Assembly of MTs, zinc-sheets and zinc macrotubes

Three types of tubulin polymers were assembled and examined by electron microscopy (EM) (Fig. 1C). Protofilament arrangements were also confirmed by their diffraction patterns (Fig. 1C). Based on EM images, the width of a zinc-sheet was 200±63 nm (± SD) (Fig. 1D) and the length was 3.3±1.6 µm (± SD) (Fig. 1E). The width of a MT was 30±2 nm (± SD, N = 20) and that of a zinc-macrotube was 293±18 nm (± SD, N = 20).

In a previous study, zinc-sheets under certain conditions often possessed a MT-like structure along one edge [28]. Therefore, we took utmost care so that polymerized zinc-sheets did not contain MT structures by reviewing the polymerizing conditions of Zn²⁺ and GTP concentrations, pH, incubation time and tubulin freshness. We mostly observed thin and folded ribbon-like structures, but did not find MTs or MT-like structures associated at the zinc-sheet edges in EM general views (Fig. 1C). As a control, authentically polymerized MTs were confirmed by EM (Fig. 1C).

However, there was a possibility that MTs polymerized in our zinc-sheet preparations and rarely resided on the carbon surface of an EM grid. To investigate this issue, we performed the following experiment. Cy5-labeled zinc-sheets and BODIPY FL-labeled MTs were polymerized individually, and then both polymers were mixed so that similar numbers of each were recognized by fluorescence microscopy (Fig. S1 A). The mixed solution was deposited onto a carbon grid for EM, and we observed a similar number of each polymer (Fig. S1B). Therefore, the possibility of hidden MTs was eliminated. Thus, we concluded that our zinc-sheet preparations did not contain MTs.

Another detail we addressed was zinc-sheet stability, because zinc-sheets tended to depolymerize in the motility assay solution with tubulin and Zn²⁺ concentrations lower than those under the polymerization condition. To prevent depolymerization, we stabilized zinc-sheets by fixation with glutaraldehyde (GA) for analysis in motility assays if necessary. The tubulin cross-linking ratio in the GA-treated zinc-sheets [zinc-sheets (GA)] was 46%, and GA-treated MTs [MTs (GA)] was 39% (Fig. S2). The formation of zinc-sheets (GA) was confirmed as identical to that of zinc-sheets in EM images.

Gliding movements of MTs and zinc-sheets

In motility experiments, we examined whether tubulin polymers exhibited gliding on a glass surface coated with cytoplasmic dynein molecules purified from porcine brain or recombinant rat kinesin-1 dimers fused with an Avi-tag (RK430-Avi) (Fig. S2). Zinc-sheets as well as MTs moved on both surfaces (Fig. 2, and Videos S1, S2, S3, S4, S5, S6, S7, S8). Although MT gliding trajectories were mostly straight (Fig. 2A, upper), numerous zinc-sheet gliding trajectories were curved or circular (Fig. 2A, lower). Mean curvatures of the zinc-sheet trajectories on dynein and kinesin (Fig. S3C and G) were slightly larger compared with those of the MTs (Fig. S3A and E). The curved zinc-sheet trajectories were distinct from the straight MT trajectories, suggesting that the flexible ribbon-like structure of zinc-sheets can be driven by dynein and kinesin. Both GA-fixed polymers were confirmed to move on dynein- and kinesin-coated surfaces, but with decreased gliding velocities (Fig. 2B). Notably, the gliding fraction and velocity of a zinc-sheet (GA) on the dynein surface was greatly decreased, probably due to the folded and tangled structures of the zinc-sheet (GA) (Fig. 2B). In contrast, zinc-macrotubes bound to the dynein- and kinesin-coated surfaces, but did not show gliding movement. These results indicate that protofilament arrangements on the zinc-macrotube surface cannot support dynein and kinesin movement. Thus, the observed zinc-sheet gliding movement should be supported by a single protofilament at one edge of the sheet that exposes the motor domain binding sites.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. MT and zinc-sheet gliding movements on dynein- and kinesin-coated glass surfaces.

(A) Movement trajectories of tubulin polymers plotted by positions marked every 3 s from the starting points (open circles). MT movements on dynein (upper left) and RK430-Avi (upper right). Zinc-sheet movements on dynein (lower left) and RK430-Avi (lower right). (B) Velocities of the gliding movement. Bars represent the mean ± SEM, and (N) is the number of measured gliding tubulin polymers.

https://doi.org/10.1371/journal.pone.0042990.g002

To examine whether a zinc-sheet provides only one protofilament as a track for motor proteins, we observed zinc-sheets (GA) decorated with monomeric kinesin molecules [RK354 (G234A)] (Fig. S2), which bind to microtubules with high affinity, but is deficient in ATP hydrolysis and does not move on microtubules [29]. As shown in Figure S4A, kinesins appeared in a line at the far left protofilament. To confirm the kinesin repeat, diffraction patterns were obtained from the areas containing several protofilaments (Fig. S4B–G). Previously, MTs decorated with kinesins have shown a clear patterns of 8 nm longitudinal repeats [30]–[33]. Although the diffraction patterns of MTs do not show spots at the 8 nm layer line, those of zinc-sheets show weak spots at the 8 nm layer line (Fig. 1C) [25]. The areas containing the left edge of the kinesin-decorated zinc-sheet showed strong spots at the 8 nm layer line (Fig. S4B and C), and these 8 nm spots represented the kinesin repeat on the far left protofilament. The other areas without the left edge (Fig. S4D and E) or containing the right edge (Fig. S4F and G) showed obscure 8 nm spots, as did both edge areas of a non-decorated zinc-sheet (Fig. S4H and I). These results demonstrate that kinesin dominantly binds to one protofilament at one edge of a zinc-sheet, suggesting that a zinc-sheet contributes a single protofilament track for kinesin. Because dynein and kinesin binding sites on the tubulin surface overlap around the helix 12 of β-tubulin [21], a zinc-sheet also contributes a single protofilament as a track for dynein.

Single molecule motility on MTs and zinc-sheets

Next, a single molecule motility assay with dynein and kinesin was performed on the tubulin polymers. For this purpose, we used a recombinant yeast dynein dimer (GST-Dyn1_331kDa) and a rat kinesin dimer (RK430) fused with GFP to visualize under a total internal reflection fluorescence (TIRF) microscope (Fig. S2). GFP-GST-Dyn1_331kDa is truncated at the tail domain, dimerizes via GST and walks on MTs processively at a similar velocity as that of full-length yeast dynein [20]. In gliding assay, porcine brain dynein was used, however, single molecules of porcine brain dynein do not walk processively and unidirectionally on MTs [34]–[36]. Therefore, this dynein cannot be used for a single molecule motility assay. Furthermore, because the zinc-sheets were liable to depolymerize during the TIRF microscopy observation, zinc-sheets (GA) were used. Under these conditions, single molecules of both dynein and kinesin moved processively and repeatedly on zinc-sheets (GA) (Fig. 3A, and Videos S9 and S10), and approximately half of the zinc-sheets (GA) in views contributed as tracks for the processive movement of dynein and kinesin. This track frequency (∼50%) was much higher than that of possible imperceptible MT contamination in our zinc-sheet preparations mentioned above. Therefore, dynein and kinesin were considered to move on zinc-sheets. Meanwhile, dynein and kinesin bound to one tip of a zinc-macrotube and were not observed to move (Fig. 3A). The result is consistent with that an exposed helix 12 protofilament is estimated to be approximately 70 nm at one tip of the zinc-macrotube, and that this length is not long enough to detect the processive dynein and kinesin movement in our assay system. Occasionally, dynein and kinesin also nonspecifically bound to some parts of zinc-macrotubes and did not move (Fig. 3A). Therefore, it appeared that the surface lattices of zinc-macrotubes and zinc-sheets did not support dynein and kinesin movement. From these observations, we concluded that dynein and kinesin walked on only one protofilament at the edge of zinc-sheets.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Single molecule motility of dynein and kinesin on tubulin polymers.

(A) Tubulin polymers (left panels) and kymographs of GFP-GST-Dyn1_331kDa and RK430-GFP movements (right panels). (B) Mean-square displacement (MSD) plots of GFP-GST-Dyn1_331kDa (left) and RK430-GFP (right) movements on MTs, MTs (GA) and zinc-sheets (GA) with fitted quadratic curves (solid line). Each plot represents the mean ± SEM. (C) Velocities of GFP-GST-Dyn1_331kDa (left) and RK430-GFP (right) movements on MTs, MTs (GA) and zinc-sheets (GA) determined from MSD plots. Bars represent the mean ± SEM.

https://doi.org/10.1371/journal.pone.0042990.g003

The velocities of the dynein movements on MTs, MTs (GA) and zinc-sheets (GA) were very similar [MT: 36±1 nm/s, MT (GA): 35±2 nm/s, zinc-sheet (GA): 35±2 nm/s] (Fig. 3C, and Table S1). These values were lower than that reported previously on MTs (102±32 nm/s) [20], because we used an assay buffer with a lower pH (pH 6.8) and lower salt concentration (25 mM potassium acetate) to improve the zinc sheet stability and the frequency of dynein movement. Additionally, it is known that the gliding velocity of MTs on yeast dyneins [20] is much slower than that on porcine brain dynein (Fig. 2B). Therefore, the dynein velocities obtained here are considered to be reasonable values, and the change of the assay conditions (lower pH and lower salt concentration) did not appear to significantly impair the dynein walking mechanism. It is note worthy that the dynein velocities on these three types of tracks were almost the same, suggesting that dynein motility is unaffected by GA fixation and the single protofilament available for movement.

We could not determine the statistical walking length of dynein and kinesin on zinc-sheets because EM showed infrequent tears and breaks in zinc-sheets, resulting in difficulty to assess whether dynein and kinesin were stochastically dissociated or structurally interrupted. However, the observed walking lengths were ∼500 nm (Table S1), indicating that both dynein and kinesin took more than 50 steps on a single protofilament.

Although the velocities of dynein were similar on MTs, MTs (GA) and zinc-sheets (GA), those of kinesin decreased with GA fixation (Fig. 3B and C). This difference may be attributed to dynein and kinesin binding sites on the tubulin surface, which overlap but are not identical. The velocity data suggest that kinesin is susceptible to GA fixation of the tubulin polymers. Moreover, it is possible that cross-linked sites in zinc-sheets after GA fixation are different from those in MTs, because their tubulin arrangements are different from each other. This difference in cross-linked sites might enable kinesin to move faster on zinc-sheets (GA) than on MTs (GA).

The velocities of dynein and kinesin on zinc-sheets are considered to be comparable to those on MTs. These properties of single molecule velocities were different from those of gliding velocities, which was probably due to the gliding movement being affected by the stiffness and distortion of the polymers and the number of motor proteins.

Protein Preparation

Porcine brain as starting material to recover tubulin and dynein was purchased from Tokyo Shibaura Organ Co., Ltd. Other proteins were all recombinantly expressed and purified from E. Coli or S. Cerevisiae.

Purification methods are described in Materials and Methods S1.

Assembly of MTs, zinc-sheets and zinc-macrotubes

MTs were polymerized from purified tubulin (3 mg/ml) as described previously [39]. Zinc-sheets were polymerized from 1 mg/ml tubulin in Zn²⁺ polymerization buffer (100 mM Pipes-KOH pH 6.1, 1 mM MgSO₄, 0.5 mM EGTA, 4 mM GTP, 0.1 mM DTT, and 0.8 mM ZnSO₄) by incubation for 1 h at 32°C. Zinc-sheets were stabilized with 20 µM paclitaxel. At pH 6.1, zinc-sheets are formed first and then roll up to form zinc-macrotubes after incubation for 10 h [26], [27]. Although zinc-sheets at pH 6.1 have a tendency to bend rather than remain straight, they were more resistant to depolymerization as compared with those at pH 5.5–5.9. Zinc-macrotubes were polymerized in the same manner as zinc-sheets, except the incubation was performed at 30°C for 20 h. Fluorescently labeled tubulin polymers were prepared by copolymerizing fluorescent tubulin and unlabelled tubulin at ratios of 1∶15–30.

The tubulin polymers were fixed with GA to maintain their stability for observation. Tubulin polymers were diluted with the polymerization buffer or Zn²⁺ polymerization buffer to 10%, and then the diluted solution was combined with 0.1% GA (G5882, Sigma). After incubation for 5 min at room temperature (RT), 100 mM Tris-Acetate (pH 7.5) was added to stop the cross-linking reaction. Fixed tubulin polymers were layered on a 20% sucrose cushion (10 mM Pipes-KOH pH 6.8, 50 mM potassium acetate, 4 mM MgSO₄, 1 mM EGTA, 20 µM paclitaxel, 0.1 mM DTT, and 20% sucrose) and centrifuged at 14,000 ×g for 10 min at 27°C. The pellet was rinsed with 50 KAce assay buffer (10 mM Pipes-KOH pH 6.8, 50 mM potassium acetate, 4 mM MgSO₄, 1 mM EGTA, 20 µM paclitaxel, and 0.1 mM DTT) and resuspended in the same buffer. Tubulin cross-linking ratios in tubulin polymers were determined by densitometry of SDS-PAGE using Quantity One (Bio-Rad).

Electron Microscopy

Ten microliters of the tubulin polymers was applied to carbon-coated copper grids, stained with 1.5% uranyl acetate, and then observed under a transmission electron microscope (H-7500, HITACHI) with an accelerating voltage of 80 kV. Film images were scanned with a film scanner (LS-9000 ED, Nikon). The diffraction patterns were obtained by Image J.

In the zinc-sheet and kinesin binding experiment, polymerized zinc-sheets (GA) (10% final dilution) were mixed with RK354 (G234A) (0.13 µM final concentration) in 50 KAce assay buffer for 3 min.

Gliding Assay of Tubulin Polymers

Gliding assays were performed in 50 KAce assay buffer supplemented with 1 mM ATP under a dark field microscope, as described previously [40]. Glass surface were coated with porcine brain cytoplasmic dynein and blocked by 5 mg/ml bovine serum albumin (BSA) (A0281, Sigma), or coated with biotinylated BSA (A6043, Sigma) to tether streptavidin (Wako) to bind RK430-Avi and then blocked with 7 mg/ml casein.

The ends of tubulin polymers were manually marked every 3 s using “Mark 2” software [41]. The curvatures of the gliding trajectories (µm⁻¹, κ) were calculated according to κ = 1/R (where R represents a radius of curvature) in 5 µm (Fig. S3A–C and E–H) or 2 µm (Fig. S3D) fractions of each trajectory. The mean curvature was determined using nonlinear fitting of the cumulative probability distribution [41].

Single Molecule Motility Assay on Tubulin Polymers

Fluorescently labeled tubulin polymers and GFP fusion motor proteins were observed by TIRF microscopy as described previously [41] at 23–25°C. The coverslips were silanized [42], and flow cells were constructed [41]. The flow cells were first coated with a 0.5% anti-β-tubulin monoclonal antibody solution (T5201, Sigma), allowed to absorb for 5 min, and then blocked with 1% Pluronic F-127 (P2443, Sigma) for 5 min, followed by 7 mg/ml casein (2242, Merck) for 5 min. Fluorescent tubulin polymers were then introduced and incubated for 5 min. After washing with 25 KAce assay buffer-(10 mM Pipes-KOH pH 6.8, 25 mM potassium acetate, 4 mM MgSO₄, 1 mM EGTA, 20 µM paclitaxel, and 0.1 mM DTT), flow cells were filled with 25 KAce assay buffer containing 2–10 pM GFP fusion motor protein, 0.7 mg/ml casein, 0.1% Tween 20, 0.2% glucose, 1% 2-mercaptoethanol, 85 U/ml glucose oxidase (G2133, Sigma), 1,300 U/ml catalase (106810, Roche), and 1 mM ATP.

The positions of GFP fusion motor proteins were determined by a two-dimensional Gaussian fitting algorithm with “Mark 2” software [41]. For RK430-GFP, all runs were classified into moving and non-moving fractions by the variance per duration of interaction (VD; nm²/s) of 1,500 nm²/s for each run as described previously [43]. For GFP-GST-Dyn1_331kDa, the VD was 700 nm²/sec calculated by tracking GFP-GST-Dyn1_331kDa absorbed non-specifically on the glass surface. Mean-square Displacement (MSD; ρ(τ)) was plotted by averaging the squared displacement for non-overlapping intervals τ. The velocity and the diffusion coefficient were determined by fitting ρ(τ) with ρ(τ)  = 2Dτ + v2τ2 (where D represents the diffusion coefficient, and v is the drift velocity) [41].

Microscopy image acquisition

A dark-field microscope (BX-51, Olympus) equipped with a 403/0.75 objective lens (UPlanFl, Olympus) was used for image acquisition. Images were magnified by a 53 TV adaptor, acquired with an intensified charge-coupled device (CCD) detector (ICD-6100, Ikegami), and recorded onto a hard disk via an analog/digital video converter (ADVC300, Canopus).

The TIRF microscope (IX71, Olympus) was equipped with a 1003/1.45 objective lens (PlanApo, Olympus). Images were magnified by IX2-SPT (Olympus), and captured by a back-illuminated CCD detector (DV887DCS-BV, Andor). GFP, BODIPY-FL were excited with a 488 nm argon ion laser (IMA101, Melles Griot) and Cy5 were excited with a 632.8 nm He-Ne laser (05LHP991, Melles Griot). The exposure time was 140 ms.