The Dis1/XMAP215 family of microtubule-associated proteins conserved from yeast to mammals is essential for cell division. XMAP215, the Xenopus member of this family, has been shown to stabilize microtubules in vitro, but other members of this family have not been biochemically characterized. Here we investigate the properties of the Saccharomyces cerevisiae homologue Stu2p in vitro. Surprisingly, Stu2p is a microtubule destabilizer that binds preferentially to microtubule plus ends. Quantitative analysis of microtubule dynamics suggests that Stu2p induces microtubule catastrophes by sterically interfering with tubulin addition to microtubule ends. These results reveal both a new biochemical activity for a Dis1/XMAP215 family member and a novel mechanism for microtubule destabilization.
Microtubules are semirigid tube-like polymers built up from tubulin subunits composed of a 1:1 heterodimer of α- and β-tubulin. Microtubules in vitro and in vivo show an assembly property known as dynamic instability in which microtubule ends stochastically switch from states of growing to shrinking (an event called catastrophe) and from shrinking to growing (an event called rescue). Microtubule polarity is reflected in the difference in polymerization dynamics at the two ends, with the plus end growing faster than the minus end (for review see Desai and Mitchison, 1997).
Although dynamic instability is an intrinsic property of microtubules growing in vitro from pure tubulin, there are significant differences in the dynamic properties of microtubules in vitro and in vivo. For a given concentration of tubulin, the growth rate and the catastrophe frequency are much higher in vivo than in vitro. Furthermore, the growth rate and the catastrophe frequency of microtubules vary both during development and through the cell cycle. Thus, it is thought that microtubule dynamics in vivo are influenced by additional cellular factors (for reviews see Desai and Mitchison, 1997; Kinoshita et al., 2002). Such factors can be roughly placed in one of two functional families: those that promote microtubule assembly, such as XMAP215, tau, or MAP4, and those that inhibit microtubule assembly, such as members of the KinI family, Op18/Stathmin, or katanin. The mechanisms by which microtubule stabilizers and destabilizers modulate microtubule dynamics are different. Neuronal microtubule-associated proteins like tau, MAP1, or MAP2 are thought to cross-link adjacent tubulin subunits along the microtubule lattice and thereby suppress microtubule destabilization (for review see Desai and Mitchison, 1997). Other microtubule-associated proteins such as EB1 or CLIPs bind to microtubule ends where they directly or indirectly stabilize microtubules by acting as anticatastrophe factors (Brunner and Nurse, 2000; Akhmanova et al., 2001; Nakamura et al., 2001; for review see Schuyler and Pellman, 2001). Some destabilizers such as Op18/Stathmin work by sequestering free tubulin or through directly modulating microtubule structure (Belmont and Mitchison, 1996; Curmi et al., 1997; Jourdain et al., 1997; Howell et al., 1999). Others such as the KinI family of destabilizers or katanin require the energy of ATP hydrolysis to destabilize microtubules (McNally and Vale, 1993; Desai et al., 1999).
Recently, one family of essential microtubule stabilizers, the Dis 1/TOG family, has emerged as a putative master regulator of microtubule dynamicity (for reviews see Ohkura et al., 2001; Kinoshita et al., 2002). Members have been found in every organism where investigated, and it is currently the only known family of microtubule-associated proteins common to plants, fungi, and animals. XMAP215, from Xenopus laevis, is the best studied member of this family (Gard and Kirschner, 1987; Vasquez et al., 1994; Tournebize et al., 2000; Cassimeris et al., 2001; Kinoshita et al., 2001; Popov et al., 2001, 2002). In cytoplasmic egg extracts, XMAP215 stabilizes microtubules primarily by antagonizing the activity of the microtubule destabilizer XKCM1, a member of the KinI family (Tournebize et al., 2000). Thus, immunodepletion of XMAP215 decreases the steady-state length of microtubules. Similarly, zyg-9 mutants in Caenorhabditis elegans (Matthews et al., 1998) and msps mutants in Drosophila also seem to decrease the steady-state length of microtubules (Cullen et al., 1999; Lee et al., 2001). Together, these studies suggest an evolutionarily conserved role of this family in stabilizing microtubules. However, analyses of Stu2p, the Saccharomyces cerevisiae member of this family, have complicated the general conclusion that proteins from this family are microtubule stabilizers. In in vivo, run-down experiments strongly reduced Stu2p levels lead to less dynamic microtubules, with a lower catastrophe frequency (Kosco et al., 2001). These results suggest that the role of Stu2p in yeast is to destabilize microtubules. On the other hand, a more indirect study showed Stu2p to be required for spindle elongation (Severin et al., 2001).
It is unclear whether this difference in activity between the S. cerevisiae and the Xenopus members of the family represent fundamental differences in the biochemical properties of the proteins or whether the proteins are acting in different cellular contexts. The stabilizing activity of XMAP215 in vivo is reflected in its effect on microtubule dynamics in vitro where it acts as a microtubule stabilizer: when added to pure tubulin, XMAP215 stimulates the plus end growth rate 7–10-fold (Gard and Kirschner, 1987; Vasquez et al., 1994). However, the biochemical activity of Stu2p on purified tubulin growing in vitro has not been measured.
To investigate the biochemical activities of Stu2p, we sought to make pure recombinant protein. Stu2p has previously been expressed in reticulate lysates (Wang and Huffaker, 1997), but the amounts obtained were too small for tubulin-dependent assays. Although Escherichia coli–expressed Stu2p is apparently soluble, it aggregated in the presence of microtubules (unpublished data). Therefore, we expressed and purified Stu2p using recombinant baculovirus. Typically, a 500-ml insect cell culture yielded 4 mg protein that was over 90% pure (Fig. 1 A).
To characterize the recombinant Stu2p, we determined its Stokes radius using gel filtration on a Superose 6 column. Fig. 1 B (top) shows a Western blot of the collected fractions of the Stu2p run. Arrows indicate the position of marker proteins used to calibrate the column. Stu2p exhibited a symmetric peak, and none was found in the void volume of the column, confirming that the recombinant protein is soluble. From these data, we derived a Stokes radius of 11.1 nm for the recombinant Stu2p. To compare the Stokes radius of the recombinant Stu2p with endogenous Stu2p, yeast extracts were run on the same gel filtration column and Stu2p was monitored by Western blotting (Fig. 1 B, bottom). The Stokes radius of endogenous Stu2p was found to be ∼12.3 nm. Thus, the Stokes radius of endogenous and recombinant Stu2p can be considered almost identical given the very different nature of pure recombinant proteins and yeast extracts. Such high Stokes radii suggest that Stu2p forms a homomultimer, is a very elongated molecule, or both. To differentiate between these possibilities, we obtained the S value of Stu2p using sucrose gradients. Recombinant Stu2p or yeast extracts were spun through a linear 5–40% sucrose gradient. Fig. 1 C shows the Western blot for recombinant Stu2p (top) and for yeast extracts (bottom). From these gradients, we derived an apparent S value for both endogenous and recombinant Stu2p of 4.6. Combining S value and Stokes radius provides a native molecular weight of 199.6 kD for the recombinant and 221.2 kD for the endogenous Stu2. Since the calculated molecular weight of Stu2p is 100.9 kD, this suggests that it forms a stable dimer in solution. We confirmed that Stu2p forms a stable complex with itself in vivo by coimmunoprecipitating endogenous Stu2p together with enhanced GFP (EGFP)*-tagged Stu2p in yeast extracts (Fig. S1, available at http://www.jcb.org/cgi/content/full/jcb.200211097/DC1). Therefore, we conclude that in contrast to Stu2p, XMAP215, its Xenopus homologue, remains monomeric in solution (Gard and Kirschner, 1987; Cassimeris et al., 2001). Furthermore, Stu2p has a very high Stokes radius for a protein of 200 kD. A globular protein of this size would be expected to have a Stokes radius of ∼5 nm. Thus, like XMAP215 (Gard and Kirschner, 1987; Cassimeris et al., 2001), the Stu2p dimer has a very elongated shape.
Many microtubule-associated proteins like MAP1, MAP2, and tau (Serrano et al., 1985), but not the human Stu2p homologue ch-TOG (Spittle et al., 2000), require the COOH-terminal tail of β-tubulin for microtubule binding. To determine the involvement of the β-tubulin COOH terminus in binding of Stu2p to microtubules, we made microtubules without the COOH terminus by limited subtilisin proteolysis of taxol-stabilized microtubules. Fig. 2 A shows a Coomassie-stained gel of mock-digested and digested microtubules and a Western blot probed with an antibody specific for the COOH terminus of β-tubulin. We determined the relative extent of binding of Stu2p to these microtubules by incubating increasing amounts of them with 18 nM Stu2p at room temperature. Bound Stu2p was separated from unbound Stu2p by centrifugation, supernatants and pellets were analyzed by Western blotting (Fig. 2 B), and the extent of binding was subsequently quantitated (Fig. 2 C). Over a wide concentration range of microtubules, Stu2p bound to a similar extent to the undigested and the digested microtubules. Thus, Stu2p, like its human homologue ch-TOG but unlike many other microtubule-associated proteins so far studied, does not seem to have a strong requirement of the COOH terminus of β-tubulin for microtubule binding.
We tested the effects of recombinant Stu2p on the length of microtubules growing in vitro from purified centrosomes using rhodamine tubulin to monitor microtubule length. Microtubule growth was initiated at 29°C at 26 μM tubulin in the presence of increasing amounts of Stu2p. After 10 min, the reactions were fixed, quenched, and analyzed microscopically. Fig. 3 A shows a representative aster for each tested Stu2p concentration, and Fig. 3 B is a plot of the aster size distribution at the different Stu2p concentrations. Stu2p addition was found to lead to a clear reduction in microtubule length. This effect starts at 0.1 μM Stu2p and saturates below 0.33 μM. Thus, we conclude that Stu2p acts to destabilize microtubules in vitro.
To further confirm that Stu2p can act in vivo as a destabilizer of cytoplasmic microtubules, we used the temperature-sensitive mutant stu2–10 to abolish Stu2p function (Severin et al., 2001). To avoid complications from a stu2–10–dependent cell cycle arrest (Severin et al., 2001), we arrested both wild-type and stu2–10 cells in S phase at the permissive temperature with hydroxyurea (HU). After HU arrest, the cultures were shifted to 37°C, and the length of cytoplasmic microtubules were determined. Fig. 4 A shows a FACS® analysis of the cells throughout the experiment, demonstrating a complete cell cycle block at S phase. Fig. 4 B shows a projected antitubulin immunofluorescence of wild-type or stu2–10 cells arrested in HU at the permissive temperature before (top) or after the shift to the restrictive temperature (bottom and representative close ups in Fig. 4 C). Fig. 4 D shows the respective length distribution of cytoplasmic microtubules, and Fig. 4 E shows the average cytoplasmic microtubule lengths. Very similar in length before the temperature shift (wild type, 1.03 ± 0.07 [SEM; p < 0.05]) μm; stu2–10, 1.07 ± 0.07 μm), the length of cytoplasmic microtubules of stu2–10 and wild-type cells increases after the shift to the restrictive temperature (Fig. 4, D and E), reflecting the effect of the higher temperature and the larger cell size (see Materials and methods). However, after the temperature shift cytoplasmic microtubules of stu2–10 cells are on average ∼30% longer than the ones of wild-type cells (wild-type, 1.70 ± 0.13 μm; stu2–10, 2.29 ± 0.20 μm) (Fig. 4, D and E). Thus, Stu2p can act as a microtubule destabilizer in vivo and in vitro.
We wanted to further understand the mechanism by which Stu2 destabilizes microtubules. Stu2p could destabilize microtubules by inhibition of growth rate, promotion of microtubule shrinkage rate, through induction of catastrophes or prevention of rescues. To distinguish between these possibilities, we used video-enhanced differential interference contrast microscopy (VE-DIC). Microtubules were nucleated from purified centrosomes adsorbed to the coverslips of small perfusion chambers. Fig. 5 A shows three typical microtubule traces for three different Stu2p concentrations. Fig. 5 B shows the average growth rate, Fig. 5 C shows the catastrophe rate, and Fig. 5 D displays the average shrinkage rate as a function of Stu2p concentration. These data are summarized in Fig. 5 E. Rescue events occurred too rarely to make a statement about Stu2p's influence on the rescue frequency. Stu2p has a very minor effect on microtubule shrinkage rate that is only seen at high Stu2p concentrations. In contrast, Stu2p shows a marked inhibition of microtubule growth rate that starts at very low concentrations (0.03 μM) and begins to saturate at ∼0.1 μM. Similarly, Stu2p shows a marked promotion of catastrophes that begins at 0.03 μM and saturates at ∼0.1 μM. Thus, Stu2p destabilizes microtubules by decreasing their growth rate and increasing their catastrophe rate. This is in direct contrast to its Xenopus homologue XMAP215 that under very similar buffer conditions in vitro acts as a microtubule stabilizer (Vasquez et al., 1994; Kinoshita et al., 2001).
We wondered whether the stimulation of catastrophes by Stu2p was a direct effect on the catastrophe rate or indirectly caused by its inhibition of microtubule growth rate. To test this we first obtained the relation between microtubule growth rate and catastrophe frequency in the absence of Stu2p under our buffer conditions. We recorded microtubule growth at different tubulin concentrations using VE-DIC, binned the microtubule traces according to their growth rates, and calculated the corresponding catastrophe frequencies. The graph is shown in Fig. 5 F (varying tubulin). In agreement with published data (Drechsel et al., 1992), the catastrophe frequency increases asymptotically with decreasing growth rate. To this plot we added the growth rate catastrophe frequency values of the experiment in the presence of Stu2p (Fig. 5 F, varying Stu2p). At each given growth rate of microtubules in the presence of Stu2p the corresponding catastrophe frequency is almost identical to the one at a similar growth rate in the absence of Stu2p. Thus, Stu2p probably promotes microtubule catastrophes by inhibiting the microtubule growth rate.
Microtubule-associated proteins are known to be strong microtubule nucleators under certain conditions as shown recently for XMAP215 (Popov et al., 2002). One possibility is that the decreased growth rate could result from a reduction in the free tubulin pool due to nucleation of noncentrosomal microtubules by Stu2p. To test the nucleation ability of Stu2p, we mixed Stu2p and tubulin in the absence of centrosomes and recovered nucleated microtubules by centrifugation. 22 μM tubulin was incubated with increasing amounts of Stu2p for 10 min at 29°C. The reactions were spun through a 40% glycerol cushion at 29°C, and supernatant and pellet were analyzed by SDS-PAGE and Coomassie staining. No significant nucleation was seen over the wide concentration range of Stu2p tested (Fig. 3 C). In agreement, we observed no spontaneously nucleated microtubules in the VE-DIC assays (unpublished data). Thus, we believe that the destabilization of microtubules is a direct effect of Stu2p on microtubule dynamics and not an indirect effect on the effective tubulin concentration available for polymerization.
How does Stu2p inhibit microtubule growth? Interestingly, the inhibition of the microtubule growth rate by Stu2p saturated at relatively low Stu2p concentrations (∼0.1 μM Stu2p with 21.5 μM tubulin in the VE-DIC assay, below 0.33 μM Stu2p with 26 μM tubulin in a fixed time point assay). Mechanistically, that argues in favor of a model in which a limited number of microtubule-binding sites necessary for efficient growth become preferentially bound and saturated by Stu2p. An obvious location of such sites is the microtubule end. Indeed, there is evidence for Stu2p binding in vivo to MT ends (He et al., 2001; Kosco et al., 2001). Therefore, we determined the relative affinities of Stu2p for microtubule ends versus the lattice. We took advantage of the fact that taxol-stabilized microtubules can be sheared so that the concentration of polymeric tubulin remains constant with a significant increase in the concentration of microtubule ends. If Stu2p binds microtubule ends preferentially, more Stu2p should bind to sheared microtubules. Examples of sheared and unsheared microtubules are shown in Fig. 6 A. Sheared microtubules were on average fivefold shorter then the unsheared microtubules (see Materials and methods). Increasing amounts of microtubules were incubated with 19 nM Stu2p at room temperature. Microtubule-bound Stu2p was separated from unbound Stu2p by centrifugation, and the supernatant and pellet were analyzed by Western blotting. Fig. 6 B shows the Western blot probed for Stu2p and tubulin. Fig. 6 C shows a graph with the percent of microtubule-bound Stu2p plotted against the microtubule concentrations. At each given concentration of microtubules, more Stu2p bound to the sheared than to the nonsheared microtubules. Thus, Stu2p binds preferentially to microtubule ends as opposed to lateral microtubule binding sites. Under less stringent conditions, some end binding has been reported previously in vitro for the human Stu2p homologue ch-TOG (Spittle et al., 2000).
Binding of Stu2p to sheared microtubules is an indirect measurement of the affinity of Stu2p to microtubule ends. To directly visualize Stu2p binding to microtubule ends, we labeled Stu2p with the fluorescent dye Cy3. Cy3-Stu2p was incubated with taxol-stabilized microtubules labeled with Oregon green. To mark the microtubule polarity, the microtubule minus ends were brighter than the plus ends (Fig. 7 A; see Materials and methods). Reactions were fixed, quenched, and analyzed microscopically. Fig. 7 A shows representative Stu2p-bound microtubules. Microtubules are in green, and Stu2p is in red. Stu2p can be seen preferentially at the microtubule plus end. The distribution of Stu2p binding to microtubules was quantified by scanning the intensities of the Stu2p signal along the length of individual microtubules. Signals higher than 10% of the median intensity along the given microtubule were scored with their location. Fig. 7 B shows the distribution of the number of Stu2p spots per microtubule. Fig. 7 C shows a histogram with the frequency at which Stu2p signals are found at 10% distance intervals from the minus end to the plus end of the microtubules. Almost 40% of the Stu2p signals were found at microtubule plus ends. The remainder of the signals are relatively uniformly distributed with a frequency of ∼7% per distance interval along the rest of the microtubule, arguing against any preference of Stu2p among these lateral sites. To eliminate the possibility that Stu2p binding to the microtubule minus ends was inhibited by the relatively high levels of labeling there, we made nonpolarity marked microtubules and examined the end binding of Stu2p. Under these conditions as with polarity marked microtubules, ∼40% of the microtubules had Stu2p bound to one end only (unpublished data).
In this report, we have investigated the activity of Stu2p in vitro and how this activity relates to its known function in vivo. In contrast to the commonly held belief that members of the Dis1/XMAP215 family of microtubule-associated proteins all act to stabilize microtubules, we find that Stu2p is a microtubule destabilizer. Furthermore, we show that Stu2p is to our knowledge unique among the few known microtubule end–binding proteins in recognizing and preferentially binding microtubule plus ends in vitro. Our hydrodynamic analysis shows that Stu2p, like XMAP215, is an elongated molecule. However, in contrast to XMAP215, which is a monomer in solution (Gard and Kirschner, 1987; Cassimeris et al., 2001), Stu2p seems to be a dimer in vitro and in vivo. The real molecular weight of recombinant and endogenous Stu2p is ∼200 and ∼220 kD, respectively. Given the different nature of yeast extracts and pure proteins, a difference in the derived real molecular weight of 10% is within the experimental error. Thus, although we cannot rule out that Stu2p might be stably complexed with a small protein of ∼20 kD in vivo, we consider this to be unlikely.
The question we set out to answer was whether the difference in activity between the S. cerevisiae and the Xenopus members of the family in vivo represent fundamental differences in the biochemical properties of the proteins or whether the proteins are acting in different cellular contexts. In vitro, XMAP215 stabilizes microtubules by increasing the growth and shrinkage rate without affecting catastrophe frequency (Gard and Kirschner, 1987; Vasquez et al., 1994). Here we show that in contrast to XMAP215, Stu2p destabilizes microtubules in vitro.
Stu2p constitutes a new class of microtubule destabilizers. The other well-characterized microtubule destabilizers differ from Stu2p in their effect on microtubule dynamics. The KinI family of microtubule destabilizers bind to both plus and minus ends and are thought to trigger catastrophes by bending protofilaments into a structure that is found in a microtubule undergoing a catastrophe (Desai et al., 1999; Moores et al., 2002). They need the energy of ATP hydrolysis to recycle for multiple rounds of action. The destabilizing mechanism of katanin also depends on the hydrolysis of ATP. It disrupts noncovalent bonds between tubulin dimers within the microtubule lattice presumably by preferentially acting on microtubule lattice defects (McNally and Vale, 1993; Davis et al., 2002). Op18/Stathmin stimulates catastrophes directly at tips (Belmont and Mitchison, 1996; Howell et al., 1999) or acts through sequestering tubulin (Curmi et al., 1997; Jourdain et al., 1997; Howell et al., 1999). In contrast, Stu2p does not need energy to operate nor directly promotes catastrophes but rather decreases microtubule growth rate with a resultant increase in catastrophe frequency.
Stu2p's mechanism of destabilization cannot be explained by direct sequestering, given the relative concentrations of Stu2p and tubulin (0.1 and 21.5 μM, respectively) and the saturation of destabilization. How then do the microtubule-binding characteristics of Stu2p relate to its destabilizing activity? We propose that Stu2p destabilizes microtubules by preferentially binding to their plus ends and sterically hindering tubulin addition. This would lead to a decrease in microtubule growth rate, an increased probability of hydrolysis of the stabilizing GTP cap at microtubule plus ends, and an indirect increase in catastrophe frequency. Another possibility is that Stu2p changes the structure of the microtubule end, for instance, by affecting the length of the protofilament sheet characteristic for growing ends (Chretien et al., 1995; Arnal et al., 2000).
How does the in vitro activity of Stu2p relate to its function in vivo? Using a temperature-sensitive allele of Stu2, we were able to show that Stu2p can act in vivo as a microtubule destabilizer as it does in vitro. This is in agreement with an earlier in vivo study (Kosco et al., 2001). Here, using a copper run-down, the authors showed that removal of Stu2p resulted in less dynamic microtubules, by decreasing the catastrophe and rescue frequency without affecting the growth rate. How does our in vitro data compare with the studies on dynamics in vivo in the absence of Stu2p? In vitro, we also saw a Stu2p-dependent increase in the catastrophe frequency consistent with the in vivo data. However, although Stu2p decreased the growth rate in vitro, no effect on the growth rate in the absence of Stu2p was noticed by Kosco et al. (2001). Several reasons can be imagined for this discrepancy. In vivo, Stu2p is operating in the context of other microtubule-associated proteins that influence microtubule dynamics, e.g., Bim1 or Bik1 (Berlin et al., 1990; Pellman et al., 1995; Tirnauer et al., 1999). As a comparison, XMAP215 depletion from cytoplasmic egg extracts does not drastically alter the growth rate, although it is a very potent effecter of growth rate in vitro (Gard and Kirschner, 1987; Vasquez et al., 1994; Tournebize et al., 2000). Most importantly, the microtubule-destabilizing activity of Stu2p might be regulated in a cell cycle–dependent manner in vivo. Cell cycle regulation could also account for the fact that in another in vivo study (Severin et al., 2001), based on its role in spindle elongation during anaphase, Stu2p was proposed to act as a microtubule stabilizer. An intriguing model that could reconcile the differences concerning the in vivo role of Stu2p is one in which Stu2p could change in a regulated manner from a stabilizer to a destabilizer.
The use of polarity-marked microtubules allowed us to demonstrate that the end binding of Stu2p is specific to plus ends, in agreement with in vivo data that showed Stu2p to be colocalized with microtubule ends of cytoplasmic microtubules (He et al., 2001; Kosco et al., 2001). Our results suggest that the binding of Stu2p to plus ends in vivo may be direct and not mediated through other proteins. Other end-binding proteins, such as Bim1 and Bik1, involved in attachment of microtubules to cortical sites have been suggested to interact with Stu2p (Chen et al., 1998; Miller et al., 2000; for review see Schuyler and Pellman, 2001). One possibility is that Stu2p binds to microtubule ends and recruits other end-binding factors. To our knowledge, Stu2p is the first described microtubule-associated protein that recognizes and binds specifically to the microtubule plus end in vitro. Other plus end binders, such as CLIP-170, have been described in the literature (Perez et al., 1999; for review see Schuyler and Pellman, 2001). However, these do not recognize microtubule ends directly but rather transiently associate with growing microtubule ends by copolymerization with tubulin (Diamantopoulos et al., 1999; for review see Schuyler and Pellman, 2001).
Materials And Methods
Chemicals were from Sigma-Aldrich unless stated otherwise. The following reaction buffers were used: BRB80 (80 mM Pipes 6.8, 1 mM EGTA, 1 mM MgCl2); SRB1 (BRB80, 80 mM KCl, 0.5 mg/ml casein, 1 mM GTP, 1 mM DTT); SRB2 (SRB1, 0.25% Brij-35); and SRB3 (BRB80, 100 mM KCl, 0.25% Brij-35, 5% glycerol).
Stu2p antibody production
Polyclonal Stu2p antibodies were raised against a COOH-terminal peptide of Stu2p in rabbits and subsequently affinity purified (Eurogentec).
Purification of tubulin and centrosomes
Tubulin was purified to homogeneity from porcine brain by three cycles of microtubule assembly before phosphocellulose chromatography (Ashford et al., 1998). Rhodamine or Oregon green labeling of tubulin was done as described previously (Hyman et al., 1991). Centrosomes were purified from human KE37 cells using a low ionic strength lysis step and subsequent sucrose gradients (Moudjou and Bornens, 1998).
Microtubules were spotted at appropriate dilutions on poly-lysine microscope slides (Sigma-Aldrich), fixed for 3 min in −20°C methanol, and incubated in PBN (PBS, 2% BSA, 0.1% NP-40) for 5 min at RT. Microtubules were subsequently incubated directly with FITC-labeled DM1α tubulin antibody (Sigma-Aldrich) in PBN for 20 min at RT. After several washes, the slides were mounted in antifade-mounting medium. Images were acquired using a Axioplan 2 microscope (Carl Zeiss MicroImaging, Inc.) with a 63× Plan-Apochromat lens (NA 1.40; Carl Zeiss MicroImaging, Inc.) together with the MetaMorph Imaging System (version 1.4.2.) and processed using Adobe Photoshop®.
Preparation of yeast extracts
Yeast extracts were prepared in SRB3, 1 mM DTT, and protease inhibitor cocktail (Complete Mini, EDTA free; Roche) by bead beating using a W303 (mating type a) strain. Extracts were spun in a Beckman Coulter TLA 100 rotor at 100 krpm for 5 min at 4°C, and aliquots of the supernatant were snap frozen in liquid nitrogen and kept at −80°C. The protein concentration was between 10 and 20 mg/ml.
Purification of recombinant Stu2p as baculovirus-derived protein
A 500-ml suspension culture of Sf+ cells was infected at a density of 106 cells per ml with 2 ml of recombinant baculovirus at ∼108 pfu/ml. Recombinant baculovirus expressing full-length Stu2p was a gift from Peter Sorger (Massachusetts Institute of Technology, Cambridge, MA) and prepared using the pFAST Bac system (Invitrogen). Cells were harvested 52 h after infection. Cell extracts were prepared by resuspension of the cell pellet in 50 mM Hepes, pH 7.5, 5% glycerol, 0.1% Triton X-100, 150 mM NaCl, and 1 mM DTT, and supplemented with protease inhibitors. The extracts, cleared by ultracentrifugation and ultrafiltration, were loaded onto a cation exchange column (HiTrap SP, 5 ml; Amersham Biosciences) equilibrated in CB (6.7 mM Hepes, 6.7 mM MES, 6.7 mM sodium acetate, pH 7.2) and 150 mM NaCl. A linear gradient from CB, 150 mM NaCl to CB, 1 M NaCl was developed over 10 column vol. Stu2p eluted between 420 and 490 mM NaCl. These fractions were pooled and run on a gel filtration column (Superose 6, prep grade; Amersham Biosciences) in BRB80, 75 mM KCl, 1 mM DTT, and 10 μM ATP, and supplemented with protease inhibitors. Fractions containing the Stu2p peak were pooled and loaded onto a cation exchange column (HiTrap SP, 1 ml; Amersham Biosciences), equilibrated, and run in BRB80, 100 mM KCl. Stu2p was eluted in BRB80, 450 mM KCl and desalted using a PD10 column equilibrated and run in SRB3 supplemented with protease inhibitors and (optionally) 0.25% Brij-35. This preparation was aliquoted, snap frozen, and kept in liquid nitrogen.
Direct labeling of Stu2p
7.5 μM recombinant Stu2p in SRB3 was incubated with 180 μM monoreactive Cy3 (Amersham Biosciences) for 1 h on ice. After quenching the reaction by addition of potassium glutamate to 220 mM, labeled Stu2p was desalted into SRB3 and spun at 100 krpm for 5 min at 4°C in a Beckman Coulter TLA 100 rotor. DTT was added to 1 mM, and the supernatant was aliquoted, snap frozen, and kept in liquid nitrogen. The labeling stoichiometry (dye to protein) was determined spectrophotometrically and found to be 1.05.
Gel filtration was done on a Superose 6 PC 3.2/30 column (2.4 ml; Amersham Biosciences), equilibrated, and run in SRB3 at a flow rate of 0.02 ml/min. The column was calibrated in several runs with the following markers (Amersham Biosciences) of known Stokes radius: albumin, thyroglobulin, chymotrypsinogen A, ferritin, ovalbumin, catalase, aldolase, ribonuclease A. The void volume was determined using Blue Dextran. Recombinant Stu2p or yeast extracts cleared by ultrafiltration were subsequently run, 100-μl fractions were collected, and the peak position of Stu2p was determined using the chromatogram (absorption at 280 nm) or by SDS-PAGE and subsequent Western blotting of the collected fractions using a polyclonal Stu2p antibody.
Sucrose step gradients were prepared from 5 to 40% sucrose (in SRB3) in five equal steps and allowed to diffuse into a linear gradient overnight at 4°C. Precleared (100 krpm in a Beckman Coulter TLA 100, at 4°C for 5 min) recombinant Stu2p or yeast extracts, and in parallel a mixture of BSA, aldolase, catalase, and thyroglobulin (Amersham Biosciences) as markers of known S value, were spun through the gradients (4°C for 4 h at 50 krpm in a Beckman Coulter TLS 55 rotor), and gradients were fractionated by taking 16 equal fractions from the top. The position of proteins within the gradients was determined on Coomassie-stained SDS-PAGE gels (markers) or by Western blotting using a polyclonal Stu2p antibody (recombinant Stu2p and yeast extracts).
Microtubule nucleation assay
All protein preparations were precleared before usage (100 krpm in a Beckman Coulter TLA 100, at 4°C for 5 min). 22 μM tubulin was incubated in SRB2 with varying concentrations of Stu2p (0–2.5 μM) for 10 min at 29°C. Nucleated microtubules were separated from soluble tubulin by centrifugation through a 40% glycerol cushion (70 krpm for 5 min at 29°C in a Beckman Coulter TLA 100 rotor). Equivalent amounts of supernatant and pellet were analyzed by SDS-PAGE and Coomassie staining.
Determination of microtubule length by a fixed time point assay
Microtubules were grown from purified centrosomes at 26 μM tubulin (1:10 labeled with rhodamine tubulin) and various concentrations of Stu2p (0–1 μM Stu2p) in SRB2 at 29°C for 10 min. Prior to usage, protein preparations were precleared by centrifugation (100 krpm at 4°C for 5 min in a Beckman Coulter TLA 100 rotor). Reactions were fixed with 0.75% glutaraldehyde for 3 min and subsequently quenched for 5 min with 0.1% sodium borohydride. Part of the reaction was squashed under a coverslip and analyzed microscopically. Images were acquired using a Axioplan 2 Microscope (Carl Zeiss MicroImaging, Inc.) with a 63× Zeiss Plan-Apochromat lens (NA 1.40; Carl Zeiss MicroImaging, Inc.) together with the MetaMorph Imaging System (version 1.4.2.). The radii of at least 39 asters were measured for each Stu2p concentration using the MetaMorph Imaging software, and images were subsequently processed using Adobe Photoshop®.
Measurement of cytoplasmic microtubule length in vivo
Wild-type cells or stu2–10 cells (both GFP-tub1–His3 and derivatives of W303) were grown in YP, 2% raffinose at 25°C to an OD600 of 0.4, arrested with 100 mM HU at 25°C for 140 min, and shifted to 37°C for another 140 min. A FACScan® was used as described (Epstein and Cross, 1992). Cells were prepared for indirect immunofluorescence as published (Piatti et al., 1996) using the mouse monoclonal antitubulin antibody DM1α (Sigma-Aldrich) and a FITC-conjugated secondary antibody. DNA was visualized by DAPI. Imaging was done using a 100×, 1.35 NA objective on an Olympus DeltaVision microscope using SoftWorx (Applied Precision). 40 0.2-μm stacks were taken for each field, and cytoplasmic microtubule length in single cells was measured three-dimensionally from the tip of the microtubule to the spindle pole body as multiple segments (SoftWorx; Applied Precision). Between 153 and 175 cytoplasmic microtubules were quantitated. Images depicted in this paper were projected using SoftWorx (Applied Precision) and processed using Adobe Photoshop®. To determine cell sizes, DIC images of at least 68 cells were acquired using a Axioplan 2 microscope (Carl Zeiss MicroImaging, Inc.) with a 63× Plan-Apochromat lens (NA 1.40; Carl Zeiss MicroImaging, Inc.) together with the MetaMorph Imaging System (version 1.4.2.). Cells were analyzed by measuring the length of a line drawn from the tip of the mother cell through the bud neck to the tip of the bud using the MetaMorph Imaging software. The average mother cell–bud distance was for the cells arrested at the permissive temperature with HU before the shift to the restrictive temperature: 6.5 ± 1.1 μm (standard deviation) for the wild-type and 6.5 ± 1.0 μm for the stu2–10 cells. After the temperature shift, this distance was 7.5 ± 1.1 μm for the wild-type and 7.5 ± 1.1 μm for the stu2–10 cells. Thus, stu2–10 and wild-type cells did not differ in their size.
The VE-DIC assay was adapted from Kinoshita et al. (2001). Small (∼12 μl) home made perfusion chambers were put on a metal block on ice, and purified centrosomes were perfused into the chambers and allowed to adsorb to the coverslips for 5 min. The glass surface was blocked three times for 5 min with 5 mg/ml casein in BRB80, and free casein was removed by four washes with 20 μl 0.25% Brij-35 in BRB80. 30-μl mixtures of prespun (4°C for 5 min at 100 krpm in a Beckman Coulter TLA 100 rotor) tubulin (21.5 μM) and various levels of prespun Stu2p (0–1 μM) in SRB2 was perfused at 4°C into the chambers. Reactions were started by transferring the chambers to a metal block at RT and then immediately onto the stage of a VE-DIC microscope kept in a temperature control box at 29°C. For each experiment, one aster was analyzed between 5 and 12 min after the start of the reaction. At least four experiments were done at each concentration of Stu2p. Videos were acquired on an Olympus BX50 microscope using an Olympus PlanApo 60× lens (NA 1.40) and a Hamamatsu 2400 Newvicon video-camera. Images were enhanced with an Argus 20 (Hamamatsu). Microtubules were tracked using ScionImage (version 1.62). The parameters of microtubule polymerization were subsequently determined from the entire dataset at each concentration of Stu2p. Rescue events occurred too rarely to be reported as a rescue frequency.
To determine the relationship of growth rate to catastrophe frequency, we used the same set-up with minor changes: smaller chambers (∼6 μl) were employed and axonemes (gifts from Denis Chrétien, Université de Rennes, Rennes, France) instead of centrosomes were used as microtubule nucleators. The tubulin concentration ranged from 14.3 to 27.5 μM in SRB2. Three films from ∼1.5 to ∼13 min after the start of the reaction were obtained at each concentration of tubulin. Microtubule growth rates were linear with tubulin concentration. Microtubules were grouped according to their growth rates in groups of 0.06-μm/min increments, and the average growth rate and the catastrophe frequency of the groups were subsequently calculated. Groups contained between 9 microtubules with a total growth time of 19 min and 44 microtubules with a total growth time of 104 min.
Taxol-stabilized microtubules were sheared on ice 10 times for 1 s with 3 min on ice in between with a tip sonicator at 40% amplitude. The lengths of at least 91 microtubules were determined by immunofluorescence using the MetaMorph Imaging software. The average microtubule length was 16.6 μm for the unsheared (SEM [P < 0.05]; 2.5 μm) and 3.4 μm for the sheared microtubules (SEM [P < 0.05]; 0.5 μm).
Microtubule copelleting assay
Increasing amounts of taxol-stabilized microtubules were incubated in SRB2: 20 μM taxol with prespun (4°C for 5 min at 100 krpm in a Beckman Coulter TLA 100 rotor) Stu2p for 5 min at RT. Microtubule-bound protein was separated from unbound protein by centrifugation (25°C for 5 min at 70 krpm in a Beckman Coulter TLA 100 rotor). For the shearing experiment, the reactions were spun through a 40% glycerol BRB80 cushion. Equivalent amounts of supernatants and pellets were subsequently analyzed by Western blotting using a polyclonal Stu2p antibody, the monoclonal anti–α-tubulin antibody DM1α (Sigma-Aldrich), or the monoclonal anti–β-tubulin antibody TUB2.1 (Sigma-Aldrich). The percentage of binding was measured using the Histogram function in Adobe Photoshop®.
Production of polarity-marked microtubules
Bright microtubule seeds were made by incubating 75 μM of a prespun (4°C for 5 min at 100 krpm in a Beckman Coulter TLA 100 rotor) 1.8:1 mixture of Oregon green–labeled tubulin and unlabeled tubulin in BRB80, 1 mM DTT with 0.1 mM GMPCPP at 37°C for 15 min. The seeds were diluted 1:125 in a prewarmed mixture containing 20 μM of a 1:5 mixture of Oregon green–labeled and unlabeled tubulin in BRBR80, 1 mM GTP, 1 mM DTT, and 12.5 μM NEM-treated tubulin (to inhibit minus end polymerization). After a 30-min incubation at 37°C, the mixture was diluted in prewarmed BRB80, 20 μM taxol, and 1 mM DTT. Taxol-stabilized microtubules were reconcentrated by centrifugation (25°C for 5 min at 50 krpm in a Beckman Coulter TLA 100 rotor) and resuspended in BRB80, 20 μM taxol, and 1 mM DTT.
Removal of the COOH terminus of β-tubulin in microtubules
Taxol-stabilized microtubules were incubated in BRB80, 1 mM DTT, and 20 μM taxol for 15 min at 37°C with 30 μg/ml subtilisin (Fluka). The digestion was stopped by adding PMSF to 5 mM. Completion of digestion was checked by SDS-PAGE and Coomassie staining or by Western blotting using the β-tubulin–specific mAb TUB2.1 (Sigma-Aldrich) whose epitope is situated in the COOH terminus of β-tubulin.
Visualization of Stu2p binding on microtubules
Prespun (4°C for 5 min at 100 krpm in a Beckman Coulter TLA100 rotor) 16 nM directly labeled Stu2p in SRB3, 1 mM DTT, and 20 μM taxol was incubated for 5 min at RT with 11.5 μM taxol-stabilized polarity marked microtubules. The reaction was fixed with 0.75% glutaraldehyde for 3 min, quenched for 5 min with 0.1% sodium borohydride, and analyzed microscopically after a 1:200 dilution in SRB3, 20 μM taxol. Images were acquired using a Axioplan 2 microscope with a 63× Plan-Apochromat lens (NA 1.40; Carl Zeiss MicroImaging, Inc.) together with the MetaMorph Imaging System (version 1.4.2.). The intensities of the signals in the rhodamine (Stu2p) channel were scanned along the length of microtubules using the Linescan function of the MetaMorph Imaging software with a pixel width of 5. Peaks higher then 10% of the median intensity along the microtubule were scored with their location along the microtubule. Microtubules were divided in 10% intervals from their minus ends to their plus ends. 124 polarity marked microtubules were quantitated. In the absence of directly labeled Stu2p, no significant number of peaks were detected. The median intensity in the absence and presence of directly labeled Stu2p was found to be virtually identical. Images were processed using Adobe Photoshop®.
Online supplemental material
Fig. S1 demonstrates that Stu2p interacts with itself in vivo. The figure is available online at http://www.jcb.org/cgi/content/full/jcb.200211097/DC1.
We acknowledge Peter Sorger for the kind gift of recombinant baculovirus, Denis Chrétien for the kind gift of axonemes, Susi Berthold for the help with expressing and purifying Stu2p, and the members of the Hyman lab for their comments on this manuscript.
The online version of this article includes supplemental material.
Abbreviations used in this paper: HU, hydroxyurea; VE-DIC, video-enhanced differential interference contrast microscopy.