You are on page 1of 16

Received: 24 February 2019 Revised: 26 August 2019 Accepted: 26 August 2019 Uncorrected manuscript published: 28 August 2019 Published

on: 2 October 2019


DOI: 10.1111/tra.12692

ORIGINAL ARTICLE

A novel strategy to visualize vesicle-bound kinesins reveals


the diversity of kinesin-mediated transport

Rui Yang1,2 | Zoe Bostick3 | Alex Garbouchian3 | Julie Luisi1 | Gary Banker1 |
Marvin Bentley3

1
The Jungers Center for Neurosciences
Research, Oregon Health & Science University, Abstract
Portland, Oregon In mammals, 15 to 20 kinesins are thought to mediate vesicle transport. Little is
2
Department of Biochemistry, Duke
known about the identity of vesicles moved by each kinesin or the functional signifi-
University, Durham, North Carolina
3
Department of Biological Sciences and the cance of such diversity. To characterize the transport mediated by different kinesins,
Center for Biotechnology and Interdisciplinary we developed a novel strategy to visualize vesicle-bound kinesins in living cells. We
Sciences, Rensselaer Polytechnic Institute,
Troy, New York applied this method to cultured neurons and systematically determined the localiza-
tion and transport parameters of vesicles labeled by different members of the
Correspondence
Marvin Bentley, Department of Biological Kinesin-1, -2, and -3 families. We observed vesicle labeling with nearly all kinesins.
Sciences and the Center for Biotechnology Only six kinesins bound vesicles that undergo long-range transport in neurons. Of
and Interdisciplinary Sciences, Rensselaer
Polytechnic Institute, 110 8th Street, Troy, NY these, three had an axonal bias (KIF5B, KIF5C and KIF13B), two were unbiased
12180. (KIF1A and KIF1Bβ), and one transported only in dendrites (KIF13A). Overall, the
Email: bentlm3@rpi.edu
trafficking of vesicle-bound kinesins to axons or dendrites did not correspond to their
Funding information motor domain preference, suggesting that on-vesicle regulation is crucial for kinesin
National Institute of Mental Health, Grant/
Award Number: MH066179 targeting. Surprisingly, several kinesins were associated with populations of
somatodendritic vesicles that underwent little long-range transport. This assay should
Peer Review
The peer review history for this article is be broadly applicable for investigating kinesin function in many cell types.
available at https://publons.com/publon/
10.1111/tra.12692/ KEYWORDS
Kinesin, membrane trafficking, motor protein, neuron, polarity, transport, vesicle

1 | I N T RO D UC T I O N large number of kinesin genes raises a fundamental question: Why are


so many motors required to fulfill what appears to be the same basic
Neurons are highly polarized cells that require proper localization of function of moving vesicles towards the plus-end of microtubules?
membrane proteins for nearly every aspect of their function. These Some of the kinesin diversity may provide redundancy,7-9 but that
membrane proteins are synthesized and processed in the rough endo- alone is unlikely to explain the size of the kinesin family. Instead, the
plasmic reticulum and the Golgi apparatus, which are located in the diversity could allow different kinesins to confer different transport
cell body and proximal dendrites. Proteins are then incorporated into parameters, targeting vesicles to specific destinations. Individual
vesicles that are transported to their appropriate destination by kinesins may also be subject to vesicle-dependent differential regula-
microtubule motor proteins.1,2 Microtubule plus-end-directed trans- tion, allowing them to confer a variety of transport parameters depen-
3
port is mediated by kinesins and minus-end directed transport largely dent upon the vesicle to which they are bound.
by dynein. Whereas mammalian cells express only a single dynein Identifying all of the vesicles transported by a given kinesin is an
4,5
motor domain, the human genome contains 45 kinesins, some 15 to important first step towards understanding the functional diversity
20 of which are thought to mediate neuronal vesicle transport.6 The within the kinesin family and defining the mechanisms that underlie

© 2019 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd

Traffic. 2019;20:851–866. wileyonlinelibrary.com/journal/tra 851


852 YANG ET AL.

the accuracy of vesicle transport. Tracking vesicles labeled with fluo- multiple vesicle populations in axons and dendrites.3,22-24 Kymo-
10-14
rescently tagged cargo proteins is common practice, but labeling graphs from the axon and dendrites of a cell expressing KIF5B-GFP
vesicles by expressing fluorescent kinesins has proven challenging. were devoid of transport events (Figure 1A). Kymograph lines with a
Some Kinesin-3 and -4 family members7,12,14-16 have been used to positive slope indicate anterograde transport and those with a nega-
track neuronal vesicles, but we know of no examples of successful tive slope indicate retrograde transport. This convention is followed in
vesicle labeling with fluorescently tagged Kinesin-1 and -2 family all figures. Expressing members of the Kinesin-3 family sometimes
members. Here we describe technical improvements to enhance vesi- results in vesicle labeling,7,15 but labeling is often inconsistent. Expres-
cle labeling by fluorescent kinesins. Instead of expressing full-length sion of full-length GFP-tagged Kinesin-3 family member KIF13B in
kinesins, we express only the vesicle-binding tail region of the motor cultured neurons resulted in the labeling of a few punctate structures
protein. To further minimize background fluorescence, we record from that underwent transport (Figure 1C). Presumably full-length GFP-
transfected cells at early expression times and focus on low- tagged kinesins bind to the vesicles normally transported by these
expressing cells. For Kinesin-1 second, we also employ an expression motors, but overexpression results in a pool of unbound cytosolic
control system to minimize the level of unbound cytosolic kinesins17 kinesins that obscures vesicle labeling.
and use fluorescent probes with high quantum efficiency and low To reduce the pool of unbound fluorescently-tagged kinesins we
extinction coefficients.18 Because the expressed kinesin constructs employed a series of strategies. First, instead of expressing tagged
lack a motor domain, vesicle movement is mediated only by endoge- full-length kinesins we only expressed the vesicle-binding tail domain.
nous motors. This report focuses on the role of kinesins in neuronal The majority of full-length kinesin molecules are cytoplasmic and pre-
transport but the labeling strategy we describe should be widely appli- sumably in an autoinhibited state rather than bound to vesicles.25 We
cable as the questions raised in this study are equally relevant in all reasoned that expressing constructs lacking the motor domain may
eukaryotic cells. make the tail domain more accessible to bind vesicles. We previously
We utilized this novel labeling strategy to complete the first sys- found that expression of the kinesin C-terminal domain alone is suffi-
tematic analysis of the localization and transport parameters of vesi- cient for vesicle binding7,26 and engineered kinesins that lack the
cles labeled with members of the Kinesin-1, −2, and − 3 families in motor and first predicted coiled-coil that mediates dimerization.20
cultured hippocampal neurons. Vesicles bound by different kinesins Expressing this kinesin “tail” construct alone markedly improved label-
displayed a variety of distributions and transport parameters, differing ing of all kinesins (Figures 1–5).
in their velocity, polarity (ie, their relative preference for entering axon For Kinesin-1s, we implemented a transcription control system
or dendrite), and the directionality of their movements within axons previously developed by Arnold and coworkers.17 Here the kinesin
or dendrites. Six kinesins labeled vesicles undergoing long-range tail was fused to two other domains: (a) A nuclear localization signal
transport. Of these, three had an axonal bias (KIF5B, KIF5C and that causes kinesins not bound to vesicles to translocate into the
KIF13B), two moved equally in axons and dendrites (KIF1A and nucleus; (b) A DNA-binding zinc finger domain that represses tran-
KIF1Bβ), and one was restricted to dendrites (KIF13A). Other kinesins scription of the kinesin expression vector. Kinesin tails that are not
labeled distinct populations of somatodendritic organelles that seldom bound to vesicles are targeted to the nucleus where they inhibit
underwent long-range movements, perhaps indicating a function for transcription of further kinesin tail mRNA from the plasmid. Nuclear
these kinesins beyond long-range transport. Notably, the preference pores can only accommodate proteins smaller than 90-110 kDa,27 a
of vesicle-attached kinesins for axons or dendrites often differed from limit most full-length kinesins exceed. Kinesin-1 tails are ~400 resi-
the transport preference of their motor domains, as defined in motor- dues long, which puts them below the nuclear pore size limit. This
19-21
domain expression assays. This observation suggests that the transcription control system improved Kinesin-1 tail labeling, but it
identity of the vesicle that a kinesin binds is a crucial determinant of was not effective for other kinesins, presumably due to their larger
that kinesin's transport parameters, an idea we refer to as “on-vesicle sizes.
regulation”. To further reduce expression levels, we deliberately recorded
cells after short expression times, starting as early as 3.5 hours
after transfection, and focused on low expressing cells which
2 | RESULTS
exhibited maximal labeling of discrete structures with low cyto-
plasmic background. Finally, for some constructs we used HaloTag
2.1 | Novel strategies to enhance vesicle labeling
visualized with dye JF549 developed by the Lavis lab.18 The quan-
with fluorescent kinesin tails
tum yield and extinction coefficient of JF549 are significantly
An obvious strategy to evaluate a kinesin's transport parameters is to higher than GFP, resulting in improved labeling of dim organelles,
express a fluorescently tagged version of that kinesin and monitor the thus enabling imaging at earlier expression times. Specifically,
vesicles it labels. With most kinesins, including Kinesin-1 family mem- Kinesin-1 tails benefitted from using HaloTag dyes. For all other
bers, this approach does not result in consistent vesicle labeling. For kinesin tails, GFP and HaloTag gave comparable results. This is
example, expression of GFP-tagged KIF5B in cultured hippocampal useful for two-color experiments in which with HaloTags can be
neurons resulted in diffuse fluorescence without any visible organelle visualized together with a GFP-tagged cargo protein (Figure 6).
labeling (Figure 1A) even though this kinesin is known to transport For kinesins that have known accessory subunits, these were co-
YANG ET AL. 853

F I G U R E 1 Strategies to enhance vesicle labeling with fluorescent kinesins. Representative images of seven DIV hippocampal neurons
expressing full-length GFP-KIF5B (A), HaloTag-KIF5B tail visualized with JF549 (B), full-length GFP-KIF13B (C) and GFP-KIF13B tail (D) for 6 to
8 hours at comparable levels. In (A) and (B), KLC1a was co-expressed. Yellow lines indicate the axons and dendrites from which high magnification
views and kymographs were generated. For clarity, transport events from kymographs were redrawn as solid black lines. Kymograph lines with a
positive slope indicate anterograde transport and those with a negative slope indicate retrograde transport. This convention is followed in all
figures. Scale bars: low magnification = 20 μm; high magnification = 5 μm

expressed to enhance vesicle labeling, as endogenous expression This combination of strategies markedly improved the ability to
levels of vesicle adaptors were not always sufficient to target record transport of structures labeled with exogenous kinesins.
overexpressed kinesin tails to vesicles. Expression of HaloTag-KIF5B revealed extensive vesicle labeling in
854 YANG ET AL.

F I G U R E 2 Kinesin-1-labeled vesicles move bidirectionally in dendrites, but are biased to the axon. Representative images of seven DIV
hippocampal neurons expressing HaloTag-KIF5B tail (A) or HaloTag-KIF5C tail (B), each co-expressed with KLC1a and visualized with JF549.
Yellow lines indicate the axons and dendrites from which high magnification views and kymographs were generated. Kymographs show that both
KIF5B- and KIF5C-labeled vesicles underwent short, bidirectional movement in dendrites and mostly long-range anterograde movements in
axons. Scale bars: low magnification = 20 μm; high magnification = 5 μm. Graphs show quantification of the number of transport events per
100 μm (C), the average velocity of moving vesicles (D) and their average run length (E). KIF5B: 107 events in 17 cells. KIF5C: 386 events in
17 cells. Error bars show the SEM

axons and dendrites. Kymographs from these regions showed numer- range transport in the axon and dendrites that were not present after
ous transport events (Figure 1B). Similarly, expression of fluorescent expression of full-length KIF13B.
KIF13B tail (Figure 1D) enhanced vesicle labeling over full-length This new labeling strategy enabled visualization of vesicles by
GFP-KIF13B (Figure 1C). Tail-labeled structures underwent long- most members of the Kinesin-1, -2 and -3 families. We obtained
YANG ET AL. 855

F I G U R E 3 Kinesin-2-labeled vesicles are largely restricted to the somatodendritic domain and undergo minimal long-ranged transport.
Representative images of seven DIV hippocampal neurons expressing GFP-KIF3B (A) or GFP-KIF3C (B) tail together with KIF3A tail. Yellow lines
indicate the axons and dendrites from which high magnification views and kymographs were generated. Kymographs show that these vesicles did
not undergo long-range transport events. Scale bars: low magnification = 20 μm; high magnification = 5 μm

consistent labeling of structures with 11 of the 15 kinesins we tested. dendrites. Kymographs show robust vesicle movement in the axon
For simplicity, we use the term “vesicle” to describe these structures with many more anterograde than retrograde events. Quantitative
although kinesin labeling alone cannot differentiate between transport kymograph analysis of transport events showed that there were fewer
vesicles, RNA granules, neurofilaments, or other structures kinesins transport events in dendrites, and that these events were divided
might interact with. Two-color experiments (Figure 6) show that at equally between anterograde and retrograde directions (Figure 2C).
least some of the kinesin tail-labeled structures are vesicles that con- Transport velocities were comparable in axons and dendrites
tain cargo proteins en route to the plasma membrane. The principal (Figure 2D). Vesicles moving in dendrites exhibited slightly shorter ret-
properties of different kinesin-labeled vesicles are summarized in rograde than anterograde run lengths, but this difference was not sta-
Table 1. Only KIF5A, KIF17, and two members of the Kinesin-4 family, tistically significant (P > .14) (Figure 2E). Most cells had a notably
KIF21A and KIF21B, failed to label vesicles. higher labeling intensity in the axon than in any of the dendrites
(Figure 2A and Video S1). In low expressing cells the axonal enrich-
ment was not as obvious, but the transport characteristics of moving
2.2 | Kinesin-1 family members preferentially traffic
vesicles were the same. Co-expression of KLC2 yielded similar results
in axons
(data not shown); little or no vesicle labeling was observed without
Kinesin-1 family members are thought to transport a wide range of co-expression of KLC. KIF5C labeling and transport behavior mim-
neuronal cargoes. These include vesicle populations that carry pro- icked that of KIF5B (Figure 2B-E and Video S1). Labeled vesicles were
teins destined for the axonal or dendritic plasma membranes23,28,29 as similar in size, shape and distribution. The most notable difference
well as mitochondria,22,30 lysosomes,31 neurofilaments32 and was that KIF5C-labeled about three times as many vesicles as KIF5B
33
mRNAs. The majority of Kinesin-1 exists as a tetramer with two (Figure 2A-C). KIF5C expression is enriched in brain,35 and neurons
heavy chains and two light chains (KLCs) that act as vesicle adap- may have KIF5C-specific vesicle populations.
tors.34 Figure 2A shows a cell expressing HaloTag-KIF5B tail along Previous analyses of Kinesin-1 motor domains showed that truncated
with untagged KLC1. KIF5B-labeled vesicles in the axon and KIF5B and KIF5C motor domains both have an axonal preference.19,20,36
856 YANG ET AL.

F I G U R E 4 Kinesin-3-labeled vesicles exhibit a diverse range of transport parameters. Representative images of seven DIV hippocampal
neurons expressing GFP-KIF1A tail (A), GFP-KIF1Bβ tail (B), GFP-KIF13A tail (C) and GFP-KIF13B tail (D). Yellow lines indicate the axons and
dendrites from which high magnification views and kymographs were generated. Kymographs show that KIF1A- and KIF1Bβ-labeled vesicles
undergo short, bidirectional movement in dendrites and axons. Vesicles labeled by KIF13A are restricted to dendrites. Vesicles labeled by KIF13B
move bidirectionally in dendrites and have a strong preference for anterograde movement in the axon. Scale bars: low magnification = 20 μm;
high magnification = 5 μm. Graphs show quantification of the number of transport events per 100 μm (C), the average velocity of moving vesicles
(D) and their average run length (E). KIF1A: 430 events in 22 cells. KIF1Bβ: 683 events in 18 cells. KIF13A: 181 events in 290 cells. KIF13B:
649 events in 15 cells. Error bars show the SEM

The labeling and transport parameters of Kinesin-1-labeled vesicles shows 2.3 | Kinesin-2 family members KIF3A/B/C label
that most vesicles move in the axon (Figure 2A-C), matching the prefer- organelles that are restricted to the somatodendritic
ence of the motor domains. This supports the hypothesis that the KIF5B region
and KIF5C motor domains inherently confer axon-selective transport in
The Kinesin-2 family consists of two KIF3 heterodimers, KIF3AB and
the absence of on-vesicle regulation.19 Among the vesicles labeled in den-
KIF3AC, and the homodimeric KIF17.41,42 Kinesin-2s have been pro-
drites could be the previously proposed Kinesin-1-mediated dendrite-
posed to participate in axonal and dendritic transport,43,44 but their
selective carrriers that transport AMPA receptors.28,37 This observation
neuronal cargoes and transport characteristics are not well under-
may suggest that dendritic transport by Kinesin-1 requires on-vesicle reg-
stood. Figure 3A shows a cell co-expressing GFP-KIF3B tail along with
ulatory mechanisms.28,38
unlabeled KIF3A tail (Video S2). GFP-KIF3B tail labeled a population
In contrast to KIF5B and KIF5C, we did not observe vesicle label-
of vesicles that were restricted to the soma and dendrites. Expression
ing with Kinesin-1 family member KIF5A. KIF5A may be specialized
of GFP-KIF3C tail with KIF3A tail gave similar results (Figure 3B and
for transport of other structures, including mRNA39 and neuro-
Video S2). Quantification of fluorescence intensities resulted in a den-
filaments.32,40 If these cargoes bind only a small number of kinesin drite to axon ratio of 4.7 ± 0.7 (n = 18) for KIF3B and 6.3 ± 1.0
dimers or use a different kinesin adaptor, they may not be visible over (n = 18) for KIF3C. Surprisingly, these organelles seldom underwent
cytoplasmic background in our assays. long-range transport. The identity of these organelles is unknown, but
YANG ET AL. 857

(A) GFP-KIF1Bα tail


Axon Dendrite Axon Dendrite Axon Dendrite

Axon

Dend

(B) GFP-KIF1C tail


Axon Dendrite Axon Dendrite Axon Dendrite

Dend

Axon

(C) GFP-KIF16B tail


Axon Dendrite Axon Dendrite Axon Dendrite

Dend

Axon

10 μm
30 s

F I G U R E 5 Some Kinesin-3 family members label vesicles that do not undergo long-range transport. Representative images of seven DIV
hippocampal neurons expressing GFP-KIF1Bα tail (A), GFP-KIF1C tail (B) and GFP-KIF16B tail (C). Yellow lines indicate the axons and dendrites
from which kymographs were generated. Kymographs show that vesicles labeled by these kinesins underwent few long-range transport events.
Low magnification scale bar: 20 μm; high magnification scale bar: 5 μm

they are apparently distinct from the dendritic endosomes that bind 2.4 | Kinesin-3 family members display a diverse
KIF16B16 since they could not be labeled with antibodies against range of localization and transport parameters
EEA1 (data not shown). KIF3AB relies on the accessory protein KAP3
The Kinesin-3 family is large, and its members have been implicated in
for vesicle binding in cilia,45,46 and co-expression of exogenous
many different aspects of vesicle transport. KIF1A and KIF1Bβ are the
KAP3B resulted in comparable labeling (data not shown). The associa-
principal motors associated with synaptic vesicles.12 KIF1A also
tion of KIF3 motors with dendritic vesicles is unexpected, since these
moves secretory granules, alphaherpesvirus particles, and some
motors prefer axonal microtubules or are unpolarized in assays utiliz-
20 dendritically polarized vesicles.7,15,52,53 KIF1Bβ moves dendritically
ing constitutively active motor domains.
polarized mRNPs,54 while KIF13A and KIF13B both play a role in
Expression of the other Kinesin-2 family member, GFP-KIF17 tail,
dendrite-selective transport.7
failed to label vesicles although previous results suggest that it is a
GFP-tagged KIF1A tail labeled vesicles in both axons and den-
motor for dendritic cargoes.44,47-51 One possible reason for the lack
drites (Figure 4A,E and Video S3). Kymographs show that these vesi-
of vesicle labeling is that the expression level of endogenous KIF17
cles were transported efficiently and bidirectionally in both axons and
adaptor proteins is not sufficient to target overexpressed KIF17. Co-
expression of the KIF17 adaptor mLin10 47
or identification of other dendrites. KIF1Bβ also labeled vesicles in axons and dendrites

KIF17 vesicle adaptors may enhance labeling in this assay. (Figure 4B,E and Video S3), similar to KIF1A. In some cells, KIF1Bβ
858 YANG ET AL.

F I G U R E 6 Two-color live-cell imaging can identify the cargos moved by different kinesins. (A) Representative high magnification images and
kymographs of cells expressing the indicated kinesin tails and cargoes. Yellow lines indicate transport events in both channels. HaloTags were
visualized with JF549. Cells with KIF5B tails were cotransfected with KLC1. (B) A table indicating the percentage of kinesin-labeled vesicles that
co-transported with various cargo markers. Anterograde axonal transport and bidirectional dendritic transport were analyzed separately

labeling was enriched in the proximal axon, suggesting that this axon and dendrites (Figure 4F,G). The fact that KIF1A is known to
kinesin may play a role in targeting proteins to the axon initial seg- bind vesicles containing the low-density lipoprotein receptor,7 whose
ment. Kymographs showed robust transport in both axon and den- transport is restricted to dendrites, indicates that this motor binds at
drites. While the transport characteristics of KIF1Bβ vesicles were least two distinct vesicle populations with different transport
similar to those labeled with KIF1A, KIF1Bβ labeled about two to parameters.
three times as many vesicles as KIF1A (Figure 4E). Velocities and run KIF13A-labeled vesicles underwent long-range bidirectional trans-
lengths of KIF1A- and KIF1Bβ-labeled vesicles were comparable in port in soma and dendrites but were not found in the axon beyond
YANG ET AL. 859

TABLE 1 Kinesin localization in cultured hippocampal neurons

Long-range Tail transport Motor domain


Kinesin family Kinesin tail Accessory subunit Vesicle labeling movement Distribution preference preferencea
Kinesin-1 KIF5A KLC1 or KLC2 No Axonal
KIF5B KLC1 or KLC2 Yes Yes Axon and dendrites Axonal Axonal
KIF5C KLC1 or KLC2 Yes Yes Axon and dendrites Axonal Axonal
Kinesin-2 KIF3AB Yes No Somatodendritic Axonal
KIF3AC Yes No Somatodendritic Weakly axonal
KIF17 No Axonal
Kinesin-3 KIF1A Yes Yes Axon and dendrites No preference Non-selective
KIF1Bα No Soma and proximal dendrites Non-selective
KIF1Bβ Yes Yes Axon and dendrites No preference Non-selective
KIF1C Yes No Soma and proximal dendrites
KIF13A Yes Yes Somatodendritic Dendritic Axonal
KIF13B Yes Yes Axon and dendrites Axonal Non-selective
KIF16B Yes No Somatodendritic
Kinesin-4 KIF21A No Axonal
KIF21B No Non-selective
a
Motor domain preferences are based on Huang and Banker study.20

the initial segment (Figure 4C and Video S3). This distribution is remi- in anchoring vesicles and that long-range transport is mediated by
niscent of vesicles labeled with prototypical dendritically polarized another motor, such as dynein.
7,55-58
membrane proteins. We previously implicated KIF13A as a
motor on dendritically polarized vesicles7 and this result supports that
2.5 | Identifying cargoes carried by specific kinesins
hypothesis. Interestingly, the KIF13A motor domain is axon
selective,20 indicating that on-vesicle regulation is crucial for its polar- The identity of the protein content in vesicles that are bound by spe-
ized transport to dendrites. The closely related KIF13B-labeled vesi- cific kinesins is crucial information for understanding the biological

cles that moved in all neurites, but more vesicles entered the axon function of these motors. To test the viability of kinesin tail labeling

than dendrites. In dendrites KIF13B-labeled vesicles moved bidirec- for identifying cargoes, we systematically co-expressed kinesin tails
with candidate fluorescent cargo proteins and imaged their move-
tionally (Figure 4D,E and Video S3). KIF13B and KIF13A both labeled
ments (Figure 6). If two proteins move coincidently—in the same
vesicles that exhibit high velocities and run lengths (Figure 4F,G).
direction and at the same velocity—it strongly suggests that they are
KIF13B plays a role in axon formation,59 possibly by transporting PIP3
associated with the same vesicles. Figure 6A shows example kymo-
vesicles into the axon,60 and is also involved in dendrite-selective
graphs illustrating the movements of two different kinesins (KIF5B
transport of transferrin receptor and low-density lipoprotein receptor
and KIF13A), each co-expressed with the axonal plasma membrane
vesicles.7 The mechanisms of the on-vesicle regulation that allow
protein neuron-glia cell adhesion molecule (NgCAM).10,55,61,62 Previ-
KIF13B to mediate such distinctly different transport patterns are
ous work19,21,63 and the behavior of KIF5 tails shown here (Figure 2)
unknown.
support the hypothesis that members of the Kinesin-1 family
Three other Kinesin-3 family members, KIF1Bα-, KIF1C- and
mediate the transport of axonally polarized proteins. However, this
KIF16B-labeled organelles that were restricted to the somatodendritic
kinesin-cargo interaction has never been shown directly. NgCAM-
domain, but these organelles seldom underwent long-range transport
GFP-labeled vesicles that moved bidirectionally in dendrites, but
(Figure 5). KIF1Bα- and KIF1C-labeled small punctate structures in the
preferentially entered the axon, where they underwent mostly antero-
cell body and proximal dendrites, although the identity of the labeled grade transport (Figure 6A). KIF5B-labeled vesicles overlapped with
structures was not apparent. KIF16B labeled large, round structures NgCAM in axon and dendrites, but KIF13A-labeled vesicles did not.
that extended throughout the dendritic tree, and had previously been We systematically analyzed the amount of overlap between three
identified as EEA1-positive endosomes.16 These vesicles were largely kinesins (KIF5B, KIF13A and KIF1A) and cargo proteins thought to
stationary; only a small number occasionally underwent movements at move in distinct vesicle populations, including transferrin receptor
very slow velocities, as seen previously.16 One possible explanation (a dendritically polarized protein), brain-derived neurotrophic factor
for the sporadic transport is that kinesins are bound in an inactive (BDNF; a marker of secretory granules), and NgCAM (Figure 6B).
state and that occasional transport events are initiated by temporary More than half of KIF5B-labeled vesicles undergoing anterograde axo-
motor activation. A second possibility is that these kinesins play a role nal transport also carried NgCAM and 30% of dendritic KIF5B-labeled
860 YANG ET AL.

TABLE 2 Quantitative analysis of kinesin vesicle movement

Axon Dendrites

Anterograde events Retrograde events Anterograde events Retrograde events

Velocity Velocity Velocity Velocity


Kinesin N (μm/s) Events/100 μm (μm/s) Events/100 μm (μm/s) Events/100 μm (μm/s) Events/100 μm
KIF5B 17 1.08 ± 0.07 3.11 ± 0.58 1.19 ± 0.22 0.84 ± 0.33 1.24 ± 0.12 0.89 ± 0.22 1.07 ± 0.13 0.68 ± 0.18
KIF5C 17 1.17 ± 0.07 10.61 ± 1.36 1.18 ± 0.08 3.22 ± 0.58 1.17 ± 0.10 2.08 ± 0.35 1.03 ± 0.07 2.00 ± 0.42
KIF1A 22 1.02 ± 0.12 2.84 ± 0.76 0.89 ± 0.07 3.83 ± 0.65 1.17 ± 0.09 3.69 ± 0.42 1.12 ± 0.07 3.48 ± 0.36
KIF1Bβ 18 1.04 ± 0.05 9.42 ± 1.15 1.07 ± 0.06 9.74 ± 0.79 1.19 ± 0.07 7.09 ± 0.65 1.14 ± 0.06 6.05 ± 0.73
KIF13A 20 NA NA NA NA 1.56 ± 0.06 1.89 ± 0.27 1.60 ± 0.13 1.26 ± 0.18
KIF13B 16 1.23 ± 0.07 12.06 ± 1.75 1.62 ± 0.15 3.47 ± 0.55 1.89 ± 0.13 4.48 ± 0.73 1.81 ± 0.10 4.65 ± 0.52

Note: N, number of cells. Numbers are mean ± SEM, Event number, events/50 μm/min. For each motor between 107 and 683 events were analyzed for
each kinesin.

vesicles exhibited co-movement with NgCAM. Vesicles labeled with type. In the axon, kinesins mediate anterograde movement, and there
KIF5B did not overlap with vesicles labeled with transferrin receptor. were no significant differences in the anterograde velocities of differ-
KIF13A did not co-transport with NgCAM or BDNF granules but over ent kinesin-labeled vesicles in the axon (Figures 2D and 4D). All aver-
30% of dendritic KIF13A tail-labeled vesicles co-transported with aged slightly more than 1 μm/s, consistent with many earlier
transferrin receptor (Figure 6B), consistent with previous reports.7 observations. There were, however, significant differences in the den-
A significant fraction of KIF1A tail-labeled vesicles were also labeled dritic velocities of vesicles bound by different kinesins (Table 2).
with BDNF (15% of axonal vesicles, nearly 40% of dendritic vesicles), KIF13A- and KIF13B-labeled vesicles moved with velocities of ~1.6
also consistent with prior results.15 Together, these results show that and ~1.8 μm/s which were significantly higher (P < .03) than all other
tail labeling is a powerful tool for identifying the cargoes that are kinesins (which averaged between 1.0 and 1.2 μm/s). Notably, the
moved by a particular kinesin and confirm that kinesin-vesicle interac- velocity of KIF13B-labeled vesicles was significantly greater in den-
tions are highly specific. drites than in the axon. The fact that KIF13A and KIF13B are particu-
Expression of kinesin tails could cause a dominant negative effect. larly effective dendritic transporters is consistent with their role in
To address this, we expressed GFP-NgCAM together with Halo- dendrite-selective transport.7 Run lengths for vesicles labeled with
tagged KIF5B, KIF13A or KIF13B tails and measured the transport different kinesins averaged between 5 and 10 μm in axons and
parameters of vesicles undergoing anterograde transport in the axon. dendrites.
The number of transport events observed when labeling vesicles with There were pronounced differences in the number of transport
fluorescent cargo proteins varies considerably from cell to cell, but events among vesicles bound by different kinesins, including differ-
cells expressing KIF5B—and hence subject to any potential dominant ences in the relative balance between anterograde and retrograde
negative effects—showed more NgCAM transport events than those transport and in the relative number of events observed in axons and
expressing KIF13s (data not shown). Vesicle velocity may be a more dendrites. To determine if the transport of vesicles labeled with some
sensitive measure of dominant negative effects, since a reduction in kinesins exhibited a bias for the anterograde or retrograde direction,
the number of motors bound to each vesicle could result in a reduc- we calculated a directionality index for each kinesin (DI: the number
tion in velocity. NgCAM vesicles moved with an average velocity of of anterograde transport events divided by the sum of anterograde
1.2 μm/s when KIF13A tail was expressed (n = 68 events from eight and retrograde events). If the number of anterograde and retrograde
cells), 1.1 μm/s when KIF13B tail was expressed (n = 168 events from transport events within either axons or dendrites are the same, the DI
10 cells), and 1.4 μm/s when KIF5B tail was expressed (n = 217 events is 0.5. Values above 0.5 indicate an anterograde, and values below 0.5
from 14 cells). These results show that expression of the KIF5B tail a retrograde bias (Figure 7A,B). In the axon, KIF5B, KIF5C and KIF13B
domain does not inhibit axonal transport of NgCAM vesicles and sup- all exhibited a strong anterograde bias. In contrast, the number of
ports the hypothesis that short-term, low-level expression of kinesin anterograde and retrograde movements was roughly equal for both
tails does not cause a dominant negative effect. KIF1A and KIF1Bβ. More than 90% of axonal microtubules are ori-
ented with their plus ends away from the cell body.64,65 The retro-
grade movement of kinesin-labeled vesicles implies that kinesins
2.6 | Quantitative analysis reveals diverse transport
remain bound during dynein-mediated transport. This observation is
parameters of different kinesin-labeled vesicles
consistent with models in which the direction of vesicle transport is
One of the principal advantages of this labeling strategy is that it regulated by activation and deactivation of opposing motors, which
allows for the quantitative analysis of kinesin-labeled vesicle transport remain bound to vesicles regardless of the direction of their trans-
and the systematic comparison of different kinesins within a single cell port.23,29,66 Our findings also suggest that kinesin-dynein interactions
YANG ET AL. 861

are differentially regulated for different kinesins because the relative


directional biases of labeled vesicles are quite different. KIF5B, KIF5C
and KIF13B showed a strong anterograde bias in the axon, whereas
KIF1A and KIF1Bβ moved roughly equally in both directions. In den-
drites, no kinesins labeled vesicles with a strong directionality prefer-
ence (Figure 7B). This is consistent with the mixed orientation of
dendritic microtubules, which could enable kinesins to mediate both
anterograde and retrograde transport.64,67,68
There were also important differences among different kinesins in
the number of transport events observed in axons or dendrites. To
capture these differences, we calculated a polarity index (PI) for each
kinesin, which is a measure of the preference for transport from the
cell body into axons or dendrites (Figure 7C). The PI is defined as the
number of anterograde transport events in the axon divided by the
sum of anterograde evens in axons and dendrites. Several kinesins
labeled vesicles that exhibited polarized transport (Figure 7C). Both
Kinesin-1 motors, KIF5B and KIF5C, were biased to the axon
(PI = 0.68 and 0.82, respectively). A third kinesin with an axonal pref-
erence was the Kinesin-3 family member KIF13B (PI = 0.72). Only one
kinesin was strongly polarized to dendrites, the Kinesin-3 family mem-
ber KIF13A (PI = 0.12). KIF1Bβ did not show a strong preference for
either axon or dendrite, indicated by PI = 0.55. The closely related
KIF1A had a slight preference for dendrites (PI = 0.35).

3 | DISCUSSION

Here we describe a combination of methodological improvements


that enable detection of kinesins bound to vesicles in living cells. To
illustrate the application of this new approach, we systematically char-
acterized the localization and transport parameters of the principal
members of the Kinesin-1, -2, and -3 families that are expressed in
cultured hippocampal neurons. Our experiments reveal a wide range
of transport parameters among different kinesins. Only six of these
kinesins exhibited long-range transport. Surprisingly, several other
kinesins labeled vesicles that underwent little long-range transport.
Co-labeling experiments of kinesin tails with cargo proteins show the
viability of this method for identifying the cargoes moved by each
kinesin. Our results also show that the transport parameters exerted
by a particular kinesin are not defined solely by the properties of its
motor domain but can be modified by on-vesicle regulation. The novel
methods described here will be useful in future studies to address the
mechanisms of differential kinesin regulation by interactions with tar-
get vesicles.

F I G U R E 7 Quantitative analysis of kinesin-labeled vesicles that 3.1 | Labeling vesicles with expressed kinesin tails
undergo long-ranged transport. (A) The directionality of axonal transport
events of vesicles labeled with different Kinesin-1 and Kinesin-3 family To define the role of different kinesins in vesicle transport, we
members. (B) The directionality index of dendritic transport events of describe an improved method for visualizing the vesicles moved by a
vesicles labeled with different kinesins. (C) The polarity index of vesicles particular kinesin. The kinesin tail labeling strategy we describe is a
labeled with these same kinesins. Indices were derived from kymograph
considerable improvement over expressing fluorescently tagged full-
analysis of seven to nine DIV hippocampal neurons expressing the
designated kinesin tail labels. No axonal directional index was generated for length kinesins. In principle, this labeling method is capable of visualiz-
vesicles labeled by KIF13A, as those vesicles were restricted to dendrites ing all of the vesicles that are moved by a particular kinesin, as long as
862 YANG ET AL.

TABLE 3 Kinesin tail constructs and accessory proteins used to label vesicles

Construct Construct design Accession number


Halo-KIF5A tail pCb_CAG-HaloTag-LYGAGADLGAGAGAGAGAG-ZincFinger-LPGI-KRAB NM_008447.4
(A)-GGGSGGGSGGGLYI-KIF5A378-1027
Halo-KIF5B tail pCb_CAG-HaloTag-LYGAGADLGAGAGAGAGAG-ZincFinger-LPGI-KRAB NM_008448.3
(A)-GGGSGCGSGGGLYI-KIF5B376-963
Halo-KIF5C tail pCb_CAG-HaloTag-LYGAGADLGAGAGAGAGAG-ZincFinger-LPGI-KRAB NM_001107730
(A)-GGGSGCGSGGGLYKGGGSGGGSGGGP-KIF5C378-955
KIF3A tail FRB-3myc-GGGSGGGSGGGLYK-KIF3A419-702 NM_001300792.1 variant 2
GFP-KIF3B tail GFP-GGGSGGGSGGGLYK-KIF3B391-747 NM_008444.4
411-796
GFP-KIF3C tail GFP-GGGSGGGSGGGLYR-KIF3C NM_008445.2
GFP-KIF17 tail GFP-GGGSGGGSGGGLYKRTGASGGGSGGGSGGGSGGGTSYPYM- NM_010623.4
KIF17400-1038
GFP-KIF1A tail GFP-3myc-YIDITDTNTVPGGPKL-KIF1A395-1698 NM_001294149.1
GFP-KIF1Bα tail GFP-LYKGGSGG-KIF1Bα387-1150 NM_008441.2
387-1770
GFP-KIF1Bβ tail GFP-LYKGGSGG-KIF1Bβ NM_207682.2
GFP-KIF1C tail GFP-GGGSGGGSGGGLYS-KIF1C400-1103 NM_006612.5
361-1749
GFP-KIF13A tail GFP-GGGSGGGSGGGLYKYSDLELKLRILQSTVPRAR-KIF13A NM_010617.2
GFP-KIF13B tail GFP-GGGSGGGSGGGLYKGGGSGGGSGGG-KIF13B442-1843 NM_001081177.1
458-1317
GFP-KIF16B tail GFP-LYKGGGSGGGSGG-KIF16B -RTV NM_024704.4
GFP-KIF5B full-length GFP-TREFGKLAAIRT-KIF5B NM_008448
GFP-KIF13B full-length GFP-VHDAWRSWNR-KIF13B NM_001081177.1
KLC1a FRB-3myc-LYKGGSGG-mmKLC1a NM_008450.2
55
NgCAM-GFP Sampo et al. Z75013
TfR-GFP TfR-GDPPVAT-GFP M11507
BDNF-mRFP BDNF-GGGPRdPPVAT-mRFP NM_001270630.1

there are enough binding sites available on vesicles. We previously As with any assay, this approach has limitations, particularly when
described a “split-kinesin” assay to determine interactions between a negative results are obtained. One concern is technical: even in neu-
particular kinesin and any labeled vesicle population of interest.7,8,26 rons with successful labeling, vesicles are usually dim. Modern sCMOS
That assay is based on altering the transport properties of a given ves- or EMCCD cameras can detect dim signals, but there may be vesicle
icle population if and only if it binds an expressed kinesin tail. While populations that bind so few kinesin molecules that they are not
that assay has proved useful in identifying kinesins that transport dif- detectable above background. A second concern is that in some cases,
ferent vesicles populations, it is essentially binary, that is, it shows co-expression of accessory or adaptor proteins may be required for
definitively that some labeled vesicles bind a candidate kinesin but is adequate vesicle labeling. For example, KIF5B and KIF5C only labeled
difficult to know what fraction of labeled vesicles interact with the vesicles when a KLC isoform was also expressed. In addition, we failed
kinesin. Thus, that assay provides no information about the transport to detect mitochondrial labeling with these kinesins, probably because
properties of the vesicles that bind a specific kinesin. The tail labeling the endogenous levels of Trak/Milton adaptor protein22,30 were insuffi-
strategy described here offers a complementary approach that cient. Less is known about the vesicle-binding interactions for other
enables characterization of the transport properties of the vesicles kinesins, and identification of other vesicle adaptors may reveal addi-
bound by a given kinesin. Tail labeling can also identify cargo proteins tional vesicle populations for specific kinesins.
in two-color experiments and can provide quantitative information Another concern is that overexpression of kinesin tail constructs
about the percentage of overlap between different cargoes and could cause a dominant negative effect by dimerization with endoge-
kinesins. nous kinesins to form single-headed motors or through competition
Another advantage of the tail-labeling strategy is that it allows the with endogenous motors for a limited number of vesicle binding sites.
direct comparison of all relevant kinesins within a single cell type. We We sought to avoid these problems by using constructs that lacked
found in neurons that even closely related kinesins can exhibit quite the first coiled-coil which is important for dimerization and by record-
different transport parameters depending on the vesicles they bind. ing at early expression times. Another potential technical problem is
While our experiments were focused on neurons, we expect this that the specific label, such as GFP, may alter the specific kinesin
method will provide similar information about the role of kinesins in a motor behavior, yet in other cases be a good reporter. Experiments
broad range of eukaryotic cell types. with NgCAM and different kinesin tails (discussed above) showed that
YANG ET AL. 863

at least for Kinesin-1 dominant negative effects are unlikely. We which such regulation is achieved is not known, but several models
expressed comparable tail constructs in previous studies and detected have been proposed. One possibility is that multiple motors—dynein,
no change in the extent of vesicle transport and localization.7,8,26 Pre- myosins, or other kinesins—bind vesicles together and that transport
sumably, endogenous full-length kinesins can still attach to vesicles behavior is the sum of their activities,75-79 which may be regulated in
and mediate transport in the presence of overexpressed tails. a vesicle-specific way. A second possibility is that adaptor proteins
Finally, labeling a vesicle with a given kinesin does not formally selectively activate or inhibit kinesins22,23 and that the regulatory pro-
prove that this kinesin mediates that vesicle's transport. That said, the teins that associate with a given kinesin are vesicle-specific.
different kinesins we expressed labeled vesicle populations with dis- A crucial next step towards understanding on-vesicle regulation will
tinctly different transport parameters, suggesting that tail binding is be the identification of the regulatory proteins that interact with vesicle-
specific for a subset of neuronal vesicles. bound kinesins. The ability to visualize vesicle-associated kinesins directly
is an important tool towards this goal. For example, two-color experi-
ments with a kinesin and a candidate regulatory protein can confirm their
3.2 | Neuronal kinesins exhibit diverse transport
localization on the same vesicles. Manipulation of candidate regulatory
parameters
proteins followed by assessment of changes in the transport parameters
Much is known about how kinesin motor domains behave in vitro and of relevant kinesin-labeled vesicle populations may provide insight into
in living cells,21,36,69-71 but relatively little is known about the behavior the regulation of specific kinesins. Because of the complexity of kinesin-
of vesicle-bound kinesins. One surprising outcome from this study is mediated transport and its regulation, we expect labeling of vesicle-bound
that only six kinesins, all members of the Kinesin-1 and Kinesin-3 fam- kinesins to be a useful tool in many future applications.
ilies, exhibited long-range transport. Moreover, the transport parame-
ters of the vesicles labeled by these kinesins often did not match the
4 | MATERIALS AND METHODS
preference of their motor domains. For example, KIF13A labeled
dendritically polarized vesicles even though its motor domain prefers
4.1 | Cell culture
axonal microtubules in cell expression assays.20 Even the Kinesin-1
family members, which we show move axonally polarized NgCAM Primary hippocampal neurons were prepared as described previ-
vesicles, have been implicated in the transport of dendritically polar- ously.80-82 Hippocampi were dissected from E18 rats, trypsinized, dis-
28,38
ized vesicles. Given that cells generate a multitude of vesicles that sociated, and plated on poly-L-lysine-treated 18 mm glass coverslips.
need to be delivered to distinct locations, on-vesicle regulation may Cultures were grown in minimal essential medium with N2 supple-
be the mechanism that allows for this. ments and maintained at 37 C in an incubator with a 5% CO2 atmo-
A further surprising result of this study was that several kinesins sphere. Constructs were transfected into stage 4 hippocampal
labeled vesicles that underwent little long-range transport. These vesi- neurons (8-12 days in culture) with Lipofectamine 2000 (Thermo
cles appear to fall into three distinct categories: (a) small vesicles in Fisher) and allowed to express for 5-18 hours before imaging. For sev-
the soma and proximal dendrites that are labeled by KIF1Bα and KIF1C; eral difficult-to-image kinesins it was necessary to focus on the lowest
(b) large vesicles in soma and proximal dendrites labeled by KIF3AB and expressing cells, which was accomplished by recording as early as
KIF3AC and (c) large endosomal vesicles labeled by KIF16B that are 5 hours after transfection. This was done to minimize the background
present in the soma and throughout the dendritic tree. This is surpris- fluorescence caused by soluble kinesins.
ing, because the constitutively active motor domains of these kinesins
move efficiently in neurons when expressed alone20 or cross-linked to
4.2 | DNA constructs
peroxisomes.36,72 However, recent experiments from the Verhey lab
found that monomeric kinesin motor domains cause movement when a Kinesin tail constructs were generated by replacing the N-terminal motor
sufficient number of them are targeted to peroxisomes.73 Therefore, domain and the dimerization domain with a fluorescent tag. A full list of
the peroxisome assay, while powerful, may not reflect physiological kinesin tails used in this study and their correlated amino acid residue
vesicle transport behavior. This fact supports the hypothesis that sequence is in Table 3. For Kinesin-1 family members, a zinc-finger DNA-
vesicle-bound kinesins can be regulated. binding domain and a KRAB(A) nuclear localization and transcription
One possibility may be that these kinesins act as tethers that main- repressor domain17 were fused to the N-terminus of kinesin tails. The
tain organelles at a particular location within the cell. In non-neuronal zinc-finger domain (CCR5ZFL) that recognizes the CCR5 gene83 was
74
cells, KIF1C interacts with Rab6 to mediate Golgi tethering. It is pos- followed by the KRAB(A) domain from rat Kid-1. The DNA recognition
sible that similar mechanisms regulate the tethering function of kinesins site of CCR5ZFL (50 -gtcatcctcatc-30 ) was inserted upstream of the CAG
in neurons. Another possibility is that these kinesins are on the vesicles promoter to allow for self-regulation by CCR5ZFL/KRAB(A).
in an inactive state, awaiting a signal to trigger short-range transport. In Each kinesin construct was designed to exclude the motor and dimer-
the latter case, there must be regulatory mechanisms that maintain the ization domains as we had previously established that the tail domain
inactive state until activation by a particular signal. alone is sufficient for vesicle binding.7,8,26 Every part of the molecule after
We conclude that the transport parameters of most, if not all, the first coiled-coil domain (as defined by the coils predictive algorithm84)
kinesins are governed by on-vesicle regulation. The mechanism by was fused to an N-terminal fluorescent probe. Removal of the first coiled-
864 YANG ET AL.

coil is intended to minimize dimerization with endogenous kinesins, which primary dendrite and the axon, distal to the initial segment. P-values
could otherwise result in a dominant negative effect. A heterodimer con- mentioned in the text were determined by Student's t tests.
sisting of a full-length endogenous kinesin and a motorless tail would not
be able to processively walk along microtubules.
ACKNOWLEDG MENTS

We thank Dr Stefanie Kaech for her advice on imaging and Swagatha


4.3 | Imaging
Dey for her help in generating the KIF3 constructs. We also thank Drs
For live imaging, cells were maintained in Hibernate E medium without Susan Gilbert, Catherine Drerup, Ines Hahn, and André Völzmann and
phenol red (Brain-Bits). Images were acquired using a microscope (model members of the Bentley lab for their comments on the manuscript.
Ti-E; Nikon) equipped with a spinning-disk confocal-head (model CSU- We thank Geraldine Quinones for excellent technical assistance. We
W1, Yokogawa), and images were captured with an sCMOS camera (Zyla, thank the BioResearch facility at Rensseler for help with tissue collec-
Andor). KIF3AB and KIF3AC images were acquired with an Andor Drag- tion for neuronal cultures. This work was supported by National Insti-
onfly built on a Ti2 (Nikon) with a CFI Apo 60× 1.49 objective (Nikon) tutes of Health grant MH066179.
and captured with an sCMOS camera (Zyla 4.2+, Andor). The entire imag-
ing stage and objectives were maintained at 37 C in a warmed enclosure
CONFLIC T OF INT ER E ST
(full lexan incubation ensemble; OkoLab). A Plan-Apo 100× 1.49 NA
objective (Nikon) was used with 2 × 2 binning to acquire image streams. The authors declare no competing financial interests.
During image acquisition, z-axis movement was controlled by the Perfect
Focus system on the Ti-E microscope (Nikon). Recordings were acquired
OR CID
at two frames per second. For further details, see Kaech et al.82
Axons were identified with anti-neurofascin antibody (NeuroMab, Marvin Bentley https://orcid.org/0000-0002-1754-0332
Cat #: 75-027) conjugated to CF405 (Mix-n-Stain CF405S Antibody
Labeling Kit; Biotum, Cat#: 92231) in the imaging medium. Cells
expressing constructs with HaloTag were treated with 50 nM Janelia RE FE RE NCE S

Farm 549 dye18 for 10 minutes and washed with conditioned medium 1. Xiao Q, Hu X, Wei Z, Tam KY. Cytoskeleton molecular motors: struc-
for 10 minutes prior to imaging. tures and their functions in neuron. Int J Biol Sci. 2016;12(9):1083-
1092.
2. Bhabha G, Johnson GT, Schroeder CM, Vale RD. How dynein moves
4.4 | Image analysis along microtubules. Trends Biochem Sci. 2016;41(1):94-105.
3. Hirokawa N, Noda Y, Tanaka Y, Niwa S. Kinesin superfamily motor
MetaMorph software (Molecular Devices) was used to generate kymo- proteins and intracellular transport. Nat Rev Mol Cell Biol. 2009;10
graphs from movies. Transport events were identified on the kymographs (10):682-696.
and the coordinates were exported to Excel (Microsoft) for analysis. It can 4. Reck-Peterson SL, Redwine WB, Vale RD, Carter AP. The cytoplasmic
dynein transport machinery and its many cargoes. Nat Rev Mol Cell
be difficult to identify dim transport events with complete certainty. To
Biol. 2018;149:1.
ensure consistency, a single reviewer identified transport events on all 5. Sweeney HL, Holzbaur ELF. Motor proteins. Cold Spring Harb Perspect
kymographs and scored only clear and “indisputable” transport events, Biol. 2018;10(5):a021931.
avoiding those that were borderline. Reviewer was blinded to the kinesin 6. Silverman MA, Kaech S, Ramser EM, et al. Expression of kinesin
superfamily genes in cultured hippocampal neurons. Cytoskeleton.
expressed in a given cell and kymographs from all conditions were ran-
2010;67(12):784-795.
domly mixed together to prevent observer bias. 7. Jenkins B, Decker H, Bentley M, Luisi J, Banker G. A novel split
To ensure only microtubule-based long-range transport events kinesin assay identifies motor proteins that interact with distinct vesi-
were included in the analysis, excursions <3 μm were excluded. Each cle populations. J Cell Biol. 2012;198(4):749-761.
contiguous line with a constant slope was scored as a single transport 8. Bentley M, Banker G. A novel assay to identify the trafficking proteins
that bind to specific vesicle populations. Curr Protoc Cell Biol. 2015;
event and its velocity and run length (distance moved along the long
69:13.8.1-13.8.12.
axis of the neurite) were calculated. A single vesicle could undergo 9. Lim A, Rechtsteiner A, Saxton WM. Two kinesins drive anterograde
multiple transport events if there was a distinct pause between them. neuropeptide transport. Mol Biol Cell. 2017;28(24):3542-3553.
Between 17 and 22 cells were evaluated for each condition, including 10. Burack MA, Silverman MA, Banker G. The role of selective transport
in neuronal protein sorting. Neuron. 2000;26(2):465-472.
cells from at least two independent cultures. For KIF13A tail-labeling
11. Silverman MA, Kaech S, Jareb M, et al. Sorting and directed transport
we excluded the first 50 μm of axon from the transport analysis, as of membrane proteins during development of hippocampal neurons
this region has dendritic characteristics and stains positively for the in culture. Proc Natl Acad Sci. 2002;98(13):7051-7057.
dendritic marker MAP2. 12. Niwa S, Tanaka Y, Hirokawa N. KIF1Bβ- and KIF1A-mediated axonal
KIF3AB and KIF3AC dendrite to axon ratios were calculated from transport of presynaptic regulator Rab3 occurs in a GTP-dependent
manner through DENN/MADD. Nat Cell Biol. 2008;10(11):1269-
the first frame of each cell recoding. After a background subtraction,
1279.
the average intensity of one-pixel wide line scans was used to calcu- 13. Maday S, Holzbaur ELF. Compartment-specific regulation of
late the ratio. Linescans were at least 10 μm long and measured on the autophagy in primary neurons. J Neurosci. 2016;36(22):5933-5945.
YANG ET AL. 865

14. Ghiretti AE, Thies E, Tokito MK, et al. Activity-dependent regulation 35. Kanai Y, Okada Y, Tanaka Y, Harada A, Terada S, Hirokawa N. KIF5C,
of distinct transport and cytoskeletal remodeling functions of the a novel neuronal kinesin enriched in motor neurons. J Neurosci. 2000;
dendritic kinesin KIF21B. Neuron. 2016;92(4):857-872. 20(17):6374-6384.
15. Lo KY, Kuzmin A, Unger SM, Petersen JD, Silverman MA. KIF1A is 36. Lipka J, Kapitein LC, Jaworski J, Hoogenraad CC. Microtubule-binding
the primary anterograde motor protein required for the axonal trans- protein doublecortin-like kinase 1 (DCLK1) guides kinesin-3-mediated
port of dense-core vesicles in cultured hippocampal neurons. Neurosci cargo transport to dendrites. EMBO J. 2016;35(3):302-318.
Lett. 2011;491(3):168-173. 37. Rathgeber L, Gromova KV, Schaefer I, Breiden P, Lohr C, Kneussel M.
16. Farkhondeh A, Niwa S, Takei Y, Hirokawa N. Characterizing KIF16B GSK3 and KIF5 regulate activity-dependent sorting of gephyrin
in neurons reveals a novel intramolecular “stalk inhibition” mechanism between axons and dendrites. Eur J Cell Biol. 2015;94(3–4):173-178.
that regulates its capacity to potentiate the selective somatodendritic 38. Kelliher MT, Yue Y, Ng A, et al. Autoinhibition of kinesin-1 is essential
localization of early endosomes. J Neurosci. 2015;35(12):5067-5086. to the dendrite-specific localization of Golgi outposts. J Cell Biol.
17. Gross GG, Junge JA, Mora RJ, et al. Recombinant probes for visualiz- 2018;217(7):2531-2547.
ing endogenous synaptic proteins in living neurons. Neuron. 2013;78 39. Ma B, Savas JN, Yu MS, Culver BP, Chao MV, Tanese N. Huntingtin
(6):971-985. mediates dendritic transport of β-actin mRNA in rat neurons. Sci Rep.
18. Grimm JB, English BP, Chen J, et al. A general method to improve 2011;1:140.
fluorophores for live-cell and single-molecule microscopy. Nat 40. Wang L, Brown A. A hereditary spastic paraplegia mutation in
Methods. 2015;12(3):244-250. kinesin-1A/KIF5A disrupts neurofilament transport. Mol Neu-
19. Nakata T, Hirokawa N. Microtubules provide directional cues for rodegener. 2010;5(1):52.
polarized axonal transport through interaction with kinesin motor 41. Scholey JM. Kinesin-2: a family of heterotrimeric and homodimeric
head. J Cell Biol. 2003;162(6):1045-1055. motors with diverse intracellular transport functions. Annu Rev Cell
20. Huang C-F, Banker G. The translocation selectivity of the kinesins Dev Biol. 2013;29(1):443-469.
that mediate neuronal organelle transport. Traffic. 2012;13(4): 42. Gilbert SP, Guzik-Lendrum S, Rayment I. Kinesin-2 motors: kinetics
549-564. and biophysics. J Biol Chem. 2018;293(12):4510-4518.
21. Yang R, Bentley M, Huang CF, Banker G. Analyzing kinesin motor 43. Yamazaki H. KIF3A/B: a heterodimeric kinesin superfamily protein
domain translocation in cultured hippocampal neurons. Methods Cell that works as a microtubule plus end-directed motor for membrane
Biol. 2016;131:217-232. organelle transport. J Cell Biol. 1995;130(6):1387-1399.
22. Wang X, Schwarz TL. The mechanism of Ca2+-dependent regulation 44. Guillaud L, Setou M, Hirokawa N. KIF17 dynamics and regulation of
of kinesin-mediated mitochondrial motility. Cell. 2009;136(1): NR2B trafficking in hippocampal neurons. J Neurosci. 2003;23(1):
163-174. 131-140.
23. Fu MM, Holzbaur ELF. JIP1 regulates the directionality of APP axonal 45. Cole DG, Chinn SW, Wedaman KP, Hall K, Vuong T, Scholey JM.
transport by coordinating kinesin and dynein motors. J Cell Biol. 2013; Novel heterotrimeric kinesin-related protein purified from sea urchin
202(3):495-508. eggs. Nature. 1993;366(6452):268-270.
24. Raynaud F, Homburger V, Seveno M, et al. SNAP23-Kif5 complex 46. Yamazaki H, Nakata T, Okada Y, Hirokawa N. Cloning and characteri-
controls mGlu1 receptor trafficking. J Mol Cell Biol. 2018;10(5): zation of KAP3: a novel kinesin superfamily-associated protein of
423-436. KIF3A/3B. Proc Natl Acad Sci. 1996;93(16):8443-8448.
25. Hollenbeck PJ. The distribution, abundance and subcellular localiza- 47. Setou M, Nakagawa T, Seog D-H, Hirokawa N. Kinesin superfamily
tion of kinesin. J Cell Biol. 1989;108(6):2335-2342. motor protein KIF17 and mLin-10 in NMDA receptor-containing vesi-
26. Bentley M, Decker H, Luisi J, Banker G. A novel assay reveals prefer- cle transport. Science. 2000;288(5472):1796-1802.
ential binding between Rabs, kinesins, and specific endosomal sub- 48. Chu PJ, Rivera JF, Arnold DB. A role for Kif17 in transport of Kv4.2.
populations. J Cell Biol. 2015;93(10):4604. J Biol Chem. 2006;281(1):365-373.
27. Wang R, Brattain MG. The maximal size of protein to diffuse through 49. Yin X, Feng X, Takei Y, Hirokawa N. Regulation of NMDA receptor
the nuclear pore is larger than 60 kDa. FEBS Lett. 2007;581(17): transport: a KIF17-cargo binding/releasing underlies synaptic plastic-
3164-3170. ity and memory in vivo. J Neurosci. 2012;32(16):5486-5499.
28. Setou M, Seog D-H, Tanaka Y, et al. Glutamate-receptor-interacting 50. Bai X, Karasmanis EP, Spiliotis ET. Septin 9 interacts with kinesin
protein GRIP1 directly steers kinesin to dendrites. Nature. 2002;417 KIF17 and interferes with the mechanism of NMDA receptor cargo
(6884):83-87. binding and transport. Mol Biol Cell. 2016;27(6):897-906.
29. Maday S, Twelvetrees AE, Moughamian AJ, Holzbaur ELF. Axonal 51. Janssen AFJ, Wulf PS, Kapitein LC, Hoogenraad CC. Three-step
transport: cargo-specific mechanisms of motility and regulation. Neu- model for polarized sorting of KIF17 into dendrites. Curr Biol. 2016;
ron. 2014;84(2):292-309. 26(13):1705-1712.
30. van Spronsen M, Mikhaylova M, Lipka J, et al. TRAK/Milton motor- 52. Kramer T, Greco TM, Taylor MP, Ambrosini AE, Cristea IM,
adaptor proteins steer mitochondrial trafficking to axons and den- Enquist LW. Kinesin-3 mediates axonal sorting and directional trans-
drites. Neuron. 2013;77(3):485-502. port of alphaherpesvirus particles in neurons. Cell Host Microbe. 2012;
31. Farías GG, Guardia CM, De Pace R, Britt DJ, Bonifacino JS. BORC/- 12(6):806-814.
kinesin-1 ensemble drives polarized transport of lysosomes into the 53. Kratchmarov R, Kramer T, Greco TM, et al. Glycoproteins gE and gI
axon. Proc Natl Acad Sci U S A. 2017;114:E2955-E2964. are required for efficient KIF1A-dependent anterograde axonal trans-
32. Uchida A, Alami NH, Brown A. Tight functional coupling of Kinesin- port of alphaherpesvirus particles in neurons. J Virol. 2013;87(17):
1A and dynein motors in the bidirectional transport of neuro- 9431-9440.
filaments. Mol Biol Cell. 2009;20(23):4997-5006. 54. Charalambous DC, Pasciuto E, Mercaldo V, et al. KIF1Bβ transports
33. Kanai Y, Dohmae N, Hirokawa N. Kinesin transports RNA: isolation dendritically localized mRNPs in neurons and is recruited to synapses
and characterization of an RNA-transporting granule. Neuron. 2004; in an activity-dependent manner. Cell Mol Life Sci. 2012;70(2):
43(4):513-525. 335-356.
34. Schnapp BJ. Trafficking of signaling modules by kinesin motors. J Cell 55. Sampo B, Kaech S, Kunz S, Banker G. Two distinct mechanisms target
Sci. 2003;116(11):2125-2135. membrane proteins to the axonal surface. Neuron. 2003;37(4):
611-624.
866 YANG ET AL.

56. Farías GG, Cuitino L, Guo X, et al. Signal-mediated, AP-1/clathrin- 73. Schimert KI, Budaitis BG, Reinemann DN, Lang MJ, Verhey KJ. Intra-
dependent sorting of transmembrane receptors to the cellular cargo transport by single-headed kinesin motors. Proc Natl
somatodendritic domain of hippocampal neurons. Neuron. 2012;75 Acad Sci U S A. 2019;116(13):6152-6161.
(5):810-823. 74. Lee PL, Ohlson MB, Pfeffer SR. The Rab6-regulated KIF1C kinesin
57. Petersen JD, Kaech S, Banker G. Selective microtubule-based trans- motor domain contributes to Golgi organization. Elife. 2015;4:
port of dendritic membrane proteins arises in concert with axon spec- e06029.
ification. J Neurosci. 2014;34(12):4135-4147. 75. Lewis TL, Mao T, Svoboda K, Arnold DB. Myosin-dependent targeting
58. Farías GG, Guardia CM, Britt DJ, Guo X, Bonifacino JS. Sorting of of transmembrane proteins to neuronal dendrites. Nat Neurosci.
dendritic and axonal vesicles at the pre-axonal exclusion zone. Cell 2009;12(5):568-576.
Rep. 2015;13(6):1221-1232. 76. Al-Bassam S, Xu M, Wandless TJ, Arnold DB. Differential trafficking
59. Yoshimura Y, Terabayashi T, Miki H. Par1b/MARK2 phosphorylates of transport vesicles contributes to the localization of dendritic pro-
kinesin-like motor protein GAKIN/KIF13B to regulate axon forma- teins. Cell Rep. 2012;2(1):89-100.
tion. Mol Cell Biol. 2010;30(9):2206-2219. 77. Kapitein LC, Schlager MA, Kuijpers M, et al. Mixed microtubules steer
60. Horiguchi K, Hanada T, Fukui Y, Chishti AH. Transport of PIP3 by dynein-driven cargo transport into dendrites. Curr Biol. 2010;20(4):
GAKIN, a kinesin-3 family protein, regulates neuronal cell polarity. 290-299.
J Cell Biol. 2006;174(3):425-436. 78. Kuijpers M, van de Willige D, Freal A, et al. Dynein regulator NDEL1
61. Yap CC, Wisco D, Kujala P, et al. The somatodendritic endosomal reg- controls polarized cargo transport at the axon initial segment. Neuron.
ulator NEEP21 facilitates axonal targeting of L1/NgCAM. J Cell Biol. 2016;89(3):461-471.
2008;180(4):827-842. 79. Stucchi R, Plucinska G, Hummel JJA, et al. Regulation of KIF1A-driven
62. Wisco D, Anderson ED, Chang MC, et al. Uncovering multiple axonal dense core vesicle transport: Ca2+/CaM controls DCV binding and
targeting pathways in hippocampal neurons. J Cell Biol. 2003;162(7): Liprin-α/TANC2 recruits DCVs to postsynaptic sites. Cell Rep. 2018;
1317-1328. 24(3):685-700.
63. Jacobson C, Schnapp B, Banker GA. A change in the selective translo- 80. Kaech S, Banker G. Culturing hippocampal neurons. Nat Protoc. 2006;
cation of the Kinesin-1 motor domain marks the initial specification 1(5):2406-2415.
of the axon. Neuron. 2006;49(6):797-804. 81. Kaech S, Huang CF, Banker G. General considerations for live imaging
64. Baas PW, Deitch JS, Black MM, Banker GA. Polarity orientation of of developing hippocampal neurons in culture. Cold Spring Harb Pro-
microtubules in hippocampal neurons: uniformity in the axon and toc. 2012;7(3):312-318.
nonuniformity in the dendrite. Proc Natl Acad Sci U S A. 1988;85(21): 82. Kaech S, Huang CF, Banker G. Short-term high-resolution imaging of
8335-8339. developing hippocampal neurons in culture. Cold Spring Harb Protoc.
65. Leterrier C, Dubey P, Roy S. The nano-architecture of the axonal 2012;7(3):340-343.
cytoskeleton. Nat Rev Neurosci. 2017;32:435. 83. Mani M, Kandavelou K, Dy FJ, Durai S, Chandrasegaran S. Design,
66. Hancock WO. Bidirectional cargo transport: moving beyond tug of engineering, and characterization of zinc finger nucleases. Biochem
war. Nat Rev Mol Cell Biol. 2014;15(9):615-628. Biophys Res Commun. 2005;335(2):447-457.
67. Yau KW, Schatzle P, Tortosa E, et al. Dendrites in vitro and in vivo 84. Lupas A, Van Dyke M, Stock J. Predicting coiled coils from protein
contain microtubules of opposite polarity and axon formation corre- sequences. Science. 1991;252(5009):1162-1164.
lates with uniform plus-end-out microtubule orientation. J Neurosci.
2016;36(4):1071-1085.
68. Tas RP, Chazeau A, Cloin BMC, Lambers MLA, Hoogenraad CC, SUPPORTING INF ORMATION
Kapitein LC. Differentiation between oppositely oriented microtu-
bules controls polarized neuronal transport. Neuron. 2017;96(6): Additional supporting information may be found online in the
1264-1271.e5. Supporting Information section at the end of this article.
69. Vale RD. The molecular motor toolbox for intracellular transport. Cell.
2003;112(4):467-480.
70. Verhey KJ, Kaul N, Soppina V. Kinesin assembly and movement in
How to cite this article: Yang R, Bostick Z, Garbouchian A,
cells. Annu Rev Biophys. 2011;40(1):267-288.
Luisi J, Banker G, Bentley M. A novel strategy to visualize
71. Bentley M, Banker G. The cellular mechanisms that maintain neuronal
polarity. Nat Rev Neurosci. 2016;17(10):611-622. vesicle-bound kinesins reveals the diversity of kinesin-
72. Guardia CM, De Pace R, Sen A, et al. Reversible association with mediated transport. Traffic. 2019;20:851–866. https://doi.
motor proteins (RAMP): a streptavidin-based method to manipulate org/10.1111/tra.12692
organelle positioning. PLoS Biol. 2019;17(5):e3000279.

You might also like