Stu2p/XMAP215 proteins are essential microtubule polymerases that use multiple αβ-tubulin-interacting TOG domains to bind microtubule plus ends and catalyze fast microtubule growth. We report here the structure of the TOG2 domain from Stu2p bound to yeast αβ-tubulin. Like TOG1, TOG2 binds selectively to a fully ‘curved’ conformation of αβ-tubulin, incompatible with a microtubule lattice. We also show that TOG1-TOG2 binds non-cooperatively to two αβ-tubulins. Preferential interactions between TOGs and fully curved αβ-tubulin that cannot exist elsewhere in the microtubule explain how these polymerases localize to the extreme microtubule end. We propose that these polymerases promote elongation because their linked TOG domains concentrate unpolymerized αβ-tubulin near curved subunits already bound at the microtubule end. This tethering model can explain catalyst-like behavior and also predicts that the polymerase action changes the configuration of the microtubule end.https://doi.org/10.7554/eLife.03069.001
Dynamic filaments of proteins, called microtubules, have several important roles inside cells. Microtubules provide structural support for the cell; they help to pull chromosomes apart during cell division; and they guide the trafficking of proteins and molecules across the cell.
The building blocks of microtubules are proteins called αβ-tubulin, which are continually added to and removed from the ends of a microtubule, causing it to grow and shrink. Other proteins that interact with the microtubules can help to speed up these construction and deconstruction processes. Ayaz et al. took a closer look at the structure of one particular family of proteins that make it easier for the microtubules to grow, using a technique called X-ray crystallography. The resulting images show two sites—called TOG1 and TOG2—on the enzymes that attach to the αβ-tubulin proteins. Ayaz et al. found that this binding can only occur when αβ-tubulin has a curved shape, which only happens when the tubulins are not included in, or are only bound weakly to the end of, a microtubule.
Previous research suggested that the two binding sites might work together to provide ‘scaffolding’ that stabilizes the microtubule. However, genetic experiments by Ayaz et al. show that microtubules will grow even if one of the binding sites is missing. Both TOG1 and TOG2 bind to αβ-tubulin in the same way, and by using computer simulations Ayaz et al. found that this helps to speed up the growth of microtubules. This is because the enzyme's two sites concentrate the individual tubulin building blocks at the ends of the filament. For example, TOG2 could bind to the end of the microtubule, while TOG1 holds an αβ-tubulin protein nearby and ready to bind to the filament's end. This tethering allows the microtubules to be assembled more efficiently.https://doi.org/10.7554/eLife.03069.002
Microtubules are dynamic polymers of αβ-tubulin that have critical roles in chromosome segregation and intracellular organization (reviewed in Desai and Mitchison, 1997). The polymerization dynamics of microtubules are regulated by multiple cellular factors. Evolutionarily conserved proteins in the Stu2p/XMAP215 family (Gard and Kirschner, 1987; Ohkura et al., 1988; Wang and Huffaker, 1997) regulate microtubule dynamics by promoting fast microtubule elongation. These essential proteins use multiple αβ-tubulin binding tumor overexpressed gene (TOG) domains to selectively recognize the growing microtubule end and promote its elongation (Al-Bassam et al., 2006; Brouhard et al., 2008; Widlund et al., 2011; Al-Bassam et al., 2012).
A significant advance in understanding occurred with a landmark study of XMAP215 in which in vitro reconstitution assays demonstrated that the polymerase affected the rate of microtubule elongation without affecting the apparent equilibrium (Brouhard et al., 2008). This and other observations (e.g., Shirasu-Hiza et al., 2003; van Breugel et al., 2003) formed the basis for describing XMAP215 as a catalyst for microtubule elongation. Catalytic action in turn led to the concept that TOG-containing polymerases might stabilize an otherwise rate-limiting intermediate along the polymerization pathway (Brouhard et al., 2008). However, an understanding of mechanism has been limited because the nature of this intermediate, and how TOG domains might selectively promote it, remained unclear.
A recent study from our group revealed that the TOG1 domain from Stu2p binds preferentially to a ‘curved’ conformation of αβ-tubulin that cannot be incorporated into the body of the microtubule (Ayaz et al., 2012). Our study also showed that a TOG1-TOG2 construct could bind two αβ-tubulins (Ayaz et al., 2012). This latter observation suggested that two TOG domains might cooperate to stabilize an αβ-tubulin:αβ-tubulin interface, and consequently that cooperative binding to αβ-tubulin might contribute to polymerase activity. Our study did not determine how the polymerase recognizes the extreme microtubule end, but we speculated based on apparent biochemical differences between the TOG1 and TOG2 domains (Al-Bassam et al., 2006; Ayaz et al., 2012) that selective interactions between TOG2 and a different, end-specific conformation of αβ-tubulin might be important.
In the present study we sought to gain insight into the mechanism of end recognition by determining the conformation of αβ-tubulin recognized by TOG2 and by testing whether cooperative TOG:αβ-tubulin interactions contributed to polymerase activity. We first determined the crystal structure of a TOG2:αβ-tubulin complex. This structure reveals that TOG2 binds to the same curved conformation of αβ-tubulin that TOG1 does. Our biochemical experiments underscore this structural similarity by demonstrating that the two TOG domains have comparable affinities for αβ-tubulin, with KD ∼100 nM. Next, we used analytical ultracentrifugation to show that in TOG1-TOG2 the two linked TOG domains bind non-cooperatively to two αβ-tubulins. Non-cooperative binding indicates that TOG1-TOG2 does not stabilize an αβ-tubulin:αβ-tubulin interface. Together with biochemical and genetic experiments, our results lead to a model that explains how the polymerase activity can emerge from the action of two tethered TOG domains that each bind independently to a conformation of αβ-tubulin that is incompatible with the microtubule lattice. We propose that the polymerase activity arises because linked TOG domains selectively increase the effective concentration of αβ-tubulin near weakly bound, curved αβ-tubulins already on the microtubule end. A computational realization of this model supports our proposal by recapitulating catalyst-like behavior. The model further suggests that the polymerase achieves its effect in part by transiently altering the configuration of the growing end.
We previously showed that the TOG1 domain from Stu2p binds preferentially to a curved conformation of αβ-tubulin (Ayaz et al., 2012). However, TOG1 is dispensable for the plus-end binding of Stu2p (Al-Bassam et al., 2006), and because of apparent differences in the biochemical behavior of TOG1 and TOG2 (Al-Bassam et al., 2006; Ayaz et al., 2012) we speculated that TOG2 might bind to a different, lattice-induced conformation of αβ-tubulin (Ayaz et al., 2012). We solved the crystal structure of a TOG2:αβ-tubulin complex (Figure 1) to resolve this ambiguity. The structure was determined by molecular replacement from crystals that diffracted anisotropically to a minimum Bragg spacing of 2.8 Å (Table 1). The final model has good geometry (Table 1; Molprobity [Chen et al., 2010] clash score 1.79; 95.6% favored residues in Ramachandran plot) and has been refined to an Rfree of 0.259 (Rwork = 0.217).
The structure of the TOG2:αβ-tubulin complex is remarkably similar to that of the TOG1:αβ-tubulin complex (Figure 1; Ayaz et al., 2012). Conserved residues like W341 and R519 on the tubulin-binding surface of TOG2 make very similar contacts with αβ-tubulin as their equivalents in TOG1 (W23 and R200, respectively) (Figure 1A,B). The structure of TOG2 bound to αβ-tubulin superimposes with the structure of ‘free’ TOG2 (Slep and Vale, 2007) with 0.7 Å rmsd over 237 Cα atoms (Figure 1C). Because the TOG2 domains pack differently in the ‘bound’ and ‘free’ crystals, the similarity in structure suggests that these TOG domains are rigid modules that do not change conformation when they bind αβ-tubulin. Even though the structure reported here was obtained from a new crystal form, the conformation of αβ-tubulin in complex with TOG2 is nearly identical to that seen for TOG1-bound αβ-tubulin (Figure 1D) (12.3° of curvature in the TOG2 complex vs 13.1° in the TOG1 complex), and is characteristically distinct from the straight conformation of αβ-tubulin (1° of curvature by our measure) (Löwe et al., 2001; Figure 1E). Underscoring this similarity, individual tubulin chains between the TOG1 and TOG2 complexes superimpose on each other with 0.4 Å rmsd over Cα atoms. Thus, both TOG1 and TOG2 bind preferentially to the same, curved conformation of αβ-tubulin. By extension, other TOGs in this family probably also bind to curved αβ-tubulin.
The shared preference of TOG1 and TOG2 for curved αβ-tubulin has implications for the mechanism of end recognition. Because TOG:αβ-tubulin interactions are required for plus-end localization of Stu2p/XMAP215 polymerases (Al-Bassam et al., 2006), preferential binding of TOGs to curved αβ-tubulin suggests that curved αβ-tubulin itself is the distinctive end-specific feature the polymerase recognizes. In contrast, other plus-end tracking proteins like Eb1 recognize lattice-specific features and consequently show ‘comet-like’ localization that extends into the microtubule body (Nakamura et al., 2012; Maurer et al., 2014). By invoking binding to an epitope that cannot exist in the body of the microtubule, our model explains how Stu2p/XMAP215 polymerases localize to the extreme microtubule end (Nakamura et al., 2012; Maurer et al., 2014).
The shared preference for fully curved αβ-tubulin also poses an apparent paradox, because it indicates that the polymerase is constructed from ‘parts’ that bind most strongly to a conformation of αβ-tubulin that cannot exist in the microtubule lattice. We explore and propose a resolution for this apparent contradiction in later sections.
To determine if the structural similarity between the TOG1 and TOG2 complexes with αβ-tubulin extends to a biochemical similarity, we measured the affinity of TOG:αβ-tubulin interactions. Previously we had used polymerization-blocked αβ-tubulin mutants (Johnson et al., 2011) for TOG binding assays. To eliminate the possibility that blocking mutations and/or fluorescent labeling might mask or alter αβ-tubulin:αβ-tubulin interactions, we developed a label-free assay in which polymerization-competent αβ-tubulin could be used. Analytical ultracentrifugation monitored by absorbance at 230 nm (A230) allowed us to work at concentrations of αβ-tubulin at which higher-order oligomers of αβ-tubulin were nearly undetectable (Figure 2A). Indeed, at αβ-tubulin concentrations ranging from 0.08 to 1 μM there is a single dominant peak at 5.8 S with only 1.7 and 2.8% of the material sedimenting faster at the lowest and highest concentration tested. This uniform sedimentation behavior made it possible for us to use analytical ultracentrifugation as a quantitative binding assay.
We measured the affinity of TOG1 and TOG2 for αβ-tubulin by separately titrating variable amounts of each TOG domain into a constant amount of αβ-tubulin and analyzing the resulting sedimentation behavior (Figure 2B,C). We fit the concentration-dependent sedimentation profiles using single-site binding isotherms (Dam and Schuck, 2005; Figure 2B,C). The fits indicate that both TOG domains form relatively tight, 1:1 complexes with αβ-tubulin: TOG1 binds with KD = 70 nM and TOG2 binds with KD = 160 nM. The KD for the TOG2:αβ-tubulin complex reported here is consistent with our prior measurement using fluorescence anisotropy and polymerization blocked αβ-tubulin (Ayaz et al., 2012). Thus, in addition to a shared preference for the same curved conformation of αβ-tubulin, TOG1 and TOG2 also bind αβ-tubulin with comparable affinities. From both a biochemical and a structural perspective, the TOG1 and TOG2 domains are remarkably similar to each other.
When present together in a TOG1-TOG2 construct, each of the TOG domains can engage its own αβ-tubulin (Ayaz et al., 2012). We speculated that the two TOG-bound αβ-tubulins might also interact with each other, and that this cooperativity might make important contributions to polymerase function (Figure 3A).
We used analytical ultracentrifugation to determine if cooperative interactions stabilized the TOG1-TOG2:(αβ-tubulin)2 complex. First, using mutations on TOG1-TOG2 that disrupt the slightly weaker TOG2:αβ-tubulin interactions, we determined that TOG1-TOG2 bound to a single αβ-tubulin sediments at 7.1 S (Figure 3B). Next, using a mixture in which αβ-tubulin was superstoichiometric with respect to TOG1-TOG2 (Figure 3C), we observed a larger species sedimenting at 9.1 S. This value places a lower limit on the sedimentation coefficient for the TOG1-TOG2:(αβ-tubulin)2 complex (Figure 3C, inset) that is consistent with hydrodynamic calculations (García De La Torre et al., 2000) using extended models in which the two αβ-tubulins do not contact each other and the TOG domains are separated by varying distances. Hydrodynamic calculations predict that compact models containing longitudinally or laterally of αβ-tubulins should sediment around 10.5–10.9 S (Figure 3A,C). Knowledge about the sedimentation behavior of the one and two tubulin complexes allowed us to analyze a titration of TOG1-TOG2 into αβ-tubulin under different assumptions about cooperativity (Figure 3D). The resulting signal weighted average sedimentation coefficients are not consistent with models that assume even modest 10-fold positive or negative cooperativity (Figure 3D). Instead, the data are much better described by a model in which each TOG domain in TOG1-TOG2 interacts independently with its αβ-tubulin (the best fit was obtained from a model invoking less than twofold negative cooperativity, not shown). Noncooperative binding contradicts our initial expectation and indicates that αβ-tubulin:αβ-tubulin contacts do not provide additional stability to the TOG1-TOG2:(αβ-tubulin)2 complex.
The lack of positive cooperativity stabilizing the TOG1-TOG2:(αβ-tubulin)2 complex is striking because in this complex the two bound αβ-tubulins are physically constrained to occupy a relatively small volume and therefore are effectively at quite high concentration relative to each other. Indeed, allowing 55 Å for the length of each TOG domain and assuming that the ∼75 amino acid linker is flexible and can maximally span 220 Å we estimate that the two TOG1-TOG2-bound αβ-tubulins are minimally at an effective concentration of roughly 200 μM, likely higher because the linker will rarely be fully extended. This effective concentration of TOG1-TOG2-bound αβ-tubulin is in the range of current estimates for the longitudinal KD for αβ-tubulins (e.g., Gardner et al., 2011). Observing non-cooperative binding therefore suggests that some property of the TOG1-TOG2 linker antagonizes or counterbalances longitudinal interactions between αβ-tubulins (see below for experiments concerning the linker sequence). Our data do not rule out that linked TOG domains might stabilize lateral interactions between αβ-tubulins because these lateral interactions are thought to be much weaker (only molar affinity [Gardner et al., 2011]): even at effective αβ-tubulin concentrations approaching 1 mM a weak interface like this would not be populated and therefore would contribute little additional stability to the TOG1-TOG2:(αβ-tubulin)2 complex. Whatever the underlying mechanism, the unexpected observation that the two linked TOG domains behave as independent αβ-tubulin binding modules places significant constraints on biochemical models for the mechanism of these polymerases.
Stu2/XMAP215 family polymerases contain at least two different TOG domains. This property might indicate that the two domains have different functional specialization, but the structural and biochemical similarity of TOG1 and TOG2 (described above) does not seem consistent with separation of function. To investigate if different TOG domains are required for polymerase function we constructed a ‘TOG-swapped’ and other variants of Stu2p and assayed their ability to rescue the conditional depletion of endogenous Stu2p using a previously described assay (Kosco et al., 2001; Al-Bassam et al., 2006; Ayaz et al., 2012).
We first performed rescue assays using a ‘full-length’ Stu2p construct that retains the ability to dimerize and thus that contains two identical TOG1-TOG2 segments linked through the coiled-coil dimerization interface. A construct (TOG2-TOG2) in which TOG1 was replaced by a second copy of TOG2 rescued as well as did the wild-type (TOG1-TOG2) (Figure 4A). We also tried to replace TOG2 with a second copy of TOG1, but control experiments indicated that this variant was not stable (data not shown) and we have not yet pursued it further. Instead, we used site-directed mutagenesis as an alternate way to ablate the αβ-tubulin binding activity of TOG1 or TOG2. Constructs in which either TOG was impaired for αβ-tubulin binding (TOG1(R200A)-TOG2 or TOG1-TOG2(R519A)) also gave full rescue (Figure 4A). Ablating the function of individual TOGs is not without functional consequences under more stringent conditions, because these same constructs do show compromised rescue under conditions of microtubule stress (Ayaz et al., 2012). These data indicate that under normal conditions, the polymerase can function with only TOG1 or only TOG2 domains. Thus, having different TOG domains does not appear to be essential for polymerase function.
Using dimeric rescue constructs did not allow us to test Stu2p variants that contained only a single functional TOG domain of either kind. We therefore introduced the same mutations into a previously characterized dimerization-impaired variant of Stu2p in which the coiled-coil dimerization element had been deleted (Stu2p-Δcc) (Al-Bassam et al., 2006; Figure 4B). Consistent with prior observations (Al-Bassam et al., 2006), dimerization impaired Stu2p only partially compensated for the depletion of endogenous Stu2p. Even in this sensitized background, however, we found that as long as there were two functional TOG domains they did not need to be different. Stu2p(TOG2-TOG2)-Δcc showed rescue efficiency very similar to that of Stu2p(TOG1-TOG2)-Δcc (Figure 4B). Dimerization impaired variants of Stu2p in which TOG1 or TOG2 was defective for αβ-tubulin binding did not show any rescue activity (Figure 4B), consistent with a prior in vitro study of XMAP215 that demonstrated a requirement for at least two TOG domains (Widlund et al., 2011). More importantly, the ability of TOG2 to substitute for TOG1 in this more stringent, dimerization-impaired background strengthens the conclusion that the polymerase does not require different TOG domains for its function.
The dimeric Stu2p(TOG1-TOG2(R519A)) and Stu2p(TOG1(R200A)-TOG2) variants rescued the depletion of endogenous Stu2p in spite of the fact that the two functioning TOG domains were linked through the coiled-coil dimerization segment (and in the case of TOG1-TOG2(R519A) with a defective TOG domain in the linking sequence). This result suggested that how two TOG domains were linked was relatively unimportant, as long as they were linked. To explore this more systematically we determined how randomizing and/or shortening the TOG1-TOG2 linker in dimerization impaired Stu2p affected its rescue activity.
To test if the primary sequence of the TOG1-TOG2 linker was important for function, we prepared two variants of Stu2pΔcc in which the order of the central 65 amino acids of the TOG1-TOG2 linker (residues 252–316) was randomized (Figure 5A). This ‘shuffling’ strategy preserves the overall amino acid composition but should disrupt any local features specific to the natural sequence. Both shuffled linkers gave rescue activity nearly indistinguishable from the wild-type linker (Figure 5A), consistent with the robust rescue activity of alternatively linked functional TOGs in dimeric Stu2p(TOG1-TOG2(R519A)) and Stu2p(TOG1(R200A)-TOG2) (Figure 4A). Thus, the sequence linking the two functional TOG domains tolerates significant variation.
We also made a series of internal deletions in the TOG1-TOG2 linker to determine if shortening the linker affected rescue activity. Deleting 14 amino acids had little effect on rescue activity (Figure 5B). Rescue activity was mildly compromised by deletions of 22, 32, 40, and 50 amino acids (Figure 5B). Deleting 60 amino acids from the TOG1-TOG2 linker substantially abolished rescue activity (Figure 5B). These data suggest that Stu2p function is compromised when the sequence linking two TOG domains becomes too short. To examine the question of linker length more generally, we analyzed the distribution of TOG1-TOG2 linker length in ∼300 Stu2p/XMAP215 orthologs. This analysis showed that linker lengths below ∼40 amino acids occur very rarely (Figure 5C), supporting the notion that there is a minimal linker length below which polymerase function is compromised.
Any model for the polymerase function must recapitulate the catalyst-like properties documented previously (Brouhard et al., 2008) while also accounting for the seemingly paradoxical observation that both TOG1 and TOG2 from Stu2p bind preferentially to a curved conformation of αβ-tubulin that cannot exist in the body of the microtubule. A good model must likewise be consistent with the lack of detectable positive cooperativity stabilizing the TOG1-TOG2:(αβ-tubulin)2 complex, and should not require different TOG domains or a non-spacer-like role for the sequence linking the two TOG domains.
In Figure 6 we cartoon a simple ‘tethering’ mechanism that incorporates these new constraints, using a minimal ‘two TOG’ polymerase (Widlund et al., 2011) to simplify the representation. The central elements of this model are: (i) plus-end localization of the polymerase occurs by a TOG domain binding to curved αβ-tubulin at the microtubule end; (ii) full curvature of αβ-tubulin at the microtubule end requires that it not have any lateral neighbors; (iii) αβ-tubulin bound to the other TOG domain is physically tethered to the microtubule end and therefore associates with it more frequently. When not engaged with αβ-tubulin at the microtubule end, we assume based on results from others (Brouhard et al., 2008) that the polymerase diffuses on the microtubule lattice by virtue of its basic domain, thereby rapidly ‘searching’ for another curved αβ-tubulin.
Can this tethered delivery model recapitulate the catalyst-like action of these polymerases? To address this question, we took advantage of a kinetic model for microtubule assembly that we have been developing as part of other ongoing work.
Our model for microtubule elongation is very similar to (and was inspired by) one previously developed by Odde and Cassimeris (VanBuren et al., 2002). Alternative and more complicated models exist (e.g., VanBuren et al., 2005; Margolin et al., 2012), but this relatively simple one captures the essence of the biochemistry—that the microtubule end presents a mix of high and low affinity binding sites, depending on neighbor context–in a reasonable way and using a small number of adjustable parameters (‘Materials and methods’). Our model simulates microtubule polymerization one subunit association or dissociation at a time using kinetic Monte Carlo simulations on a two-dimensional lattice with staggered periodic boundary conditions to mimic the cylindrical structure of a 13 protofilament microtubule (Figure 7A). Because XMAP215 has been shown to promote the elongation of microtubules in the presence of a hydrolysis resistant GTP analog (Brouhard et al., 2008), we performed ‘GTP-only’ simulations (no GTP hydrolysis) to simplify the behavior by eliminating catastrophe. We performed a manual grid search of longitudinal and corner interactions (in the model these interactions are the dominant contributors to growth rate) to identify parameters that reproduced experimentally observed microtubule growth rates. To validate our program, we first fit the same experimental data (Walker et al., 1988) as did Odde and Cassimeris (VanBuren et al., 2002), obtaining similar parameters (Figure 7—figure supplement 1). Other measurements of microtubule dynamics (e.g., Mitchison and Kirschner, 1984; Drechsel et al., 1992; Hyman et al., 1992; Brouhard et al., 2008; Gardner et al., 2011]) have shown much lower critical concentrations than the Walker et al. data (Walker et al., 1988) do, so we also searched for parameters to approximate the growth rates observed in the presence of GMPCPP (Brouhard et al., 2008; Gardner et al., 2011). This procedure identified a different set of parameters (Figure 7C; VanBuren et al., 2002). These models provide a starting point for exploring potential polymerase mechanisms.
An explicit model for the polymerase requires too many adjustable parameters to be feasible at this time. To address the question ‘can tethering give catalyst-like activity’, we made a number of simplifying assumptions to create an implicit representation of polymerase action that isolates the ‘tethered delivery’ steps while ignoring other aspects of polymerase function and biochemistry. We assumed that every curved, longitudinally-bound ‘singleton’ αβ-tubulin at the MT end is engaged by one TOG domain of a polymerase (Figure 7D, left panel). This assumption provides end-localized polymerase in the model without having to parameterize/describe its diffusive movement on the microtubule lattice (this movement is mediated by the basic domain, which is required for polymerase function), or if it arrives at the microtubule end with its TOGs pre-loaded with αβ-tubulin (XMAP215 has been observed to carry αβ-tubulin while diffusing on the lattice). To mimic the tethering-induced local increase in αβ-tubulin concentration, in the model we accelerated the rate of αβ-tubulin association to the site neighboring the polymerase-bound singleton (Figure 7D, arrow). We do not assume that the linker is directing the tethered subunit to a particular site–separate simulations showed that accelerating the rate of αβ-tubulin association on top of the polymerase-bound singleton had no effect on growth rates (data not shown). We also assumed that when the polymerase bound singleton acquired a lateral neighbor it became straight (Rice et al., 2008; Buey et al., 2006), and that this straightening was linked to TOG disengagement (Ayaz et al., 2012).
To explore if enhanced trapping of singleton αβ-tubulins could recapitulate catalyst-like activity, we simulated microtubule elongation rates using a range of accelerated tethered association rates. In reality, the tethered association rate is likely to be more or less constant, with maximal polymerase activity instead determined by the degree to which it saturates its binding sites on the plus end. Varying the tethered association rate provides a way to mimic this ‘saturation’ effect without having to model it explicitly. Strikingly, the simulated microtubule elongation rates increase with the tethered αβ-tubulin association rate while the apparent equilibrium constant for elongation (x-intercept) remains approximately constant (Figure 7D, middle; Figure 7—figure supplement 1). Affecting the rate but not the apparent equilibrium indicates that tethering-enhanced associative trapping of weakly-bound singletons can recapitulate the essence of catalyst-like activity. The increase in simulated growth rate (less than 10-fold) is significantly less than the maximal increase in tethered αβ-tubulin association rate we explored (Figure 7D, right; Figure 7—figure supplement 1). Thus, in addition to capturing catalyst-like action, our simple model also shows the modest overall stimulation that is characteristic of these polymerases.
Why did the model only stimulate elongation modestly, as the polymerases do? We speculated that because longitudinally associated αβ-tubulin is weakly/transiently bound at the microtubule end (∼mM affinity in our fitted parameters and those of comparable models (VanBuren et al., 2002; Gardner et al., 2011); this compares to working concentrations of αβ-tubulin that are typically at most 20 μM), the low population and lifetime of this ‘substrate’ was limiting the activity obtainable in our model.
We probed the linkage between ‘substrate scarcity’ and polymerase activity in two ways. We first created an alternative model in which the polymerase accelerated delivery of αβ-tubulin adjacent to higher affinity and much longer-lived ‘corner’ (longitudinal + lateral interactions) sites, at least one of which is always present at the microtubule end (Figure 7E, left). In this alternative model the simulated elongation rates increase linearly with the increased effective concentration (Figure 7E, right), yielding much stronger stimulation that also lacks catalyst-like characteristics (Figure 7E, middle). This result is consistent with the idea that catalyst-like activity is related to the low affinity/lifetime of the end-bound singleton substrate. To explore this implication more directly we obtained parameter sets with higher affinity longitudinal interactions by using different assumed values of kon. As observed previously (VanBuren et al., 2002), these alternative parameterizations can approximate the experimental rates of (uncatalyzed) microtubule elongation by compensating for faster/slower kon with weaker/stronger longitudinal affinity (but retaining the same ‘corner’ [longitudinal + lateral] affinity). We observed in simulations that stronger longitudinal affinity gave higher maximal polymerase activity (Figure 7D, right; see also Figure 7—figure supplement 1). These results support the notion that the polymerase activity in our model is related to the population/lifetime of longitudinally-associated αβ-tubulin singleton at the microtubule end.
The tethering model suggests a plausible transition state/collision complex that limits the rate of microtubule polymerization in the absence of a polymerase (Figure 8). The idea is that at the MT end, longitudinally-bound αβ-tubulins tend to dissociate much faster than the rate at which unpolymerized αβ-tubulins associate at neighboring sites. Accordingly, most of the ‘singleton’-type associations are unproductive for elongation, and the MT grows primarily by additions into ‘corner’ sites. In the model we propose, Stu2p/XMAP215 polymerases enhance elongation by selectively concentrating αβ-tubulin near singleton sites, thereby specifically promoting their associative trapping. We propose that the ‘trapping complex’—two side-by-side αβ-tubulins without additional neighbors to either side–may be the transition state. In our model the polymerase catalyzes formation of this ‘side-by-side’ state by selectively concentrating (tethering) the reactants that lead to it. Associative trapping of singletons also ‘roughens’ the microtubule end through the creation of additional, longer-lived ‘corner’ sites. By virtue of having more favorable sites of interaction, these rougher microtubule ends can transiently capture unpolymerized subunits more efficiently and independently of the polymerase, through ‘filling in’ (a contribution from filling-in was anticipated previously [Brouhard et al., 2008]).
Because our model does not explicitly represent the polymerase, it cannot explain the molecular mechanisms underlying processive action. Perhaps basic-domain mediated rapid diffusion on the microtubule lattice, as has been measured for XMAP215 (Brouhard et al., 2008), is sufficient for the liberated polymerase to engage a newly added αβ-tubulin.
In summary, we propose that Stu2p/XMAP215 polymerases localize to the plus end by virtue of preferential interactions between TOG domains and curved αβ-tubulin. Our proposal is consistent with the view that curved, GTP-bound αβ-tubulin is an on-pathway polymerization intermediate (e.g., Buey et al., 2006; Rice et al., 2008; Nawrotek et al., 2011). It also explains the need for at least two TOG domains (Widlund et al., 2011) because one TOG is required to bind curved αβ-tubulin on the microtubule and another linked TOG is required for tethering an unpolymerized αβ-tubulin. In this model, side-by-side trapping and the release of TOG domains will be linked because of the polymerization-associated conformational changes that occur in αβ-tubulin. The model provides a simple molecular explanation for the catalytic-like action of the polymerase, recapitulates the modest maximal stimulation, and rationalizes how a polymerase can be built from domains that bind preferentially to a microtubule-incompatible conformation of αβ-tubulin.
Plasmids to express TOG2 and wild-type or polymerization-blocked yeast αβ-tubulin have been described previously (Johnson et al., 2011; Ayaz et al., 2012) and were used without further modification. Some plasmids (to express ‘full-length’ Stu2p and its R200 and R519 mutants) for the rescue assay were also previously described (Ayaz et al., 2012). ‘Dimerization-impaired’ plasmids were constructed using mutagenesis primers to delete the region that codes for the coiled-coil dimerization element (corresponding to amino acids 658–761). These primers contained two regions that were complementary to the coding sequences just upstream and downstream of the deleted region. ‘Linker truncation’ plasmids were constructed using a similar strategy to remove increasingly large sections from the middle of the sequence linking TOG1 to TOG2 (deleted regions are: Δ14:273-286; Δ22:277-298; Δ32:273-304; Δ40:257-296; Δ50:251-300; Δ60:251-310). The ‘linker shuffled’ and ‘TOG swapped’ plasmids were made using GeneArt seamless cloning (Invitrogen, Carslbad, CA). Shuffled linkers were inserted starting from synthetic, codon-optimized cDNA (IDT, Coralville, IA), so we also prepared a codon optimized natural linker sequence to control for possible effects on expression levels. DNA coding for shuffled versions of the 65 central amino acids (residues 252–316) of the TOG1-TOG2 linker was obtained from IDT and amplified using primers to append flanking nucleotides of wild-type upstream and downstream flanking sequence. The parent plasmid, excluding the region coding for the to-be-replaced linker sequence, was amplified in a separate reaction. The two products were mixed, incubated, and transformed according to the manufacturer's instructions. A similar strategy was used to replace TOG1 (residues 11–245) with a second copy of TOG2 (residues 318–560). The entire coding region of all plasmids was verified by DNA sequencing.
The TOG1(1–317), TOG2(318–560), and TOG1-TOG2(1–560) were expressed in bacteria with C-terminal His6 tags and purified using Ni-affinity and ion exchange chromatography (Ayaz et al., 2012). Wild-type or polymerization-blocked yeast αβ-tubulin (β:T175R,V179R) were overexpressed in Saccharomyces cerevisiae (Johnson et al., 2011). All proteins were purified as previously described (Johnson et al., 2011; Ayaz et al., 2012).
For crystallization, pure polymerization blocked yeast αβ-tubulin (β:T175R,V179R) and TOG2 domain were dialyzed into RB150 (25 mM Tris pH 7.5, 150 mM NaCl, 1 mM MgCl2, 1 mM EGTA) and then mixed at equimolar stoichiometry. One tenth volume of 2 M L-proline was added before concentrating the protein using 30 kDa cutoff Amicon Ultra concentrators. When the desired concentration was reached (∼2–3 mg/ml), GTP was added from a 100 mM stock to achieve a final concentration of 1 mM. Sparse-matrix crystallization screening (typically mixing protein with precipitants at 1:1 and 2:1 protein:reservoir ratios, using 200 nl of reservoir solution) was performed using a Phoenix DT Drop Setter (Rigaku, The Woodlands, TX). Needle- and blade-like crystals were obtained in multiple PEG-containing conditions, and these hits were optimized using finer screens. The crystal used for data collection was grown from 25% (vol/vol) PEG 3350, 0.25 M (NH4)2SO4, 0.1 M MES pH 6.1). For harvesting, cryoprotectant (reservoir solution supplemented with 100 mM NaCl and 40% [vol/vol] glycerol) was added directly to the drop before looping crystals for freezing in liquid nitrogen in liquid nitrogen.
Diffraction data were collected at Argonne National Laboratory using APS beamline 19ID using remote data collection. Diffraction data were processed using HKL2000 (Otwinowski and Minor, 1997). Crystals adopt space group C2 with one complex in the asymmetric unit. The diffraction was anisotropic (Table 1). We used 2.81 Å as the high-resolution cut-off to avoid excessive loss of completeness.
All crystallographic calculations after diffraction data processing were performed using Phenix (Adams et al., 2010). Phases were obtained by molecular replacement, using as search models the yeast α- and β-tubulin chains from PDB 4FFB (Ayaz et al., 2012) with GTP removed, and the structure of TOG2 from Stu2p (PDB 2QK1 [Slep and Vale, 2007]). Model building was performed in Coot (Emsley et al., 2010). Disordered regions were removed, and GTP and new sidechains were manually placed where the electron density indicated that was appropriate. The model was refined conservatively, and based on the behavior of Rfree and other indicators we decided that a combination of TLS with two grouped B-factors for each residue represented an optimal strategy. We also ran tests to optimize the relative weighting of the X-ray and chemical restraints, consistently observing that optimal results (low Rfree and low Rfree-Rwork) were obtained when the covalent geometry was tightly restrained to ideal values.
Samples for analytical ultracentrifugation (TOG1, TOG2, TOG1-TOG2, and polymerization-competent yeast αβ-tubulin) were dialyzed into RB100 (25 mM Tris pH 7.5, 1 mM MgCl2, 1 mM EGTA, 100 mM NaCl) containing 20 μM GTP. Samples were mixed and incubated at 4°C for at least one hour prior to the experiment. All analytical ultracentrifugation experiments were carried out in an Optima XL-I centrifuge using an An50-Ti rotor (Beckman–Coulter, Brea, CA). Approximately 390 μl of each sample were placed in charcoal-filled, dual-sector Epon centerpieces. Sedimentation (rotor speed: 50,000 rpm) was monitored using absorbance optics, and centrifugation was conducted at 20°C after the centrifugation rotor and cells had equilibrated at that temperature for at least 2.5 hr. Protein partial-specific volumes, buffer viscosities, and buffer densities were calculated using SEDNTERP (Laue et al., 1992).
The c(s) distributions were generated using SEDFIT (Schuck, 2000; Schuck et al., 2002). Isotherms for TOG1, TOG2, and TOG1-TOG2 binding to αβ-tubulin were assembled by importing c(s) distributions into GUSSI (available at http://biophysics.swmed.edu/MBR/software.html) and integrating them to obtain the weighted-average sedimentation coefficients as a function of TOG concentration. These isotherms were exported to SEDPHAT (Dam and Schuck, 2005) for evaluation. For the binding of TOG1 and TOG2 to αβ-tubulin, a simple bimolecular binding model was used. For the binding of TOG1-TOG2 to αβ-tubulin, a two-site binding model (i.e., two αβ-tubulins binding to one TOG1-TOG2) with two microscopic association constants was used. Quantities measured in separate experiments (s-values of the individual components, s-values of TOG1-TOG2:(αβ-tubulin)1, and the association constants derived with the single-domain TOG constructs) were treated as fixed parameters. We used a range of possible s-values for the TOG1-TOG2:(αβ-tubulin)2, informed by different assumptions about cooperativity as shown in Figure 5C and by hydrodynamic calculations (García De La Torre et al., 2000). To calculate sedimentation coefficients, we first prepared compact ‘longitudinal’ and ‘lateral’ models by docking the TOG1 and TOG2 complexes onto the αβ-tubulins in the stathmin complex (pdb 1SA0 [Ravelli et al., 2004]) and a section of microtubule (coordinates provided by Ken Downing, Lawrence Berkeley lab), respectively. Extended models were generated by translating one TOG complex relative to the other. To calculate coordinates for the linker, we used the Modeller (Sali and Blundell, 1993) plugin to UCSF Chimera (Pettersen et al., 2004), calculating five different linker traces for each model in an attempt to capture some of the likely variability. Each resulting model was passed to HYDROPRO (García De La Torre et al., 2000), and sedimentation coefficients were calculated using a solvent density of 0.99823 g/ml and otherwise default settings.
The genetic rescue assays were adapted from Al-Bassam et al. (2006) and performed as previously described (Ayaz et al., 2012). The parent strain was CUY1147ΔLEU2 (CUY1147 (Kosco et al., 2001) in which the LEU2 gene was replaced with a marker coding for G418 resistance). This strain expresses a tagged version of Stu2p that is selectively degraded upon exposure to copper. The parent plasmid was pWP70 (Wang and Huffaker, 1997), a CEN plasmid that expresses (non-degradable) HA-tagged Stu2p from its endogenous promoter. For rescue assays CUY1147ΔLEU2 transformed with rescue plasmid or empty vector was grown overnight in CSM-Leu media, normalized to A600 = 1, and plated at serial 10-fold dilutions onto CSM-Leu plates with or without 500 μM CuSO4. Plates were incubated at 30°C or room temperature and imaged after several days.
To collect a large number of TOG-domain containing sequences, we performed psiblast (Altschul et al., 1997) (non-redundant [nr] database, with e-value = 0.005 over three iterations) on amino acid sequences from TOG domains of known structure: the TOG1 and TOG2 domains from Stu2p (S. cerevisiae), the TOG2 domain from XMAP215 (Xenopus laevis) (Slep and Vale, 2007), and the TOG3 domain from Zyg9 (Caenorhabditis elegans) (Al-Bassam et al., 2007). The full protein sequences for all hits from each of the four runs were combined and identical hits were removed. They were then clustered using CLANS (Frickey and Lupas, 2004) (default parameters) over 130 iterations. The Stu2 cluster was extracted and aligned using mafft (Katoh and Standley, 2013) and passed to Jalview (Waterhouse et al., 2009). Proteins that were missing residues in highly conserved regions of TOG1 and TOG2, in particular in the boundaries adjacent to the start and end of the linker, were manually pruned. Based on the structures, we defined the boundaries for the end of TOG1 and beginning of TOG2 as K243 and L319, respectively. A linker for a given protein was measured counting these boundary residues and all residues in between. To reduce ‘overcounting’ artifacts, before making the histogram we thinned the set of sequences using the ‘remove redundancy’ (threshold = 0.99) feature of Jalview.
We wrote a computer program to perform kinetic Monte Carlo simulations of microtubule elongation. We implemented an algorithm similar to one described previously (VanBuren et al., 2002). In the program, the microtubule lattice is represented by a two dimensional lattice with a staggered periodic boundary condition to mimic the cylindrical microtubule structure. A 5 × 13 section of the lattice was designated as a ‘seed’ and considered to be permanently occupied. The program simulates microtubule elongation one biochemical reaction (in this case subunit addition or dissociation, GTP hydrolysis is ignored) at a time. A lattice site is considered available for subunit addition if it is empty and a neighboring site is occupied. The rate of subunit addition into any available site is given by kon*[αβ-tubulin]. Occupied sites (excepting the seed) are considered available for dissociation, with the rate of dissociation given by kon*KD where KD is the affinity of interaction, which is determined by the neighbor state (number and type of lattice contacts) and obtained from longitudinal and corner affinities through thermodynamic coupling (Erickson and Pantaloni, 1981; VanBuren et al., 2002). Simulations begin with only 13 possible events, each on an association onto the end of a protofilament. Possible events are stored in an indexed priority queue, sorted by their ‘execution times’ (Gibson and Bruck, 2000). Execution times are determined by first sampling a random number x between 0 and 1, and then calculating the time as −(1/rate)*ln(x), where rate gives the appropriate first- or pseudo-first order rate constant. At each step the event with the shortest execution time is implemented, the simulation time is advanced accordingly, and the list of possible events and their associated rates is updated to account for changes in subunit neighbor state. To obtain the length (in μm) of a simulated microtubule at a given time, we divide the number of subunits by 1625, the number of αβ-tubulin subunits in 1 μm of microtubule.
To model our proposed mechanism for polymerase action, we accelerated the rate at which unpolymerized subunits add adjacent to longitudinally associated ‘singletons’ at the microtubule end (or next to more tightly bound ‘corner’ subunits) (Figure 7). This acceleration mimics a ‘tethering’ effect. Control experiments showed that accelerating the rate for adding on top of a longitudinally-associated singleton had negligible effect (not shown). To keep the model as simple as possible, we did not account for any potential stabilization of TOG-bound singletons at the microtubule end.
PHENIX: a comprehensive Python-based system for macromolecular structure solutionActa Crystallographica Section D, Biological Crystallography 66:213–221.https://doi.org/10.1107/S0907444909052925
Fission yeast Alp14 is a dose-dependent plus end-tracking microtubule polymeraseMolecular Biology of the Cell 23:2878–2890.https://doi.org/10.1091/mbc.E12-03-0205
Stu2p binds tubulin and undergoes an open-to-closed conformational changeThe Journal of Cell Biology 172:1009–1022.https://doi.org/10.1083/jcb.200511010
Gapped BLAST and PSI-BLAST: a new generation of protein database search programsNucleic Acids Research 25:3389–3402.https://doi.org/10.1093/nar/25.17.3389
MolProbity: all-atom structure validation for macromolecular crystallographyActa Crystallographica Section D, Biological Crystallography 66:12–21.https://doi.org/10.1107/S0907444909042073
Modulation of the dynamic instability of tubulin assembly by the microtubule-associated protein tauMolecular Biology of the Cell 3:1141–1154.https://doi.org/10.1091/mbc.3.10.1141
Efficient exact stochastic simulation of chemical systems with many species and many channelsAmerican Chemical Society pp. 1876–1889.
A microtubule-associated protein from Xenopus eggs that specifically promotes assembly at the plus-endThe Journal of Cell Biology 105:2203–2215.https://doi.org/10.1083/jcb.105.5.2203
Role of GTP hydrolysis in microtubule dynamics: information from a slowly hydrolyzable analogue, GMPCPPMolecular Biology of the Cell 3:1155–1167.https://doi.org/10.1091/mbc.3.10.1155
MAFFT multiple sequence alignment software version 7: improvements in performance and usabilityMolecular Biology and Evolution 30:772–780.https://doi.org/10.1093/molbev/mst010
Control of microtubule dynamics by Stu2p is essential for spindle orientation and metaphase chromosome alignment in yeastMolecular Biology of the Cell 12:2870–2880.https://doi.org/10.1091/mbc.12.9.2870
90–125, Cambridge, The Royal Society of Chemistry90–125, Cambridge, The Royal Society of Chemistry.
Refined structure of alpha beta-tubulin at 3.5 A resolutionJournal of Molecular Biology 313:1045–1057.https://doi.org/10.1006/jmbi.2001.5077
The mechanisms of microtubule catastrophe and rescue: implications from analysis of a dimer-scale computational modelMolecular Biology of the Cell 23:642–656.https://doi.org/10.1091/mbc.E11-08-0688
The determinants that govern microtubule assembly from the atomic structure of GTP-tubulinJournal of Molecular Biology 412:35–42.https://doi.org/10.1016/j.jmb.2011.07.029
Cold-sensitive and caffeine-supersensitive mutants of the Schizosaccharomyces pombe dis genes implicated in sister chromatid separation during mitosisThe EMBO Journal 7:1465–1473.
Methods in Enzymology307–326, Methods in Enzymology, Elsevier, vol 276.
UCSF Chimera–a visualization system for exploratory research and analysisJournal of Computational Chemistry 25:1605–1612.https://doi.org/10.1002/jcc.20084
The lattice as allosteric effector: structural studies of alphabeta- and gamma-tubulin clarify the role of GTP in microtubule assemblyProceedings of the National Academy of Sciences of USA 105:5378–5383.https://doi.org/10.1073/pnas.0801155105
Comparative protein modelling by satisfaction of spatial restraintsJournal of Molecular Biology 234:779–815.https://doi.org/10.1006/jmbi.1993.1626
Identification of XMAP215 as a microtubule-destabilizing factor in Xenopus egg extract by biochemical purificationThe Journal of Cell Biology 161:349–358.https://doi.org/10.1083/jcb.200211095
Mechanochemical model of microtubule structure and self-assembly kineticsBiophysical Journal 89:2911–2926.https://doi.org/10.1529/biophysj.105.060913
Estimates of lateral and longitudinal bond energies within the microtubule latticeProceedings of the National Academy of Sciences of USA 99:6035–6040.https://doi.org/10.1073/pnas.092504999
Dynamic instability of individual microtubules analyzed by video light microscopy: rate constants and transition frequenciesThe Journal of Cell Biology 107:1437–1448.https://doi.org/10.1083/jcb.107.4.1437
Stu2p: a microtubule-binding protein that is an essential component of the yeast spindle pole bodyThe Journal of Cell Biology 139:1271–1280.https://doi.org/10.1083/jcb.139.5.1271
XMAP215 polymerase activity is built by combining multiple tubulin-binding TOG domains and a basic lattice-binding regionProceedings of the National Academy of Sciences of USA 108:2741–2746.https://doi.org/10.1073/pnas.1016498108
Stephen C HarrisonReviewing Editor; Harvard Medical School, United States
eLife posts the editorial decision letter and author response on a selection of the published articles (subject to the approval of the authors). An edited version of the letter sent to the authors after peer review is shown, indicating the substantive concerns or comments; minor concerns are not usually shown. Reviewers have the opportunity to discuss the decision before the letter is sent (see review process). Similarly, the author response typically shows only responses to the major concerns raised by the reviewers.
Thank you for sending your work entitled “A tethered delivery mechanism explains the catalytic action of a microtubule polymerase” for consideration at eLife. Your article has been favorably evaluated by Randy Schekman (Senior editor) and 4 reviewers, one of whom, Steve Harrison, is a member of our Board of Reviewing Editors, and one of whom, William Hancock, has agreed to reveal his identity.
The Reviewing editor and the other reviewers discussed their comments before we reached this decision, and the Reviewing editor has assembled the following comments to help you prepare a revised submission.
This manuscript reports the structure of αβ-tubulin-bound Stu2 TOG2 and shows that the tubulin heterodimer is in the “curved” configuration, like the tubulin bound to TOG1. Further experiments show that tubulin binding to two linked domains is not cooperative. A series of experiments with “designed” Stu2 fragments then shows that any two TOGs on one molecule (either monomer or dimer) will give nearly normal catalytic activity. The authors find a minimum linker length for full activity of about 40 residues (consistent with the ca.300 TOG-TOG sequences in the database); the linker sequence is irrelevant. The manuscript then goes on to suggest a specific model for polymerization and to report results of simulations that embody the model. The notion is that one of the two required TOGs binds an αβ-tubulin that has associated longitudinally with the end of a MT, rather than at a “corner”, where the helical array creates lateral as well as longitudinal contacts. The former contact is, of course, weaker than the latter, but the model postulates that the high local concentration of αβ-tubulin created by the relatively weakly bound heterodimer on the second TOG domain (or presumably on any of the additional ones in a dimer or in a many-TOG monomer) promotes addition adjacent to the first heterodimer. Lateral contacts then enforce a straight conformation for both, releasing the catalyst and stabilizing the added pair as part of the growing end.
The reviewers agreed that the experimental data are sound and that they impose interesting constraints on the mechanism of polymerization. We invite a revision that attends to the following questions and reservations.
1) The authors used HYDROPRO for the prediction of the sedimentation coefficients of the extended and laterally and longitudinally associated tubulin subunits with Stu2. As this information was used to interpret the AUC binding experiments of TOG1-TOG2 with tubulin using different assumption about cooperativity, the authors should provide more details about the exact way these calculations were performed. In this context, it is a bit surprising that the sedimentation coefficients for the two putative laterally and longitudinally arranged tubulin subunits in complexes with Stu2 are similar. Do the authors have an explanation for this similarity?
2) Modeling. The reviewers found the explanation of the rationale for the model – and even its specific details – hard to follow, although there was agreement that the model is interesting and plausible enough to be included. Because the data will stand, even if future work alters the model, the authors might want consider moving the entire model description and outcome into the Discussion, or in some other way underscore the distinction. A revision of the manuscript should in any case respond to the following points:
a) The authors considered only one possible mode of action, in which one TOG domain of Stu2 binds to a tubulin subunit at the distal end of a microtubule while the other TOG domain binds to soluble tubulin. Moreover, they did not consider the role of the basic microtubule binding domain, which is well known to be crucial for the XMAP215 polymerase activity. They should explain why they simplified their system to such a degree and why they did not consider additional possibilities.
b) The model assumes that when a singleton tubulin adds, Stu2p instantaneously binds to it. Brouhard et al. measured D=0.3 µm2/s for XMAP215, which means it would take 13 µsec to diffuse 8 nm. That seems pretty fast. But singletons also dissociate very rapidly. For kon=4 x 106 M-1s-1 and 8 mM KD of longitudinal affinity used by Ayez, koff = 3.2 x 104 s-1, meaning that singletons have a 30 µsec residence time if not stabilized. So it is likely that some singletons will not have a bound Stu2p, and association is not actually instantaneous. This effect will not fundamentally change the model performance – it is one of degree, and it will be diminished with stronger longitudinal associations simulated in Figure 7 – but the authors should comment.
c) The model assumes that the second TOG domain will come with an attached tubulin. If the model were that one TOG domain binds the singleton tubulin and then the second binds a free tubulin, that would be a problem – even with a 107 M-1s-1 on-rate and 1 µM tubulin concentration, the wait would be 100 µsec (longer than the association time of the singleton). Earlier in the paper, the authors find the Kd for tubulin binding to the TOG domains to be 100nM. That means that at 1µM free tubulin, 90% of TOG domains will have tubulin bound. So the assumption that the second TOG comes with a bound tubulin seems valid. One corollary, however, is that with µM free tubulin concentrations in solution, the TOG domains should all have a bound tubulin when the Stu2p is diffusing on the microtubule, held there by its basic domain, and the singleton will also be delivered by the Stu2p in most cases. It would have to be, because less than 10% of the Stu2p diffusing on the microtubule will have a free site that can bind the singleton. This point doesn't really change the model behavior, but it does give a picture of tubulin-bound TOGs diffusing on the microtubule that isn't conveyed in the paper. (See also comment a above).
d) The corner model, which is not consistent with the data, is a nice control. It was, however, difficult to clarify the differences between the two models – in both of them there is a “corner effect”, but one is with a curved singleton and the second is with a straight and stabilized subunit. The authors could do some work to clarify this difference.
e) The y-axis of the growth rate plots in Figure 7 are in units µm/µM-min. Unless the reviewers are missing something, the units should be µm/min. When measuring the slope, the concentration comes in, but that's a different calculation.https://doi.org/10.7554/eLife.03069.013
- Luke M Rice
- Elisabeth A Geyer
- Felipe-Andrés Piedra
- Zbyszek Otwinowski
- Emily S Vu
- Nick V Grishin
- Luke M Rice
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank D Borek for assistance with diffraction data processing, D Tomchick and Z Chen in the UT Southwestern Structural Biology Core facility for help and advice. H Yu, D Borek, X Zhang, and E Goldsmith gave comments on the manuscript. LMR is the Thomas O Hicks Scholar in Medical Research. Results shown in this report are derived from work performed at Argonne National Laboratory, Structural Biology Center at the Advanced Photon Source, beamline 19-ID. Argonne is operated by UChicago Argonne, LLC, for the US Department of Energy, Office of Biological and Environmental Research under contract DE-AC02-06CH11357. F-AP and EG were supported by NIH T32 GM008297, and ESV was supported by an NSF predoctoral fellowship. This work was supported by grants NIH GM098543, and NSF MCB1054947. Coordinates have been deposited in the Protein Data Bank, accession code 4U3J.
- Stephen C Harrison, Reviewing Editor, Harvard Medical School, United States
© 2014, Ayaz et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.