EBs Promote Catastrophe Induction by MT-Destabilizing MTAs.

To examine the effect of different MTAs on MT polymerization, we used an in vitro MT assembly assay in which MTs are grown from guanylyl (α,β)methylenediphosphonate (GMPCPP)-stabilized MT seeds immobilized on coverslips (13, 17) (Fig. S1 A–C). MT growth was initiated by the addition of a 15 μM mixture of unlabeled and rhodamine-conjugated tubulin (at a ratio 30:1) and observed by total internal reflection fluorescence (TIRF) microscopy within ∼30 min after tubulin addition. In these conditions, the formation of polymer mass is not sufficient to cause a significant change in the concentration of soluble tubulin, making this assay optimal for observing direct effects of different agents on MT plus-end dynamics. We examined how MT polymerization is affected by MT-destabilizing drugs, including the colchicine site-binding compounds colchicine, combretastatin A4, and podophyllotoxin, and the vinca site-binding drugs vinblasine, vincristine, vinorelbine, and dolastatin 15. We found that all these MTAs could prevent sustained MT elongation in the presence of tubulin alone (Fig. 1 A and B).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Effects of MT-destabilizing MTAs on MT dynamics. (A) Representative kymographs illustrating the dynamics of rhodamine-labeled MTs grown in the absence of GFP-EB3 (Left) or GFP-EB3 labeled MTs (Right) in the presence of MT-destabilizing MTAs. The positions of MT seeds are indicated by black lines above the kymographs. (B) MTA concentrations sufficient to inhibit MT outgrowth from the seed, so that no MT elongation exceeding 0.5 μm was observed within a 10-min period. (C–H) Length of MT growth episodes, MT growth rate, and catastrophe frequency at increasing concentrations of the indicated MTAs. Error bars represent SEM. Abbreviations: CA4, combretastatin A4; Col, colchicine; Dola, dolastatin 15; Podo, podophyllotoxin; VB, vinblastine; VC, vincristine; VR, vinorelbine.

To test how EBs modulate MTA activity, we next performed the same experiments in the presence of 10 nM GFP-tagged EB3. As readout, we used either rhodamine-labeled tubulin, as described above, or GFP-EB3 in the absence of fluorescent tubulin. Similar MT dynamics parameters were obtained in both conditions (Table S1), indicating that they are not affected by the use of different fluorescent markers. In agreement with our previous observations, the addition of 10 nM EB3 elevated both the MT growth rate and the catastrophe frequency by a factor of ∼1.5 (Table S1) (16). Next, we added destabilizing MTAs to the assay and found that the concentration of the drugs sufficient to inhibit sustained MT elongation was by a factor of 2–3 lower when MTs were grown in the presence of EB3 (Fig. 1 A and B), indicating that EB3 sensitizes MTs to the action of MTAs.

To understand how EB3 cooperates with MTAs, we measured parameters of MT dynamics using an automated kymograph analysis algorithm, which allowed us to analyze a large number of MT growth events (17, 18). We found that, in the absence of EB3, increasing concentrations of destabilizing MTAs decreased MT elongation primarily by reducing MT growth rate (Fig. 1 C, E, and G), as previously shown (19⇓⇓–22). In contrast, the catastrophe frequency increased only mildly, if at all (Fig. 1 C, E, and G, and Table S2). In the presence of EB3, only a mild change in MT growth rate was observed at drug concentrations close to the ones completely inhibiting MT outgrowth (Fig. 1 D and F and Table S3); instead, we saw a dramatic increase in catastrophe frequency (Fig. 1 A and H and Table S3). When the drug concentration was further elevated, very high catastrophe frequency induced a state whereby only extremely short MT outgrowth episodes at the plus end of the GMPCPP-stabilized seed could be seen (Fig. 1A and Fig. S1D). The same effect was observed in the presence of another member of the EB family, EB1 (Fig. S2A). These results demonstrate that destabilizing MTAs, irrespective of their binding site, increase the catastrophe frequency of EB-bound MTs at concentrations that do not affect MTs when the EBs are absent. Thus, EBs sensitize MTs by evoking MTA-induced catastrophes.

Next, we tested whether this effect can also be observed in cells. Using live imaging of MT plus-end dynamics in HeLa cells stably expressing EB3-GFP, we observed strong catastrophe induction by nanomolar concentrations of vinblastine in conditions when the MT growth rate was only mildly reduced (Fig. 2 and Table S4). This indicates that our in vitro observations reflect a physiologically relevant phenomenon.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Vinblastine increases MT catastrophe frequency in cells in a concentration-dependent manner. HeLa cells stably expressing EB3-GFP were treated with the indicated concentrations of vinblastine for 30 min and imaged with a 0.5-s time interval. (A) Live images of HeLa cells expressing EB3-GFP, either in control conditions or after 30 min of treatment with 5 nM vinblastine. Examples of MT tips used for making kymographs are indicated by arrows. (B) Kymographs illustrating the behavior of individual MT plus ends. (C and D) Quantifications of MT growth rate and catastrophe frequency. Error bars represent SEM.

MT-Destabilizing MTAs Accelerate MT Aging.

Previous work has shown that the catastrophe rate depends on MT age, as “younger” MTs have a lower probability to undergo a catastrophe than “older” ones (7, 8). This suggests that not one but multiple intermediate catastrophe-promoting steps have to occur during MT polymerization before the transition from growth to shrinkage takes place (7, 8). To test how MTAs affect MT aging in the absence or presence of EB3, we analyzed the precise distribution of MT growth times. If catastrophes ensue after several randomly occurring steps of equal rate, durations of MT growth episodes should follow a gamma distribution described by two parameters: the minimum number of intermediate steps (n) and the rate at which these steps occur (r).

The distributions of MT growth times in our assays could indeed be fitted by gamma distributions (Fig. 3 A and B). We determined the parameters n and r for different conditions and found that they were only mildly affected by the addition of EB3 alone or by the addition of destabilizing drugs to MTs grown in the absence of EBs (Fig. 3 C–F and Table S5). The number of steps was between 3 and 4.5 in different conditions, which is comparable with the values determined previously for tubulin alone (7). In the presence of EB3, increasing concentrations of MT-destabilizing MTAs had no strong effect on the step parameter, whereas the value of the rate parameter was elevated (Fig. 3 D and F, and Table S5). Thus, EB3-bound MTs age more rapidly in the presence of MTAs.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Analysis of the duration of MT growth episodes. (A and B) Cumulative distributions of MT growth times (times from MT outgrowth to catastrophe) in the presence of different concentrations of podophyllotoxin with and without GFP-EB3 (solid lines). Gamma fits of the distributions are shown by stippled lines. (C–F) Gamma distribution parameters n (number of intermediate states) and r (rate of occurrence of intermediate states) in different conditions, as indicated. Abbreviations are the same as in Fig. 1C. Error bars indicate SEM.

Accumulation of EB Proteins at the MT Tips Is Affected by MT Aging.

Although the mechanistic details of the catastrophe-promoting steps are unknown, one possibility is that MT lattice defects arise during polymerization and are propagated so that the MT tip structure is altered (Fig. 4A). Because EB proteins bind to the stabilizing cap (13, 14), EB accumulation at the MT end could potentially reflect such age-related changes. To test this hypothesis, we measured the intensity of GFP-EB3 signal at MT plus ends during MT growth and observed that, in addition to the increase during growth initiation and a steep decrease before the catastrophe, EB3 comet intensity fluctuated substantially (Fig. 4 B and C). These fluctuations could reflect the occurrence of transient MT lattice defects or other alterations in the MT tip structure, which get corrected while the MT grows. Importantly, in addition to these fluctuations, we also observed that the EB3 signal at the MT plus end gradually decreased during the growth event, which could be observed both in control conditions (relatively long growth episodes) and was even more obvious in the presence of MTAs (substantially shorter growth episodes; Fig. 4 B–D). A similar decrease in signal intensity at the MT end before the catastrophe has previously been observed in reconstitution experiments with the fission yeast homolog of EB3, Mal3 (14). The gradual decrease of the EB comet intensity at growing MT ends supports the view that catastrophe induction is a gradual multistep process.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Analysis of EB3 comet intensity and the effects of MT-stabilizing MTAs on MT dynamics. (A) Scheme illustrating the model of MT aging by propagating MT lattice defects. (B–D) Analysis of GFP-EB3 intensity at MT plus ends in control samples and in the presence of 150 nM dolastatin. (B) Representative MT growth events. (C) Normalized GFP-EB3 intensity profiles for the events shown in B. (D) Averaged profiles of normalized GFP-EB3 intensity (n = 16 for both plots) where time was rescaled so that the duration of each growth episode is equal to 1 (black lines; error bars represent SEM). To illustrate the general trend, the mean values were smoothened using a sliding window of size 10 (green lines). (E and F) Gamma distribution parameters n (number of intermediate states) and r (rate of occurrence of intermediate states) in different conditions, as indicated. Catastrophe times were calculated separately for MT growth events originating from the seed or a stable rescue point in the presence of paclitaxel, and after a random MT rescue. Stable rescue points were defined as MT lattice sites where multiple rescues (more than three) occurred. Error bars represent SEM. (G) Representative kymographs illustrating MT dynamics in the absence (Left) or presence of GFP-EB3 (Right) and MT-stabilizing MTAs. The stippled white lines indicate the positions of stable rescue points. (H and I) MT catastrophe and rescue frequency as a function of concentration of the indicated MTAs. Error bars indicate SEM. (J) Schematic summary of the effects of MTAs on MT dynamics in the absence and presence of GFP-EB3.

Fewer Steps Are Needed to Cause a Catastrophe of a MT After a Rescue.

A consequence of the “MT defect propagation” model is that, during MT depolymerization, the “memory” of the MT defects will be obliterated. However, if a MT is rescued, some of the propagated defects could persist (Fig. 4A), and fewer steps would be required to cause a new catastrophe. To test this prediction, we examined the distribution of growth times upon rescues and found that the number of steps occurring before rescued MTs underwent a catastrophe was approximately twofold lower than for MTs growing from the GMPCPP-stabilized MT seed in the same conditions (Fig. 4 E and F). This result demonstrates that rescued MTs display distinct properties with respect to catastrophe: they behave as if they are more aged than those originating from the seed.

MT-Stabilizing MTAs Promote Both Catastrophes and Rescues in the Presence of EBs.

To obtain additional insights into the relationship between catastrophes and rescues, we turned to MT-stabilizing drugs, which we expected to be rescue inducers. We used paclitaxel, which binds to the taxane site, and peloruside A, which binds to a so far poorly characterized tubulin site (23). At low concentrations, the two drugs had no strong effect on the MT growth rate (Table S2). In the absence of EBs, both drugs suppressed catastrophes: no catastrophes at all could be detected in the presence of 100 nM paclitaxel and 50 nM peloruside A (Table S2). However, upon the addition of EB3, the drugs acted in the opposite way: a mild but significant increase in catastrophe frequency was observed (Fig. 4 G and H and Tables S3 and S6). These results are consistent with a previous study of the effect of paclitaxel and epothilone B on EB3-decorated MTs (24). For paclitaxel, catastrophe induction was more obvious for lower concentrations such as 100 nM than for higher concentrations (Fig. 4H). Thus, EB3 alters the MT growth properties in such a way that even MT-stabilizing agents promote rather than suppress catastrophes.

Despite the slightly increased catastrophe frequency in the presence of EB3, the MT-stabilizing drugs did not inhibit MT elongation because they also strongly stimulated rescues (Fig. 4 G and I). This effect was not specific to EB-bound MTs as it was also observed at 50 nM paclitaxel in the absence of EB3 (Table S2; at higher concentrations catastrophes, and hence rescues, were absent). Interestingly, in addition to randomly occurring rescues, the drugs also induced the appearance of “stable rescue points”—MT sites located at a significant distance from the seed that initiated repeated rescue events (Fig. 4G and Fig. S2B). To examine the difference between stable and random rescue points, we measured the number of steps before catastrophe (Table S7). We found that, for MTs originating from stable rescue points, this was similar to the number of steps observed for MTs growing from seeds, whereas catastrophes upon random rescue required fewer steps (Fig. 4 E and F). Thus, besides promoting random rescues, MT-stabilizing agents can introduce into the MT lattice specific sites where the consequences of the propagating MT aging process are eliminated.