To study how Xenopus Cdk5rap2 affects γTuRC’s activity, we added its purified γTuNA domain (Figure 1A–B; Figure 1—figure supplement 1; aa 56–89, isoform X1) to Xenopus laevis egg extract and assessed its impact on microtubule (MT) nucleation (Figure 1C). Using total internal reflection (TIRF) microscopy and fluorescent end binding protein 1 (EB1) to label growing MT plus ends, we quantified individual MT nucleation events (Figure 1C–D). In the control reaction, the egg extract showed a typical low level of MT nucleation (Figure 1C, ‘buffer’,~3 MTs per field). In contrast, addition of wildtype γTuNA triggered an increase in MT nucleation of up to ~75-fold in a titration series (Figure 1C–D). The Xenopus F71A mutant equivalent to the human F75A mutant (Figure 1B), hereafter referred to as ‘F75A’, did not significantly increase MT number even at the highest concentration (3.6 µM, Figure 1C–D). Thus, the γTuNA domain activates MT nucleation in extract and requires the F75 residue, validating prior studies (Fong et al., 2008; Choi et al., 2010). Using sucrose gradients to fractionate mock and γTuNA-treated extracts, we also conclude that the γTuNA domain has no effect on γTuRC assembly, ruling out one possible explanation for this increase in MT number (Figure 1—figure supplement 2). While we cannot rule out that full-length Cdk5RAP2 might affect γTuRC assembly, we believe this is also unlikely as recent work has demonstrated that γTuRC can be assembled via heterologous expression of just γTuRC components and the RUVBL1-RUVBL2 AAA ATPase complex, without addition of a CM1-containing protein (Zimmermann et al., 2020).

We next confirmed whether the γTuNA domain’s ability to increase MT nucleation in extract was dependent on the known MT nucleator, γTuRC. To do this, we first attempted depleting γTuRC from extract using our previously published rabbit-derived, anti-gamma tubulin antibody (Thawani et al., 2020). This γTuRC-depleted extract would then be assayed in the presence of γTuNA in our TIRF assay. However, due to low antibody yields and batch-to-batch variability, we were unable to generate γTuRC-depleted extract at consistent levels via this method. As an alternative, we instead depleted extracts of γTuRC via pulldown of γTuNA-coupled beads. With a single round of depletion, we observed a loss of >75% of γ-tubulin signal indicating a depletion of γTuRC (Figure 1E). In the mock-treated extract where γTuRC was not depleted, the γTuNA domain’s ability to increase MT nucleation levels remained unchanged (Figure 1F). By contrast, exogenous γTuNA no longer activated MT nucleation in γTuRC-depleted extracts (Figure 1F). Hence, the γTuNA domain requires the universal MT template, γTuRC, to activate MT nucleation.

As γTuNA co-depletes γTuRC, we wondered whether this interaction would be sufficient to generate artificial MT asters (Figure 2A). To that end, we coated micron-scale beads with wildtype or mutant γTuNA domains and added them to extract. After a pulldown step, we assayed these beads for MT aster formation in vitro in the presence of purified fluorescent tubulin and GTP under oblique TIRF (Figure 2B). We found that wildtype γTuNA-coated beads formed large MT asters mimicking the potent MT nucleation of the centrosome (Figure 2B). In contrast, the F75A mutant beads formed severely impaired asters (Figure 2B). Mock-treated beads did not form asters. To confirm the stable presence of γTuRC, we repeated the bead pulldown from extract and attached the beads via an antibody against Mzt1, a γTuRC subunit, to surface-treated coverslips (Figure 2A). We then added fluorescent tubulin and GTP before live imaging via TIRF microscopy (Figure 2C). Critically, we observed that wildtype γTuNA beads attached and formed large MT asters in vitro, indicating that these beads had retained γTuRC and any other necessary MT nucleation factors (Figure 2C, Video 1).

Figure 2 with 3 supplements see all

Download asset Open asset

Figure 2—source data 1

Uncropped images for Figure 2.

https://cdn.elifesciences.org/articles/80053/elife-80053-fig2-data1-v2.zip

Download elife-80053-fig2-data1-v2.zip

Video 1

Download asset

From this we conclude that γTuNA domains are sufficient to specify new sites of γTuRC-mediated MT nucleation. Critically, this finding allowed us to develop a new γTuRC purification scheme based on scaled-up pulldowns with Halo-human γTuNA (outlined in Figure 2D–F), which we discuss in more detail later. Finally, the ability of F75A beads to weakly nucleate asters points to residual, but persistent, binding of γTuRC. We believe this is due to the low stringency wash particular to this experiment, as we do not detect γTuRC on F75A beads after higher stringency washes in subsequent experiments (western blots in Figure 2—figure supplement 1C, Figure 3D–E).

Figure 3 with 1 supplement see all

Download asset Open asset

Figure 3—source data 1

Numerical data for Figure 3, includes normalized size-exclusion chromatography traces for Figure 3B, quantified pulldowns in Figure 3F, and quantified pulldowns in Figure 3G.

https://cdn.elifesciences.org/articles/80053/elife-80053-fig3-data1-v2.zip

Download elife-80053-fig3-data1-v2.zip

Figure 3—source data 2

Labeled and raw blots used in Figure 3D and E.

https://cdn.elifesciences.org/articles/80053/elife-80053-fig3-data2-v2.zip

Download elife-80053-fig3-data2-v2.zip

Having confirmed that the γTuNA domain strongly recruits γTuRC from extract, we next investigated the γTuNA-γTuRC interaction. In a recent structural study, the authors generated a model of a parallel coiled-coil that directly interacts with γTuRC (Wieczorek et al., 2020a). The authors suggested that this coiled-coil is in fact a γTuNA dimer, although biochemical validation of this dimer state and its effect on γTuRC activity were not provided (Wieczorek et al., 2020a). To that end, we selectively mutated hydrophobic residues found within a heptad-repeat region of γTuNA. Specifically, we mutated the hydrophobic residues F63, I67, L70, and L77 to either alanine or aspartate (Figure 3A). To validate the well-conserved nature of this domain, we generated both human and Xenopus versions, referred to here by the residue position in the human sequence (Figure 1B).

We initially focused on the double, triple, and quadruple mutants for both human and Xenopus γTuNAs. We performed size-exclusion chromatography (SEC) and compared the peak retention volumes of wildtype and mutated γTuNAs. Our SEC data revealed that wildtype γTuNA is a dimer (Figure 3B–C). By comparing the SEC traces for the double, triple, or quadruple mutants from both Xenopus and human γTuNAs, we found that γTuNA dimerization was dependent on residues I67, L70, and L77 (Figure 3B). The double hydrophilic mutants (I67D/L70D) from both human and Xenopus versions were entirely monomeric. This was also true for the human double-alanine mutant, I67A/L70A (Figure 3B).

To resolve each residue’s individual contribution to γTuNA dimerization, we generated alanine point mutants for F63, I67, L70, and L77 in Xenopus γTuNA. We also tested the F75A mutant of Xenopus γTuNA, as we wanted to know whether its loss-of-function coincided with loss of dimerization. We compared the SEC traces for these point mutants and found that mutating residues F63 or F75 to alanine had no deleterious effect on γTuNA dimerization (Figure 3B). By contrast, individually mutating residues I67, L70, or L77 increasingly interfered with dimerization, resulting in intermediate populations between full dimer and full monomer (Figure 3B–C). Mutation of the L70 or L77 residues resulted in the most drastic impairment, further confirming that this central region is crucial for γTuNA dimerization.

With the insight that the γTuNA domain is an obligate dimer, we next asked whether dimerization was required to bind γTuRC. We performed pulldowns of N-terminally Halo-tagged γTuNA mutants from Xenopus egg extract. We determined the amount of γTuRC bound for each γTuNA construct by probing for the γTuRC components GCP5 and γ-tubulin (Figure 3D–G). We found that both human and Xenopus double aspartate mutants (I67D/L70D), as well as the human triple mutant (I67D/L70D/L77D) did not bind γTuRC, indicating that loss of dimerization results in loss of γTuRC binding (Figure 3D and F). Interestingly, we found that the intermediate dimer mutants (I67A, L70A, or L77A) had correspondingly intermediate levels of γTuRC binding ability (Figure 3E). The I67A mutant, for example, was only weakly impaired in terms of dimerization (Figure 3B) and subsequently retained its ability to bind γTuRC (Figure 3E and G). As dimerization was increasingly impaired in the L70A and L77A mutants, γTuRC binding became increasingly weaker (Figure 3G). In the most extreme example, the L77A mutant, which had the most substantial dimerization defect, had complete loss of γTuRC binding (Figure 3G). Critically, the known F75A mutant did not bind γTuRC, as expected (Figure 3D–G). As our SEC data shows that F75A does not affect γTuNA dimerization, we conclude that both γTuNA dimerization and the F75 residue are required for binding γTuRC (Figure 3B and D–G). Finally, we found that forcing γTuNA dimerization via the addition of a constitutively dimeric coiled-coil domain (GCN4) did not rescue the ability of the intermediate dimer mutants to bind γTuRC (Figure 3—figure supplement 1). This suggests that simply bringing intermediate dimer mutants within tight proximity is not enough to induce restoration of the proper γTuRC binding interface.

Having identified specific mutations that impaired γTuNA’s ability to dimerize and bind γTuRC, we next asked what effect these mutants had on MT nucleation in extract. We added wildtype or mutant Xenopus γTuNA to freshly prepared extracts and again tracked MT plus-ends via fluorescent EB1 as a measure of MT number (Figure 4). As before, wildtype γTuNA triggered an increase in MT nucleation, when compared to the buffer control (Figure 4). The F75A mutant had little effect on extract MT levels (Figure 4). Similarly, the L77A mutant, which cannot bind γTuRC in extract (Figure 3G), did not increase MT nucleation (Figure 4). Intriguingly, when we examined the intermediate γTuRC-binding mutants I67A and L70A, we found that the I67A mutant activated MT nucleation to ~50% of wildtype levels, but L70A had no activity (Figure 4B). This was surprising as I67A’s activation effect was on the order of its γTuRC-binding ability (~50% vs~67%; compared to wildtype), suggesting binding ability was predictive of the activation effect in extract (Figure 3G). However, because the L70A mutant had little activity in extract (~6%, Figure 4B) but retained ~35% binding ability (Figure 3G), it appears that there is a threshold to γTuRC’s activation in extract. We further analyze the implications of this divergent behavior between γTuNA mutants in our Discussion.

Figure 4

Download asset Open asset

Figure 4—source data 1

Numerical data used in Figure 4.

https://cdn.elifesciences.org/articles/80053/elife-80053-fig4-data1-v2.zip

Download elife-80053-fig4-data1-v2.zip

While we had explored the effect of wildtype γTuNA and its dimer mutants on MT nucleation in extract, we had yet to determine if γTuNA directly increased γTuRC’s activity in vitro. As we briefly mentioned (Figure 2D–F), we used beads coupled to a Halo-human γTuNA construct to purify endogenous Xenopus γTuRC from extract (Figure 2D–F), similar to previous work (Wieczorek et al., 2020b). Mass spectrometry confirmed that the dominant co-precipitant was indeed Xenopus γTuRC (Figure 2—figure supplement 1). We also confirmed the presence of fully assembled γTuRC rings via negative-stain electron microscopy (Figure 2E–F and Figure 2—figure supplement 3). Using this purified γTuRC, we investigated the effect of wildtype and mutant γTuNAs on γTuRC’s activity in vitro via in vitro TIRF assays (Figure 5). In these assays, biotinylated γTuRCs were attached to passivated coverslips before imaging with TIRF microscopy (schematized in Figure 5—figure supplement 1). This not only offers high signal-to-noise but also allows tracking of individual γTuRC-mediated MT nucleation events.

Figure 5 with 3 supplements see all

Download asset Open asset

Figure 5—source data 1

Numerical data from Figure 5’s in vitro TIRF assays with purified γTuRC and γTuNA: including MT mass measurements, MT number, MT growth speed, and MT lengths.

https://cdn.elifesciences.org/articles/80053/elife-80053-fig5-data1-v2.zip

Download elife-80053-fig5-data1-v2.zip

We started by first comparing total MT mass generated in our assay (Figure 5B). Strikingly, the addition of γTuNA triggered a 5-fold increase in MT mass as compared to the buffer control (Figure 5B, Video 2 and Video 3). To determine if this was a direct stimulation of γTuRC’s activity, we then quantified the number of γTuRC-nucleated MTs within the first 150 s (Figure 5C), the MT nucleation rate (Figure 5D), the mean MT growth speed (Figure 5E), and the mean maximum MT length (Figure 5F). These quantifications revealed that wildtype γTuNA sharply increased the γTuRC nucleation rate from 1.2 MTs/s to 24.5 MTs/s (~20-fold increase, Figure 5C–D). While there was a slight increase in mean MT growth speed (+0.2 µm/min), this did not translate into a significant effect on MT length (Figure 5E–F). We also found that wildtype γTuNA saturated our assay within 30 s (Figure 5C), with a decreased nucleation rate of 0.15 MTs/s that remained constant for the remainder of the experiment (Figure 5—figure supplement 1). Thus, we conclude that γTuNA’s effect is almost exclusively due to a direct ~20-fold increase in γTuRC activity and not due to altered MT dynamics.

Video 2

Download asset

Video 3

Download asset

Incidentally, we also found that bulky N-terminal tags on γTuNA completely ablated its ability to activate γTuRC, instead turning it into a specific γTuRC repressor (see Figure 5—figure supplement 2).

As expected, the F75A mutant did not increase MT mass in vitro (Figure 5B). Furthermore, both the F75A and L77A mutants had no significant effect on the initial γTuRC nucleation rate (Figure 5D). However, we did observe that L77A caused a weakly significant increase in MT number beginning at 150 s (Figure 5C, p ~ 0.03). Similarly, we found that around 150 s L77A increased γTuRC’s nucleation rate 1.4-fold when compared to buffer (Figure 5—figure supplement 1). Beyond 150 s, the L77A mutant triggered a delayed increase in MT mass (Figure 5B), a behavior not observed in extract (Figure 4). From this discrepancy, we inferred that the presence of other factors in extract blocks impaired dimer mutants like L77A from interacting with or stimulating γTuRC.

To confirm that the γTuNA domain’s activation effect could be modeled simply as a change in γTuRC activity, we simulated the above experiments using our experimentally determined nucleation rates. We found that a simple deterministic model based on the initial nucleation rates was sufficient to capture most of the behavior in our system, except for the L77A mutant of γTuNA (Figure 5—figure supplement 3). Using a single constant growth speed for all conditions, we found that the nucleation rate completely captured the effect of wildtype and F75A γTuNA on MT number and mass (Figure 5—figure supplement 3). For L77A, our simulation suggests a two-phase behavior where its effect on nucleation rate increases at some late stage, possibly due to a shift from an impaired dimer state to complete dimer when bound to γTuRC (Figure 5—figure supplement 3F). Regardless, both the in vitro and simulated data demonstrate that wildtype γTuNA’s effect on MT nucleation is due to a direct increase in γTuRC activity without altering MT dynamics.

As there appeared to be an activation barrier in extract, but not in vitro, we investigated whether known negative regulators of MT nucleation were responsible. We focused on the tubulin-sequestering protein stathmin (or op18), which regulates the available tubulin pool for nucleation and polymerization ( Belmont and Mitchison, 1996; Gavet et al., 1998). For every mole of stathmin present (1.5 µM endogenous concentration), two moles of tubulin are removed (Gigant et al., 2000; Wühr et al., 2014).

We sought to first confirm that stathmin negatively regulated MT nucleation by γTuRC (as in our previous work; Thawani et al., 2020), and then determine whether γTuNA had any effect on this. To that end, we added increasing amounts of exogenous stathmin to extract and measured its effect on MT nucleation (Figure 6A–B). At double and triple the endogenous concentration of stathmin in extract (~3 µM or ~4.2 µM final), we observed a drastic loss of MT nucleation and polymerization (Figure 6A–B). Surprisingly, γTuNA was still able to activate MT nucleation (Figure 6A), even at the highest concentration of stathmin tested, with a ~ five-fold increase in the number of MTs (Figure 6B, 2.7 µM stathmin).

Figure 6

Download asset Open asset

Figure 6—source data 1

Numerical data used in Figure 6: including raw EB1 counts for stathmin/γTuNA experiments in extract, and MT mass measurement for in vitro TIRF assays with purified γTuRC, γTuNA, and stathmin.

https://cdn.elifesciences.org/articles/80053/elife-80053-fig6-data1-v2.zip

Download elife-80053-fig6-data1-v2.zip

Next, we assessed whether the γTuNA-γTuRC complex could also overcome stathmin’s effect in vitro (Figure 6C). For simplicity, we tracked the total MT signal (or MT mass) produced after 250 s in the TIRF-based nucleation assay (Figure 6D). We first observed the activity of γTuRC alone at 15 µM tubulin (Figure 6C, upper left panel). We next tested γTuRC in the presence of either 2.5 or 4 µM stathmin and found that stathmin resulted in losses of 59% and 81% MT mass, respectively (Figure 6C, ‘No γTuNA’ conditions). Interestingly, addition of γTuNA to stathmin and γTuRC rescued MT mass (Figure 6C–D, ‘+ γTuNA’). In fact, at 4 µM stathmin with γTuNA, MT mass levels were restored to the level of the γTuRC control (no stathmin, no γTuNA). This suggested that γTuNA indirectly counteracts the effect of stathmin in extract by increasing the efficiency of γTuRC-mediated MT nucleation at lower tubulin concentrations.

To test this possibility, we assayed γTuRC activity in vitro at its previously reported critical concentration of 7 µM tubulin; at or below this concentration γTuRC activity is minimal to non-existent (Consolati et al., 2020; Thawani et al., 2020). This tubulin concentration is also equivalent to that in our assay in Figure 6C, as 4 µM stathmin will remove 8 µM tubulin, leaving only 7 µM tubulin free in a 15 µM reaction. As expected, at this critical concentration γTuRC had little detectable activity with only ~1–2 MTs nucleated over a five-minute time course (Figure 6E, buffer). By contrast, γTuRC activity increased twenty-five-fold when in the presence of γTuNA (Figure 6E). Thus, γTuNA decreases γTuRC’s critical tubulin concentration, or restated, increases its ability to nucleate MTs at lower tubulin concentrations. We believe this result explains how γTuNA indirectly counteracts the inhibitory effect of stathmin on γTuRC.

In this work, we investigated how MT nucleation is regulated by Cdk5rap2’s γTuNA-mediated activation of γTuRC (Figure 6F). We showed that the γTuNA domain is an obligate dimer and that both dimerization and the F75 residue are crucial for binding γTuRC, providing the first biochemical validation for a recent structure of γTuRC bound to a putative γTuNA dimer (Wieczorek et al., 2020a). Moreover, we defined other core residues required for both γTuNA dimerization and subsequent γTuRC binding. We found that γTuNA dimers directly activate γTuRC-dependent MT nucleation in extract and in vitro. Finally, we uncovered that γTuNA dimers overcome barriers to MT nucleation posed by the tubulin sequestrator stathmin by enhancing γTuRC’s activity at low tubulin concentrations. Because γTuNA domains are also found in myomegalin and the branching factor TPX2, among others, our findings are broadly applicable to multiple MTOCs and model eukaryotes from yeast to humans.

It remains an open question how the γTuNA dimer directly enhances γTuRC’s nucleation activity. It is tempting to speculate that γTuNA binding triggers a conformational change of γTuRC from its wide diameter lattice to a closed state (Liu et al., 2020; Wieczorek et al., 2020a; Consolati et al., 2020). Because previous work has characterized the structures of γTuNA bound to human and Xenopus γTuRC and observed no obvious structural change (Liu et al., 2020; Wieczorek et al., 2020a; Wieczorek et al., 2020b), it is possible that binding a γTuNA dimer only transiently biases γTuRC toward the closed ring conformation. Furthermore, our own prior modeling suggested that the free energy provided by stochastic binding and lateral association of the first 3–4 tubulin dimers is sufficient to overcome the energy barrier between γTuRC’s open and closed states (Thawani et al., 2020). The γTuNA domain possibly lowers this energy barrier for ring closure, as evidenced by its ability to lower γTuRC’s critical tubulin concentration. Ultimately, detection of an activated state of vertebrate γTuRC may require the presence of tubulin, GTP, and the γTuNA domain combined with sophisticated structural methods.

Intriguingly, recent work in yeast suggests that near-closure, or biasing, of the γTuRC ring might require simultaneous binding by multiple CM1/γTuNA domains present in proteins like Spc110p (Brilot et al., 2021). Unlike in vertebrate structures of γTuRC, yeast γTuRC appears to have six well-defined CM1/γTuNA binding sites, where overhanging CM1 domains facilitate the formation of the ring by reinforcing lateral interactions between its constituent γ-tubulin small complex (γTuSC) subunits (Brilot et al., 2021). This might suggest that additional γTuNA binding sites may also be present in vertebrate γTuRCs, and simultaneous binding of multiple γTuNAs could further enhance γTuRC activity. As of yet, any additional γTuNA binding sites have not been detected in vertebrate γTuRC structures, but this possibility is exciting and can actively be investigated with single molecule imaging of fluorescently labeled γTuRC and γTuNA.

Altogether, our current model for γTuNA activation of γTuRC involves a direct binding event between a dimerized γTuNA-containing protein (e.g. Cdk5rap2) and γTuRC, which activates γTuRC’s MT nucleation ability ~20-fold. This activation overcomes negative regulation by stathmin in the cytoplasm or low local tubulin concentrations (Figure 6F). Once bound, these activated γTuRCs are enriched by specific MTOCs, like the centrosome, via additional localization motifs present in the γTuNA-containing protein. As Cdk5rap2 is also present in the cytoplasm it may be possible that some level of activation also occurs outside MTOCs, although recent work hints at how both access to the γTuNA domain and Cdk5rap2 localization is regulated via phosphorylation to prevent ectopic activation (Conduit et al., 2010; Conduit et al., 2014; Hanafusa et al., 2015; Feng et al., 2017; Tovey et al., 2021).

We note that the γTuNA domain’s enhancement of γTuRC activity may be greater than our reported 20-fold. Based on our analysis, an estimated 33% of our purified γTuRCs likely retain a γTuNA dimer at the end of the purification (Figure 2—figure supplement 2C). This means that our baseline (‘buffer’) level of γTuRC activity is likely higher than might otherwise be observed with a γTuNA-independent purification. We do not believe this slightly elevated baseline level interferes with demonstrating that γTuNA dimers activate γTuRC, but rather stress that γTuNA dimers might have an even more potent effect on γTuRC activity than we report here.

In the experiments presented in Figures 3 and 4, we found that the L70A and L77A mutants similarly formed intermediate SEC peaks between full dimer and monomer (Figure 3C) but had divergent γTuRC binding ability in extract pulldowns (Figure 3E). This suggests that the region of the γTuNA coil from position 75 to position 77 (F75, L77) is the core γTuNA-γTuRC binding interface, within which mutations are not well-tolerated for stable γTuRC interaction in extract. We observed that mutations at positions moving from this core towards the N-terminus (L70 to I67 to F63) had less and less impact on both dimerization and γTuRC binding ability in extract (Figure 3). Thus, the divergent behavior for L70A and L77A appears to be a result of L70’s position outside the most critical region, retaining a small amount of γTuRC binding. However, as our extract assays demonstrate in Figure 4, this small amount of binding by L70A is not sufficient to significantly activate γTuRC in extract.

Interestingly, we also found that the double I67A/L70A mutant had a strong loss of dimerization but still retained some γTuRC binding, just below the level of the L70A single point mutant (Figure 3E and G). By comparing this to the human I67D/L70D mutant (Figure 3F) we found that double substitution to aspartate, instead of alanine, completely removed this residual γTuRC binding. This suggests that retaining some hydrophobicity at these positions might preserve enough of the coil structure to allow for a weak interaction with γTuRC, despite lacking the required hydrophobicity to form a stable coiled-coil dimer (Figure 3A). In support of this, closer inspection of the peak SEC retention volumes (Figure 3B–C) reveals that human I67D/L70D is eluted ahead of human I67A/L70A (~15.0 vs~15.2 mL), indicating that I67D/L70D has a larger hydrodynamic radius despite differing in only two residues. We believe this difference is reflective of changes in the γTuNA coil structure, where the hydrophilic aspartate residues now cause the coil to extend, kink, or otherwise deform in a way that increases the hydrodynamic radius of the I67D/L70D protein. This drastic change in the local coil structure, in addition to blocking coiled-coil dimerization, also likely prevents even weak interactions with γTuRC.

Yet, I67A/L70A still retains a small amount of residual binding to γTuRC. How might this I67A/L70A mutation be overcoming the loss of dimerization to weakly bind γTuRC? It is possible that two separate monomeric coils of mutant γTuNA might bind the same γTuRC and form a weak complex. In this scenario, the interaction with the γTuRC would stabilize the γTuNA dimer, overcoming the loss of the strongly hydrophobic contacts normally present in the coiled-coil dimer interface. We believe that we have observed a related phenomenon with the Xenopus L77A mutant in our in vitro reactions (Figure 5B), where late in the assay L77A can begin to increase γTuRC activity despite lacking strong dimerization and strong γTuRC binding ability (Figure 3B–E). We hypothesize that this late effect is reflecting mutant L77A monomers that are stochastically stabilized into a dimer on γTuRC (Figure 5—figure supplement 3). We further predict that this is also a function of the in vitro environment, which is more permissive of these types of interactions, as L77A does not display this behavior in extract. The fact that I67A/L70A can weakly bind in extract is likely due to the fact that I67 and L70 are outside the critical core region discussed above (aa 75–77). Furthermore, we predict that this weak binding can only occur in hydrophobic-to-weaker-hydrophobic substituted versions of γTuNA, like I67A/L70A or L77A. These types of substitutions are not as likely to cause drastic changes to the overall coil structure of a γTuNA monomer, which might allow for two of these monomers to be stabilized into a dimer on γTuRC.

Finally, our γTuNA-GCN4 fusion constructs were our attempt to rescue the coiled-coil dimer and subsequent γTuRC binding ability (Figure 3—figure supplement 1). While this did rescue dimerization in an SEC assay (Figure 3—figure supplement 1), these fusion constructs did not rescue γTuRC binding. This divergent result is likely because the fused GCN4 domain did not restore the local coil structure of γTuNA (if impacted). Our GCN4 fusion also has no impact on the hydrophobic character of the core region (aa 75–77) an aspect which appears to be most critical for γTuRC binding in extract. Also, we suggest that specific residues might be required for both dimerization and for making specific contacts with γTuRC. For these cases, inducing dimerization would never be sufficient to restore wildtype levels of γTuRC binding as the specific residue enabling stable interaction would still be missing. We imagine that the core residues, like L77, have twin impacts on both dimerization and stable γTuRC binding.

We propose that dimerization is a key component of how γTuNA interacts with γTuRC (supported by the cryo-EM structure by Wieczorek et al., Cell Reports, 2020), but dimerization on its own is not sufficient. Indeed, the F75 residue, which would be located on the outer surface of the dimer (Figure 3A), was required for activation and strong binding in all our assays. We hypothesize that this is likely due to a stabilizing or docking role where this outer surface residue helps ‘lock’ the γTuNA domain into γTuRC.

This study was partly motivated by an apparent discrepancy between the original reports of γTuNA’s ability to activate γTuRC in vivo and in vitro (Fong et al., 2008; Choi et al., 2010; Muroyama et al., 2016) and more recent in vitro data from our group and others that found little to no effect (Liu et al., 2020; Thawani et al., 2020).

In their recent structural study of antibody-purified Xenopus γTuRC, Liu and colleagues concluded that the N-terminal region of Cdk5rap2 (or CEP215) containing the γTuNA domain had little to no effect on γTuRC MT nucleation in vitro (Extended Data Fig. 9b from Liu et al., 2020). They did, however, report that wildtype γTuNA (CEP215N) co-precipitated γ-tubulin, while the F75A mutant did not (Extended Data Fig. 9c from Liu et al., 2020). Liu and colleagues used N-terminally GST tagged versions of γTuNA (CEP215N). Like our colleagues, we find that the presence of a large N-terminal tag does not interfere with γTuNA’s ability to bind γTuRC (Figure 3). However, our studies revealed that a bulky N-terminal Halo-tag on γTuNA turns this activator into a specific inhibitor of γTuRC-mediated MT nucleation in extract and in vitro (Figure 5—figure supplement 2). This is likely due to the steric clash produced by two copies of the bulky N-terminal tag in proximity to the critical nucleation interface on the γ-tubulin ring.

Slightly confounding, in a previous fixed in vitro assay (Muroyama et al., 2016) Muroyama and colleagues did observe an activation effect with an N-terminally GST-tagged truncation of CDK5RAP2. This suggests that differences in the distance between the bulky tag and γTuNA, as well as the ratio of γTuNA to γTuRC tested, determines whether an activation effect is possible. If it is true that multiple γTuNA binding sites exist in vertebrate γTuRCs (as in yeast; Brilot et al., 2021), then this N-terminal tag effect might be further compounded as multiple steric clashes could be present. We note that the original reports from the Qi group used the small FLAG tag (Choi et al., 2010), and our work is based on the small Strep-His tag at the N-terminus.

Finally, in our prior work (Thawani et al., 2020), we had established an antibody-based Xenopus γTuRC purification, albeit with limited yield and batch-to-batch variability. Although N-terminally 6xHis-tagged γTuNA activates MT nucleation in extract (as presented in Figure 1 of this study), it had little to no effect on the original antibody-purified Xenopus γTuRC in vitro (Figure 6 in Thawani et al., 2020). This inability to activate antibody purified γTuRC puzzled us. We initially thought that an additional factor might be required for γTuNA-mediated activation of γTuRC. However, even with mass spectrometry data from our group and others (Liu et al., 2020; Consolati et al., 2020; Wieczorek et al., 2020a), we did not find an obvious target. Since then, we developed the Halo-γTuNA purification method described here, which is routinely at least 20-fold higher yield, higher purity, and ultimately has more robust activity. This resulted in increased density of nucleation competent γTuRCs present in our single molecule assays, as well as better detection of γTuNA’s activation effect. Silver staining the peak γTuRC fraction for our new prep (Figure 2—figure supplement 2B) showed the same banding pattern as that published with our previous antibody prepped γTuRC, indicating that aside from γTuRC components, there was no obvious major factor present to explain the response to γTuNA. Rather, we believe the difference can be explained by the greater yield and consistent quality of γTuRC provided by the Halo-γTuNA prep. As such, we validate and extend the original γTuNA studies by the Qi group.

The fragment of human CDK5RAP2 (51-100) containing the CM1 motif/γTuNA domain was sub-cloned into a bacterial expression pST50 vector using Gibson cloning. This vector was engineered with N-terminal Strep-TagII, 6xHis, TEV cleavage, HaloTag, and PreScission 3 C protease cleavage sites. This vector was then truncated to human CDK5RAP2 aa 53–98 (see Table 1). The resulting construct, Strep-His-TEV-HaloTag-3C-human γTuNA, (Halo-human γTuNA), was expressed in Rosetta 2 (DE3) E. coli cells. Rosetta 2 cells were grown in 2 L terrific broth (TB) cultures to O.D.=0.7 and induced with 0.5 mM IPTG at 16 °C for 18 h. The cultures were pelleted, snap-frozen, and stored at –80° C.

The CM1/γTuNA-motif in Xenopus laevis Cdk5rap2 was confirmed via sequence alignment to the human version (Figure 1B) and inserted using Gibson cloning into the same pST50 bacterial construct as above. This generated Strep-His-TEV-HaloTag-3C-Xenopus γTuNA (Halo-Xenopus γTuNA; Xenopus Cdk5rap2 aa 56–101). The loss-of-function mutation, F75A, first identified by Fong et al., 2008, was introduced into the Halo-Xenopus γTuNA sequence at the equivalent, conserved phenylalanine at position 71 to make Strep-His-TEV-HaloTag-3C-Xenopus γTuNA “F75A” (Halo-Xenopus γTuNA F75A). Both these constructs were expressed as described above with the human version.

The amino acid sequence of the human γTuNA used is: SPTRARNMKDFENQITELKKENFNLKLRIYFLEERMQQEFHGPTEH. The sequence for the Xenopus γTuNA (wildtype) used in this work is: MKDFEKQIAELKKENFNLKLRIYFLEEQVQQKCDNSSEDLYRMNIE. All γTuNA constructs generated in this study are listed in Table 1. For GCN4 C-terminal fusions, we used pET28a-Hook3 aa 1–160-GCN4 plasmid, which was a gift from Dr. Ron Vale (Addgene plasmid # 74608; RRID:Addgene_74608).

To purify the human and Xenopus Halo-γTuNA constructs, 2 L TB cell pellets were thawed on ice and resuspended into 50 mL of Strep Lysis Buffer (50 mM TRIS, pH = 7.47, 300 mM NaCl, 6 mM β-mercaptoethanol, 200 µM PMSF, 10 µg/mL DNase I), and a single dissolved cOmplete EDTA-free Protease Inhibitor Cocktail tablet (cat # 11873580001, Roche). Cells were resuspended using a Biospec Tissue Tearor (Dremel, Racine, WI) and lysed in an Emulsiflex C3 (Avestin, Ottawa, Canada) by processing four times at 10,000–15,000 psi. Cell lysate was spun at 30,000 rpm for 30 min, 2 °C in a Beckman Optima-XE 100 ultracentrifuge, 45Ti rotor. Supernatant was then passed twice through a 15 mL column volume (CV) of Strep-Tactin Superflow resin (IBA, Goettingen, Germany). The column was then washed with 10 CV of Strep Bind buffer (50 mM TRIS, pH = 7.47, 300 mM NaCl, 6 mM β-mercaptoethanol (BME), 200 µM PMSF). The γTuNA proteins were then eluted with 1.5 CV of Strep Elution buffer (Strep Bind Buffer with 3.3 mM D-desthiobiotin (cat. #2-1000-005, IBA)). Yield and purity were assessed via SDS-PAGE gel and Coomassie stain. Concentration was assessed via Bradford assay. All γTuNA constructs yielded between 40 and 60 mg of protein (per 2 L TB culture) at >98% purity.

For all size-exclusion assays, we used an ÄKTA Pure system with a Superdex 200 increase 10/300 GL column (cat. #2, Cytiva, Marlborough, MA), with a 500 µL manual injection loop. All assays were done in CSFxB, 6 mM BME, without sucrose at 4 °C. Strep-His-TEV-Halo-3C-γTuNA constructs shown in Figure 3 were run at ~50 µM final concentration in a total volume of 550 µL Strep bind buffer (see above), at a 0.7 mL/min flow rate. Absorbance at 280 nm was used to track the protein peak. Each trace was normalized by the maximum peak for that run, prior to combined plotting with all other constructs in MATLAB (ver. R2019a, MathWorks, Natick, MA; RRID:SCR_001622).

Xenopus laevis egg extracts were prepared as previously described (Good and Heald, 2018). For assaying γTuNA’s effect on MT nucleation levels, 7.5 µL of extract were incubated on ice with 0.5 µL 10 mM Vanadate (0.5 mM final), 0.5 µL 1 mg/mL end-binding 1 (EB1)-mCherry protein (0.05 mg/mL final), 0.5 µL of 1 mg/mL Cy5-tubulin (0.05 mg/mL final), 0.5 µL of CSFxB (10% sucrose), and 0.5 µL of TRIS control buffer (50 mM TRIS, pH = 7.47, 300 mM NaCl, 6 mM β-mercaptoethanol) or 0.5 µL of γTuNA protein (previously diluted in Tris control buffer such that final concentrations are as stated in Figure 1). Reactions (10 µL total) were gently mixed by pipetting once, before adding to a channel on a 6-channel slide at 18–20°C. All γTuNA concentrations for each condition (wildtype or F75A mutant) were imaged in parallel on the same slide. Multi-channel images were acquired sequentially and at 1 min intervals using the NIS-Elements AR program (NIKON, ver. 5.02.01-Build 1270; RRID:SCR_014329). The 647 nm/Cy5 channel (excitation: 678 nm, emission: 694 nm) was used for microtubules (MT) and the 561 nm channel for EB1 MT plus-tips (ex: 587 nm, em: 610 nm). The images were captured on a Nikon Ti-E inverted system (RRID:SCR_021242), with an Apo TIRF 100 x oil objective (NA = 1.49), and an Andor Neo Zyla (VSC-04209) camera with no binning and 100 ms exposures. Resulting images were 2048 by 2048 pixels (132.48 µm x 132.48 µm). The 561 nm channel was pseudo-colored green.

To quantify MT nucleation levels, we extracted the number of MT plus ends (tracked by EB1 spots) for each condition using the 561 nm/EB1-mCherry channel. We wrote a macro in FIJI (ImageJ; Schindelin et al., 2012; RRID:SCR_002285) to automate counting EB1 spots. Briefly, 50 µm by 50 µm (800x800 pixels²) representative windows were cropped from each field of view, smoothed using the FIJI function (‘Process—>Smooth’), and thresholded using the Yen option. The built-in FIJI functions ‘Find ‘dges” and ‘Analyze particles’ were then used to count the number of thresholded spots. These values, representing the number of MT plus ends, were then normalized to the buffer control for each condition to obtain the fold change in MT number. These fold changes were then averaged across four biological replicates (independent extract preps) and plotted using Prism GraphPad 7 (GraphPad Software, San Diego, CA; RRID:SCR_002798). Representative images are shown in Figure 1. The 95% confidence intervals and SEM are also shown.

To purify native Xenopus γTuRC from Xenopus egg extract, we employed a strategy similar to previous work (Wieczorek et al., 2020b) and originally observed by Choi et al., 2010. Here we similarly use γTuNA as a bait for γTuRC, except we use the human version of Halo-γTuNA directly coupled to beads via the affinity of the Halo-tag for its substrate. These beads are then used to pulldown γTuRC from Xenopus laevis egg extract.

Xenopus laevis egg extracts were prepared as described previously. Extracts were snap-frozen in liquid nitrogen and stored at –80 °C. One day prior to purification, 15 mg of Halo-human γTuNA were thawed and diluted to ~1 mg/mL in a 15 mL conical tube with Coupling Buffer (20 mM HEPES, pH = 7.5 with KOH, 75 mM NaCl) to a final NaCl concentration of 100–135 mM NaCl. Next, 2 mL of Halo Magne bead slurry (cat. #G7287, Promega, Madison, WI) were washed with MilliQ water and 3 CV of modified CSF-XB, 2% sucrose buffer (10 mM HEPES, pH = 7.7, 100 mM KCl, 1 mM MgCl2, 0.1 mM CaCl2, 5 mM EGTA, 2% sucrose). Washed beads were then incubated under constant rotation with the 15 mL of Halo-human γTuNA for 16 hr overnight at 4 °C. The beads were collected next day via a magnetic stand and washed with 3 CV of CSF-XB and resuspended to 2 mL with CSF-XB, 2% sucrose. Beads were either used fresh or stored for up to a week at 4 °C.

The next day, frozen extract (4 mL) was thawed in a room-temperature water bath. A total of 500 µL of clean, uncoupled Halo Magne beads were washed with MilliQ water and 3 CV of CSF-XB, 2% sucrose for pre-clearing of extract. Thawed extract was then incubated with the 500 µL of uncoupled Halo beads for 30 min at 4 °C with rotation to remove non-specific binders. The pre-clearing beads were removed with a magnet, and the now ‘pre-cleared’ extract was used to resuspend the 2 mL equivalent of coupled Halo-human γTuNA beads. Halo-human γTuNA beads were incubated with the extract for 2 hr, 4 °C, while rotating. The beads were then removed from the extract, washed with 4.5 CV of CSF-XB, and if required, incubated with 40 µM NHS-PEG4-Biotin (cat. #A39259, ThermoFisher) for 1 hr on ice and then resuspended with 2 mL of “3C” Elution buffer (600 µg PreScission “3C” protease diluted in CSF-XB, 2% sucrose). Proteins were eluted from the beads overnight at 4 °C, 15–18 hr, with rotation.

The elution containing cleaved γTuNA, γTuRC, and other factors, was concentrated to ~300 µL in a 100 kDa MWCO Amicon 4 mL spin concentrator. This was then spun through a 10–50% w/w sucrose gradient (in CSFxB buffer) using a TLS55 rotor in a Beckman Coulter Optima MAX-XP ultracentrifuge at 200,000 g, 2 °C for 3 hr. The gradient was manually fractionated such that the first fraction was the same size as the input (~300 µL), and each subsequent fraction was 140 µL in size. Samples of each fraction were run on a 4–12% Bis-TRIS SDS-PAGE gel for 10 min, 140 V. The SNAP i.d. 2.0 rapid Western blotting system (cat. #SNAP2MM, EMD-Millipore, Burlington, MA) was used to determine the peak γTuRC fraction (usually Fraction 7; γ-tubulin, GTU-88 mouse antibody, cat. # T6557, Sigma, St. Louis, MO). A 5 µL sample of the peak fraction was then added to glow discharged copper CF400-CU EM grids (cat. # 71150, Electron Microscopy Sciences, Hatfield, PA) for 1 min, and stained with 0.75% Uranyl Formate (UF) for 40 s. The grid was then imaged on a Philips CM100 transmission electron microscope at ×64,000 magnification, 80 kV, to verify the presence of intact γTuRCs, as well as to assess purity. A representative image is shown in Figure 2. The peak γTuRC fraction was stored on ice and used within 24 h or snap-frozen and stored at –80 °C. Our prep yielded an average peak concentration between 150–200 nM γTuRC (per 140 µL fraction). This was determined via Western blots probing for GCP4 in our peak fraction, which was compared against a known standard of recombinant GCP4. As there are two copies of GCP4 in each γTuRC, we divided our measured [GCP4] by two to get peak [γTuRC]. Via a fused Halo-AU1 reporter version of human γTuNA, we measured 50 nM γTuNA dimer in the peak γTuRC fraction, suggesting 67% of purified γTuRCs lost their γTuNA dimer bait at the end of the sucrose gradient step (Figure 2—figure supplement 2). This loss of γTuNA bait has been previously observed (Choi et al., 2010; Muroyama et al., 2016).

Data processing of negative-stain EM data was done using Relion 3.1 (Zivanov et al., 2018). We manually picked 4866 total particles from 59 micrographs, followed by particle extraction and 5 rounds of 2D class averaging for particle sorting prior to 3D reconstructions. 4,593 particles were used for ab initio, non-templated reconstructions. Particles were further sorted with 3D class averaging, and 2,692 selected particles were used for structure refinement. Final refinement iterations were done using a 100 Å low-pass filtered mask with a contour cutoff of 0.007. This gave a final mask-sharpened map at 28 Å resolution as determined by the gold-standard FSC cutoff of 0.143 (Figure 2—figure supplement 3). CryoEM movie data was aligned and CTF corrected using CryoSparc 3.2 (Punjani et al., 2017), followed by subsequent manual picking of particles on CTF-corrected micrographs. A total of 800 particles were manually picked, extracted, and used to calculate 2D class averages.

To compare γTuRC binding across mutant versions of γTuNA, we performed pulldowns from Xenopus egg extract with Halo-Magne beads (cat # G7281, Promega) coupled to N-terminally Halo-tagged versions of either human wildtype, Xenopus wildtype, or Xenopus dimer mutant γTuNA (as in Figure 3). Mock beads (blocked with bovine serum albumin) were used to assess background levels of non-specific precipitants. For each condition, we diluted 2 mg total of bait protein in 1 mL of Coupling Buffer (see section 6.5). Next, we resuspended 170 µL worth of Halo-Magne beads (washed 3 x with Coupling Buffer) with each protein mix. These were incubated under rotation at 4 °C for 1 h. Beads were then collected via a magnetic stand, washed three times with Coupling Buffer, and then resuspended to 1 mL final volume with CSFxB (2% sucrose).

Next, 2 mL total of frozen Xenopus egg extract were thawed in a room-temperature water bath. Beads for each condition were then collected via a magnetic stand and resuspended with 200 µL of extract. To each, we then added 800 µL of CSFxB (2% sucrose), and incubated under rotation at 4 °C for 2 hr. Beads were then washed twice with 1 mL of CSFxB (2% sucrose), prior to resuspension in 120 µL of 1 x SDS-PAGE sample buffer (6 mM DTT). Beads were then boiled at 95 °C for 5 min. After magnetic removal of beads, the elutions were spun at 17,000 g for 1 min to remove aggregates or beads.

We ran 40 µL of each elution per lane in 4–12% Bis-Tris SDS-PAGE gels at 140 V for 1 hr. Proteins were then transferred to nitrocellulose membranes and probed for γTuRC components via Western blot: GCP5 (1:250 dilution of mouse anti-GCP5 antibody (E-1), # sc-365837, Santa Cruz Biotechnology, Santa Cruz, CA; RRID:AB_10847352) and γ-tubulin (1:1000 dilution of GTU-88 mouse antibody, cat. # T6557, Sigma; RRID:AB_2863751). To confirm equal coupling of bait to beads, we also probed for the Strep-tag on γTuNA (1:1000 dilution of Strep-tag mouse monoclonal antibody, cat. # 34850, Qiagen, RRID:AB_2810987). Band intensities were measured in ImageJ and normalized in Prism 7 by the wildtype γTuNA band (positive control). Independent experiments, after normalization, were averaged together into the charts seen in Figure 3 (N=3 for Figure 3F, N=2 for Figure 3G).

To simulate the effect of γTuNA in our single molecule TIRF assay, we wrote a deterministic simulation in MATLAB based on the measured nucleation rates for each condition. For simplicity, we assume no spontaneous MT nucleation and no MT catastrophes. This system was modeled by Equation 2, as follows:

where N(t)=the current number of nucleated MTs (or active γTuRCs), k=the nucleation rate (MTs nucleated per second), and N_max is the total number of γTuRCs activatable by that condition. At N(t=0), the number of active γTuRCs or nucleated MTs is zero. At each time step (1 s), new MTs are added to the system according to the nucleation rate measured experimentally (k), which decreases until reaching saturation. These new MTs were then randomly placed on a simulated 40x40 pixel² plane with a random initial orientation based on one of eight discrete conditions. At each new time step, the length of previous MTs is incremented by a constant growth speed (the mean speed from all conditions in Figure 5E, as µm/s). This process of nucleation and growth is repeated until the end of the simulation, generating simulated movies of this process. For the L77A mutant, we also generated a second two-step simulation where, after 150 s, k_L77A is arbitrarily redefined as the wildtype k (k_WT), and N_max in Equation 2 is redefined as (wildtype N_max – L77A N_(150s)), where L77A N_(150s)=the number of MTs already present at 150 s. Plots and simulated frames are shown in Figure 5—figure supplement 3.

Two-sample, unpaired Student’s t-tests were used with significance declared when p-values were less than p=0.05. No a priori sample size or power analysis calculations were performed. For extract or γTuRC in vitro experiments, the number of datasets (or ‘N’) refers to biological replicates, performed with either independent Xenopus egg extracts or independent Halo-γTuRC preps from different Xenopus egg extracts. For all main figure data, standard error of the mean (SEM) is used to indicate uncertainty in our measurement of each condition’s mean, except for Figure 5E and F, where the complete distributions of growth speed and MT length measurements are instead shown for clarity.

Experimental use of Xenopus laevis frogs was done in strict accordance with our approved Institutional Animal Care and Use Committee (IACUC) protocol # 1941–16 (Princeton University).

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

The authors thank all members of the Petry lab, past and present. In particular, we would like to thank Dr. Raymundo Alfaro-Aco. We also thank Prof. Fred Hughson (Princeton University), Prof. Paul Conduit (Institut Jacques Monod), and Bernardo Gouveia (Princeton University) for their insightful comments on our manuscript. We also thank Dr. Jodi Kraus for sharing reagents. We especially thank both the Princeton Mass Spectrometry core facility and the ThermoFisher Center for Multiplexed Proteomics (TCMP) at Harvard Medical School. We also thank the Imaging and Analysis Center (IAC) at Princeton University, which is partially supported by the Princeton Center for Complex Materials (PCCM), a National Science Foundation (NSF) Materials Research Science and Engineering Center (MRSEC; DMR-2011750), and in particular, Dr. John Schreiber. We also thank Dr. Ron Vale (UCSF) for the plasmid containing GCN4 (Addgene plasmid# 74608).

Experimental use of Xenopus laevis frogs was done in strict accordance with our approved Institutional Animal Care and Use Committee (IACUC) protocol # 1941-06 (Princeton University).

1. Preprint posted: April 11, 2022 (view preprint)
2. Received: May 13, 2022
3. Accepted: December 7, 2022
4. Accepted Manuscript published: December 14, 2022 (version 1)
5. Version of Record published: January 20, 2023 (version 2)

© 2022, Rale 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.