Acetylated Microtubules Are Preferentially Bundled Leading to Enhanced Kinesin-1 Motility

Abstract

The motor proteins kinesin and dynein transport organelles, mRNA, proteins, and signaling molecules along the microtubule cytoskeleton. In addition to serving as tracks for transport, the microtubule cytoskeleton directs intracellular trafficking by regulating the activity of motor proteins through the organization of the filament network, microtubule-associated proteins, and tubulin posttranslational modifications. However, it is not well understood how these factors influence motor motility, and in vitro assays and live cell observations often produce disparate results. To systematically examine the factors that contribute to cytoskeleton-based regulation of motor protein motility, we extracted intact microtubule networks from cells and tracked the motility of single fluorescently labeled motor proteins on these cytoskeletons. We find that tubulin acetylation alone does not directly affect kinesin-1 motility. However, acetylated microtubules are predominantly bundled, and bundling enhances kinesin run lengths and provides a greater number of available kinesin binding sites. The neuronal MAP tau is also not sensitive to tubulin acetylation, but enriches preferentially on highly curved regions of microtubules where it strongly inhibits kinesin motility. Taken together, these results suggest that the organization of the microtubule network is a key contributor to the regulation of motor-based transport.

Introduction

Motor proteins navigate a dense and complex cytoskeletal network to transport organelles, proteins, and vesicular cargoes (1). The cytoskeleton does not merely act as a passive track for motor proteins; instead it actively modulates motor activity through posttranslational modifications (PTMs), microtubule-associated proteins (MAPs), and its network organization. This tubulin code directs trafficking to target cargoes to specific destinations in space and time (2, 3).

Microtubules (MTs) are subject to a range of posttranslational modifications that correlate with their polymerization dynamics and subcellular organization. Many PTMs alter the C-terminal tails of tubulin, including tyrosination/detyrosination, polyglutamylation, and polyglycylation (3). In contrast, acetylation of α-tubulin occurs at lysine 40 (K40) in the lumen of the microtubule. Acetylation is a marker of long-lived, stabilized microtubules (4). In neurons, axonal microtubules are highly acetylated, whereas more dynamic tyrosinated microtubules are enriched in dendrites and growth cones (5). Tracking of fluorescently labeled kinesin motors in living fibroblast cells shows that kinesin-1 moves preferentially along acetylated microtubules (6, 7), and tubulin acetylation contributes to the localization of kinesin-1 and kinesin-1-driven cargoes to the distal axon in neurons (8, 9). However, the motility and binding of kinesin-1 are not affected by tubulin acetylation in purified in vitro systems (10, 11). The discrepancy between kinesin-1 behavior in living cells and in vitro leads to the question: how is the tubulin code read and interpreted to regulate motor protein motility? What cellular components are required to preferentially recruit kinesin-1 to acetylated microtubules?

The organization of the microtubule network may play a role in regulating motor transport. Filament organization affects the localization of actin-based motors, where myosin V and VI motors preferentially associate with cortical meshworks of actin filaments, whereas myosin X is targeted to actin bundles in filopodia (12, 13, 14, 15). Similarly, the motility of microtubule motors is influenced by bundling. Due to its extended structure, cytoplasmic dynein can bind multiple microtubules simultaneously (16), and thus is likely to be sensitive to the spacing between microtubules. Although kinesin-1 moves along a single protofilament (17, 18), its run lengths are enhanced on microtubules bundled in the presence of MAP65 in comparison to single microtubules (19).

Microtubule-associated proteins control microtubule organization and alter the association of motor proteins with microtubules, suggesting they may act as readers of the tubulin code. MAPs have specific subcellular localizations that correlate with tubulin posttranslational modifications (2). MAP2 localizes to dendrites in neurons, whereas tau stabilizes and cross-links microtubules in neuronal axons (20). In addition, MAPs perturb the interaction of motor proteins with microtubules to regulate their motility. Tau strongly inhibits the motility of kinesin-1 in vitro, whereas dynein and kinesin-2 are less sensitive to tau (18, 21). Other MAPs like MAP7 (ensconsin) activate kinesin-1 motility (22, 23). In this way, MAPs act as both positive and negative regulators of motor proteins.

To determine how posttranslational modifications, microtubule organization, and microtubule-associated proteins contribute to regulating motor protein motility, we performed kinesin motility assays on isolated native cytoskeletal networks (Fig. 1 A). Briefly, we permeabilize the cell membrane with a mild detergent while stabilizing microtubules with Taxol or GMPCPP following protocols similar to those developed for extracting actin cytoskeletons (Figs. S1 and S3; Materials and Methods) (13, 24). Extracted cytoskeletal networks model the native microtubule organization and posttranslational modifications while enabling control over the biochemical conditions such as motor density and MAPs. However, the system is limited to examining static microtubule networks and cannot address the effect of microtubule dynamics and remodeling. We find that tubulin acetylation does not directly affect kinesin-1 motility on isolated cytoskeletons. However, more kinesin-1 motors bind to bundles compared to single microtubules, and kinesin-1 run lengths are increased on microtubule bundles. Interestingly, acetylated microtubules are highly bundled in COS-7 cells, suggesting that microtubule bundling contributes to kinesin-1’s preference for acetylated microtubules. The microtubule-associated protein tau binds both acetylated and nonacetylated microtubules with similar affinity. Tau is enriched on curved microtubules, where it strongly inhibits kinesin motility. Our findings suggest that the organization of microtubules influences the motility and localization of motor proteins and microtubule-associated proteins to direct intracellular transport.

Figure 1.

The motility of single kinesin motors is enhanced by MT bundling. (A) To isolate endogenous MT networks, cells were treated with detergent to remove the membrane whereas the MT network was stabilized with Taxol. Purified fluorescently tagged motor proteins were added with ATP onto the isolated cytoskeleton. (B) (Left) Time-lapse movies of GFP-conjugated kinesin-1 motility are captured with TIRF microscopy (Movie S1). (Middle) After the motility assay, MTs were fixed and stained with tubulin antibodies (acetylated tubulin shown in red and α-tubulin in green). Note that the 6-11B-1 antibody recognizes both acetylated and deacetylated MTs, but not filaments that have never been acetylated (59). (Right) Trajectories of kinesin-1 were tracked with subpixel resolution (∼25 nm) using TrackMate. (C) The MT cytoskeleton is shown in a living cell (pre-extraction) and after the extraction treatment, using SiR-tubulin (Spirochrome). (D) Zoomed images from yellow box in (B). The SD map of kinesin-1 was generated from the time-lapse movie. Single-molecule tracking produced the trajectories of kinesin (shown in different colors). Yellow lines on the SD map represent the regions of interest for drawing kymographs shown in (E). (E) Kymographs show a higher number of binding events for bundled MTs that are either acetylated or nonacetylated. (F) Kinesin-1 motors show increased run lengths, (G) binding times, and (H) number of runs on acetylated and nonacetylated bundles compared to the single MTs. For each cell, the values are normalized to the motility on nonacetylated single MTs (baseline). F–H) (Top plots) Colors represent different cells for a total of eight cells over seven experiments. Large black circles and error bars indicate the mean ± SE for all cells. The raw means for each cell for all plots are shown in Fig. S4A. (F and G) (Bottom) The normalized distribution of kinesin run lengths (F) and binding times (G) for all cells (mean in red, SEM in yellow). The grayscale represents the density of points with black as the highest density. (I) Off-axis displacement of runs of kinesin-1 does not differ on the single versus bundled MTs (p = 0.08) (SD). Acetylated single: n = 362, acetylated bundled: n = 4648, nonacetylated single: n = 2070, nonacetylated bundled: n = 1976 total runs (all cells, runs > 50 nm). Full-factorial two-way ANOVA was used to determine the effect of acetylation (p = 0.09) and bundling (p < 0.001) on kinesin-1 log-transformed run lengths (interaction term: p = 0.46), as well as on binding times. Posthoc Tukey test then compared the differences among the four MT populations as shown on the bottom plots: ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.001; N.S., no significance. To see this figure in color, go online.

Materials and Methods

Purified proteins

Recombinant rat kinesin-1 GFP430 protein was a gift from Dr. Gary Brouhard’s lab (Department of Biology, McGill University). Tau-3RS maleimide Alexa 568 was expressed in Escherichia coli and labeled as described (18).

Cell culture

COS-7 cells (American Type Culture Collection, Manassas, VA) were plated in MatTek glass-bottom dishes (coverslip 1.0) at low dilution in DMEM media (Thermo Fisher Scientific, Waltham, MA), supplemented with 10% (v/v) FBS (Thermo Fisher Scientific) and 1% (v/v) Glutamax (Thermo Fisher Scientific) and incubated for 24–48 h in 37°C incubator at 5% CO₂ to reach ∼40% confluence. Cells were washed with PBS 1× prewarmed at 37°C before detergent extraction.

Isolated microtubule networks stabilized by Taxol

Cells were treated with Extraction Buffer (1% (v/v) Triton X-100, 4% (w/v) PEG (MW: 35 000), 20 μM Taxol (Cytoskeleton, Denver, CO) in BRB100 buffer (100 mM PIPES, pH 6.9, 1 mM MgCl₂, 1 mM EGTA) for 5 min at room temperature (Fig. 1, A and C; Fig. S1). The extraction protocol was adapted from (25). After Extraction Buffer, the isolated cytoskeletons were washed with Taxol-BRB80 before performing the motility assay.

Isolated microtubule networks stabilized by GMPCPP

COS-7 cells were treated with GMPCPP Extraction Buffer (1% (v/v) Triton X-100, 8% (w/v) PEG, 200 μM GMPCPP in BRB100) for 5 min at room temperature. Note that the PEG concentration was increased from 4 to 8% and no Taxol was introduced in this system, replaced by GMPCPP (Jena Bioscience, Jena, Germany). In some experiments, Alexa 647-labeled tubulin (0.25 μM) purified from cow brain was included during the extraction step to image the incorporation of tubulin during extraction (Fig. S3 A). The Extraction Buffer was then removed partially from the sample, replaced with 8% PEG in BRB80 with 200 μM GMPCPP, and incubated for 1 h at 37°C. A second Extraction Buffer step (1% Triton X-100, 4% PEG in BRB100) followed for 3 min at room temperature. The samples were then washed with BRB80 before performing motility assays.

Microtubule imaging in living cells

Cells were incubated for 3 h in FluoroBrite DMEM (Thermo Fisher Scientific) complete media with 1 μM SiR-tubulin (Spirochrome; http://spirochrome.com/) and 20 μM verapamil (Spirochrome). Cells were imaged using epifluorescence before and after extraction.

Motility assay

After preparing isolated cytoskeletons, the buffer was replaced with motility mix composed of 0.2 mg/mL BSA, 10 mM DTT, 15 mg/mL glucose, ≥2000 units/g glucose oxidase, ≥0.6 units/g catalase, 2 mM ATP, and 200 pM kinesin-1 GFP430 in Taxol-BRB80 (for Taxol-microtubules) or in BRB80 (for GMPCPP-stabilized microtubules). All reagents were obtained from Sigma-Aldrich (St. Louis, MO) unless otherwise specified. Motility assays did not exceed 1 h in duration. For tau experiments, motility assays were first performed in the absence of tau. Then, the motility mix was removed and 1 nM tau-3RS Alexa 568 in Taxol-BRB80 was added onto the sample for an incubation time of 30 min to allow tau to bind (21). Motility mix with 1 nM tau was added on the extracted microtubules for imaging of kinesin-1 and tau molecules. Imaging was performed on an Eclipse Ti-E inverted microscope (Nikon, Melville, NY) with custom optics for total internal reflection fluorescence (TIRF), and imaged using EMCCD camera (iXon U897; Andor Technology, South Windsor, CT).

Retrospective immunostaining

After performing the motility assay, the microtubules were fixed using 4% paraformaldehyde for 10 min, blocked with blocking buffer (2% BSA in PBS 1×) for 15 min, and treated with antibodies for posttranslationally modified tubulin (6-11B-1 acetyl-α-tubulin; Thermo Fisher Scientific) and total tubulin (α-tubulin, Sigma-Aldrich) for 1 h. Alexa Fluor secondary antibodies (Thermo Fisher Scientific) were incubated for 30 min. Cells were washed between antibody changes with blocking buffer.

Determination of microtubule posttranslational modifications and bundling

Microtubules were categorized as 1) acetylated single, 2) acetylated bundled, 3) nonacetylated single, and 4) nonacetylated bundled microtubules. Briefly, the microtubules were initially classified as single or bundled based on the intensity in the TIRF images, where bundles are ∼2–4 times more intense than single microtubules (Fig. S2 D, see inset). Then, superresolution reconstructions were generated using kinesin localizations. These reconstructions were used to confirm the classifications of single and bundled microtubules, as the improved resolution allowed us to identify individual microtubule tracks within the bundles. In superresolution reconstructions, the intensity and width of microtubule bundles is significantly greater than for single microtubules (Fig. S2, A–C). To quantify the fraction of bundled microtubules in nonextracted and extracted cells, we applied a threshold to the TIRF images of microtubules at 1.5 the intensity of single microtubules and calculated the fraction of the area above this threshold. Further details are provided in Fig. S2.

Single-molecule tracking

Kinesin-1 GFP molecules were tracked using TrackMate (26). Using this plug-in, the fluorescent signal emitted from each motor molecule was first identified using a Laplacian of Gaussian filter, and then localized by fitting the point spread functions to a 2D Gaussian. Runs of single-molecules >50 nm were considered for further analysis. Thresholds for detection were set appropriately to capture signal from single molecules and minimize background signal. Then, the detected spots were tracked with the Linear Assignment Problem tracker and linked from frame to frame to trace trajectories of kinesin-1. The tracking parameters were set to assume a maximal velocity of kinesin-1 of 1 μm/s (Fig. S5), with gaps of two frames allowed to account for molecules momentarily not appearing within a 300-ms frame, estimated by cross-verification on kymographs. For assessment of tracking results, kinesin molecules were tracked on reconstituted microtubules to compare run length mean among similar systems from previous studies (Fig. S5).

Trajectory analysis

Kinesin-1 trajectories and the xy coordinates of microtubules of interest were imported into the software MATLAB (The MathWorks, Natick, MA) to analyze trajectories located on the different microtubule populations using code written in our laboratory. Run lengths were defined as the total displacement of kinesin upon binding until unbinding from the microtubule surface. The binding time was calculated as the total duration of the binding event, and the number of runs is defined by the number of binding events over a defined microtubule area over time. Off-axis displacement indicates the deviation of the signal’s position compared to the central axis (at 0 nm on x axis). Minimum value thresholds were set to filter trajectories with run lengths of at least 50 nm and minimum of four points in trajectories (equivalent to three frames). Run-length and binding-time distributions were fit to single exponentials using maximum likelihood estimates (27). To investigate the effect of acetylation, bundling, and tau, statistical analysis was performed with ANOVA designs and posthoc Tukey tests using RStudio (https://www.rstudio.com/). The exponential distributions for kinesin run lengths and binding times were log-transformed to normal distributions to meet the normality assumption of ANOVA.

Results

Microtubule bundling enhances kinesin motility

We isolated microtubule networks (Fig. 1 A) that retain many aspects of the organization and posttranslational modifications of the native cytoskeleton (Fig. 1 C; Fig. S1), and performed single-molecule tracking to examine how kinesin-1 navigates on different microtubule subpopulations (Fig. 1, A, B, and D; Movies S1, S2, S3, and S4). We compared kinesin-1 trajectories on acetylated and nonacetylated single microtubules on isolated cytoskeletal networks stabilized with Taxol, and find kinesin-1 run lengths and binding are largely unaffected by acetylation alone (p > 0.80; Fig. 1, F and H) in agreement with previous observations in purified in vitro assays (10, 11) and structural studies comparing acetylated and deacetylated microtubules (28).

In contrast, microtubule organization affects kinesin-1 motility when comparing trajectories on microtubule bundles to single filaments. Kinesin-1 exhibits longer run lengths and a larger number of runs on bundles of two or more microtubules compared to single microtubules (Fig. 1, E and H). Kinesin-1 showed higher mean run lengths (p < 0.05) and a greater than twofold increase in the number of runs (p < 0.01) on bundled microtubules relative to motility on single filaments (Fig. 1, F–H). The increase in the number of runs is consistent with the greater number of available binding sites in bundles of 2–3 microtubules (Fig. S2). We hypothesized that longer runs might be due to the ability to switch between multiple microtubules in a bundle. However, we do not observe an increase in off-axis displacements on microtubule bundles (Fig. 1 I), suggesting that switching between microtubules is rare. Previous studies suggest that the spacing of microtubules in a bundle affects transport; kinesin run lengths increased on MAP65-induced microtubule bundles but not bundles assembled randomly in the presence of crowding agents (19). Thus, our observations suggest that the microtubule bundles formed by cells are organized to allow robust motor protein transport.

To examine the effect of Taxol on kinesin-1 motility, we developed an alternative method to extract microtubule networks stabilized with GMPCPP (Fig. S3; Materials and Methods). Taxol binding to the microtubule filaments induces small conformational changes in the lattice (29, 30). As such, Taxol could potentially affect the influence of PTMs on kinesin motility and MAP binding (31). By stabilizing the microtubules with GMPCPP, this method aims to mostly retain the native structure and nucleotide state along the length of the microtubule while stabilizing the plus-ends (Fig. S3 A). We find that kinesin-1 assays on extracted GMPCPP-stabilized microtubules agree with our observations on Taxol-stabilized networks, showing enhanced motility of kinesin-1 on microtubule bundles (Figs. S3 and S4).

Acetylated microtubules are predominantly bundled

Kinesin-1’s preference for acetylated microtubules on isolated cytoskeletons is driven by bundling rather than a direct effect of acetylation on kinesin affinity. The subpixel resolution trajectories of single motor molecules were used to generate superresolution reconstructions of the microtubule network (Fig. 2 A; Fig. S2 A). Tubulin immunostaining and superresolution reconstructions were used to differentiate single and bundled microtubules (Fig. S2). Costaining of acetylated and α-tubulin demonstrates that most acetylated microtubules are localized in bundles (Fig. 2, A and B, red arrows). We find that 95 ± 2% of the acetylated microtubules are bundled (in extracted and nonextracted cells), compared to 43 ± 3% of nonacetylated microtubules (Fig. 2 C). Taken together, our observations of kinesin motility and microtubule organization indicate that although tubulin acetylation alone does not impact kinesin-1 run lengths or binding (Fig. 1), acetylated microtubules are often bundled, which results in enhanced kinesin-1 recruitment and run lengths.

Figure 2.

Acetylated MTs are often bundled in COS-7 fibroblast cells. (A) Acetylated MTs (left) colocalize with MT bundles as indicated by intense tubulin staining (right). Single MTs show low signal intensity (see Fig. S2 for detailed description of characterizing single and bundled MTs). The majority of acetylated MTs are bundled (red arrows indicate bundles). The superresolution image generated by kinesin binding on MTs shows that bundles are composed of two or more MT filaments (Fig. S2). (B) Lines were drawn on acetylated and nonacetylated MTs using the images in (A), which were then categorized into single or bundled MTs. (C) The fraction of acetylated and nonacetylated MTs in bundles (by length) were quantified as described in (B) for nonextracted cells (n = 3) and extracted cells (n = 3). An average of 95 ± 2 (mean ± SE)% of all acetylated MTs were found to be bundled in cells, compared to 43 ± 3% of nonacetylated MTs. The fraction of bundled MTs is similar in nonextracted and extracted cells (ANOVA, acetylated: p = 0.96, nonacetylated: p = 0.06). To see this figure in color, go online.

The microtubule-associated protein tau inhibits kinesin-1 motility on acetylated and bundled microtubules

To investigate whether MAPs might act as readers of the tubulin code by altering their binding in response to tubulin posttranslational modifications or bundling, we conducted motility assays on extracted microtubule networks in the absence and presence of the neuronal MAP tau. Tau stabilizes axonal microtubules and acts as an obstacle to inhibit kinesin-1 motility (18, 21, 32). In agreement with previous studies (33, 34), we observed that tau-3RS bound to extracted microtubule networks in both static and dynamic populations (Fig. 3 A). Tau binding was not affected by tubulin acetylation or bundling, and kinesin-1 run lengths decreased on all microtubule populations to a similar degree in the presence of tau (p < 0.01) (Fig. 3, B and C). We observe that binding times increased (p < 0.01; Fig. 3 D) and velocities decreased (Fig. 3 F) on tau-decorated microtubules. Kymographs indicate that motor proteins pause more often in the presence of tau (Fig. 3 B). These results suggest that inhibition of kinesin-1 motility by tau is not sensitive to microtubule bundling or acetylation (p > 0.25).

Figure 3.

Tau inhibits kinesin-1 motility and induces more frequent pausing. (A) At a concentration of 1 nM, individual tau-3RS patches (left) are seen binding and diffusing along MTs as shown on the SD map (middle) (Movie S5). Kymographs (right) show that tau exists in stable (kymograph 1 generated from yellow line 1) and dynamic populations (kymographs 1 and 2). (B) Kinesin-1 motility assays were performed on extracted MT networks in the absence (− tau) and presence of 1 nM tau (+ tau). Kinesin-1 kymographs show decreased displacement on both single and bundled tau-decorated MT populations. Kymographs also show a higher occurrence of pausing (yellow arrows) and immobile kinesin motors (red arrows) in the presence of tau. (C) Kinesin run lengths are decreased in the presence of tau for all four MT populations (p < 0.01), whereas (D) binding times are increased (p < 0.01). Colors represent different cells (four cells over three experiments, which were also used for Fig. 1 represented with the same colors). The mean of all experiments and the SE of the mean is shown in black symbols. For each cell, experimental means (C–E, left side) and distributions (C and D, right side) are normalized to nonacetylated single MTs without tau (baseline). Raw means are shown in Fig. S4E. (E) The number of kinesin binding events was unaffected by tau (p > 0.40). (F) Kinesin average velocities decreased in the presence of tau (p < 0.01). (G) Off-axis displacement of kinesin-1 on tau-decorated MTs does not differ on single versus bundled MTs (p = 0.12) (SD). No tau, acetyl single: n = 148; acetyl bundled: n = 1359; nonacetyl single: n = 748; nonacetyl bundled: n = 1158 total runs (all cells, runs > 50 nm). With tau, acetyl single: n = 89; acetyl bundled: n = 1493; nonacetyl single: n = 1214; nonacetyl bundled: n = 1932 total runs (all cells, runs > 50 nm). Three-way ANOVA tested the effect of tau, acetylation, and bundling as well as their interactions (p > 0.25 for all interactions) on log-transformed normalized kinesin run length and binding time and posthoc Tukey test assessed the differences between the MT populations. To see this figure in color, go online.

Tau is enriched on curved microtubule segments

Tau preferentially localizes to highly curved segments of microtubules in extracted cytoskeletal networks. At high concentrations of tau (10 nM), we observe increased tau localization at curved microtubule segments (Fig. 4 A). At low concentrations (1 nM), tau exhibits more static binding at highly curved regions of microtubules (Fig. 4 B). Tau has been shown to localize to curved microtubules in living cells (35), and other MAPs like doublecortin also preferentially decorate curved microtubules (36). Tau might recognize microtubule curvature, or enrichment might be due to the increased lattice defects or GTP islands present at bent regions (37). Kinesin-1 runs often end when they reach highly curved segments of microtubules that are heavily decorated by tau (Fig. 4 C), suggesting tau locally inhibits kinesin-1 motility in response to specific microtubule geometries.

Figure 4.

Tau accumulates on curved regions of MTs. (A) Tau enriches itself on MT segments with high curvature. (B) Time-lapse frame of 1 nM tau and SD map of time-lapse movie. (Left) Given here is zoom of area in yellow square and its kymograph. At low concentrations, tau exhibits longer residence times on curved MT segments of MTs (yellow arrows). (C) Given here are kymographs of kinesin-1 (green) and 1 nM tau (red) representing the area of MT filaments left and right of the curvature’s peak at the center (shown by yellow arrow). Kinesin-1 runs (shown with red arrow) end near the apex of the curve, corresponding with stably bound tau patches. To see this figure in color, go online.

Discussion

The microtubule cytoskeleton is organized into distinct subpopulations in the cell, defined by posttranslational modifications, microtubule-associated proteins, and network architecture. The signals encoded into the cytoskeleton regulate the motility of motor proteins to direct intracellular trafficking. Through analyzing the motility of single motors as they navigate isolated microtubule networks, we find that microtubule bundling enhances kinesin-1 motility. Kinesin exhibits longer run lengths on bundled microtubules, and microtubule bundles recruit more motors due to the increased number of binding sites. In agreement with previous in vitro studies, kinesin-1 is not appreciably affected by microtubule acetylation (10, 11). However, most acetylated microtubules are bundled, resulting in enhanced kinesin-1 transport, suggesting that microtubule bundling contributes to the selectivity of kinesin-1 for acetylated microtubules observed in cells (6, 7, 8).

Isolated cytoskeletons provide a useful tool to investigate the influence of the organization of the cytoskeletal network on motor motility, by allowing tight control over biochemical components and stoichiometry while closely approximating the native architecture of the cytoskeleton (Fig. 1; Fig. S1) (13, 24). The reduced background signal and ability to control the density of motor proteins allows high-resolution single-molecule tracking and determination of individual runs, which is challenging in living cells. However, detergent extraction requires that microtubules are stabilized with either Taxol or GMPCPP, which may alter kinesin motility (31) and results in increased levels of microtubule acetylation (Fig. S1) (8, 38). The increased levels of acetylation observed in the presence of Taxol and GMPCPP may reflect cellular mechanisms, where stabilized microtubules are preferentially acetylated (4). Microtubule stabilization by MAP overexpression in fibroblast cells was also shown to increase tubulin K40 acetylation (39). Although our observations indicate that microtubule bundling contributes to the selectivity of kinesin-1 for acetylated microtubules, it is likely other factors not present in the isolated cytoskeleton assays are also involved. Microtubule dynamics may play a role, as acetylated microtubules are less dynamic and exhibit more lateral fluctuations. Katrukha et al. (7) suggest that lateral fluctuations are due to recruitment of force generators to acetylated microtubules, which may also act to bundle these microtubules. A diverse set of microtubule-associated proteins, not present in these assays, may also contribute.

The greater number of kinesin-1 runs on microtubule bundles we observe is consistent with a greater number of available binding sites in bundles compared to single microtubules. However, the mechanism for increased run lengths along microtubule bundles is less clear, as kinesin-1 moves along a single protofilament on single microtubules (17, 18). Whereas off-axis displacements suggest that kinesin also moves along a single protofilament in microtubule bundles (Fig. 1 I), the tracking resolution does not allow us to rule out rare events where kinesin dissociates and rapidly rebinds to an adjacent microtubule. Further, Conway et al. (19) observed that kinesin-1 occasionally switched direction on mixed-polarity microtubule bundles formed in vitro, leading to longer runs. Alternatively, the microenvironment around microtubule bundles may be crowded, slowing kinesin detachment. Bundled microtubules may also be marked by posttranslational modifications in addition to acetylation that alter kinesin-1 affinity (8, 40, 41). Another possibility is that microtubules undergo subtle conformational changes when bundled, which in turn alter kinesin binding and motility.

The neuronal MAP tau enriches on segments of microtubules with high curvature. Static patches of tau locally inhibit kinesin-1 motility, and may act to direct kinesin-1 cargoes to straight bundled microtubules and promote efficient transport toward the cell periphery. Tau binding may also stabilize curved microtubules, which have been shown to be less stable than straight microtubules (42) or protect microtubules from severing by katanin (43).

Tubulin posttranslational modifications, the association of microtubule-associated proteins, and the architecture of the microtubule network are highly interdependent factors that interact to modulate microtubule stability and motor protein motility. MAPs such as tau cross-link microtubules into bundles (44). Microtubule stabilization by bundling and MAP binding then triggers posttranslational modifications (4, 45, 46). Bundled, acetylated microtubules often perform specialized functions in the cell such as in neuronal axons and primary cilia (41, 47), and acetylation has been proposed to protect long-lived microtubules from mechanical aging (48, 49). Feedback mechanisms among MAPs, PTMs, and microtubule bundling may enable the cell to control its cytoskeletal organization and the localization of microtubule motors and associated proteins.

We performed motor protein motility assays on isolated microtubule cytoskeletons to provide a bridge between in vitro assays on reconstituted microtubules and in living cells. We used these isolated microtubule networks to investigate how microtubule bundling, posttranslational modifications, and MAPs regulate kinesin transport. Here, we show that the motility of single kinesin-1 motors is strongly dependent on microtubule bundling. Further, in cells, vesicular cargoes and organelles are transported by teams of kinesin and dynein motors, and interact with multiple microtubules (50). In addition, dynein often switches protofilaments (51, 52) and is able to cross-link microtubules (16). Taken together, these results suggest that microtubule bundling and network organization regulate the motility of motor proteins and may provide a mechanism for the cell to establish preferred routes for transport.

Author Contributions

L.B. and A.G.H. designed research. C.L.B. contributed new reagents. L.B. performed the experiments. L.B. and A.G.H. analyzed data. L.B., A.G.H., and C.L.B. wrote the manuscript.

Editor: Steven Rosenfeld.

Supporting Citations

References (53, 54, 55, 56, 57, 58) appear in the Supporting Material.

Supporting Material

Movie S1. Time-Lapse of Kinesin-1 GFP Motility Assay on Taxol-Stabilized Extracted Microtubule Networks

See Fig. 1. 10× real-time.

Movie S2. Time-Lapse of Kinesin-1 GFP Motility Assay on Taxol-Stabilized Extracted Microtubule Networks

See Fig. 1. 10× real-time.

Movie S3. Time-Lapse of Kinesin-1 GFP Motility Assay on Taxol-Stabilized Extracted Microtubule Networks

See Fig. 1. 10× real-time.

Movie S4. Time-Lapse of Kinesin-1 GFP Motility Assay on Taxol-Stabilized Extracted Microtubule Networks

See Fig. 1. 10× real-time.

Movie S5. Time-Lapse of tau-3RS Alexa 568 on Taxol-Stabilized Extracted Microtubules

See Fig. 3 A. 10× real-time.