Kinesin-1 heavy chain mediates microtubule sliding to drive changes in cell shape

Lateral Microtubule Motion Is Independent of Microtubule Dynamics.

It is well established that microtubule dynamics can produce sufficient force to buckle a microtubule (21). To investigate whether the observed movement was caused by microtubule polymerization or depolymerization, we blocked microtubule dynamics by treating cells with paclitaxel. As long-term paclitaxel treatment can dramatically disrupt the microtubule network, we imaged cells between 5 and 30 min following the addition of 5 μM paclitaxel. This treatment did not cause microtubule reorganization, but was sufficient to block microtubule polymerization (Fig. S1). Strikingly, the lateral motion of microtubules was clearly observed when microtubule dynamics were suppressed, and was qualitatively similar to the motion observed in untreated cells (Movie S2).

Marking Microtubules with a Fiducial to Quantify Movement.

To elucidate the precise mechanism of microtubule movement, we developed a method to quantify the movement by applying a fiduciary mark to microtubules in Drosophila S2 cells. We created a fusion of Drosophila α-tubulin and an N-terminal fluorescent tag, photoconvertible protein Dendra2 (22), under the control of an inducible metallothionein promoter (pMT). Before photoconversion, the emission of Dendra2 has a characteristic peak at 505 nm, but following conversion with blue light the emission peak shifts to 575 nm. Using a line-scan confocal LSM510 microscope (Zeiss), and converting with a 405-nm diode laser, we limited the photoconversion to a small circular area approximately 5 μm in diameter between the cell nucleus and the periphery.

Performing the photoconversion in S2 cells allowed us to follow the movement of microtubule segments outside of the photoconverted area (Fig. S2). However, the fluorescence was rapidly lost from labeled segments as the photoconverted (red) tubulin was incorporated into newly forming microtubules elsewhere in the cytoplasm as a result of the microtubule dynamics. In paclitaxel-treated cells, the loss of fluorescence as a result of microtubule dynamics was prevented, allowing the observation of sliding over an extended period (Fig. 2 A and C and Movie S3). Our photoconversion experiments revealed that, rather than being the transport of small microtubule fragments along long filaments, entire long microtubule filaments undulate and buckle.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Microtubule sliding visualized and quantified using photoconversion. (A) S2 cell stably expressing pMT-Dendra2-α-tubulin before photoconversion shown in the green channel. The same cell in the red channels immediately after conversion (B) and 6 min 29 s later (C) demonstrates movement of the photoconverted segments beyond the converted area. (D–F) Cell treated with 5 μM cyto D 1 h before imaging and imaged 1 h after plating. Other conditions are identical to A–C, respectively. (G) Distribution of microtubule segment sliding velocities of 23 tracked segments from eight cells. Each point is equal to one velocity vector (i.e., the displacement between two consecutive frames). (H) Quantification of motile fraction of converted segments in 20 frames (15 s intervals) with or without cyto D treatment. Error bars indicate SEM. All cells were treated with 5 μM paclitaxel. (Scale bars, 5 μm.)

We determined the degree of microtubule movement relative to the total number of converted microtubule segments by calculating the motile fraction of the converted regions at the end of each 7-min time-lapse sequence (Materials and Methods). Photoconverting a circular region eliminated any angle bias in the analysis of microtubule movement. The mean motile fraction of converted microtubule segments was 0.36 (n = 10 cells), indicating that, on average, 36% of the total amount of fluorescent microtubule segments moved outside of the converted region. Tracking of the leading end of 23 microtubule segments from eight cells revealed that many segments spent most of the time not moving, but underwent sudden long-distance travel and were capable of moving during these bursts as fast as 13 μm/min (Fig. 2G). Other segments moved more constantly. The mean peak velocity of the 23 trajectories analyzed was 4.2 ± 0.5 μm/min (SEM).

Microtubule Movement Is Independent of Other Filament Networks.

Retrograde actin flow could be a potential source of pushing force driving microtubule movement in the cytoplasm. However, treatment of S2 cells with the actin-depolymerizing drug cytochalasin D (cyto D) did not inhibit microtubule motility (Fig. 2 D–F). Untreated S2 cells and those treated with cyto D had the same mean motile fraction (Fig. 2H). S2 cells lack an intermediate filament network (23), and when treated with cyto D, the microtubule network is the only remaining intact cytoskeletal network. Therefore, our results show that the microtubules are not moving against any other cytoskeletal filaments.

Microtubule Movement Is Not a Reaction to Cargo Transport.

We used the RNAi method to identify motor proteins that might directly move microtubules or indirectly cause reactionary microtubule buckling by moving cargo. Successful protein knockdown was confirmed by Western blot (Fig. 3A). Cytoplasmic dynein is involved in the microtubule-based transport of the majority of cargoes studied, from nuclear envelope components (24) to RNA complexes (25). Kinesin-1 is also involved in a significant amount of microtubule based transport, and binds the majority of its cargoes through the light chain. If the lateral microtubule movement we observe were a reactionary force to the movement of cargo, RNAi of either dynein or the kinesin light chain (KLC) would inhibit the movement. In fact, knockdown of KLC had no effect on the amount of microtubule motility observed (Fig. 3A and Movie S4), although this treatment inhibited the movement of membrane organelles along microtubules (26). Similarly, Klp68D (a Drosophila kinesin II subunit), another motor involved in cargo transport, did not influence microtubule motility (Fig. 3A). RNAi of the dynein heavy chain led to a small but statistically significant (P = 0.0406) increase in the motile fraction (Fig. 3A and Movie S5). As it is well established that dynein knockdown stops bidirectional cargo transport (13, 25, 26), these results show that cargo transport contributes very little, if at all, to the movement of microtubules in the cytoplasm.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

RNAi of conventional kinesin, but not other motors, results in the cessation of microtubule–microtubule sliding and prevents the formation of microtubule bundle–filled processes. (A) Motile fraction of converted microtubule segments in S2 cells subjected to RNAi-mediated depletion of motors and associated proteins. Time-lapse sequences taken every 15 s for 20 frames. *P < 0.05, **P < 0.00001. Student two-tailed t test was performed for independent samples assuming variance differs for each sample. Error bars indicate SEM. (Inset) representative Western blots. Dilutions of untreated (mock) cells provided to estimate degree of knockdown. (B) Replacement of endogenous KHC with exogenous KHC results in rescue of sliding. S2 cells stably expressing pMT-Dendra2-α-tubulin/pMT-KHC (full length) were induced for 1 wk and subjected to RNAi of endogenous KHC by targeting the UTR of the mRNA (3′UTR KHC) or RNAi of both endogenous and exogenous KHC by targeting the coding region (CDS KHC) for 96 h before microscopy. ‘WT’ cells are equivalent to mock cells in A. The motile fraction was determined after photoconversion. (Inset, Right) representative Western blot showing extent of KHC knockdown. Exogenous (pMT-KHC) kinesin is present after treatment with dsRNA against the 3′UTR. (C and D) Microtubule networks visualized in mCherry tubulin expressing S2 cells plated in 5 μM cyto D for 1 h. Control cells (C) form long processes containing microtubule bundles whereas KHC knockdown cells form irregularly shaped lamellas filled with straight and less bundled microtubules (D). (Scale bars, 10 μm.)

Together, these findings suggest that the movement we observe is explained by microtubules sliding against one another. To test this hypothesis, we treated stable mCherry tubulin–expressing S2 cells with the microtubule-depolymerizing drug colcemid to create short, linear microtubule fragments. We consistently observed the movement of apparently overlapping microtubule fragments slide against, rotate, and realign parallel to one another (Fig. 1 D and E and Movie S6). Long-term treatment of these cells with colcemid led to the formation of a few microtubule bundles including the entire set of original microtubule fragments (Fig. 1F). We therefore conclude that the microtubule movement we observe is the result of microtubule–microtubule sliding.

RNAi of Conventional KHC Eliminates Microtubule–Microtubule Sliding.

Recent studies suggest a role for molecular motors in driving microtubule buckling, and implicate the minus-end directed motor dynein in the majority of microtubule bending events (3, 5). Our results indicate dynein does not drive microtubule sliding, and suggest another motor may be responsible. In particular, the in vitro ability for kinesin-1 to slide microtubules against one another (16) and the existence of an ATP-independent C-terminal microtubule-binding domain in the kinesin heavy chain (18, 19) indicate that conventional kinesin may play a role in interphase microtubule sliding.

Accordingly, we knocked down the kinesin-1 heavy chain, which resulted in a dramatic inhibition of microtubule–microtubule sliding (Fig. 3A and Movie S7). Specificity of knockdown was verified using two independent dsRNA sequences against the 3′UTR region of KHC and the N-terminal coding region. In both cases, the motile fraction was dramatically reduced (P = 1.3 × 10⁻⁶) from 0.36 ± 0.03 in WT cells to 0.03 ± 0.02 in the 3′UTR KHC RNAi cells and 0.04 ± 0.02 in the KHC coding region RNAi cells (Fig. 3A). To verify specific knockdown, exogenous full-length KHC was stably expressed before RNAi of endogenous kinesin-1 using dsRNA against the 3′UTR region. The sliding phenotype was rescued to near-WT levels (0.29 ± 0.04) in these cells (Fig. 3B).

KHC Drives Process Formation in S2 Cells Treated with Cyto D.

We previously demonstrated that S2 cells plated on Con A–coated substrate in the presence of cyto D form long processes (27) (Fig. S3). Microtubules in the processes form parallel bundles with the plus ends at the process tips (27) (Fig. 3C). After 9 h, these processes are between 10 and 60 μm in length (n = 40). Time-lapse fluorescent imaging of S2 cells expressing mCherry tubulin plated in cyto D–containing media reveal the rapid formation of these processes, which grow at a maximal rate of 3.7 ± 0.6 μm/min (SD) just as the cells adhere to the surface (n = 10). Microtubule looping and entry into the growing processes in the boundary region apparently contributes to the growth of these processes (Movie S8).

Knockdown of KHC in S2 cells prevents the formation of these microtubule bundle–filled processes (Fig. S3). Microtubules do not form bundles or processes in KHC knockdown cells treated with cyto D. Instead, the cells remain spread; but, unlike an untreated WT cell, the microtubule network appears straight and absent of lateral motion (Fig. 3D). Furthermore, KHC RNAi prevents microtubule bundle formation in cells treated with colcemid (Fig. S4). Instead, the short microtubule fragments concentrate around the nucleus in an unorganized fashion. Kinesin-1 is known to be able to cross-link microtubules in vitro (16, 18) and localizes to microtubule bundles in cells (3). Together, these results indicate a role for KHC in sliding microtubule filaments against one another to drive the formation of the microtubule bundle–filled processes we observe in cyto D–treated cells. Moreover, the amount of force generated by kinesin-1 driven microtubule sliding is sufficient to deform the cell membrane and drive the formation of processes, suggesting that microtubule sliding provides more force than polymerization.

KHC-Driven Microtubule Sliding in Amphibian and Mammalian Cells.

Drosophila S2 cells have unique cytoskeletal networks. In addition to lacking intermediate filaments, they also lack interphase microtubule organizing centers (28). We wondered if the microtubule–microtubule sliding phenomenon we observed in S2 cells was a consequence of having free microtubules unhindered by “normal” cytoskeleton organizing structures and centrosome attachment of the minus ends. We demonstrate the occurrence of sliding in Xenopus fibroblasts by transiently transfecting with Dendra2-Xenopus α-tubulin and following the movement of photoconverted fragments outside of the converted region (Fig. 4 A and B and Movie S9). Additionally, time-lapse microscopy of GFP-tubulin–expressing rat kangaroo epithelial (Ptk2) cells treated with paclitaxel to inhibit microtubule dynamics reveals dramatic microtubule sliding and looping (Movie S10). Ptk2 cells were microinjected with a function-blocking SUK-4 antibody against the kinesin-1 heavy chain (29), along with a dextran tracer. SUK-4 injection caused all microtubule motion to cease, even in the absence of paclitaxel treatment (Movie S11). Six percent of injected cells displayed microtubule movement compared with uninjected or cells injected with control IgG (anti-myc antibody; Fig. 4C and Movie S12). We conclude that kinesin-1 drives sliding in mammalian cells in addition to Drosophila.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Microtubule sliding occurs in a variety of cell types, and is driven by KHC in Ptk2 cells. (A) Xenopus fibroblast transiently expressing Dendra2-α-tubulin immediately after photoconversion in a narrow bar; (B) the same cell approximately 5 min later. (Scale bars, 20 μm.) Arrows point to fluorescent microtubule segments that moved away from the converted region. (C) Percent of PtK2 cells injected with SUK-4 or a control antibody (mouse anti-myc) possessing moving microtubules as compared with uninjected cells (motility determined in a double-blind survey). (D) Model of kinesin-1–driven microtubule-microtubule sliding. (E) Alignment showing conservation of C-terminal microtubule binding domain (boxed region) identified in Drosophila KHC (19).