the cellbody. Moreover, even distal dendrites contain ribosomes and endoplasmic reticula.

By contrast, axons are narrow from their origin, and do not usually contain ribosomesor endoplasmic reticula. The transition from the cell body to axon is distinct; the regionof the cell body from which an axon originates is called the axon hillock and it tapers offto the axonal initial segment, where action potentials begin. Although most parts of thecell body are rich in endoplasmic reticula, the axon hillock is not. At the axon initialsegment, the plasma membrane has thick underlying structures, and there is aspecialized bundle of microtubules.

Box 2 | Microtubule organization in axons and dendrites

As discussed in the main text, the polarity of microtubules in axons and dendritesdiffers. However, the organization of microtubules also differs.

Nascent microtubules, a polymer of α- and β-tubulins, are dynamic and unstable6.Various microtubule-associated proteins (MAPs) associate with microtubules andcontrol their dynamics in vitro and in vivo. Microtubules in neurites are usually muchmore stable than those in most interphase cells, owing to the presence of neuronal MAPs.

Quick-freeze, deep-etch electron microscopy has shown that microtubules in axonsand dendrites are highly crosslinked by meshworks of fine filamentous structures3,4.These crossbridges between microtubules consist of MAPs. The spacing betweenmicrotubules in axons is typically ~20 nm, whereas in dendrites it is typically ~65 nm.

Axons and dendrites express different MAPs, and this seems to determine the spacingbetween neighbouring microtubules. When axonal MAPs, such as tau, are overexpressedin cultured fibroblasts or Sf9 cells, microtubule bundles form and the cells extend longprocesses similar to axons135,136. These microtubule bundles are like those observed inaxons (the wall-to-wall distance of neighbouring microtubules is ~20 nm). By contrast,when dendritic MAPs, such as MAP2A or MAP2B, are overexpressed, microtubulebundles like those observed in dendrites (wall-to-wall distance of neighbouringmicrotubules ~65 nm) are formed136. The selective localization of MAPs to axons anddendrites seems to be achieved partly by the selective transport and stabilization of tauin axons and MAP2 in dendrites137–139.

As well as microtubules, axons contain neurofilaments, which are among the 10-nm-diameter intermediate filaments that are expressed in neurons. Neurofilaments are notabundant in small-caliber axons, but large-caliber axons contain many neurofilaments,and microtubules tend to run as fascicles interposed with neurofilaments.



R E V I E W S

ATPase and motor activities of KIF1Bβ32. Moreover, inheterozygous KIF1Bβ mutant mice, which are used as amodel of this disease, the transport of synaptic vesicleprecursors is significantly decreased.

KIF1Bα (formerly KIF1B), an isoform that isderived from the same gene as KIF1Bβ by alternativesplicing, transports mitochondria anterogradely27. TheN-terminal motor domains of KIF1Bα and KIF1Bβ areidentical, but their C-terminal tails share no significanthomology, whereas the C-terminal tails of KIF1A andKIF1Bβ have 61% amino acid identity27,32.

KIF5 transports vesicles that contain the amyloidprecursor protein (APP) along axons46,47. It has beenreported that these vesicles also contain β-secretase andpresenilin 1 (REF. 48), which participate in the proteolyticprocessing of APP to produce amyloid-β peptide, themain constituent of extracellular amyloid plaques inAlzheimer’s disease49,50. This indicates that the prote-olytic processing of APP to amyloid-β peptide mightoccur in transport vesicles.

KIF5 also transports other cargoes, including vesiclesthat contain apolipoprotein E receptor 2 (APOER2)51.APOER2 is the receptor for REELIN, an extracellularprotein that is defective in the reeler mouse, which hasataxia, a typical reeling gait and abnormal neuronalorganization52–54. The cargo vesicles of KIF5 also containproteins such as GAP43 and VAMP2 (REF. 47). KIF5, likeKIF1Bα, also transports mitochondria45,55. The fact thatmitochondria are transported by both KIF5 andKIF1Bα might not be surprising, as mitochondria couldhave many potential binding sites for motor proteins.KIF5 also transports oligomeric tubulin in a largetransport complex that is distinct from those of stablepolymers or other cytosolic proteins56. When fluores-cently labelled tubulin is microinjected into axons itmoves at a speed that is compatible with slow axonaltransport. This movement is perturbed if an antibody isused to block kinesin function, which indicates thatKIF5 also takes part in slow axonal transport56.

The KIF3A–KIF3B–KAP3 complex transports vesi-cles with diameters of 90–160 nm that are distinct fromsynaptic vesicle precursors or vesicles carried by othermotors, such as KIF5 and KIF2. The vesicles transportedby KIF3A–KIF3B–KAP3 are associated with FODRIN

through an interaction between KAP3 and fodrin, andare important for neurite extension57. KIF3C has alsobeen implicated in anterograde axonal transport, but thecargoes for the KIF3A–KIF3C–KAP3 complex have notbeen studied in detail. However, because Kif3c-knockoutmice show no particular phenotype, indicating that thegene is not essential58, this complex might carry cargoesthat are also carried by other motors.

The partitioning-defective protein 3 (PAR3) com-ponent of the polarity complex PAR3–PAR6–atypicalprotein kinase C, which is thought to be involved inthe process by which one juvenile neurite becomesestablished as an axon while the others remain short,accumulates selectively at the tip of the establishingaxon during this process. PAR3 was recently shown tointeract with KIF3A, which indicates that the polaritycomplex might be transported by KIF3A59.

KIF1A and KIF1Bβ transport synaptic vesicle pre-cursors along axons26,31,32. Although mature synapticvesicles are relatively uniform spheres of about 50 nmin diameter, these structures are usually not found inaxons. Instead, the components of synaptic vesicles aretransported in tubulovesicular organelles as precursors,and assembled into synaptic vesicles at synaptic termi-nals34–36. The synaptic vesicle precursor that is trans-ported by KIF1A contains synaptic vesicle proteins suchas synaptotagmin, synaptophysin and RAB3A, but notpresynaptic membrane proteins such as syntaxin 1A orsynaptosomal-associated protein 25 (SNAP25)26.

When the cargoes of KIF1A and KIF1Bβ are isolatedby immunoprecipitation they are found to containsynaptic vesicle proteins26,32. Furthermore, mice thatlack either KIF1A or KIF1Bβ have reduced synapticvesicle density at synaptic terminals, and impairedsensory and motor nerve function31,32. A mutation inKIF1Bβ has been linked to a family with a form ofhereditary peripheral neuropathy, CHARCOT-MARIE-TOOTH

DISEASE (CMT) type 2A (REFS 21,32). A heterozygous A toT point mutation, which transforms glutamine toleucine at position 98 of the ATP-binding site in themotor region of KIF1Bβ, was found in all affectedfamily members, but not in control subjects. In vitro,the mutation causes a significant decrease in the

CHARCOT-MARIE-TOOTH

DISEASE

(CMT). The most commoninherited peripheral neuropathy,characterized by weakness andatrophy of distal muscles,depressed or absent deep-tendon reflexes and mild sensoryloss. Type II is an axonopathyand type I is a myelopathy.

APOER2

(Apolipoprotein E receptor 2).A member of the low densitylipoprotein receptor gene family,which binds APOE-containinglipoproteins. It is also a receptorfor the reelin ligand onmigrating neurons.

REELIN

The gene that is disrupted in thespontaneous mutant mousereeler, which shows disruptedcellular organization in the braindue to aberrant migration ofneurons. Reelin encodes anextracellular molecule thatcontrols neuronal cellpositioning.

KIF5

KIF1A

KIF1Bα

KIF2A

KIF3

KIF4

KIFC2100 nm

Figure 1 | Principal members of kinesin superfamily proteins (KIFs) observed by low-angle rotary shadowing. Diagrams, constructed on the basis of electron microscopy orpredicted from the analysis of their primary structures, are shown on the right (the largerorange ovals in each diagram indicate motor domains). KIF5 (orange) forms a homodimerand kinesin light chains (blue) associate at the carboxyl (C) terminus to form fanlike ends.KIF1A and KIF1Bα are monomeric and globular. KIF2A forms a homodimer and its motordomains are in the middle (amino (N)-terminal, non-motor domain, blue). KIF3A and KIF3B(yellow and orange) form a heterodimer and kinesin superfamily-associated protein 3 (KAP3;green) associates at the C-terminal end. KIF4 forms a homodimer. KIFC2 also forms ahomodimer, but its motor domain is on the opposite side (N-terminal tail and α-helical coiled-coil domains, blue; C-terminal motor domains, orange). Reproduced, with permission, fromREF. 5 © (1998) American Association for the Advancement of Science.



R E V I E W S

In Kif2a–/– mice, there are an abnormally largenumber of overextended axon collateral branches inthe cerebral cortex and hippocampus, and in culturedhippocampal neurons. This results from the loss ofthe microtubule-depolymerizing activity of KIF2A atgrowth cones. In wild-type mice, microtubule exten-sion near the edge of growth cones is controlled,whereas in Kif2a–/– mice microtubules near the edgecontinue to grow, and either turn back or extend intonew branches63. Microtubule depolymerizing activityis lower in the growth cones of Kif2a–/– mice than inthose of Kif2a+/+ mice. Therefore, the uncontrolledextension of collateral branches in Kif2a–/– mice is dueto the lack of microtubule depolymerizing activity atthe advancing edges of growth cones. So, as well astransporting cargoes to growth cones, KIF2A influencesthe wiring of the brain by controlling microtubuledynamics in growth cones (FIG. 3a).

Dendritic cargoes and motorsMolecules that are transported in dendrites includethose associated with postsynaptic densities64, neuro-transmitter receptors65–71, ion channels72 and specificmRNAs. These include the mRNA for microtubule-associated protein 2 (MAP2), which is specificallyexpressed in dendrites, and for the α-subunit ofcalcium/calmodulin-dependent protein kinase II(αCaMKII) and activity-regulated cytoskeleton-associated protein (ARC), both of which are involvedin long-term potentiation2,73–77. In the past, dendritictransport has been less extensively studied than axonaltransport, partly owing to technical difficulties. Axonsfollow a long path and tend to run in fascicles, whichfacilitated early experimental approaches such as theuse of radioactive tracers and NERVE LIGATION. By contrast,dendrites tend to follow a shorter path and not to formfascicles. However, with the advent of new experimentaltechniques, several important transport systems indendrites have been clarified (FIG. 3c; TABLE 3).

First, NMDA (N-methyl-D-aspartate) glutamatereceptors (NMDARs) are transported in dendrites byKIF17 (REF. 65). KIF17, which is an N-kinesin and aplus-end-directed motor that is localized mainly indendrites, co-localizes with the NR2B subunit ofNMDARs and moves from the cell body towards thepostsynaptic region at an average speed of 0.76 µm s–1

(REF. 67). The physiological importance of the transportof NMDARs by KIF17 has been shown in transgenicmice66. Overexpression of KIF17 enhances working orepisodic-like memory and spatial learning and memoryin transgenic mice. Moreover, the genes for KIF17 andNR2B are co-regulated so that overexpression of KIF17leads to the upregulation of NR2B. This process mightinvolve the increased phosphorylation of a transcrip-tion factor, cyclic AMP responsive element bindingprotein (CREB)66,67.

Second, AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid) glutamate receptors (AMPARs)are transported by KIF5s (KIF5A, KIF5B and KIF5C)68. Inthis case, the binding of AMPARs apparently steers KIF5to dendrites, as discussed below.

KIF2 and brain wiring. KIF2, an M-kinesin, is expressedabundantly in the juvenile brain, particularly at growthcones13,60. It transports vesicles that are distinct fromthose transported by KIF5 or KIF3, and is reported totransport the β

gc-subunit of the insulin-like growth

factor 1 (IGF1) receptor. This immunochemicallydistinct receptor subunit is highly abundant in axonalgrowth cones61.

M-kinesins also have a unique microtubule-depolymerizing activity, which has been shown, by X-raycrystallography, to be due to the presence of a class-specific neck and loops62. A role for this microtubule-depolymerizing activity in neurons has only recentlybecome apparent63.

GAP43

A growth-associated,membrane-boundphosphoprotein, expression ofwhich is markedly elevatedduring neuronal developmentand regeneration.

VAMP2

VAMP (vesicle-associatedmembrane protein) is also calledsynaptobrevin.VAMPs aresynaptic vesicle proteins that arerequired for calcium-dependentexocytosis at synaptic terminals,and are also present on theaxonal surface.

KIF1A�

1695�

KIF1Bα

1151�

KIF1Bβ

1771�

KIF1C�

1100�

KIF3A�

702�

KIF3B�

748�

KIF4A�

1232�

KIF5A�

1028�

KIF5B�

962�

KIF5C�

956�

KIF13A�

1749�

KIF17�

1039�

KIFC2�

792�

KIFC3

824

KIF2A�

717�

KIFC1�

609�

Figure 2 | Structures of N-kinesins, M-kinesins and C-kinesins. The domain structures of theprincipal KIFs. The motor domains are shown in pink, the ATP-binding consensus sequence by athin purple line, the microtubule-binding consensus sequence by a thick purple line, the dimerizationdomains by yellow stripes, the forkhead-associated domains by red stripes and pleckstrin homologydomains by orange stripes. The number of amino acids in each molecule is shown on the right.KIF1Bα and KIF1Bβ are alternative transcripts of KIF1B. KIF1A, KIF1Bα and KIF1Bβ, KIF1C, KIF3A,KIF3B, KIF4A, KIF5A, KIF5B, KIF5C, KIF13A and KIF17 have their motor domains in the amino (N)terminus and are therefore N-kinesins5,14,15. KIF2A has its motor domain in the middle of themolecule and is therefore an M-kinesin. KIFC1, KIFC2 and KIFC3 have their motor domains in thecarboxyl (C) terminus and are therefore C-kinesins. N-kinesins are shown in green, M-kinesin in palepink and C-kinesins in blue.



R E V I E W S

cargoes. KIF13A transports vesicles containing theMANNOSE-6-PHOSPHATE RECEPTOR (M6PR)80 (FIG. 4a). KIF13Ais ubiquitously expressed and transports vesicles con-taining M6PR from the trans-Golgi network to theplasma membrane80. Mannose-6-phosphate (M6P)serves as a recognition signal for intracellular sorting;newly synthesized lysosomal hydrolases carryingM6P bind to M6PR in the trans-Golgi network. Thecytoplasmic side of M6PR binds to the AP1 (ADAPTOR

PROTEIN 1) ADAPTOR COMPLEX and is transported byclathrin-coated vesicles, primarily to endosomes butalso to the cell surface81,82. In transporting M6PR-containing vesicles, KIF13A binds to M6PR throughthe AP1 adaptor complex80; therefore, the AP1 adaptorcomplex serves as an adaptor for both the motor andclathrin coat. The AP1 adaptor complex comprises β1-,γ-, µ1- and δ1-adaptin subunits, and the C-terminaltail of KIF13A binds to β1-adaptin80, which is also thebinding partner of clathrin. However, KIF13A andclathrin bind to different domains of β1-adaptin.

Binding of KIF17 to scaffolding proteins. Another earlyexample of the use of scaffolding proteins by molecularmotors was KIF17 (FIG. 4b). The interaction betweenKIF17 and its cargo vesicles, which contain NMDARs, ismediated by a tripartite protein complex65 that containsLIN10 (MINT1), LIN2 (CASK) and LIN7 (VELIS/MALS).This complex is involved in the localization of proteinsin polarized cells such as neurons and epithelial cells,including synaptic vesicle exocytosis and the organiza-tion of the postsynaptic density83–85. All three proteinscontain PDZ DOMAINS86,87, but these domains do not bindto each other, leaving them free to recruit other proteinsto the complex. The C-terminal tail domain of KIF17interacts directly with the first PDZ domain of LIN10,which then sequentially interacts with LIN2 and LIN7.

Third, KIF5s also transport a large multisubunitcomplex of 42 proteins that includes the mRNAs forαCaMKII and ARC77. The S VALUE of this complex isestimated to be 1,000 or more, because it sedimentsmore than 12 times as fast as the synaptic vesicle marker(115S) or the free ribosome marker (80S) in a sucrosedensity gradient centrifugation. This complex is trans-ported exclusively in dendrites at a forward speed ofabout 0.1 µm s–1, although the tug-of-war between for-ward and backward movements decreases the net speedto 0.01–0.05 µm s–1. The proteins in this complex includethose associated with RNA transport, such as Fragile Xmental retardation proteins (FMR1, FXR1 and FXR2),PURα, PURβ and STAUFEN; those associated with protein syn-thesis, such as elongation factors (for example, EF1α andEIF2α); RNA HELICASES, such as DDX1 and DDX3; HETERO-

GENEOUS NUCLEAR RIBONUCLEOPROTEINS, such as hnRNPU andhnRNPA/B; and other RNA-associated proteins, such aspolypyrimidine tract-binding protein-associated splicingfactor (PSF)77.

In addition, KIF21B, a plus-end-directed motor, isspecifically localized in dendrites, whereas the closelyrelated KIF21A is found throughout neurons78. KIFC2, aC-kinesin, transports MULTIVESICULAR BODY-like organellesspecifically to dendrites79.

Cargo recognition by motorsIt was initially assumed that transmembrane cargo pro-teins would bind directly to specific motors. However, itis now clear that KIFs tend to use an adaptor/scaffoldingprotein complex for cargo recognition and binding.

Binding of KIF13A to AP1 adaptor. Although KIF13Ais not exclusively expressed in the nervous system, itwas one of the first examples of the involvement ofadaptor/scaffolding proteins in the binding of KIFs to

FODRIN

A spectrin family protein that isexpressed in nonerythroid cells.

NERVE LIGATION

When a nerve (such as the sciaticnerve) is ligated experimentally,anterogradely transportedcargoes accumulate on theproximal side and retrogradelytransported cargoes on the distalside of the ligated point.

S VALUE

Sedimentation coefficientexpressed in Svedberg units,whichare defined as 10–13 sec.This unit is used to characterizea protein’s sedimentationbehaviour and is higher forlarger (higher molecular weight)proteins.

PUR PROTEINS

The PUR factor was originallyidentified as a protein that bindsto purine-rich, single-strandedDNA sequences adjacent to aregion of stably bent DNAupstream of the human MYCgene.

STAUFEN

Staufen was originally identifiedas a maternal factor that isrequired for the anteroposterioraxis in Drosophila melanogasterembryos. It is an RNA-bindingprotein and is responsible for thelocalization of mRNAs.

RNA HELICASES

A family of proteins that have ahelicase domain, which bindsand unwinds RNAs. They havebeen implicated in mRNAtransport, splicing andtranslation. Putative humanRNA helicases have beenassigned the gene symbol DDX.

HETEROGENEOUS NUCLEAR

RIBONUCLEOPROTEINS

(hnRNPS ). A group of proteinsthat bind to nascent RNApolymerase II transcripts andpackage heterogeneous nuclearRNAs into hnRNP particles inthe nucleus. They have an RNA-binding motif and alsofunction in cytoplasmic mRNAtranslation and turnover.

MULTIVESICULAR BODY

An organelle with amultivesicular (inner vesicleswithin a vesicle) appearance,identified by thin-sectiontransmission electronmicroscopy. It is a subset of lateendosomes that might functionin receptor downregulation.

Table 1 | Family, subfamily and member names of kinesin superfamily proteins

Family name19 Class Previous family Subfamily Member names Examples ofname14,147 name in mammals nonmammalian

members*

Kinesin 1 N-1 KIF5 (KHC, Kinesin I, KIF5 KIF5A, KIF5B ‡KHC, Conventional kinesin) KIF5C §UNC-116

Kinesin 2 N-4 KIF3 (Kinesin II) KIF3 KIF3A, KIF3B, §KRP-85/95,KIF3C ||KRP85/95

Osm3/KIF17 KIF17 KIF17 §OSM-3

Kinesin 3 N-3 Unc104/KIF1 KIF1 KIF1A, KIF1Bα, §UNC-104KIF1Bβ, KIF1C

KIF13 KIF13 KIF13A, KIF13B

Kinesin 4 N-5 KIF4 KIF4 KIF4A, KIF4B, ¶ChromokinesinKIF21A, KIF21B

Kinesin 13 M KIF2 KIF2 KIF2A, KIF2B, #XKCM1KIF2C

Kinesin 14 C-1 Ncd/Kar3/KIFC1 KIFC1 KIFC1 ‡NCD, **Kar3C-2 KIFC2/C3 KIFC2/C3 KIFC2, KIFC3

New family names to be used for all phylogenies, previous family names based mostly on murine and human KIFs (kinesinsuperfamily proteins), and principal KIFs involved in neuronal transport are shown. Member names (genes and proteins) remainunchanged. New and previous family names for other members are Kinesin 5 (N-2, BimC/Eg5/KIF11 family), Kinesin 6 (N-6,CHO1/KIF23 family and KIF20/Rab6 kinesin family), Kinesin 7 (N-7, CENP-E/KIF10 family), Kinesin 8 (N-8, KIF18 family andKid/KIF22 family). Kinesin 9 to Kinesin 12 do not have counterparts in the previous classification. *See REF. 14 for references forindividual KIFs. ‡Drosophila melanogaster; §Caenorhabditis elegans; ||Strongylocentrotus purpuratus; ¶Gallus gallus; #Xenopuslaevis; **Saccharomyces cerevisiae. KHC, kinesin heavy chain. Under class, N, M and C indicate N-, M- and C-kinesins (see text).



R E V I E W S

there has been some debate about whether cargoes bindto the C-terminal tail of KIF5 or KLC. In some species,such as fungi, KLC is absent, implying that KIF5 aloneshould be able to bind to cargoes88–90. KIF5 binds to theN-terminal α-helical domain of KLC91, but KLC also hassix TETRATRICOPEPTIDE REPEAT (TPR) MOTIFS in its C-terminalregion, which could mediate protein–protein inter-actions92,93. Mutations in KLC result in the accumulationof membranous organelles in axons, indicating that KLCmight be involved in the association between KIF5 andits cargoes94. It now seems that cargoes bind both to theC-terminal tail of KIF5 and to KLC, but binding to KLCtends to be used for transporting cargoes to axons,whereas binding to the C-terminal tail of KIF5 is used fordirecting cargoes to dendrites68.

Binding through KLC and JIP. When vesicles are trans-ported to axons by KIF5, their binding tends to bemediated by KLC. Furthermore, the binding of KLC toa cargo protein is not direct, but is mediated by bindingto JIP1, JIP2 or JIP3 — scaffolding proteins of the c-JunN-terminal kinase (JNK) SIGNALLING PATHWAY51,95–97 (FIG. 5c).JIPs can form oligomeric complexes and simultaneouslybind to three components of the JNK signalling pathway,including JNK, and are therefore regarded as scaffoldingproteins for JNK pathway kinases98,99. JIPs are abundantat the tips of neuronal processes and synaptic junctions,where the plus ends of microtubules are found100–102.

The C termini of JIP1 and JIP2, which are about 50%identical, interact with the TPR motif of KLCs51. Inaddition, JIP1 and JIP2 contain a phosphotyrosine-binding domain, which interacts with the reelin receptorAPOER2 (REF. 103). KIF5 can associate with vesicles thatcontain APOER2 (REF. 51). KLC is also necessary for thetransport of vesicles that contain APP. Initial studies thatused co-immunoprecipitation in transfected cells and in vitro binding assays with GST fusion proteins46 led tosuggestions that the TPR motif of KLC interacts directlywith the transmembrane protein APP. However, it nowappears that JIP1 might also bridge the interaction ofKLC with APP. The cytoplasmic tail of APP binds to thephosphotyrosine-binding domain of JIP1 (REFS 96,97),and JIP1 is necessary for the interaction of APP andKLC. The in vivo significance of the two proposedmodes of interaction of APP and KLC — direct orthrough JIP1 — has yet to be determined.

JIP3 (also known as SYD, for sunday driver, inDrosophila melanogaster and JSAP1 for JNK/stress-activated protein kinase-associated protein 1 in mam-mals)102,104 also binds to the TPR motif of KLC at its N-terminal domain95. Although it is structurally un-related to them, JIP3 can form an oligomeric complexwith JIP1 or JIP2. JIP3 was proposed to be a trans-membrane protein95; however, its transmembranedomain has not been determined and the protein orproteins that bind to and are transported by JIP3 havenot been identified.

Binding through KIF5. The C-terminal tail of KIF5binds to AMPARs through a scaffolding protein, gluta-mate receptor-interacting protein 1 (GRIP1), and this

The PDZ domain of LIN7 binds to the C terminus ofthe NMDAR subunit NR2B85. Therefore, NMDARs aretransported by KIF17 through its binding to the scaf-folding protein complex for NMDARs.

KIF5 or light chain? KIF5 transports many cargoes,including mitochondria45, vesicles containing APP,APOER2,VAMP2 and GAP43 (REFS 46,47,51), vesicles con-taining AMPARs68, and RNA-transporting granules77.AsKLC associates at the C-terminal tail of KIF5 (FIG. 5a),

NODE CELLS

Ciliated cells, which havemonocilia that rotate andgenerate a leftward nodal flow ofextra-embryonic fluid. They arecrucial for the formation of theleft–right axis in an earlyembryo.

Box 3 | Kinesin superfamily proteins (KIFs) and their functions

For further details of standardized kinesin nomenclature, see REF. 19.

Kinesin 1• The KIF 5 subfamily. KIF5A, KIF5B and KIF5C have similar properties and differ

mainly in their expression patterns. KIF5B is expressed in many tissues, whereas KIF5Aand KIF5C are neuron-specific55. KIF5s transport mitochondria45, lysosomes140 andtubulin oligomers56.Vesicles containing amyloid precursor protein (APP) orapolipoprotein E receptor 2 (APOER2), which are transported to axons, bind to kinesinlight chain (KLC) through JIPs (scaffolding proteins of the c-Jun amino (N)-terminalkinase (JNK) signalling pathway)46,48,51,96,97. AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid) receptor-containing vesicles and RNA-containing granulesbind to the carboxy (C)-terminal tail of KIF5 and are transported to dendrites68,77.

Kinesin 2• The KIF3 subfamily. KIF3A and KIF3B form a heterotrimeric complex with kinesin

superfamily-associated protein 3 (KAP3)28,30. This complex transports vesiclesassociated with fodrin in axons57. It also participates in intraflagellar transport toform cilia in NODE CELLS, which are fundamental for left–right determination duringdevelopment20,141,142, and for the transport of opsin at the connecting cilia ofphotoreceptor cells143.

• The KIF17 subfamily. KIF17 forms a homodimer that transports NMDA (N-methyl-D-aspartate) receptor-containing vesicles in dendrites and is important for memoryand learning65–67.

Kinesin 3• The KIF1 subfamily. This includes KIF1A (mouse homologue of Caenorhabditis

elegans UNC-104), KIF1B and KIF1C. KIF1A and KIF1B are monomeric, whereasKIF1C is dimeric. KIF1B has two alternative transcripts, KIF1Bα (formerly KIF1B)and KIF1Bβ (REF. 32), which are identical at the N-terminal motor domain, but shareno homology elsewhere. By contrast, the C-terminal tails of KIF1A and KIF1Bβ have61% amino acid identity and both KIF1A and KIF1Bβ transport synaptic vesicleprecursors26,31,32. KIF1Bα transports mitochondria27.

• The KIF13 subfamily. KIF13A transports the mannose-6-phosphate receptorthrough its interaction with the AP1 (adaptor protein 1) adaptor complex80.

Kinesin 4• The KIF4 subfamily. KIF4A, KIF4B, KIF21A and KIF21B form this family. KIF4 is

expressed mostly in the juvenile brain144. KIF21B is localized specifically todendrites, whereas KIF21A is expressed throughout the neuron78.

Kinesin 13• The KIF2 subfamily. KIF2A, KIF2B and KIF2C have their motor domain in the

middle of the molecule (M-kinesins)13 and have a unique microtubule-depolymerizing activity, which KIF2A uses to suppress axon collateralelongation62,63. KIF2A transports vesicles in the juvenile brain60,61.

Kinesin 14• This family includes the KIFC1 and KIFC2/C3 subfamilies. Both have motor

domains in the C-terminal tail (C-kinesins).

• The KIFC2/C3 subfamily. KIFC3 is a minus-end-directed motor and transportsapically transported organelles containing annexin XIIIb in polarized epithelialcells145,146. KIFC2 transports multivesicular body-like organelles in dendrites79.



R E V I E W S

which is located between the sixth and seventh PDZdomains, is overexpressed, endogenous KIF5 accu-mulates, predominantly in the somatodendritic area.By contrast, when JIP3, which binds to KLC51, is over-expressed, KIF5 accumulates in the somatoaxonalarea, indicating that cargo binding to KIF5 steers cargoesto dendrites.

RNA-containing granules, which are transported todendrites, also bind to the C-terminal tail of KIF5 (FIG. 5b,left). The minimal binding site for RNA-containing

binding transports vesicles to dendrites (FIG. 5b, right)68.GRIP1 is a cytoskeletal postsynaptic density proteinwith seven PDZ domains that can interact with variousproteins and that is involved in the clustering ofAMPARs105. The minimal GRIP1 binding site of KIF5at its C-terminal tail (amino acid residues 807–934)overlaps with the cargo-binding domain of the funguskinesin90 and is contained in all KIF5 family genes(Kif5a, Kif5b and Kif5c). When the KIF5-bindingdomain of GRIP1 (amino acid residues 753–987),

MANNOSE-6-PHOSPHATE

RECEPTOR

(M6PR). The receptor formannose-6-phosphate, whichserves as a recognition signal forthe transport of newlysynthesized lysosomal hydrolasesfrom the trans-Golgi network toendosomes, and for the uptake ofexternal ligands on the cellsurface by receptor-mediatedendocytosis with coated pits.

AP1 ADAPTOR COMPLEX

A heterotetrameric complexcomposed of β1, γ, µ1 and δ1subunits that functions as anadaptor for clathrin-mediatedtraffic.

KIF3

FodrinAPOER2APP

KIF1A/KIF1BβKIF5/KIF1Bα

Mitochondrion

a

–� +

+

KIF5

KLC

KIF5

Microtubule

RNA-containing�granule

Synaptic vesicle�precursor

JIPsb

c

–�

LIN7�LIN2�LIN10

GRIP1

AMPA�receptor

NMDA�receptor

KIF5 KIF17KIF5

Microtubule

KAP3

++

++

–�

–�–�–�

+–�+–�

+

–�+

–� +

–�+

–�

+

–�

+

–�

+

–�

+–�

+–�

Distal dendrite

Transporting�vesicle


Multivesicular�body-like organelle

Proximal dendrite

Dendritic�spine

Dendrites

Cell body

Rough endoplasmic�reticulum

Golgi apparatus

Axon hillock

Microtubule

Initial�segment

Various transporting�vesicles

KIFAxon

Synaptic�vesicle

Axon�collateral

Microtubules

Growth coneKIF2A

Synaptic�terminalNucleus

Figure 3 | Kinesin superfamily proteins (KIFs) and cargoes for axonal and dendritic transport. a | A typical neuron,extending several dendrites (left) and a single thin axon (right) from the cell body. In the axon, microtubules are unipolar, withthe plus ends pointing towards the synaptic terminal. Microtubules form special bundles at the initial segment, which mightserve as the cue for axonal transport. Tubulovesicular organelles are transported anterogradely along microtubules by KIFs. In the growth cone of an axon collateral, KIF2A controls microtubule dynamics and the extension of collaterals. Rough endoplasmic reticula are abundant in most parts of the cell body, except for the axon hillock. Dendrites contain somerough endoplasmic reticlula. Microtubules have mixed polarity in proximal dendrites, but are unipolar in distal dendrites, withthe plus end pointing away from the cell body. Membranous organelles and RNA-containing granules are transported alongmicrotubules by KIFs. b | KIF5 transports vesicles containing APP (amyloid precursor protien) and APOER2 (apolipoprotein Ereceptor 2) by interacting with KLC (kinesin light chain)46,47,51,96,97. Mitochondria are transported by KIF5 and KIF1Bα27,45. KIF3 transports vesicles associated with fodrin57. KIF1A and KIF1Bβ both transport synaptic vesicle precursors26,31,32. JIPs,scaffolding proteins of the c-Jun amino (N)-terminal kinase (JNK) signalling pathway; KAP3, kinesin superfamily-associatedprotein 3. c | In dendrites, KIF5 transports vesicles containing AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazole propionicacid) receptors through an interaction between KIF5 and GRIP1 (glutamate receptor-interacting protein 1)68. RNA-containinggranules are also transported by interacting directly with KIF5 (REF. 77). KIF17 transports vesicles containing NMDA (N-methyl-D-aspartate) receptors by interacting through the LIN complex, a tripartite protein complex containing mammalianhomologues of the Caenhorhabditis elegans presynaptic density zone (PDZ) proteins LIN-2, LIN-7 and LIN-1065.



R E V I E W S

liposomes through its PH domain, and transports theseliposomes along microtubules in vitro108. KIF1Bβ, butnot KIF1Bα, also has a PH domain near its C terminus;therefore, the PH domain might contribute to the bind-ing of synaptic vesicle precursors to KIF1A and KIF1Bβ.However, because KIF1A and KIF1Bβ transport synapticvesicle precursors that contain a specific set of proteins,binding to lipids alone cannot mediate their recognition.This issue needs to be studied further.

Uniquely among motors, KIF1A shows processivemovement as a monomer undergoing biased diffusionalmotion along microtubules by alternately using twomicrotubule-binding loop elements within a KIF1Amolecule109–111. Because clustering of Unc104 in phos-phatidylinositol 4,5-biphosphate-containing lipidrafts through PH domains augments the transportspeed, it was proposed that motors might be dimerizedon liposomes108, and it has been shown that whenUnc104 is artifically dimerized the motors move morequickly112. However, clustering of KIF1A on the surfaceof a bead without dimerization also increases thetransport velocity, indicating that clustering might alsoaugment the velocity110.

KIF13B (also named GAKIN for guanylate kinase-associated kinesin) was found to be a binding partnerfor a guanylate kinase-like domain of DLG, the humanhomologue of the D. melanogaster discs large 1 tumoursuppressor gene113, which regulates cell proliferation andis proposed to perform scaffolding functions by linkingcytoskeletal components to transmembrane proteins.KIF13B and DLG accumulate at the tips of projectionsin cultured epithelial cells, which indicates that KIF13Btransports DLG. However, the nature of the cargoes thatare transported by KIF13B in association with DLG hasnot been studied113.

Cytoplasmic dynein and cargo recognition. The retro-grade motor, cytoplasmic dynein, forms a massivemultisubunit complex composed of heavy chains,which contain motor domains, and variable numbersof associated intermediate and light chains5,9–12. It isassociated with the protein complex dynactin, whichcontains a short filament of actin-related protein,

granules is a 59-amino-acid region of KIF5 (amino acidresidues 865–923), which is conserved in KIF5A, KIF5Band KIF5C77. Among the 42 or more proteins thatconstitute the large RNA-containing granule, PURαand PURβ are two of the most strongly bound compo-nents. However, the direct binding partner of KIF5 isunknown. When cultured neurons are transfected withboth green fluorescent protein (GFP)–PURα and cyanfluorescent protein (CFP)–KIF5, PURα-containinggranules are transported exclusively to dendrites,although CFP–KIF5 is distributed to both axons anddendrites77. However, when a CFP-tagged dominant-negative KIF5 mutant (CFP–∆N1), which contains theC-terminal tail RNA granule-binding site but lacks theN-terminal motor domain, is expressed, the movementof RNA-containing granules towards distal dendrites isinhibited77. αCaMKII and ARC mRNAs co-localizewith PURα-containing granules, but tubulin mRNAdoes not77. When RNA interference (RNAi) is used tosuppress expression of component proteins of the RNA-containing granules — such as hnRNPU, staufen, PURαand PSF — mRNA transport is suppressed. These resultsshow that RNA-containing granules are transportedto dendrites as a result of their direct binding to theC-terminal tail of KIF5.

In D. melanogaster oocytes, the oskar mRNA is local-ized to the posterior pole. Its localization depends on thekinesin heavy chain, but not on the KLC, as shown by a KLC null mutant106. In addition, in an established D. melanogaster Schneider 2 (S2) cell line, GFP-taggedFMRP forms ribonucleoprotein granules and moves alonglong, thin processes that are formed when S2 cells are cultured on a ConA-coated surface and treated with cyto-chalasin D107. The movement also depends on kinesinheavy chain, but RNA-mediated interference of KLCexpression does not affect the movement. These resultsindicate that the transport of mRNAs might depend onKIF5, but not on the KLC, in cells other than neurons.

Other cargo recognition. KIF1A has a pleckstrin homol-ogy (PH) domain near its C terminus. The Dictyosteliumdiscoideum orthologue of KIF1A (Unc104) binds to phosphatidylinositol 4,5-biphosphate-containing

LIN

The Caenorhabditis elegans PDZproteins, LIN-2, LIN-7 and LIN-10 were identified by theanalysis of the genes lin-2, lin-7,and lin-10, which define thebasolateral distribution ofepithelial proteins. Themammalian homologue ofLIN-10 is also called MINT1(Munc18-interacting protein);LIN-2 is also called CASK; andLIN-7 is called VELIS (forvertebrate LIN-7 homologue) orMALS (for mammalian LIN-7protein).

PDZ DOMAIN

A peptide-binding domain thatis important for the organizationof membrane proteins,particularly at cell–cell junctions,including synapses. It can bind tothe carboxyl termini of proteins.PDZ containing proteins oftencontain multiple PDZ domains.PDZ domains are named afterthe proteins in which thesesequence motifs were originallyidentified (PSD-95, discs large,zona occludens 1).

TETRATRICOPEPTIDE REPEAT

MOTIF

(TPR motif). Tandem repeats ofdegenerate 34 amino acids, whichmediate protein–proteininteractions. TPR-containingproteins are involved in biologicalprocesses such as transcriptionalcontrol, mitochondrial andperoxisomal protein transport,and neurogenesis.

JNK SIGNALLING PATHWAY

JNK was identified as amicrotubule-associated proteinkinase. It binds to the aminoterminus of c-Jun andphosphoroylates c-Jun. JNK isactivated by various cytokinesand environmental stresses.

FMRP

A set of proteins derived fromthe alternative splicing of FragileX mental retardation gene 1;FMRP, an RNA-binding protein,and the two highly homologousproteins FXR1P and FXR2P, areproposed to participate inmRNA transport.

ConA

Concanavalin A is a member ofthe lectin family. Lectins areproteins that bind tightly tospecific sugars and are oftenused experimentally to bind tocell surface glycoproteins. ConAis purified from the plantCanavalia ensiformis and hasspecificity for α-D-mannose andα-D-glucose.

Table 2 | Motors and cargoes in axonal transport

Motor Direction Recognition Adaptor/scaffolding Organelle/molecule Referenceprotein transported

KIF1A, KIF1Bβ Plus end Not known* Not known* Synaptic vesicle 26,31,32precursors

KIF1Bα Plus end Not known Not known Mitochondria 27

KIF3A–KIF3B– Plus end KAP3 Fodrin Fodrin-associating 57KAP3 vesicles

KIF5 Plus end KLC JIP1‡ APP-containing 46,48,96,97vesicles

KIF5 Plus end KLC JIP1/JIP2 APOER2-containing vesicles 51

KIF5 Plus end Not known Not known Mitochondria 45,55

*The binding of the pleckstrin homology (PH) domain to liposomes108 and binding on another domain to liprin-α148 have been reported,but whether it is the binding site for the recognition of synaptic vesicle precursors needs to be clarified. ‡Direct binding of APP (amyloidprecursor protein) to KLC (kinesin light chain) has also been proposed46. APOER2, apolipoprotein E receptor 2; JIP1 & JIP2, proteins ofthe JNK signalling pathway; KAP3, kinesin superfamily-associated protein 3; KIF, kinesin superfamily protein.



R E V I E W S

Smart motors? Microtubules have intrinsic polarity,which differs between axons and dendrites. This differ-ence could be used to achieve polarized transport intothe axon and dendrites. It has been proposed that thepost-Golgi transport of dendritic proteins might bemediated mainly by minus-end-directed motors43.However, as discussed above, there is increasing evidencethat this is not the case.

Membrane proteins are transported to dendrites, aswell as to axons, by plus-end-directed motors. For exam-ple, NMDARs are transported by KIF17, and AMPARsand mRNA complexes by KIF5 (REFS 65–68,77). However,the use of plus-end-directed motors for both axonal anddendritic transport poses an inherent question. How domotors differentiate axons from dendrites? Somemotors, such as KIF17 and KIF21B, might be able todifferentiate them65,78. In the case of KIF5, the directionof transport might be determined by whether cargoesbind through KLC46,51,95–97 or directly to KIF5 (REFS 68,77).

Microtubules in initial segment as cue. Although axonaland dendritic transport are often dealt with as two simi-lar, alternative pathways, in reality the requirements ofthe two are very different. A typical neuron has severaldendrites, and their proximal segments have largediameters. By contrast, each neuron has only one axon,and the diameter of a typical axon from the initial seg-ment onwards is very small. If the diameter of an axon isone-tenth of that of the cell body, axonally transportedmaterials need to be propelled from the cell body inonly about 0.25% of all possible directions to enter theaxon47. Some structural components, rather than adiffusible signal, must provide a directional cue forefficient sorting and transport to the axon47.

It has recently been shown that microtubules at theinitial segment serve as the cue for the KIF5 motordomain to enter the axon47. When the GFP–KIF5motor domain is expressed in hippocampal neurons, itaccumulates at the tips of axons, indicating that pref-erential axonal transport of KIF5 is due to its motordomain. Moreover, a mutated form of KIF5 that can berecruited to microtubules, but cannot translocate alongnor dissociate from microtubules, (rigor-KIF5) accumu-lates in the initial segment of axons, indicating that, undernormal conditions, the KIF5 motor domain prefersmicrotubules in the initial segment.

ARP1, and associated proteins such as p150Glued, as wellas dynamitin, which mediates the interaction betweencytoplasmic dynein and its cargoes5,12. Cytoplasmicdynein transports various cargoes retrogradely in theaxon; however, apart from the involvement of theprotein complex dynactin, the molecular interactionthat leads to the recognition of cargoes of cytoplasmicdynein is not clearly understood. During the traffickingof vesicles from the endoplasmic reticulum to theGolgi, βIII spectrin, a Golgi-associated spectrin iso-form, binds to ARP1 (REFS 114,115). However, in photo-receptor cells, the transmembrane protein rhodopsinbinds directly to a cytoplasmic dynein light chain(TCTEX1, t-complex testis expressed 1)116.

Directional transport and sortingAs discussed above, cargo recognition seems to haveimportant roles in directional transport. However,neurons use many mechanisms to selectively sort andtransport proteins to axons and dendrites.

Table 3 | Motors and cargoes in dendritic transport

Motor Direction Recognition Adaptor/scaffolding Organelle/molecule Referencesprotein transported

KIF17 Plus end Carboxy-terminal LIN complex NMDA-receptor-containing 65vesicles

KIF5 Plus end KIF5 GRIP1 AMPA-receptor-containing 68vesicles

KIF5 Plus end KIF5 Not known RNA-containing granules 77

KIF5 Plus end Not known Not known Mitochondria 45,55

KIFC2 Minus end Not known Not known Multivesicular-body-like 79organelles

AMPA, α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid; GRIP1, glutamate receptor-interacting protein 1; KIF, kinesin superfamilyprotein; LIN complex, a tripartite protein complex containing homologues of Caenorhabditis elegans presynaptic density zone (PDZ)proteins LIN-2, LIN-7 and LIN-10; NMDA, N-methyl-D-aspartate.

a b

–� +Microtubule

NR1 NR2B

LIN2

LIN10

LIN7

KIF17KIF13A

M6PR

Scaffolding�protein�

complex

AP1�adaptor�

complex

δµγ-adaptin β1-adaptin

NMDA receptor

Figure 4 | Kinesin superfamily proteins (KIFs) bind to cargoes through adaptor orscaffolding protein complexes. a | KIF13A binds to β1-adaptin of the AP1 (adaptor protein 1)adaptor complex and the AP1 adaptor complex binds to the mannose-6-phosphate receptor(M6PR)80. The AP1 adaptor complex comprises the β1-, γ-, µ1- and δ1-adaptin subunits. β1-adaptin has three domains — trunk, hinge and ear — and the carboxy (C)-terminal tail ofKIF13A binds to the ear domain. b | the C-terminal tail of KIF17 binds to one of the PDZ domainsof LIN10 (Munc18-interacting protein, MINT1)65. LIN10, LIN2 (CASK, calcium/calmodulindependent serine protein kinase) and LIN7 (VELIS, vertebrate LIN7 homologue/MALS,mammalian LIN7 protein), all have PDZ domains and interact through regions other than the PDZdomains to form a tripartite scaffolding protein complex, which binds to the NR2B subunit ofNMDA (N-methyl-D-aspartate) receptors.



R E V I E W S

What makes microtubules at the initial segmentunique? Electron microscopy has shown that micro-tubules in the initial segment have a high density, but itis unclear how this organization is maintained. Yellowfluorescent protein (YFP)–EB1, which binds to the tips ofgrowing microtubules117, shows a particularly highaffinity for microtubules in the initial segment47, indi-cating that there is a specific property of initial segmentmicrotubules that might be recognised by both KIF5and EB1. To maintain the difference between micro-tubules in the initial segment and those in dendrites,a continuous turnover of microtubules might beneeded47. If the dynamics of microtubule turnover arechanged by treatment with a low concentration of taxol,KIF5 loses its ability to preferentially bind to micro-tubules in the initial segment and to mediate directionaltransport to the axon47.

Transport versus retention. In addition to selectivetransport, a mechanism of selective retention has beenproposed.According to this idea, cargoes are transportednonselectively to axons and dendrites, but are eliminatedat one site by selective endocytosis and retained at theother because endocytosis is prevented.

In cultured hippocampal neurons, both VAMP2 andNgCAM are highly localized to the axonal surface33.VAMP2is uniformly distributed in axons, whereas NgCAM ismore abundant in distal axons, and these two membraneproteins are targeted to the axonal surface by distinctmechanisms. NgCAM is sorted into carriers that pre-ferentially deliver their cargo proteins to the axonal mem-brane, whereas VAMP2 is delivered to the surfaces of bothaxons and dendrites, but is preferentially endocytosedfrom the dendritic membrane33 (TABLE 4).

VAMP2 has well-defined endocytosis signals in itscytoplasmic domain and the mutation of a singleamino acid (methionine to alanine at position 46) inthis domain prevents endocytosis. When the mutantconstruct is expressed, VAMP2 is evenly distributed inthe cell body and dendrites as well as the axon118. Theendocytosis signal is not homologous to other synapticvesicle membrane proteins, and it is unlikely that thissequence is directly recognized by adaptor proteins.Therefore, the sequence is probably required for bindingto other synaptic vesicle proteins that can bind toadaptor proteins for endocytosis.

NgCAM has a well-defined endocytosis signal, a TYROSINE-BASED MOTIF, on its cytoplasmic tail82. This motifassociates with AP2, a clathrin adaptor that capturesplasma membrane proteins for endocytosis by coatedpits119. However, mutations in this region that blockendocytosis do not affect axonal distribution ofNgCAM33. A mutant NgCAM protein that lacks theentire cytoplasmic domain remains polarized to axons,indicating that the axonal targeting signal is in theectodomain. The deletion of fibronectin type III-like(FnIII) repeats in the ectodomain of NgCAM results inthe loss of axonal localization without affecting endo-cytosis. Therefore, the FnIII domain serves as the axonaltargeting signal. It is hard to see how an ectodomainsequence can provide a direct signal for selective transport

–� +

KIF5

KIF5 tailKLC

KIF5

KIF5 tailKLC


mRNA

GluR2

GRIP1

Dendrite

Microtubule

–� +

KIF5

KIF5 tailKLC

KIF5

KIF5 tailKLC

Axon

Microtubule

JIP1 JIP1,2

APP APOER2

b

a

c

KIF5

KLC

Motor Neck Stalk Tail

TPR motifs

KLCGRIP1

RNA-containing granule

KIF5JIP1,2,3

KLC

KIF5

Figure 5 | Kinesin superfamily protein 5 (KIF5) and its selective transport to axons anddendrites. a | Schematic model and binding domains of KIF5 and kinesin light chain (KLC). Top, schematic model of KIF5 associating with KLC. The KIF5 dimer associates with two KLCs toform a heterotetramer24. The globular motor domains of KIF5 are shown on the left, followed bythe neck, stalk and carboxy (C)-terminal tail, with which KLC associates to form fanlike ends.Middle, the domain structures of KIF5 and binding sites for KLC, GRIP1 (glutamate receptor-interacting protein 1) and RNA-containing granules. KIF5 consists of motor, neck, stalk and taildomains. The binding site for KLC is near the C-terminal end of the stalk region (amino acids775–802)25; those for GRIP1 and RNA-containing granules are in the C-terminal tail (amino acids807–934 and 865–923, respectively)68,77. The C-terminal tail cargo-binding site of KIF5 overlapswith the cargo-binding site of fungus KIF, which lacks KLC88–90. Bottom, the domain structure ofKLC and the binding sites for KIF5 and JIPs (scaffolding proteins of the c-Jun amino (N)-terminalkinase (JNK) signalling pathway). KLC contains six TPR (tetratricopeptide repeat) motifs. The binding site for KIF5 is in the N-terminal portion of KLC and JIPs bind to the TPR motifs. b | The C-terminal tail of KIF5 binds RNA-containing granules (left) and AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid) receptor subunit GluR2-containing vesicles (right),and transports them to dendrites68,77. The direct binding partner of RNA-containing granules isunknown. Binding to the C-terminal tail of GluR2 is mediated by GRIP1 (REFS 68,105). c | The TPRmotif of KLC binds to JIPs and transports APP (amyloid precursor protein)- and APOER2(apolipoprotein E receptor 2)-containing vesicles to axons51,96,97. The phosphotyrosine-bindingdomain of JIPs interacts with APOER2 and APP. However, it has also been proposed that APPbinds directly to KLC46.



R E V I E W S

For the lipid-anchored peripheral protein GAP43,the palmitoylation motif is sufficient for axonal target-ing, whereas the N-terminal palmitoylation motif ofpostsynaptic density protein 95 (PSD-95) is necessarybut not sufficient for sorting to dendrites64. When thepalmitolylation motif of GAP43 is fused to PSD-95, thechimaera is redirected to axons. The targeting of gluta-mate decarboxylase 65 (GAD65), which synthesizesGABA (γ-aminobutyric acid), to presynaptic clustersalso depends on its palmitoylation motif120. The axon-ally targeted palmitoylation motif targets GAP43 todetergent-insoluble glycolipid-rich complexes, and thetargeting of GAD65 is sensitive to cellular cholesterolconcentrations. This indicates that these proteins mightbe transported in association with cholesterol/glyco-sphingolipid microdomains (rafts) that are assembledin the Golgi complex121.

Dendritic targeting signals. Although targeting to axonsis accomplished by complex mechanisms, transport todendrites might be analogous to basolateral transport inepithelial cells, in which proteins undergo polarized sort-ing to apical and basolateral compartments122.When thelow-density lipoprotein receptor, which is expressed inthe basolateral domain in epithelial cells, is expressed in cultured neurons using viral vectors, it is highly polar-ized to dendrites. In addition, the tyrosine-based motif atthe cytoplasmic tail, which serves as the signal for baso-lateral targeting in epithelial cells, also functions as thedendritic targeting signal42. The GFP-tagged transferrinreceptor (TfR), which is another marker for basolateralsorting in epithelial cells, is also selectively transported to dendrites43. Vesicles that have been labelled withGFP–TfR are often seen in proximal dendrites but rarelyenter the axon, which indicates that dendritic targetinginvolves selective transport. The tyrosine-based motifin the cytoplasmic N-terminal tail of TfR serves as thedendritic targeting signal123 (TABLE 5).

The Kv4 (Shal) family of voltage-gated potassiumchannels tend to localize to the somatodendritic com-partment, and a 16 amino acid dileucine-containingmotif in the cytoplasmic C-terminal tail has been identi-fied as the dendritic targeting signal72. The motif is highlyconserved from Caenorhabditis elegans to humans, and issufficient to target axonally localized channels to den-drites.Another type of potassium channel, Kv2.1, formsdense clusters on proximal dendrites, whereas the relatedKv2.2 channels are uniformly distributed124.A 26 aminoacid sequence on the cytoplasmic tail of Kv2.1 has beenidentified as the targeting signal for clustering at proximaldendrites.

Metabotropic glutamate receptors (mGluRs) are dif-ferentially targeted to dendrites and axons. In culturedhippocampal neurons, mGluR2 is targeted to dendritesand mGluR7 is targeted to axons, and a 60 amino acidC-terminal cytoplasmic domain determines whetherthe receptor is targeted to the dendrites or axon69.Furthermore, mGluR1a is targeted to dendrites, but itsalternative transcript, mGluR1b, is targeted to axons.The C-terminal tail of mGluR1a has a dendritic target-ing signal, but the alternative splicing, which forms a

to axons. One possibility is that the ectodomain ofNgCAM interacts with a protein in the same vesicle, andthat the other protein carries the axonal signal that is rec-ognized by adaptor/scaffolding proteins or motors at thecytoplasmic side for transport.

Selective elimination by endocytosis in dendrites andselective retention in axons have also been proposed tomediate localization of the voltage-gated sodium channelNav1.2 to the axon. The cytoplasmic C-terminal region ofthe channel, which includes a DILEUCINE-BASED MOTIF that isrecognized by adaptor protein (AP) complexes, is respon-sible for endocytosis of Nav1.2 and its selective distribu-tion to axons37. In addition, a cytoplasmic loop betweentransmembrane domains II and III of Nav1.2 contains asequence motif that binds to ankyrin G, which is abun-dant in the axon initial segment38,39. If this sequence isectopically expressed in a somatodendritic potassiumchannel, the channel is redirected to the axon initial seg-ment38. So, some axonal proteins are localized by selectiveretention, rather than by selective sorting and transport.

Axonal targeting signals. The axonal targeting signalsfor some proteins are described above. Other targetingsignals have also been identified. For the shaker (Kv1)family of axonal voltage-gated potassium channels, tar-geting to the axon involves a conserved T1 tetrameriza-tion domain40. Mutations that eliminate the associationof the T1 domain with the non-pore-forming auxiliarysubunit Kvβ compromise axonal targeting, and whenthe T1 domain is fused to dendritic marker proteins,they are expressed at the axonal surface.

EB1

EB1 belongs to a family ofevolutionarily conservedproteins, and binds to the plusends of microtubules in diversecell types. The end-bindingmechanism remains unknown,but might use dynamic orstructural cues found atmicrotubule ends.

NgCAM

The neuron–glia cell adhesionmolecule is a member of theimmunoglobulin (Ig)superfamily of neural celladhesion molecules, which arethought to play a role in axonalpathfinding and fasciculation.

TYROSINE-BASED AND

DILEUCINE-BASED MOTIFS

Tyrosine-based motifs (such asNPXY or YXXφ) and dileucine-based motifs (such as[DE]XXXL[LI] or DXXLL) arerecognized by components ofthe protein coats of adaptorprotein complexes.

T1 DOMAIN

A domain that is required for thetetramerization of the foursubunits that associate to formthe central pore of a potassiumchannel.

Table 4 | Mechanism and sorting signals of axonal transport

Cargo Targeting signal Mechanism* References

NgCAM Extracellular domain Selective transport 33(fibronectin-type-III-like repeats)

VAMP2 Cytoplasmic (endocytotic Selective retention 33signal) (selective endocytosis

at dendrites)

Nav1.2 Cytoplasmic (dileucine- Selective retention 37based motif) (selective endocytosis

at dendrites)

Kv1 (Shaker) T1 tetramerization domain 40

GAP43 Palmitoylation on cysteines 64

GAD65 Palmitoylation on cysteines 120

*Mechanisms are indicated only when they are explicitly discussed in the references. GAD65,glutamate decarboxylase 65; GAP43, growth-associated protein 43; Kv1, a voltage-gated potassiumchannel; Nav1.2, a voltage-gated sodium channel; NgCAM, neuron–glia cell adhesion molecule;VAMP2, vesicle-associated membrane protein.

Table 5 | Mechanism and sorting signals of dendritic transport

Cargo Targeting signal Mechanism* References

Transferrin Cytoplasmic (tyrosine- Selective transport 43,123receptor based motif)

Kv4 (Shal) Cytoplasmic (dileucine- 72based motif)

mGluR2 Cytoplasmic tail 69

*Mechanisms are indicated only when they are explicitly discussed in the references. Kv4, a voltage-gated potassium channel; mGluR2, metabotropic glutamate receptor 2.



R E V I E W S

Other proteins are transported nonselectivelythroughout the neuron before being selectively elimi-nated from inappropriate destinations by endocytosis.They are presumably retained in the desired location byassociating with cytoskeletal complexes. In this case,signals that are recognized by the endocytotic apparatusmight, in effect, serve as targeting signals.

Although many issues need to be clarified further,motors are important in all selective transportprocesses. Motors can intrinsically distinguish betweenaxons and dendrites, perhaps as a result of cues frommicrotubules. The mode of binding of cargoes tomotors can also affect the direction of transport, pre-sumably by changing the conformation of the motors.Previous structural studies have helped to clarify thecharacteristics of individual motors. For example,KIF1A is a unique monomeric motor, and single mole-cule biophysics, optical trapping, cryoelectron micro-scopy and X-ray crystallography have revealed how itmoves109–111. X-ray crystallography of KIF2 has revealedthe structural attributes that underlie its unique micro-tubule-depolymerizing activity62. Structural studies thatcompare motors in terms of axonal versus dendritictransport and their interaction with microtubules mightprovide important information.

Because of the variety of molecules that are selectivelytransported to axons and dendrites, it is not surprisingthat many mechanisms are used. Some of these mecha-nisms might be redundant, because no sorting machineryis likely to be 100% efficient. A basic understanding ofthe transport process from the viewpoint of motors andtheir association with cargoes will help to clarify thecommon principles by which cargoes are selectivelysorted and transported.

There are still motors whose functions are unknown,and little is known about how transport is regulated —for example, how the association and dissociation ofcargoes and motors is controlled131–133. These questionswill be the subject of ongoing and future studies.

shorter tail, exposes a tripeptide motif that serves as anaxonal targeting signal71.

mRNA localization in polarized cells has been studiedin oligodendrocytes125, fibroblasts126 and developingembryos127. In most cases, cis-acting sequences in mRNAsare recognized by trans-acting binding proteins. A 640nucleotide sequence in the 3′-untranslated region (3′-UTR) of MAP2 mRNA has been identified as a cis-acting dendritic targeting signal73.A 30 nucleotide cis-acting element in the 3′-UTR of αCaMKII also mediatesdendritic targeting128. In oligodendrocytes, a 21 nucleo-tide sequence in the 3′-UTR (the hnRNP A2 responseelement (A2RE)) functions as a cis-acting signal fortrafficking. When the A2RE is microinjected intohippocampal neurons, it is transported in granules todistal neurites, and the transport depends on the trans-acting factor hnRNP A2, which indicates that the A2REand hnRNP A2 might be involved in RNA targeting inneurons129. The 3′-UTR localization sequence of β-actinmRNA, which is transported by zipcode-binding proteinto the leading edge in fibroblasts, has also been shown tobe functional in hippocampal neurons130.

Conclusions and future perspectivesNeurons use many complex mechanisms to maintainthe polarity of axons and dendrites, and motor proteinsare vital for these mechanisms. Proteins are sorted andtransported in various organelles and protein complexes,and they are specifically recognized by different motorsand selectively transported to specific destinations.mRNAs are also transported in large protein–RNA com-plexes. Motors seem to recognize these cargoes throughadaptor complexes or scaffolding proteins. Some cargoproteins associate directly with adaptor complexes or scaffolding proteins, whereas other cargoes can betransported by associating with these cargo proteins.Therefore, targeting sequences can bind to adaptor orscaffolding complexes, motors, or other proteins that arenecessary for proteins to associate in organelles.

1. Grafstein, B. & Forman, D. S. Intracellular transport inneurons. Physiol. Rev. 60, 1167–1283 (1980).

2. Job, C. & Eberwine, J. Localization and translation of mRNAin dendrites and axons. Nature Rev. Neurosci. 2, 889–898(2001).

3. Hirokawa, N. Cross-linker system between neurofilaments,microtubules, and membranous organelles in frog axonsrevealed by the quick-freeze, deep-etching method. J. CellBiol. 94, 129–142 (1982).

4. Hirokawa, N. in Neuronal Cytoskeleton (ed. Burgoyne, R. D.)5–74 (Wiley-Liss Inc., New York, 1991).

5. Hirokawa, N. Kinesin and dynein superfamily proteins andthe mechanism of organelle transport. Science 279,519–526 (1998).

6. Desai, A. & Mitchison, T. J. Microtubule polymerizationdynamics. Annu. Rev. Cell Dev. Biol. 13, 83–117 (1997).

7. Burton, P. R. & Paige, J. L. Polarity of axoplasmicmicrotubules in the olfactory nerve of the frog. Proc. NatlAcad. Sci. USA 78, 3269–3273 (1981).

8. Baas, P. W., Deitch, J. S., Black, M. M. & Banker, G. A.Polarity orientation of microtubules in hippocampalneurons: uniformity in the axon and nonuniformity in thedendrite. Proc. Natl Acad. Sci. USA 85, 8335–8339(1988).

9. Vallee, R. B., Wall, J. S., Paschal, B. M. & Shpetner, H. S.Microtubule-associated protein 1C from brain is a two-headed cytosolic dynein. Nature 332, 561–563 (1988).

10. Hirokawa, N., Sato-Yoshitake, R., Yoshida, T. & Kawashima, T.Brain dynein (MAP1C) localizes on both anterogradely and

retrogradely transported membranous organelles in vivo. J. Cell Biol. 111, 1027–1037 (1990).

11. Harada, A. et al. Golgi vesiculation and lysosome dispersionin cells lacking cytoplasmic dynein. J. Cell Biol. 141, 51–59(1998).

12. Vallee, R. B., Williams, J. C., Varma, D. & Barnhart, L. E.Dynein: an ancient motor protein involved in multiple modesof transport. J. Neurobiol. 58, 189–200 (2004).

13. Aizawa, H. et al. Kinesin family in murine central nervoussystem. J. Cell Biol. 119, 1287–1296 (1992).This study identified KIFs 1–5 and reported the fullsequences of KIF2A and KIF3A in the murine centralnervous system.

14. Miki, H., Setou, M., Kaneshiro, K. & Hirokawa, N. All kinesinsuperfamily protein, KIF, genes in mouse and human. Proc.Natl Acad. Sci. USA 98, 7004–7011 (2001).The authors classified 45 genes in mice and humansas N-kinesins, M-kinesins and C-kinesins.

15. Hirokawa, N. & Takemura, R. in Molecular Motors(ed. Schliwa, M.) 79–109 (Wiley-VCH, Weinheim, 2003).

16. Hirokawa, N. & Takemura, R. Kinesin superfamily proteinsand their various functions and dynamics. Exp. Cell Res.301, 50–59 (2004).

17. Vale, R. D., Reese, T. S. & Sheetz, M. P. Identification of anovel force-generating protein, kinesin, involved inmicrotubule-based motility. Cell 42, 39–50 (1985).

18. Brady, S. T. A novel brain ATPase with properties expectedfor the fast axonal transport motor. Nature 317, 73–75(1985).

19. Lawrence, C. J. et al. A standardized kinesin nomenclature.J. Cell Biol. 167, 19–22 (2004).

20. Hirokawa, N. Stirring up development with the heterotrimerickinesin KIF3. Traffic 1, 29–34 (2000).

21. Tanaka, Y. & Hirokawa, N. Mouse models of Charcot-Marie-Tooth disease. Trends Genet. 18, S39–S44 (2002).

22. Hirokawa, N. & Takemura, R. Biochemical and molecularcharacterization of diseases linked to motor proteins. TrendsBiochem. Sci. 28, 558–565 (2003).

23. Hirokawa, N. & Takemura, R. Molecular motors in neuronaldevelopment, intracellular transport and diseases. Curr.Opin. Neurobiol. 14, 564–573 (2004).

24. Hirokawa, N. et al. Submolecular domains of bovine brainkinesin identified by electron microscopy and monoclonalantibody decoration. Cell 56, 867–878 (1989).A description of the molecular structure ofconventional kinesin.

25. Diefenbach, R. J., Mackay, J. P., Armati, P. J. &Cunningham, A. L. The C-terminal region of the stalkdomain of ubiquitous human kinesin heavy chain containsthe binding site for kinesin light chain. Biochemistry 37,16663–16670 (1998).

26. Okada, Y., Yamazaki, H., Sekine-Aizawa, Y. & Hirokawa, N.The neuron-specific kinesin superfamily protein KIF1A is aunique monomeric motor for anterograde axonal transportof synaptic vesicle precursors. Cell 81, 769–780 (1995).

27. Nangaku, M. et al. KIF1B, a novel microtubule plus end-directed monomeric motor protein for transport ofmitochondria. Cell 79, 1209–1220 (1994).



R E V I E W S

28. Yamazaki, H., Nakata, T., Okada, Y. & Hirokawa, N.KIF3A/B: a heterodimeric kinesin superfamily protein thatworks as a microtubule plus end-directed motor formembrane organelle transport. J. Cell Biol. 130, 1387–1399(1995).

29. Yang, Z. & Goldstein, L. S. Characterization of the KIF3Cneural kinesin-like motor from mouse. Mol. Biol. Cell 9,249–261 (1998).

30. Yamazaki, H., Nakata, T., Okada, Y. & Hirokawa, N. Cloningand characterization of KAP3: a novel kinesin superfamily-associated protein of KIF3A/3B. Proc. Natl Acad. Sci. USA93, 8443–8448 (1996).

31. Yonekawa, Y. et al. Defect in synaptic vesicle precursortransport and neuronal cell death in KIF1A motor protein-deficient mice. J. Cell Biol. 141, 431–441 (1998).

32. Zhao, C. et al. Charcot-Marie-Tooth disease type 2A causedby mutation in a microtubule motor KIF1Bβ. Cell 105,587–597 (2001).This study showed for the first time that a mutation ina KIF is a direct cause of human neuropathy.

33. Sampo, B., Kaech, S., Kunz, S. & Banker, G. Two distinctmechanisms target membrane proteins to the axonalsurface. Neuron 37, 611–624 (2003).This paper revealed that there are two distinctmechanisms of axonal targeting of membraneproteins.

34. Nakata, T., Terada, S. & Hirokawa, N. Visualization of thedynamics of synaptic vesicle and plasma membraneproteins in living axons. J. Cell Biol. 140, 659–674 (1998).

35. Kaether, C., Skehel, P. & Dotti, C. G. Axonal membraneproteins are transported in distinct carriers: a two-colorvideo microscopy study in cultured hippocampal neurons.Mol. Biol. Cell 11, 1213–1224 (2000).

36. Ahmari, S. E., Buchanan, J. & Smith, S. J. Assembly ofpresynaptic active zones from cytoplasmic transportpackets. Nature Neurosci. 3, 445–451 (2000).

37. Garrido, J. J. et al. Identification of an axonal determinant inthe C-terminus of the sodium channel Nav1. 2. EMBO J. 20,5950–5961 (2001).

38. Garrido, J. J. et al. A targeting motif involved in sodiumchannel clustering at the axonal initial segment. Science300, 2091–2094 (2003).

39. Lemaillet, G., Walker, B. & Lambert, S. Identification of aconserved ankyrin-binding motif in the family of sodiumchannel α subunits. J. Biol. Chem. 278, 27333–27339(2003).

40. Gu, C., Jan, Y. N. & Jan, L. Y. A conserved domain in axonaltargeting of Kv1 (Shaker) voltage-gated potassium channels.Science 301, 646–649 (2003).

41. Vogt, L. et al. Continuous renewal of the axonal pathwaysensor apparatus by insertion of new sensor molecules intothe growth cone membrane. Curr. Biol. 6, 1153–1158(1996).

42. Jareb, M. & Banker, G. The polarized sorting of membraneproteins expressed in cultured hippocampal neurons usingviral vectors. Neuron 20, 855–867 (1998).

43. Burack, M. A., Silverman, M. A. & Banker, G. The role ofselective transport in neuronal protein sorting. Neuron 26,465–472 (2000).

44. Meiri, K. F., Pfenninger, K. H. & Willard, M. B. Growth-associated protein, GAP-43, a polypeptide that is inducedwhen neurons extend axons, is a component of growthcones and corresponds to pp46, a major polypeptide of asubcellular fraction enriched in growth cones. Proc. NatlAcad. Sci. USA 83, 3537–3541 (1986).

45. Tanaka, Y. et al. Targeted disruption of mouse conventionalkinesin heavy chain, kif5B, results in abnormal perinuclearclustering of mitochondria. Cell 93, 1147–1158 (1998).

46. Kamal, A., Stokin, G. B., Yang, Z., Xia, C. H. & Goldstein, L. S.Axonal transport of amyloid precursor protein is mediated by direct binding to the kinesin light chain subunit of kinesin-I. Neuron 28, 449–459 (2000).

47. Nakata, T. & Hirokawa, N. Microtubules provide directionalcues for polarized axonal transport through interaction withkinesin motor head. J. Cell Biol. 162, 1045–1055 (2003).Research showing that the motor domain of KIF5intrinsically recognizes microtubules in the initialsegment.

48. Kamal, A., Almenar-Queralt, A., LeBlanc, J. F., Roberts, E. A.& Goldstein, L. S. Kinesin-mediated axonal transport of amembrane compartment containing β-secretase andpresenilin-1 requires APP. Nature 414, 643–648 (2001).

49. Hardy, J. & Selkoe, D. J. The amyloid hypothesis ofAlzheimer’s disease: progress and problems on the road totherapeutics. Science 297, 353–356 (2002).

50. De Strooper, B. Aph-1, Pen-2, and Nicastrin with Presenilingenerate an active γ-secretase complex. Neuron 38, 9–12(2003).

51. Verhey, K. J. et al. Cargo of kinesin identified as JIPscaffolding proteins and associated signaling molecules. J. Cell Biol. 152, 959–970 (2001).

This study shows that the TPR motif of KLC binds toJIP1, 2 and 3, and that KIF5 transports APOER2-containing vesicles.

52. D’Arcangelo, G. & Curran, T. Reeler: new tales on an oldmutant mouse. Bioessays 20, 235–244 (1998).

53. D’Arcangelo, G. et al. Reelin is a ligand for lipoproteinreceptors. Neuron 24, 471–479 (1999).

54. Willnow, T. E., Nykjaer, A. & Herz, J. Lipoprotein receptors:new roles for ancient proteins. Nature Cell Biol. 1,E157–E162 (1999).

55. Kanai, Y. et al. KIF5C, a novel neuronal kinesin enriched inmotor neurons. J. Neurosci. 20, 6374–6384 (2000).

56. Terada, S., Kinjo, M. & Hirokawa, N. Oligomeric tubulin inlarge transporting complex is transported via kinesin in squidgiant axons. Cell 103, 141–155 (2000).

57. Takeda, S. et al. Kinesin superfamily protein 3 (KIF3) motortransports fodrin-associating vesicles important for neuritebuilding. J. Cell Biol. 148, 1255–1265 (2000).

58. Yang, Z., Roberts, E. A. & Goldstein, L. S. Functionalanalysis of mouse kinesin motor Kif3C. Mol. Cell. Biol. 21,5306–5311 (2001).

59. Nishimura, T. et al. Role of the PAR-3-KIF3 complex in theestablishment of neuronal polarity. Nature Cell Biol. 6,328–334 (2004).

60. Noda, Y., Sato-Yoshitake, R., Kondo, S., Nangaku, M. &Hirokawa, N. KIF2 is a new microtubule-based anterogrademotor that transports membranous organelles distinct fromthose carried by kinesin heavy chain or KIF3A/B. J. Cell Biol.129, 157–167 (1995).

61. Morfini, G., Quiroga, S., Rosa, A., Kosik, K. & Caceres, A.Suppression of KIF2 in PC12 cells alters the distribution of agrowth cone nonsynaptic membrane receptor and inhibitsneurite extension. J. Cell Biol. 138, 657–669 (1997).

62. Ogawa, T., Nitta, R., Okada, Y. & Hirokawa, N. A commonmechanism for microtubule destabilizers — M type kinesinsstabilize curling of the protofilament using the class-specificneck and loops. Cell 116, 591–602 (2004).

63. Homma, N. et al. Kinesin superfamily protein 2A (KIF2A)functions in suppression of collateral branch extension. Cell 114, 229–239 (2003).

64. El-Husseini Ael, D., Craven, S. E., Brock, S. C. & Bredt, D. S.Polarized targeting of peripheral membrane proteins inneurons. J. Biol. Chem. 276, 44984–44992 (2001).

65. Setou, M., Nakagawa, T., Seog, D. H. & Hirokawa, N.Kinesin superfamily motor protein KIF17 and mLin-10 inNMDA receptor-containing vesicle transport. Science 288,1796–1802 (2000).One of the early reports showing that KIFs interactwith cargoes through scaffolding protein complexes.

66. Wong, R. W., Setou, M., Teng, J., Takei, Y. & Hirokawa, N.Overexpression of motor protein KIF17 enhances spatialand working memory in transgenic mice. Proc. Natl Acad.Sci. USA 99, 14500–14505 (2002).

67. Guillaud, L., Setou, M. & Hirokawa, N. KIF17 dynamics andregulation of NR2B trafficking in hippocampal neurons. J. Neurosci. 23, 131–140 (2003).

68. Setou, M. et al. Glutamate-receptor-interacting proteinGRIP1 directly steers kinesin to dendrites. Nature 417,83–87 (2002).This study shows that the interaction of cargo withthe C-terminal tail of KIF5 steers the motor todendrites.

69. Stowell, J. N. & Craig, A. M. Axon/dendrite targeting ofmetabotropic glutamate receptors by their cytoplasmiccarboxy-terminal domains. Neuron 22, 525–536 (1999).

70. Ango, F. et al. Dendritic and axonal targeting of type 5metabotropic glutamate receptor is regulated by Homer1proteins and neuronal excitation. J. Neurosci. 20,8710–8716 (2000).

71. Francesconi, A. & Duvoisin, R. M. Alternative splicingunmasks dendritic and axonal targeting signals inmetabotropic glutamate receptor 1. J. Neurosci. 22,2196–2205 (2002).

72. Rivera, J. F., Ahmad, S., Quick, M. W., Liman, E. R. &Arnold, D. B. An evolutionarily conserved dileucine motif inShal K+ channels mediates dendritic targeting. NatureNeurosci. 6, 243–250 (2003).

73. Blichenberg, A. et al. Identification of a cis-acting dendritictargeting element in MAP2 mRNAs. J. Neurosci. 19,8818–8829 (1999).

74. Blichenberg, A. et al. Identification of a cis-acting dendritictargeting element in the mRNA encoding the α subunit ofCa2+/calmodulin-dependent protein kinase II. Eur. J. Neurosci. 13, 1881–1888 (2001).

75. Guzowski, J. F. et al. Inhibition of activity-dependent Arcprotein expression in the rat hippocampus impairs themaintenance of long-term potentiation and the consolidationof long-term memory. J. Neurosci. 20, 3993–4001 (2000).

76. Krichevsky, A. M. & Kosik, K. S. Neuronal RNA granules: alink between RNA localization and stimulation-dependenttranslation. Neuron 32, 683–696 (2001).

77. Kanai, Y., Dohmae, N. & Hirokawa, N. Kinesin transportsRNA; isolation and characterization of an RNA-transportinggranule. Neuron 43, 513–525 (2004).The authors show that KIF5 transports RNA-containing granules in dendrites and that thetransport depends on the interaction of cargo and the C-terminal tail of KIF5, not on KLC.

78. Marszalek, J. R., Weiner, J. A., Farlow, S. J., Chun, J. &Goldstein, L. S. Novel dendritic kinesin sorting identified bydifferent process targeting of two related kinesins: KIF21Aand KIF21B. J. Cell Biol. 145, 469–479 (1999).

79. Saito, N. et al. KIFC2 is a novel neuron-specific C-terminaltype kinesin superfamily motor for dendritic transport ofmultivesicular body-like organelles. Neuron 18, 425–438(1997).

80. Nakagawa, T. et al. A novel motor, KIF13A, transportsmannose-6-phosphate receptor to plasma membranethrough direct interaction with AP-1 complex. Cell 103,569–581 (2000).A study showing that KIF13A recognizes its cargo, the mannose-6-phosphate receptor, through itsinteraction with an adaptor protein complex.

81. Kirchhausen, T. Adaptors for clathrin-mediated traffic. Annu.Rev. Cell Dev. Biol. 15, 705–732 (1999).

82. Bonifacino, J. S. & Traub, L. M. Signals for sorting oftransmembrane proteins to endosomes and lysosomes.Annu. Rev. Biochem. 72, 395–447 (2003).

83. Rongo, C., Whitfield, C. W., Rodal, A., Kim, S. K. & Kaplan,J. M. LIN-10 is a shared component of the polarized proteinlocalization pathways in neurons and epithelia. Cell 94,751–759 (1998).

84. Butz, S., Okamoto, M. & Sudhof, T. C. A tripartite proteincomplex with the potential to couple synaptic vesicleexocytosis to cell adhesion in brain. Cell 94, 773–782(1998).

85. Jo, K., Derin, R., Li, M. & Bredt, D. S. Characterization ofMALS/Velis-1,-2, and -3: a family of mammalian LIN-7homologs enriched at brain synapses in association with thepostsynaptic density-95/NMDA receptor postsynapticcomplex. J. Neurosci. 19, 4189–4199 (1999).

86. Garner, C. C., Nash, J. & Huganir, R. L. PDZ domains insynapse assembly and signalling. Trends Cell Biol. 10,274–280 (2000).

87. Hung, A. Y. & Sheng, M. PDZ domains: structural modulesfor protein complex assembly. J. Biol. Chem. 277,5699–5702 (2002).

88. Steinberg, G. & Schliwa, M. The Neurospora organellemotor: a distant relative of conventional kinesin withunconventional properties. Mol. Biol. Cell 6, 1605–1618(1995).

89. Kirchner, J., Woehlke, G. & Schliwa, M. Universal andunique features of kinesin motors: insights from acomparison of fungal and animal conventional kinesins. Biol. Chem. 380, 915–921 (1999).

90. Seiler, S. et al. Cargo binding and regulatory sites in the tailof fungal conventional kinesin. Nature Cell Biol. 2, 333–338(2000).

91. Verhey, K. J. et al. Light chain-dependent regulation ofkinesin’s interaction with microtubules. J. Cell Biol. 143,1053–1066 (1998).

92. Blatch, G. L. & Lassle, M. The tetratricopeptide repeat: astructural motif mediating protein–protein interactions.Bioessays 21, 932–939 (1999).

93. D’Andrea, L. D. & Regan, L. TPR proteins: the versatile helix.Trends Biochem. Sci. 28, 655–662 (2003).

94. Gindhart, J. G. Jr, Desai, C. J., Beushausen, S., Zinn, K. &Goldstein, L. S. Kinesin light chains are essential for axonaltransport in Drosophila. J. Cell Biol. 141, 443–454 (1998).

95. Bowman, A. B. et al. Kinesin-dependent axonal transport ismediated by the sunday driver (SYD) protein. Cell 103,583–594 (2000).

96. Inomata, H. et al. A scaffold protein JIP-1b enhancesamyloid precursor protein phosphorylation by JNK and itsassociation with kinesin light chain 1. J. Biol. Chem. 278,22946–22955 (2003).

97. Matsuda, S., Matsuda, Y. & D’Adamio, L. Amyloid β proteinprecursor (AβPP), but not AβPP-like protein 2, is bridged tothe kinesin light chain by the scaffold protein JNK-interactingprotein 1. J. Biol. Chem. 278, 38601–38606 (2003).

98. Yasuda, J., Whitmarsh, A. J., Cavanagh, J., Sharma, M. &Davis, R. J. The JIP group of mitogen-activated protein kinasescaffold proteins. Mol. Cell. Biol. 19, 7245–7254 (1999).

99. Davis, R. J. Signal transduction by the JNK group of MAPkinases. Cell 103, 239–252 (2000).

100. Meyer, D., Liu, A. & Margolis, B. Interaction of c-Jun amino-terminal kinase interacting protein-1 with p190 rhoGEF andits localization in differentiated neurons. J. Biol. Chem. 274,35113–35118 (1999).

101. Pellet, J. B. et al. Spatial, temporal and subcellularlocalization of islet-brain 1 (IB1), a homologue of JIP-1, inmouse brain. Eur. J. Neurosci. 12, 621–632 (2000).



R E V I E W S

102. Kelkar, N., Gupta, S., Dickens, M. & Davis, R. J. Interactionof a mitogen-activated protein kinase signaling module withthe neuronal protein JIP3. Mol. Cell. Biol. 20, 1030–1043(2000).

103. Stockinger, W. et al. The reelin receptor ApoER2 recruitsJNK-interacting proteins-1 and -2. J. Biol. Chem. 275,25625–25632 (2000).

104. Ito, M. et al. JSAP1, a novel jun N-terminal protein kinase(JNK)-binding protein that functions as a scaffold factor inthe JNK signaling pathway. Mol. Cell. Biol. 19, 7539–7548(1999).

105. Dong, H. et al. GRIP: a synaptic PDZ domain-containingprotein that interacts with AMPA receptors. Nature 386,279–284 (1997).

106. Palacios, I. M. & St Johnston, D. Kinesin light chain-independent function of the kinesin heavy chain incytoplasmic streaming and posterior localisation in theDrosophila oocyte. Development 129, 5473–5485 (2002).

107. Ling, S. C., Fahrner, P. S., Greenough, W. T. & Gelfand, V. I.Transport of Drosophila fragile X mental retardation protein-containing ribonucleoprotein granules by kinesin-1 andcytoplasmic dynein. Proc. Natl Acad. Sci. USA 101,17428–17433 (2004).

108. Klopfenstein, D. R., Tomishige, M., Stuurman, N. & Vale, R. D.Role of phosphatidylinositol(4,5)bisphosphate organizationin membrane transport by the Unc104 kinesin motor. Cell109, 347–358 (2002).

109. Okada, Y. & Hirokawa, N. A processive single-headedmotor: kinesin superfamily protein KIF1A. Science 283,1152–1157 (1999).

110. Okada, Y., Higuchi, H. & Hirokawa, N. Processivity of thesingle-headed kinesin KIF1A through biased binding totubulin. Nature 424, 574–577 (2003).

111. Nitta, R., Kikkawa, M., Okada, Y. & Hirokawa, N. KIF1Aalternately uses two loops to bind microtubules. Science305, 678–683 (2004).

112. Tomishige, M., Klopfenstein, D. R. & Vale, R. D. Conversionof Unc104/KIF1A kinesin into a processive motor afterdimerization. Science 297, 2263–2267 (2002).

113. Asaba, N., Hanada, T., Takeuchi, A. & Chishti, A. H. Directinteraction with a kinesin-related motor mediates transportof mammalian discs large tumor suppressor homologue inepithelial cells. J. Biol. Chem. 278, 8395–8400 (2003).

114. Holleran, E. A., Tokito, M. K., Karki, S. & Holzbaur, E. L.Centractin (ARP1) associates with spectrin revealing apotential mechanism to link dynactin to intracellularorganelles. J. Cell Biol. 135, 1815–1829 (1996).

115. Holleran, E. A. et al. β III spectrin binds to the Arp1 subunit ofdynactin. J. Biol. Chem. 276, 36598–36605 (2001).

116. Tai, A. W., Chuang, J. Z., Bode, C., Wolfrum, U. & Sung, C. H.Rhodopsin’s carboxy-terminal cytoplasmic tail acts as amembrane receptor for cytoplasmic dynein by binding to thedynein light chain Tctex-1. Cell 97, 877–887 (1999).

117. Tirnauer, J. S. & Bierer, B. E. EB1 proteins regulatemicrotubule dynamics, cell polarity, and chromosomestability. J. Cell Biol. 149, 761–766 (2000).

118. Grote, E. & Kelly, R. B. Endocytosis of VAMP is facilitated bya synaptic vesicle targeting signal. J. Cell. Biol. 132,537–547 (1996).

119. Kamiguchi, H. et al. The neural cell adhesion molecule L1interacts with the AP-2 adaptor and is endocytosed via theclathrin-mediated pathway. J. Neurosci. 18, 5311–5321(1998).

120. Kanaani, J. et al. A combination of three distinct traffickingsignals mediates axonal targeting and presynaptic clusteringof GAD65. J. Cell Biol. 158, 1229–1238 (2002).

121. Rodriguez-Boulan, E., Musch, A. & Le Bivic, A. Epithelialtrafficking: new routes to familiar places. Curr. Opin. CellBiol. 16, 436–442 (2004).

122. Mostov, K., Su, T. & ter Beest, M. Polarized epithelialmembrane traffic: conservation and plasticity. Nature CellBiol. 5, 287–293 (2003).

123. West, A. E., Neve, R. L. & Buckley, K. M. Identification of asomatodendritic targeting signal in the cytoplasmic domainof the transferrin receptor. J. Neurosci. 17, 6038–6047(1997).

124. Lim, S. T., Antonucci, D. E., Scannevin, R. H. & Trimmer, J. S.A novel targeting signal for proximal clustering of the Kv2.1K+ channel in hippocampal neurons. Neuron 25, 385–397(2000).

125. Carson, J. H., Kwon, S. & Barbarese, E. RNA trafficking inmyelinating cells. Curr. Opin. Neurobiol. 8, 607–612 (1998).

126. Oleynikov, Y. & Singer, R. H. Real-time visualization of ZBP1association with β-actin mRNA during transcription andlocalization. Curr. Biol. 13, 199–207 (2003).

127. Kloc, M., Zearfoss, N. R. & Etkin, L. D. Mechanisms ofsubcellular mRNA localization. Cell 108, 533–544 (2002).

128. Mori, Y., Imaizumi, K., Katayama, T., Yoneda, T. & Tohyama, M.Two cis-acting elements in the 3′ untranslated region of α-CaMKII regulate its dendritic targeting. Nature Neurosci.3, 1079–1084 (2000).

129. Shan, J., Munro, T. P., Barbarese, E., Carson, J. H. & Smith, R.A molecular mechanism for mRNA trafficking in neuronaldendrites. J. Neurosci. 23, 8859–8866 (2003).

130. Tiruchinapalli, D. M. et al. Activity-dependent trafficking anddynamic localization of zipcode binding protein 1 and β-actin mRNA in dendrites and spines of hippocampalneurons. J. Neurosci. 23, 3251–3261 (2003).

131. Morfini, G., Szebenyi, G., Elluru, R., Ratner, N. & Brady, S. T.Glycogen synthase kinase 3 phosphorylates kinesin lightchains and negatively regulates kinesin-based motility.EMBO J. 21, 281–293 (2002).

132. Morfini, G. et al. A novel CDK5-dependent pathway forregulating GSK3 activity and kinesin-driven motility inneurons. EMBO J. 23, 2235–2245 (2004).

133. Sato-Yoshitake, R., Yorifuji, H., Inagaki, M. & Hirokawa, N.The phosphorylation of kinesin regulates its binding tosynaptic vesicles. J. Biol. Chem. 267, 23930–23936(1992).

134. Peters, A., Palay, S. A. & Webster, H. The Fine Structure ofthe Nervous System: Neurons and their Supporting Cells(Oxford Univ. Press, New York, 1991).

135. Kanai, Y. et al. Expression of multiple tau isoforms andmicrotubule bundle formation in fibroblasts transfected witha single tau cDNA. J. Cell Biol. 109, 1173–1184 (1989).

136. Chen, J., Kanai, Y., Cowan, N. J. & Hirokawa, N. Projectiondomains of MAP2 and tau determine spacings betweenmicrotubules in dendrites and axons. Nature 360, 674–677(1992).This study shows that the difference in the spacingbetween microtubules in axons and dendrites isdictated by microtubule-associated proteins.

137. Okabe, S. & Hirokawa, N. Rapid turnover of microtubule-associated protein MAP2 in the axon revealed bymicroinjection of biotinylated MAP2 into cultured neurons.Proc. Natl Acad. Sci. USA 86, 4127–4131 (1989).

138. Kanai, Y. & Hirokawa, N. Sorting mechanisms of tau andMAP2 in neurons: suppressed axonal transit of MAP2 andlocally regulated microtubule binding. Neuron 14, 421–432(1995).

139. Hirokawa, N., Funakoshi, T., Sato-Harada, R. & Kanai, Y.Selective stabilization of tau in axons and microtubule-associated protein 2C in cell bodies and dendritescontributes to polarized localization of cytoskeletal proteinsin mature neurons. J. Cell Biol. 132, 667–679 (1996).

140. Nakata, T. & Hirokawa, N. Point mutation of adenosinetriphosphate-binding motif generated rigor kinesin thatselectively blocks anterograde lysosome membranetransport. J. Cell Biol. 131, 1039–1053 (1995).

141. Nonaka, S. et al. Randomization of left-right asymmetry dueto loss of nodal cilia generating leftward flow ofextraembryonic fluid in mice lacking KIF3B motor protein.Cell 95, 829–837 (1998).

142. Takeda, S. et al. Left-right asymmetry and kinesinsuperfamily protein KIF3A: new insights in determination oflaterality and mesoderm induction by kif3A– /– mice analysis.J. Cell Biol. 145, 825–836 (1999).

143. Marszalek, J. R. et al. Genetic evidence for selectivetransport of opsin and arrestin by kinesin-II in mammalianphotoreceptors. Cell 102, 175–187 (2000).

144. Sekine, Y. et al. A novel microtubule-based motor protein(KIF4) for organelle transports, whose expression is regulateddevelopmentally. J. Cell Biol. 127, 187–201 (1994).

145. Noda, Y. et al. KIFC3, a microtubule minus end-directedmotor for the apical transport of annexin XIIIb-associatedTriton-insoluble membranes. J. Cell Biol. 155, 77–88 (2001).

146. Xu, Y. et al. Role of KIFC3 motor protein in Golgi positioningand integration. J. Cell Biol. 158, 293–303 (2002).

147. Moore, J. D. & Endow, S. A. Kinesin proteins: a phylum ofmotors for microtubule-based motility. Bioessays 18,207–219 (1996).

148. Shin, H. et al. Association of the kinesin motor KIF1A withthe multimodular protein liprin-α. J. Biol. Chem. 278,11393–11401 (2003).

AcknowledgementsWe thank all members of the Hirokawa laboratory and OkinakaMemorial Institute for Medical Research. This work was supportedby a Center of Excellence grant to N.H. from the Ministry ofEducation, Culture, Sports, Science and Technology.

Competing interests statementThe authors declare no competing financial interests.

Online links

DATABASESThe following terms in this article are linked online to:Entrez Gene:http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=geneαCaMKII | APOER2 | APP | ARC | CASK | FMR1 | FXR1 | FXR2 |GAD65 | GAP43 | GRIP1 | hnRNPU | JIP1 | JIP2 | JIP3 | KAP3 |KIF3A | KIF3B | KIF3C | KIF5A | KIF5B | KIF5C | KIF17 | LIN10 |MAP2 | M6PR | PSD-95 | PURα | RAB3A | SNAP25 | VAMP2 |VELISdictyBase: http://dictybase.org/NomenclatureGuidelines.htmUnc104Access to this interactive links box is free online.

molecular motors and mechanisms of directional transport in neurons

Documents